Oral Biology - Molecular Techniques and Applications PDF

Methods in
Molecular Biology 1537
Gregory J. Seymour
Mary P. Cullinan
Nicholas C.K. Heng Editors
Oral Biology
Molecular Techniques
and Applications
Second Edition
Methods in Molecular Biology
Series Editor
JohnM. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB,UK
For further volumes:

http://www.springer.com/series/7651
Oral Biology
Molecular Techniques and Applications
Second Edition
Edited by
Gregory J. Seymour
Faculty of Dentistry, University of Otago, Dunedin, New Zealand
Mary P. Cullinan
Department of Oral Sciences, Faculty of Dentistry, University of Otago, Dunedin, New Zealand
Nicholas C.K. Heng

Faculty of Dentistry, Sir John Walsh Research Institute, University of Otago, Dunedin, New Zealand
Editors
Gregory J. Seymour Mary P. Cullinan
Faculty of Dentistry Department of Oral Sciences, Faculty of Dentistry
University of Otago University of Otago
Dunedin, New Zealand Dunedin, New Zealand
Nicholas C.K. Heng

Faculty of Dentistry, Sir John Walsh Research
Institute
University of Otago
Dunedin, New Zealand
ISSN 1064-3745 ISSN 1940-6029(electronic)

Methods in Molecular Biology
ISBN 978-1-4939-6683-7ISBN 978-1-4939-6685-1(eBook)
DOI 10.1007/978-1-4939-6685-1
Library of Congress Control Number: 9781493967384
Springer Science+Business Media LLC 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
Cover illustration: Example of a bead experiment combined with in situ hybridization (ISH) analysis to study gene
expression in embryonic tissue explants. The image shows the effects of BMP2 beads on ld1 gene expression in explants
of calvarial mesenchyme. Photograph provided by D. Rice and K. Nrhi. The bead and ISH experiments are described
in Chapter 20.
Printed on acid-free paper
This Humana Press imprint is published by Springer Nature

The registered company is Springer Science+Business Media LLC
The registered company address is: 233 Spring Street, NewYork, NY 10013, U.S.A.
Preface
It is widely accepted that evidence-based dentistry is fundamental to clinical practice and

that well-controlled randomized clinical trials followed by systematic reviews and meta-
analyses provide much of this evidence base. However, it is still the basic biological and
physical sciences that underpin advances in dentistry and form the basis for subsequent
clinical trials. It is equally true that the treatment of any disease should be based on an
understanding of the etiology and pathogenesis of that disease, and in this context, the
future of dentistry lies very much in continued research in the basic biological sciences.
This second edition of Oral Biology: Molecular Techniques and Applications continues
the approach taken in the first edition and has not attempted to cover all aspects of oral
biology, but rather to present a selection of cellular and molecular techniques that can be
adapted to cover a range of applications and diseases. The first part on saliva, for example,
has been updated and expanded to include proteomic analyses by mass spectrometry and
NMR-based metabolomics that can be used not only in the study of saliva but also in assess-
ing other oral fluids such as gingival fluid. Clearly, saliva is unique to the oral cavity but so
too is gingival fluid which, in essence, is the fluid medium of the gingiva and gingival sulcus,
and thus is the fluid environment where interactions between the plaque biofilm and the
host take place. Hence, techniques for its collection and analysis have now been included.
Although it is 6 years since publication of the first edition of this book, many of the
techniques described are still in widespread use and so have been retained, albeit updated,
in this second edition. In the part on molecular biosciences, for example, chapters on profil-
ing of oral microbial communities, quantitative real-time PCR, and adhesion of yeast and
bacteria to oral surfaces have all been retained but substantially updated.
Epigenetics is now a major theme in biology and is providing great insight into how we
interact with our environment. As DNA methylation features heavily in epigenetic studies,
new chapters on tools and strategies that facilitate the analysis of genome-wide or gene-
specific DNA methylation patterns have been included.
As in the first edition, the last part of this second edition deals with a range of approaches
that enable the behavior of cells and tissues in both health and disease to be analyzed at the
molecular level. The future of dentistry and of the profession lies in research, and it is antici-
pated that this second edition of Oral Biology: Molecular Techniques and Applications will
continue to be a useful resource for oral biologists at all levels, be they students, early career
or experienced veterans, and that it provides a ready reference enabling new techniques and
approaches to be used in answering a range of specific scientific questions that will underpin
a deeper understanding and treatment of oral diseases.
Dunedin, New Zealand GregoryJ.Seymour

Dunedin, New Zealand MaryP.Cullinan
Dunedin, New Zealand NicholasC.K.Heng
v
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Part I Saliva and Other Oral Fluids

1 Salivary Diagnostics Using Purified Nucleic Acids . . . . . . . . . . . . . . . . . . . . . . . 3
Paul D. Slowey
2 RNA Sequencing Analysis of Salivary Extracellular RNA . . . . . . . . . . . . . . . . . . 17
Blanca Majem, Feng Li, Jie Sun, and David T.W. Wong
3 Qualitative andQuantitative Proteome Analysis ofOral Fluids
inHealth andPeriodontal Disease by Mass Spectrometry . . . . . . . . . . . . . . . . . 37
Erdjan Salih
4 Antioxidant Micronutrients andOxidative Stress Biomarkers . . . . . . . . . . . . . . 61
Iain L.C. Chapple, Helen R. Griffiths, Mike R. Milward,
Martin R. Ling, and Melissa M. Grant
5 NMR-Based Metabolomics ofOral Biofluids . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Horst Joachim Schirra and Pauline J. Ford
6 Gene Therapy ofSalivary Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Bruce J. Baum, Sandra Afione, John A. Chiorini, Ana P. Cotrim,
Corinne M. Goldsmith, and Changyu Zheng
Part II Molecular Biosciences

7 The Oral Microbiota inHealth andDisease: AnOverview
ofMolecular Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Jos F. Siqueira Jr. and Isabela N. Ras
8 Microbial Community Profiling Using Terminal Restriction
Fragment Length Polymorphism (T-RFLP) andDenaturing Gradient
Gel Electrophoresis (DGGE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Jos F. Siqueira Jr., Mitsuo Sakamoto, and Alexandre S. Rosado
9 Analysis of16S rRNA Gene Amplicon Sequences
Using theQIIME Software Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Blair Lawley and Gerald W. Tannock
10 Adhesion ofYeast andBacteria toOral Surfaces . . . . . . . . . . . . . . . . . . . . . . . . 165
Richard D. Cannon, Karl M. Lyons, Kenneth Chong,
Kathryn Newsham-West, Kyoko Niimi, and Ann R. Holmes
11 Quantitative Analysis ofPeriodontal Pathogens Using Real-Time
Polymerase Chain Reaction (PCR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
M Jos Marin, Elena Figuero, David Herrera, and Mariano Sanz
vii
viii Contents
12 Methods toStudy Antagonistic Activities Among Oral Bacteria . . . . . . . . . . . . 203

Fengxia Qi and Jens Kreth
13 Natural Transformation of Oral Streptococci by Use of Synthetic
Pheromones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Gabriela Salvadori, Roger Junges, Rabia Khan, Heidi A. mdal,
Donald A. Morrison, and Fernanda C. Petersen
14 Markerless Genome Editing inCompetent Streptococci . . . . . . . . . . . . . . . . . . 233
Roger Junges, Rabia Khan, Yanina Tovpeko, Heidi A. mdal,
Fernanda C. Petersen, and Donald A. Morrison
15 Tools andStrategies forAnalysis ofGenome-Wide andGene-Specific
DNA Methylation Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Aniruddha Chatterjee, Euan J. Rodger, Ian M. Morison, Michael R. Eccles,
and Peter A. Stockwell
16 Generating Multiple Base-Resolution DNA Methylomes Using Reduced
Representation Bisulfite Sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Aniruddha Chatterjee, Euan J. Rodger, Peter A. Stockwell,
Gwenn Le Me, and Ian M. Morison
17 A Protocol fortheDetermination oftheMethylation Status
ofGingival Tissue DNA at Specific CpG Islands . . . . . . . . . . . . . . . . . . . . . . . . 299
Trudy J. Milne
18 Genome-Wide Analysis ofPeriodontal andPeri-Implant Cells andTissues . . . . 307
Moritz Kebschull, Claudia Hlsmann, Per Hoffmann,
and Panos N. Papapanou
19 Differential Expression andFunctional Analysis
of High-Throughput -Omics Data Using Open Source Tools . . . . . . . . . . . . . . 327
Moritz Kebschull, Melanie Julia Fittler, Ryan T. Demmer,
20 Exploring Genome-Wide Expression Profiles Using Machine
Learning Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Moritz Kebschull and Panos N. Papapanou
Part IIICells and Tissues

21 Embryonic Explant Culture: Studying Effects ofRegulatory
Molecules onGene Expression inCraniofacial Tissues . . . . . . . . . . . . . . . . . . . 367
Katja Nrhi
22 Oral Epithelial Cell Culture Model forStudying thePathogenesis
ofChronic Inflammatory Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Mike R. Milward, Martin R. Ling, Melissa M. Grant,
and Iain L.C. Chapple
23 Fabrication andCharacterization ofDecellularized Periodontal Ligament
Cell Sheet Constructs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
Amro Farag, Cdryck Vaquette, Dietmar W. Hutmacher, P. Mark Bartold,
and Saso Ivanovski
Contents ix
24 A Method toIsolate, Purify, andCharacterize Human Periodontal

Ligament Stem Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
Krzysztof Mrozik, Stan Gronthos, Songtao Shi, and P. Mark Bartold
25 Constructing Tissue Microarrays: Protocols andMethods Considering
Potential Advantages andDisadvantages forDownstream Use . . . . . . . . . . . . . 429
Lynne Bingle, Felipe P. Fonseca, and Paula M. Farthing
26 Growing Adipose-Derived Stem Cells Under Serum-Free Conditions . . . . . . . . 439
Diogo Godoy Zanicotti and Dawn E. Coates
27 Quantitative Real-Time Gene Profiling ofHuman Alveolar Osteoblasts . . . . . . 447
Dawn E. Coates, Sobia Zafar, and Trudy J. Milne
28 Proteomic Analysis ofDental Tissue Microsamples . . . . . . . . . . . . . . . . . . . . . . 461
Jonathan E. Mangum, Jew C. Kon, and Michael J. Hubbard
29 Characterization, Quantification, andVisualization ofNeutrophil
Extracellular Traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
Phillipa C. White, Ilaria J. Chicca, Martin R. Ling, Helen J. Wright,
Paul R. Cooper, Mike R. Milward, and Iain L.C. Chapple
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
Contributors
SandraAfione Molecular Physiology and Therapeutics Branch, National Institute

of Dental and Craniofacial Research, National Institutes of Health (NIH), Bethesda,
MD, USA
HeidiA.mdal Department of Oral Biology, Faculty of Dentistry, University of Oslo,
Oslo, Norway
P.MarkBartold Colgate Australian Clinical Dental Research Centre, Dental School,
University of Adelaide, Adelaide, Australia
BruceJ.Baum Molecular Physiology and Therapeutics Branch, National Institute
MD, USA; Molecular Physiology and Therapeutics Branch, National Institute of Dental
and Craniofacial Research, National Institutes of Health (NIH), Bethesda, MD, USA
LynneBingle Academic Unit of Oral and Maxillofacial Pathology, School of Clinical
Dentistry, University of Sheffield, Sheffield, UK
RichardD.Cannon Department of Oral Sciences, University of Otago School of
Dentistry, Dunedin, New Zealand; Faculty of Dentistry, Sir John Walsh Research
Institute, University of Otago School of Dentistry, Dunedin, New Zealand
IainL.C.Chapple School of Dentistry, Institute of Clinical Sciences, College of Medical
and Dental Sciences, University of Birmingham, Birmingham, UK
AniruddhaChatterjee Department of Pathology, Dunedin School of Medicine,
University of Otago, Dunedin, New Zealand; Maurice Wilkins Centre for Molecular
Biodiscovery, Auckland, New Zealand
IlariaJ.Chicca Institute of Clinical Sciences, College of Medical and Dental Sciences,
The School of Dentistry, University of Birmingham, Birmingham, UK
JohnA.Chiorini Molecular Physiology and Therapeutics Branch, National Institute
MD, USA
KennethChong Department of Oral Sciences, University of Otago School of Dentistry,
DawnE.Coates Faculty of Dentistry, Sir John Walsh Research Institute, University
of Otago, Dunedin, New Zealand
PaulR.Cooper Institute of Clinical Sciences, College of Medical and Dental Sciences,
AnaP.Cotrim Molecular Physiology and Therapeutics Branch, National Institute
MD, USA
RyanT.Demmer Department of Epidemiology, Columbia University Mailman School
of Public Health, New York, NY, USA
MichaelR.Eccles Department of Pathology, Dunedin School of Medicine, University
of Otago, Dunedin, New Zealand; Maurice Wilkins Centre for Molecular Biodiscovery,
Auckland, New Zealand
AmroFarag School of Dentistry and Oral Health, Regenerative Medicine Center,
Menzies Health Institute Queensland, Gold Coast, QLD, Australia
xi
xii Contributors
PaulaM.Farthing Academic Unit of Oral and Maxillofacial Pathology, School

of Clinical Dentistry, University of Sheffield, Sheffield, UK
ElenaFiguero Oral Research Laboratory, Faculty of Odontology, University
Complutense, Madrid, Spain; Etiology and Therapy of Periodontal Diseases (ETEP)
Research Group, University Complutense, Madrid, Spain; Department of Periodontology,
Faculty of Dentistry, University Complutense of Madrid, Madrid, Spain
MelanieJulia Fittler Department of Periodontology, Operative and Preventive
Dentistry, University of Bonn, Bonn, Germany
FelipeP.Fonseca Department of Oral Diagnosis, Faculty of Dentistry of Piracicaba,
FOP, UNICAMP, Piracicaba, So Paolo, Brazil
PaulineJ.Ford School of Dentistry, Oral Health Centre, The University of Queensland,
Herston, QLD, Australia
CorinneM.Goldsmith Molecular Physiology and Therapeutics Branch, National
Institute of Dental and Craniofacial Research, National Institutes of Health (NIH),
Bethesda, MD, USA
MelissaM.Grant School of Dentistry, Institute of Clinical Sciences, College of Medical
HelenR.Griffiths School of Dentistry, Institute of Clinical Sciences, College of Medical
StanGronthos Mesenchymal Stem Cell Group, Adelaide Medical School, Faculty of
Health Sciences, University of Adelaide, Adelaide, SA, Australia
DavidHerrera Etiology and Therapy of Periodontal Diseases (ETEP) Research Group,
University Complutense, Madrid, Spain; Department of Periodontology, Faculty of
Dentistry, University Complutense of Madrid, Madrid, Spain
PerHoffmann Department of Genomics, Institute of Human Genetics, University of
Bonn, Bonn, Germany; Human Genomics Research Group, Department of Biomedicine,
University of Basel, Basel, Switzerland
AnnR.Holmes Department of Oral Sciences, University of Otago School of Dentistry,
Dunedin, New Zealand; Faculty of Dentistry, Sir John Walsh Research Institute,
University of Otago School of Dentistry, Dunedin, New Zealand
MichaelJ.Hubbard Department of Pharmacology and Therapeutics, University of
Melbourne, Melbourne, VIC, Australia; Department of Pediatrics, Royal Childrens
Hospital, University of Melbourne, Melbourne, VIC, Australia
ClaudiaHlsmann Department of Periodontology, Operative and Preventive Dentistry,
Faculty of Medicine, University of Bonn, Bonn, Germany
DietmarW.Hutmacher Queensland University of Technology, Brisbane, QLD,
Australia
SasoIvanovski School of Dentistry and Oral Health, Regenerative Medicine Center,
Menzies Health Institute Queensland, Gold Coast, QLD, Australia; Menzies Health
Institute Queensland, Griffith University, Gold Coast, QLD, Australia
RogerJunges Department of Oral Biology, Faculty of Dentistry, University of Oslo, Oslo,
Norway
MoritzKebschull Department of Periodontology, Operative and Preventive Dentistry,
Faculty of Medicine, University of Bonn, Bonn, Germany; Division of Periodontics,
Section of Oral, Diagnostic and Rehabilitation Sciences, Columbia University College of
Dental Medicine, New York, NY, USA
RabiaKhan Department of Oral Biology, Faculty of Dentistry, University of Oslo, Oslo,
Norway
Contributors xiii
JewC.Kon Department of Pharmacology and Therapeutics, University of Melbourne,

Melbourne, VIC, Australia; Department of Pediatrics, Royal Childerns Hospital,
University of Melbourne, Melbourne, VIC, Australia
JensKreth Oregon Health and Science University, Portland, OR, USA
BlairLawley Department of Microbiology and Immunology, University of Otago,
FengLi Division of Oral Biology and Oral Medicine, School of Dentistry, University of
California Los Angeles (UCLA), Los Angeles, CA, USA
MartinR.Ling School of Dentistry, Institute of Clinical Sciences, College of Medical
KarlM.Lyons Faculty of Dentistry, Sir John Walsh Research Institute, University of
Otago School of Dentistry, Dunedin, New Zealand; Department of Oral Rehabilitation,
University of Otago School of Dentistry, Dunedin, New Zealand
BlancaMajem Biomedical Research Unit in Gynecology, Vall Hebron Research Institute
(VHIR) and University Hospital, University Autonoma of Barcelona (UAB),
Barcelona, Spain
JonathanE.Mangum Department of Pharmacology and Therapeutics, University of
Melbourne, Melbourne, VIC, Australia
MJosMarin Oral Research Laboratory, Faculty of Odontology, University
Complutense, Madrid, Spain
GwennLe Me Department of Pathology, Dunedin School of Medicine, University of
Otago, Dunedin, New Zealand
TrudyJ.Milne Faculty of Dentistry, Sir John Walsh Research Institute, University of
MikeR.Milward School of Dentistry, Institute of Clinical Sciences, College of Medical
IanM.Morison Department of Pathology, Dunedin School of Medicine, University of
DonaldA.Morrison Department of Biological Sciences, College of Liberal Arts and
Sciences, University of Illinois at Chicago, Chicago, IL, USA
KrzysztofMrozik Colgate Australian Dental Research Centre, Dental School,
University of Adelaide, Adelaide, SA, Australia
KatjaNrhi Institute for Molecular Medicine Finland, University of Helsinki, Helsinki,
Finland
KathrynNewsham-West Faculty of Dentistry, Sir John Walsh Research Institute,
University of Otago School of Dentistry, Dunedin, New Zealand; Department of Oral
Rehabilitation, University of Otago School of Dentistry, Dunedin, New Zealand
KyokoNiimi Department of Oral Sciences, University of Otago School of Dentistry,
PanosN.Papapanou Division of Periodontics, Section of Oral, Diagnostic and
Rehabilitation Sciences, Columbia University College of Dental Medicine, New York, NY,
USA
FernandaC.Petersen Department of Oral Biology, Faculty of Dentistry, University of
Oslo, Oslo, Norway
FengxiaQi University of Oklahoma Health Sciences Center BRC364, Oklahoma City,
OK, USA
IsabelaN.Ras Department of Endodontics and Molecular Microbiology, Estcio de S
University, Rio de Janeiro, RJ, Brazil
xiv Contributors
EuanJ.Rodger Department of Pathology, Dunedin School of Medicine, University

of Otago, Dunedin, New Zealand
AlexandreS.Rosado Institute of Microbiology Prof. Paulo de Ges, Federal University
of Rio de Janeiro, Rio de Janeiro, Brazil
MitsuoSakamoto Microbe Division/Japan Collection of Microorganisms, RIKEN
BioResource Center, Wako, Saitama, Japan
ErdjanSalih Department of Periodontology, Henry M.Goldman School of Dental
Medicine, Boston University, Boston, MA, USA
GabrielaSalvadori Department of Oral Biology, Faculty of Dentistry, University of Oslo,
Oslo, Norway
MarianoSanz Etiology and Therapy of Periodontal Diseases (ETEP) Research Group,
University Complutense, Madrid, Spain; Department of Periodontology, Faculty of
Dentistry, University Complutense of Madrid, Madrid, Spain
HorstJoachimSchirra Centre for Advanced Imaging, The University of Queensland,
Brisbane, QLD, Australia
SongtaoShi Department of Anatomy and Cell BiologySchool of Dental Medicine,
University of Pennsylvania, Philadelphia, PA, USA
JosF.Siqueira Jr. Department of Endodontics and Molecular Microbiology, Estcio de
S University, Rio de Janeiro, RJ, Brazil; Faculty of Dentistry, Estcio de S University,
Rio de Janeiro, Brazil
PaulD.Slowey Oasis Diagnostics Corporation, Vancouver, WA, USA
PeterA.Stockwell Department of Biochemistry, University of Otago, Dunedin,
New Zealand
JieSun Medical School of Shenzhen University, Shenzhen, Guangdong, China
GeraldW.Tannock Department of Microbiology and Immunology, University of Otago,
YaninaTovpeko Department of Biological Sciences, College of Liberal Arts and Sciences,
University of Illinois at Chicago, Chicago, IL, USA
CdryckVaquette Queensland University of Technology, Brisbane, QLD, Australia
PhillipaC.White Institute of Clinical Sciences, College of Medical and Dental Sciences,
DavidT.W.Wong Division of Oral Biology and Oral Medicine, School of Dentistry,
University of California Los Angeles (UCLA), Los Angeles, CA, USA; Johnson
Comprehensive Cancer Center, University of California Los Angeles (UCLA), Los
Angeles, CA, USA; Molecular Biology Institute, University of California Los Angeles
(UCLA), Los Angeles, CA, USA; Head & Neck Surgery/Otolaryngology, Henry Samuel
School of Engineering and Applied Science, University of California Los Angeles
(UCLA), Los Angeles, CA, USA
HelenJ.Wright Institute of Clinical Sciences, College of Medical and Dental Sciences,
SobiaZafar Faculty of Dentistry, Sir John Walsh Research Institute, University of Otago,
DiogoGodoyZanicotti Faculty of Dentistry, Sir John Walsh Research Institute,
University of Otago, Dunedin, New Zealand
ChangyuZheng Molecular Physiology and Therapeutics Branch, National Institute of
Dental and Craniofacial Research, National Institutes of Health (NIH), Bethesda, MD,
USA
Part I
Saliva and Other Oral Fluids

Chapter 1
Salivary Diagnostics Using Purified Nucleic Acids

PaulD.Slowey
Abstract
Saliva is an easily accessible fluid that has led to increasing interest in the development of salivary diagnos-
tics. This chapter describes some of the newer tools and procedures for collection, stabilization, and stor-
age of oral fluid matrices that aid in the successful use of saliva as a test specimen. This chapter focuses
particularly on nucleic acid components for downstream molecular diagnostic (MDx) testing, since this is
probably the area where saliva is likely to have the greatest impact in improving healthcare for the general
population.
Key words Saliva, RNA, DNA, Nucleic acids, Stabilization, Exosomes
1 Introduction
Over the last few years, the use of saliva as a noninvasive bodily fluid
for research, forensic, and clinical testing has grown tremendously
and is now in use in many areas of the global invitro diagnostic
(IVD) market.
The number of applications for saliva is growing exponentially
as evidenced by the increasing number of available tools for saliva
acquisition and subsequent testing either immediately at the point
of care or under controlled laboratory conditions.
Saliva is now used in tests for adverse responses to multiple
therapeutics, genomics for cystic fibrosis, fragile X syndrome in
autism, disorders of the salivary glands, cancers (including breast,
head and neck, and oral cancers), abused drug testing in the work-
place and other environments, as well as certain systemic diseases
including HIV, hepatitis C, and Sjgrens syndrome.
The success of any test, whether for research or diagnostic
purposes, relies on the successful harvesting of the specimen from
a subject in a standardized, repeatable fashion and careful handling
of the sample throughout the collection and downstream testing
process. This rule applies to all specimen types, but care should
Gregory J. Seymour et al. (eds.), Oral Biology: Molecular Techniques and Applications, Methods in Molecular Biology, vol. 1537,
DOI10.1007/978-1-4939-6685-1_1, Springer Science+Business Media LLC 2017
3
4 PaulD.Slowey
especially be taken with respect to processing and stabilizing saliva

samples to ensure optimum results.
The following text describes some of the newer tools and
procedures for collection, stabilization, and storage of oral fluid
matrices that aid in the successful use of saliva as a test specimen.
This chapter focuses particularly on nucleic acid components for
downstream molecular diagnostic (MDx) testing, since this is
probably the area where saliva is likely to have the greatest impact
in improving healthcare for the general population. For more
detailed information on current salivary diagnostics and available
tools, the reader is referred to several review articles on the
subject [15].
Dr Lawrence Tabak (Deputy Director of the NIH and former
head of the National Institute for Dental and Craniofacial Research,
NIDCR) characterized saliva as a mirror of the body and is
therefore reflective of disease and disease processes going on in the
human body. This precious biofluid contains many of the biomark-
ers that are indicative of disease and maladies affecting human
beings, so saliva is the ideal sample matrix for large-scale epidemio-
logical studies, population screening, and diagnosis of multiple
diseases and conditions.
Saliva is cost-effective, noninvasive, easy to transport, amenable
to simple disposal, and highly attractive in certain cultures (and
religions), which find the use of blood an unacceptable option.
More importantly, saliva contains many of the indicators of disease
found in blood, urine, and tissue samples.
Typically, levels of biomarkers in saliva are 101500 times
lower than in blood, but with the advent of newer, more sensitive
detection technologies, the analysis of salivary biomarkers has
become a much more attractive option. When patient preference
to eliminate the use of needles is considered as an additive factor,
the compelling story for saliva grows significantly stronger.
These are some of the major reasons that there has been an explo-
sion in research and development in salivary diagnostics, in the
last few years, resulting in the development of a plethora of tools
and tests using this unique bodily fluid.
A series of technological developments, which have also con-
tributed to the growing importance of saliva as a diagnostic
medium, include several high-throughput technologies such as
next-generation sequencing, proteomics, mass spectrometry,
genome wide association studies (GWAS), and genotyping, which
allow large numbers of samples to be tested in a short time. Saliva
has already been shown to be a readily adaptable specimen for use
in these high-impact technologies.
Saliva is now in routine use for the diagnosis of HIV in the
privacy of ones home [6, 7] and for the detection of multiple hor-
mones as part of a general wellness program, sold direct to the
consumer [810]. Saliva has also been used to detect drugs of
abuse [11] and in certain situations has been shown to be a
Salivary Diagnostics Using Purified Nucleic Acids 5
preferable biofluid to urine, which is currently the method of

choice. This is particularly true in the case of marijuana, when test-
ing for impairment and whether a particular individual is fit to
drive a vehicle or perform dangerous tasks.
Multiple diseases have also been detected using saliva, including
caries risk [1214]; periodontitis [15]; oral [16], breast [1722],
and head and neck cancers [23]; and salivary gland disorders
[24]. Point of care tests are now also in development looking at
viruses, bacteria [25], and difficult to measure hormones using
saliva [26].
Perhaps the area where saliva has gained the most traction is for
the collection of nucleic acids (DNA and RNA). The noninvasive
nature of saliva means that samples of DNA or RNA can be collected
at a remote site, sometimes without professional input, and trans-
ported to a laboratory where on-site testing is performed and the
results reported back to the physician, who in turn can provide rapid
feedback to the subject or patient. The elimination of the phleboto-
mist to collect a sample is the key driver in this instance.
1.1 Salivary DNA There are a number of tools available for genomic DNA collection
Collection from saliva and more are currently in development. These are
based upon the collection of whole saliva, or in some cases buccal
epithelial cells, harvested by a rinse solution or mouthwash
system.
1.2 Salivary RNA Since the discovery of RNA in saliva [16], there has been a rapid
Collection uptake in transcriptomic analysis using saliva specimens. A group
of RNAs termed core RNAs have been found to be present in
both whole saliva and saliva supernatant and verified through
experimental work [16].
The gold standard for salivary RNA collection termed
direct saliva transcriptome analysis (DSTA) [35] has been well
used routinely for collection and isolation of RNA (miRNA and
mRNA) from patients with multiple diseases. The DSTA method
involves processing salivary supernatant obtained by centrifug-
ing saliva collected by the passive drool technique at 2600g for
15min at 4C followed by aspiration from the pellet. The salivary
supernatant so obtained is stored ready for use at cool tempera-
tures, without stabilizing agents, until use. mRNAs can be isolated
by one of a number of commercial kits, but in the study by Lee
etal. [35], mRNAs were isolated using the MagMAX Viral RNA
Isolation kit (Applied Biosystems). The integrity of the mRNAs
harvested was confirmed using a series of reference genes. This
method remains the gold standard for comparative purposes.
1.3 Exosomes The discovery [27] that small microvesicles, exosomes found in
saliva, contain highly important salivary micro-RNAs (miRNAs) and
messenger RNAs (mRNAs) has spawned the development of a series
of tools to capture and interrogate microvesicles, exosomes, and
cell-free DNA (and RNA) and miRNAs for transcriptomic analysis.
6 PaulD.Slowey
A report by Gallo etal. in 2012 [27] confirming that miRNAs

in serum and saliva exist primarily inside exosomes, and that using
the exosomal fractions of these bodily fluids increases the s ensitivity
of miRNA detection, has focused a lot of attention on various
microvesicles, including exosomes.
Only recently tools for the analysis and quantification of
exosomes in blood have become available, and work has begun on
the evaluation of saliva as a readily available source of exosomes,
and early work in this area is highly promising.
The established standard for exosome isolation involves ultra-
centrifugation [41]; however, exosomes have also been isolated by
precipitation, microfiltration, and antibody-coated magnetic beads.
Saliva exosome studies have traditionally utilized ultracentrifuga-
tion for isolation [4244]; however, when exosomes were isolated
by ultracentrifugation from glandular saliva and whole saliva by
Michael etal. [42], the authors concluded that viscosity and cel-
lular contamination in whole saliva make it a less than ideal medium
for exosomal isolation, so a purified saliva specimen may be a more
advantageous specimen to use.
1.4 Cell-Free DNA Cell-free DNA (cfDNA) is an important component for evaluation
of oncological markers in various malignancies [49], for noninva-
sive prenatal testing (NIPT, [50]), and for other diseases including
rheumatoid disease, trauma, myocardial infarction, and fever and
inflammatory disease [49, 5154]. Methods for the isolation of
cfDNA again typically include blood, amniotic fluid, and other
invasive bodily fluids. While isolation of cfDNA has been carried
out using saliva, the process involves centrifugation of a whole
saliva specimen collected by the passive drool technique.
Importantly, at the heart of any successfully developed saliva
diagnostic test or procedure is the need to successfully collect, sta-
bilize, and recover the sample, so particular emphasis will be placed
on these aspects in the text to follow.
2 Materials
2.1 Salivary DNA A number of commercial tools are now available for the collection
Collection Procedures of genomic DNA from saliva specimens (see Note 1).
1. The Oragene device from DNA Genotek (Ottawa, Canada) is
the market-leading technology [28]. To collect a sample, sub-
jects expectorate (spit) into the Oragene device until a vol-
ume of 2mL of saliva has been collected. A cap on the Oragene
device containing proprietary stabilizing buffers is closed, and
this causes a stabilizing buffer to flow into the saliva sample,
resulting in a laboratory ready sample with long-term shelf life
(1 year) (see Note 2).
2. The DNASAL device (Oasis Diagnostics, Vancouver, USA)

is a raking/scraping tool that collects cells from the inside of
the oral cavity (buccal mucosa) [23, 29]. The collection head
of the DNASAL tool is rubbed gently on the inside of the
cheeks for 30s, resulting in the accumulation of cells on the
body of the DNASAL device. In addition, cells are abraded
by the mild raking action and remain free-flowing in the
saliva in the pool formed in the mouth. In order to harvest
these cells and saliva, a small amount (2.5mL) of a safe, stabi-
lizing rinse solution is taken in the mouth, swished around,
and then expectorated (spat) back into a collection tube pro-
vided. The detachable head of the DNASAL device is then
removed into the collection tube, to increase the yield of
DNA.The sample obtained is stable for up to 30 days at room
temperature.
3. Norgen Biotek (Ontario, Canada) has a device called the
Saliva DNA Collection and Preservation Device [30]. The
principles of this device are similar to the Oragene system. In
this case, the subject expectorates into a Collection Funnel
connected to a Collection Tube until a 2-mL sample of saliva
has been collected (marked by a line on the Collection
Funnel). The Collection Funnel is removed and may be recy-
cled. A preservation agent is added to the saliva sample by
means of an ampoule, and then the contents of the tube are
mixed by shaking and are now ready for analysis or transporta-
tion to a laboratory for downstream testing. The Norgen
sample is stable for up to 2 years.
4. The DNAgard Saliva device from Biomatrica is a relatively
new entrant into the field [31]. Once again, the Biomatrica
device is modeled on similar principles to the Oragene and
Norgen DNA devices. Subjects expectorate into a tube through
a removable funnel until a fill mark is reached. The contents
of a dropper bottle are then added to the saliva sample and the
mixture inverted 57 times to stabilize the sample for up to
30 months at room temperature.
5. In addition to methods using passive drool and buccal cell har-
vesting, two well-known technologies use simple swabs. Where
small to medium quantities of DNA are required, these devices
may be suitable.
(a) The Mawi Technologies iSWAB-DNA Isolation Kit [32, 33]
uses a series of routine swabs (iSWABs) for sample collec-
tion. One of the iSWABs is placed in the mouth and
rubbed against the inside of the cheek covering the whole
cheek while rotating the iSWAB.The iSWAB is then placed
into a Collection Vial with a narrow neck and screwed
down in a corkscrew-like motion until the iSWAB reaches
8 PaulD.Slowey
the bottom of the Collection Vial containing a proprietary

buffer solution. In order to mix the sample with the liquid
in the Collection Vial, the iSWAB is moved up and down
inside the Collection Vial 1015 times. The iSWAB is then
removed from the Collection Vial, and the entire proce-
dure is repeated with an additional three iSWABs, by alter-
nating between the left and right cheek. In each case, the
iSWAB samples are introduced into the same Collection
Vial in order to enrich the sample with DNA.Upon com-
pletion, a cap is placed on the Collection Vial and the sam-
ple stored or analyzed. Sample stability is several months at
ambient temperature.
(b) The Isohelix DNA Buccal Swab kit [34] is described by
the manufacturer as using a unique swab matrix design to
efficiently collect buccal cell samples. Two different swab
types are available, and in each case, samples are collected
by rubbing one of the swab types (designated SK-1 and
SK-2) firmly against the inside of the cheek or underneath
the lower or upper lip for 1min. The head of the swab is
then placed into a small Collection Tube, then the swab
head removed from the shaft of the device, either by snap-
ping the shaft at a notch etched into the side of the shaft
(SK-1) or by sliding a plastic cover over the swab head and
detaching the swab head by exerting pressure to dislodge
the swab head (SK-2). Details of sample stability are not
provided.
2.2 Salivary RNA The number of salivary RNA collection methods is fewer than for
Collection Procedures its counterpart, DNA; however, three or four technologies are
worthy of mention:
1. For the Oragene RNA device from DNA Genotek (Ottawa,
Ontario, Canada) [36, 37], subjects are asked to place a small
amount of table sugar in the palm of their hands then touch
the top of their tongue to the sugar, in order to stimulate
greater saliva flow. The sugar and pooled saliva in the mouth
are left there for 1015s without swallowing. The saliva that
pools in the oral cavity is then expectorated into the Oragene
container, a plastic Collection Tube. Expectoration is contin-
ued until a line on the Oragene device is reached (2.0mL).
The sample is then capped and tightened causing a buffer in
the cap of the Oragene device to be released into the saliva
sample causing immediate stabilization of the sample. The
mixture of sample and buffer reagent is then shaken vigorously
to mix the sample, which is reported to have a stability of 60
days at ambient temperature. The crude Oragene RNA mix-
ture may be purified using a number of kits including Qiagen
RNeasy Micro or Qiagen RNeasy Mini Kits using a centrifuga-
tion followed by pelleting step to obtain purified RNA for

downstream analysis (see Note 3).
2. Norgen Biotek (Canada) offers Saliva RNA Collection and
Purification Devices [38] based upon identical principles to
the Saliva DNA Collection Devices branded by the company
(see Subheading2.1, item 3). The only significant difference in
the collection procedure is the addition of an RNA stabilizing
reagent instead of a DNA stabilizing agent. Norgen offers spe-
cific kits for isolation of RNA from saliva samples based upon a
spin column technique.
3. Two devices are available from Oasis Diagnostics (Vancouver,
WA) for transcriptomic workup:
(a) The RNAProSAL device [39] is a system for the simul-
taneous harvesting of two cell-free samples of saliva that
may be used for both RNA and proteins or combined to
provide a higher yield of saliva for transcriptomics or pro-
teomics. In this device, saliva is collected from the pool of
saliva in the oral cavity by means of an absorbent pad con-
nected to a stem. After 13min, saliva collection is com-
plete, signified by a color change in a Sample Volume
Adequacy Indicator (SVAI), within the device, from yel-
low to bright blue. The saturated absorbent pad is squeezed
through a compression tube and then through a narrow
bore filter containing a proprietary filtration medium. The
sample is subsequently bifurcated (split into two) and col-
lected into two equivalent 2-mL Eppendorf tubes where it
may be stabilized. In the case of proteins, immediate stabi-
lization is necessary, and this is facilitated using a protein
stabilizing agent provided with the device. In the case of
RNA, the purified saliva is stable for up to 14 days but may
be stabilized as required by means of off the shelf RNA
stabilizing reagents. The total yield of purified saliva is
1.0mL.
(b) The PureSAL device [40] may be a better option if pro-
tein is required. In this RNA is required. In this case, saliva is
collected in identical fashion to the RNAProSAL device,
but a single sample of saliva is collected by squeezing the
saliva sample obtained through a compression tube into
which has been inserted a proprietary separation medium.
A minimum of 1.0mL of cell-free saliva is collected into a
single 2-mL Eppendorf tube and stabilized as above.
Two important applications have been reported for the
PureSAL device particularly, which equally apply to the sis-
ter RNAProSAL technologythese applications are for
exosomes and cell-free DNA, each of which can provide
increasingly important information on disease and disease pro-
cesses of relevance to diagnosis.
10 PaulD.Slowey
2.3 Exosomes 1. PureSAL Oral Specimen Collection Device (Catalog

Number PRSAL-401).
2. Precipitating reagent (ExoQuick-TC, System Biosciences,
Mountain View, CA).
3. EXOCET lysis buffer (System Biosciences).
2.4 Cell-Free DNA 1. PureSAL Oral Specimen Collection Device (Catalog

(cfDNA) Number PRSAL-401).
2. Falcon tubes.
3. Roche High Pure PCR Template Preparation Kit.
4. Quant-iT PicoGreen dsDNA Assay Kit (Life Technologies).
3 Methods
Recently, the PureSAL device has been compared to whole saliva

and validated for the collection of exosomes [45], quantified using
precipitating reagents (ExoQuick-TC Kits) from System
Biosciences [46]. Isolated exosomes were quantitated by a choles-
teryl ester transfer protein (CETP) assay (EXOCET, System
Biosciences) validated for the purification and quantification of
exosomes [47, 48]. It was found that using the PureSAL device
simplified collection significantly eliminated non-exosomal con-
taminating materials without loss of exosomes. A detailed descrip-
tion of the method comprising saliva collection, isolation of
exosomes, and quantification is detailed below.
3.1 Sample Collect a saliva specimen by one of the methods described above in
Collection Subheading2.2.
andStabilization
3.2 Isolation 1. Combine 1.7g of collected sample with 340L of ExoQuick-TC

ofExosomes and mix by inversion (see Note 6).
2. Incubate overnight at 4C.
3. Centrifuge sample at 16,000g for 5min.
4. Resuspend resultant pellet in EXOCET lysis buffer (85L
per tube) and incubate at 37C for 5min.
5. Centrifuge at 2000g for 5min.
6. Use resultant supernatant for analysis.
Results from the experiments are shown in Table1. The exper-
iment was repeated with a second saliva pool, and similar results
were obtained. It was noted that if whole saliva is not processed at
Table 1
Comparison of the quantity of salivary exosomes collected by the PureSAL device and whole
saliva followed by centrifugation
Process for sample isolation Number of exosomes per mL DNA (g/mL) Protein (mg/mL)
Whole salivacentrifuged 16,000g 3.10109 1.47 4.75
PureSAL device 3.25109 1.19 4.58
sufficient centrifuge speeds, non-exosomal materials remaining in

the exosome pellet will interfere with quantitation of exosomes by
the cholesteryl ester transfer protein (CETP) assay.
3.3 Cell-Free DNA 1. Sample collection.

I. PureSAL: collect a saliva specimen as described above in
Subheading2.2.
II. Whole Saliva:
(a) Collect saliva by the passive drool technique into a
50-mL Falcon tube.
(b) Centrifuge at 3000g for 20min.
(c) Take the supernatant and transfer to another centri-
fuge tube and centrifuge at 16,000g for 5min.
2. Store all samples (I) and (II) at 80C prior to DNA isolation.
3. DNA isolation.
(a) Isolate DNA with the Roche High Pure PCR Template
Preparation Kit by using 700L saliva aliquots per
isolation.
4. DNA quantification using PicoGreen.
(a) Measure DNA quantity using the Quant-iT PicoGreen
dsDNA Assay Kit (see Note 7).
Prepare a standard curve using ten different concen-
trations of lambda DNA provided in the kit. Perform
triplicate readings for increased precision.
Construct a standard curve using the values from the
ten different concentrations of lambda DNA.
Measure the samples relative to the standard curve and
present in a table format.
In the experimental work performed, it was shown that
the PureSAL device removed 98.198.2% of all DNA,
providing a total of 1.81.9% of cfDNA in comparison to
the gold standard passive drool/centrifugation method
which was effective in removing 98.999.1% of all DNA
and providing 0.91.1% of cfDNA.
12 PaulD.Slowey
4 Notes
1. DNA from samples collected using one of the above commer-

cial tools may be isolated using one of a significant number of
DNA isolation kits provided by a number of manufacturers.
The number of possibilities available is too numerous to cover
in this manuscript; however, a number of manufacturers have
developed specific saliva kits or validated certain kits to work
well for saliva specimens. The list includes Qiagen Corporation
(www.Qiagen.com), DNA Genotek (www.DNAGenotek.
com), Norgen Biotek (www.NorgenBiotek.com), Biomatrica
(www.Biomatrica.com), Oasis Diagnostics (www.4saliva.com),
Life Technologies (www.ThermoFisher.com), and others.
2. DNA Genotek received FDA 510(k) clearance for the use of
Oragene in conjunction with a test for warfarin sensitivity
developed by the company GenMark Diagnostics, so the device
may be used clinically for this single application.
3. For RNA isolation, there are fewer kits available that have been
specifically optimized for saliva specimens. The Qiagen miR-
Neasy kit has been used successfully for the isolation of purified
RNA for transcriptome work, RNA sequencing, and other
applications, as has the QIAzol lysis reagent from the same
company. Other methods that have been used include organic
extraction methods (TRIzol LS), spin filter-based methods
(QIAamp Viral (Qiagen)), NucleoSpin (Clontech), and miR-
Vana (Life Technologies) and combined method of organic
extraction and spin filter clean up (miRNeasy micro (Qiagen))
and Quick-RNA MicroPrep (Zymo Research).
4. In reference to Subheading1.4, the performance of one par-
ticular device (the PureSAL device) has been evaluated side-
by-side with the gold standard method (passive drool/
centrifugation) for cell-free DNA according to protocols out-
lined in the manuscript [55]. In the experiments performed,
the PureSAL device was found to be a superior tool for har-
vesting cfDNA.
5. In Subheading2.1, care should be taken to investigate options
for DNA purification based upon the specific application
required. These may include simple ethanol precipitation tech-
niques, spin column methods, 96-well microplates, or auto-
mated methods, such as the Promega Maxwell 16 instrument
or the Qiagen QIAsymphony equipment. Whole saliva con-
tains a significant quantity of mucinous material that can have
an impact on the quality of DNA obtained. It is recommended
that investigators contact the individual manufacturers for
details of any methods and how they may be applied to DNA
isolation from saliva, prior to the commencement of any vali-
dation studies.
6. The method used in this chapter for isolation of exosomes is

only one of a number of exosomal isolation kits now available.
These include the Exo-spin kit from Cell Guidance Systems,
Total Exosome Isolation Reagent from Thermo Fisher, miR-
CURY from Exiqon, PureExo Exosome Isolation kit from
PureExo, and ExoCap Capture Kit from JSR Biosciences.
Investigators are encouraged to validate the best method for
exosome isolation in their own laboratory.
7. The authors also carried out DNA quantification by quantita-
tive PCR (qPCR) as an alternate method of DNA assessment.
Acknowledgments
The author would like to acknowledge the support of Dr David T

Wong (UCLA) for his support and encouragement in preparing
this manuscript.
References
1. Punyadeera C, Slowey PD (2013) Saliva as an 11. Cone EJ, Huestis M (2007) Interpretation of
emerging biofluid for clinical diagnosis and oral fluid tests for drugs of abuse. Ann NY
applications of MEMS/NEMS in salivary diag- Acad Sci 1098:51103
nostics. Chapter 22in Nanobiomaterials in 12. Bradshaw DJ, Marsh PD (1998) Analysis of
Clinical Dentistry. Elsevier, pp453475, pH-driven disruption of oral microbial com-
ISBN: 978-1-4557-3127-5 munities invitro. Caries Res 32:456462
2. Streckfus C, Bigler L (2002) Saliva as a diag- 13. Bratthall D, Hansel Petersson G (2005)
nostic fluid. Oral Dis 8:6976 Cariogram-a multifactorial risk assessment
3. Slowey PD (2013) Commercial saliva collec- model for a multifactorial disease. Community
tions tools. JCalif Dent 41(2):97105 Dent Oral Epidemiol 33:256264
4. Malamud D, Tabak LA (eds) (1992) Saliva as a 14. Larmas M (1992) Saliva and dental caries:
diagnostic fluid. Vol 694, Annals NY Academy diagnostic tests for normal dental practice. Int
of Science Dent J42:199208
5. Samaranayake LP (2006) Saliva tells the bodys 15. Christodoulides N, Floriano PN, Miller CS,
health. In: Stephen Moss (ed) The benefits of Ebersole JL, Mohanty S, Dharshan P, Griffin M,
chewing. pp2435 Lennart A, Ballard KL, King CP Jr, Langub MC,
6. Zachary D, Mwenge L, Muyoyeta M, Shanaube Kryscio RJ, Thomas MV, McDevitt JT (2007)
K, Schaap A, Bond V, Kosloff B, de Haas P, Lab-on-a-chip methods for point-of-care mea-
Ayles HBMC (2012) Infect Dis 8(12):183 surements of salivary biomarkers of periodonti-
7. Testing Oral Fluid for the Presence of HIV tis. Ann NY Acad Sci 1098:411428
Antibodies February 2013. Report by the 16. Li Y, St John MA, Zhou X, Kim Y, Sinha U,
Association of Public Health Laboratories Jordan RC, Eisele D, Abemayor E, Elashoff D,
(APHL) http://www.aphl.org/AboutAPHL/ Park NH, Wong DT (2004) Salivary transcrip-
publications/Documents/ID_Feb2013_ tome diagnostics for oral cancer detection. Clin
Testing-o f-O ral-Fluid-for-the-Presence-of- Cancer Res 10:84428450
HIV-Antibodies-Brief.pdf 17. Streckfus C, Bigler L (2005) The use of solu-
8. See http://www.zrtlab.com/patients-standard- ble, salivary c-erbB-2 for the detection and
tests/saliva post-operative follow-up of breast cancer in
9. See http://www.diagnostechs.com/Pages/ women: the results of a five-year translational
Home.aspx research study. Adv Dent Res 18:1724
10. See https://www.labrix.com/SalivaryHormone 18. Streckfus C, Bigler L, Dellinger T, Dai X, Cox
Testing WJ, McArthur A, Kingman A, Thigpen JT
14 PaulD.Slowey
(2001) Reliability assessment of soluble 30. See https://norgenbiotek.com/product/

c-erbB-2 concentrations in the saliva of healthy saliva-dna-collection-preser vation-and-
women and men. Oral Surg Oral Med Oral isolation-kit
Pathol Oral Radiol Endod 91:174179 31. See http://biomatrica.com/dnagardsaliva.php
19. Streckfus C, Bigler L, Dellinger T, Dai X, 32. See http://www.mawidna.com/products/
Kingman A, Thigpen JT (2000) The presence iswab-dna-collection-kit
of soluble c-erbB-2in saliva and serum among 33. El-Hahmawi B (2014) An efficient non-invasive
women with breast carcinoma: a preliminary sample collection technology for various popu-
study. Clin Cancer Res 6:23632370 lation segments. Paper presented at the Qatar
20. Streckfus C, Bigler L, Dellinger T, Pfeifer M, Foundation Annual Research Conference, At
Rose A, Thigpen JT (1999) CA 15-3 and Doha, Qatar, Accessed 18 Nov 2014
c-erbB-2 presence in the saliva of women. Clin 34. See http://www.isohelix.com/products/
Oral Investig 3:138143 isohelix-dna-buccal-swabs
21. Streckfus C, Bigler L, Tucci M, Thigpen JT 35. Lee YH, Zhou H, Yan X, Zhang L, Chia D,
(2000) A preliminary study of CA15-3, Wong DTW (2011) Direct saliva transcriptome
c-erbB-2, epidermal growth factor receptor, analysis. Clin Chem 57:12951302
cathepsin-D, and p53in saliva among women
with breast carcinoma. Cancer Invest 36. See http://www.dnagenotek.com/US/prod-
18:101109 ucts/RE100.html
22. Streckfus C, Bigler L, Dellinger T, Kuhn M, 37. Patel RS, Jakymiw A, Yao B, Pauley BA,
Chouinard N, Dai X (2004) The expression of Carcamo WC, Katz J, Cheng JQ, Chan EK
the c-erbB-2 receptor protein in glandular sali- (2011) High resolution of microRNA signa-
vary secretions. JOral Pathol Med tures in human whole saliva. Arch Oral Biol
33:595600 56:15061513
23. Ovchinnikov DA, Cooper MA, Pandit P, 38. See https://norgenbiotek.com/product/
Coman WB, Cooper-White JJ, Keith P, saliva-rna-collection-and-preservation-devices
Wolvetang EJ, Slowey PD, Punyadeera C 39. Chiang SH, Thomas GA, Liao W, Grogan T,
(2012) Tumor-suppressor gene promoter Buck RL, Fuentes L, Yakob M, Laughlin MJ,
hypermethylation in saliva of head and neck Schafer C, Nazmul-Hossain A, Wei F, Elashoff
cancer patients. Translat Oncol 5:321326 D, Slowey PD, Wong DT (2015)
24. Hu S, Zhou M, Jiang J, Wang J, Elashoff D, RNAPro*SAL: a device for rapid and standard-
Gorr S, Michie SA, Spijkervet FK, Bootsma H, ized collection of saliva RNA and proteins.
Kallenberg CG, Vissink A, Horvath S, Wong Biotechniques 58:6976
DT (2009) Systems biology analysis of 40. See http://4saliva.com/products/
Sjogrens syndrome and mucosa-associated pure%E2%80%A2sal
lymphoid tissue lymphoma in parotid glands. 41. Thry C, Zitvogel L, Amigorena S (2002)
Arthritis Rheum 60:8192 Exosomes: composition, biogenesis and func-
25. Corstjens LAM, Abrams WR, Malamud D tion. Nat Rev Immunol 2:569579
(2012) Detecting viruses by using salivary 42. Michael A, Bajracharya SD, Yuen PS, Zhou H,
diagnostics. JAm Dent Assoc 143:12S18S Star RA, Illei GG, Alevizos I (2010) Exosomes
26. Shirtcliff EA, Buck RL, Laughlin MJ, Hart T, from human saliva as a source of microRNA
Cole CR, Slowey PD (2015)Salivary cortisol biomarkers. Oral Dis 16:3438
results obtainable within minutes of sample 43. Lau C, Kim Y, Chia D, Spielmann N, Eibl G,
collection correspond with traditional immu- Elashoff D, Wei F, Lin YL, Moro A, Grogan T,
noassays. Clin Ther 37:505514 Chiang S, Feinstein E, Schafer C, Farrell J,
27. Gallo A, Tandon M, Alevizos I, Illie GG (2012) Wong DT (2013) Role of pancreatic cancer-
The majority of MicroRNAs detectable in serum derived exosomes in salivary biomarker devel-
and saliva is concentrated in exosomes. PLoS opment. JBiol Chem 288:2688826897
One 7, e30679, http://www.Plosone.org 44. Lsser C, Alikhani VS, Ekstrm K, Eldh M,
28. See http://www.dnagenotek.com/US/pdf/ Paredes PT, Bossios A, Sjstran M, Gabrielsson
MK-006.pdf S, Ltvall J, Valadi H (2011) Human saliva,
29. Thomas GA, Oberkanins C, Berndt A, Slowey plasma and breast milk exosomes contain RNA:
PD (2014) Validation of a series of genomic uptake by macrophages. JTranslat Med 9:9
StripAssays to salivary DNA collection using 45. Laughlin MJ, Buck RL, Slowey PD (2014) A
the DNASAL Device. Paper presented at the simplified method for the isolation of exosomes
American Society of Human Genetics Meeting, from saliva using a prototype saliva collection
San Diego, October 2014 device (RPSAL-702). Paper presented at the
North American Saliva Symposium Boston, 51. Zhong XY, von Mhlenen I, Li Y (2007)
October 2014 Increased concentrations of antibody-bound
46. See https://www.systembio.com/microrna- circulatory cell-free DNA in rheumatoid arthri-
research/exoquick-exosomes/overview?gclid tis. Clin Chem 53:16091614
=CPTmjv2foMoCFQWUfgodu_gJUw 52. Lam NY, Rainer TH, Chan LY, Joynt GM, Lo
47. Savina A, Vidal M, Colombo MI (2002) The YM (2003) Time course of early and late
exosome pathway in K562 cells is regulated by changes in plasma DNA in trauma patients.
Rab11. JCell Sci 115:25052515 Clin Chem 49:12861291
48. Gupta S, Knowlton AA (2007) HSP60 traffick- 53. Chang CP, Chia RH, Wu TL, Tsao KC, Sun
ing in adult cardiac myocytes: role of the exo- CF, Wu JT (2003) Elevated cell-free serum
somal pathway. Am JPhysiol Heart Circ DNA detected in patients with myocardial
Physiol 292:H3052H3056 infarction. Clin Chim Acta 327:95101
49. Fleischhacker M, Schmidt B (2007) Circulating 54. Moreira VG, Prieto B, Rodriguez JS, Alvarez
nucleic acids (CNAs) and cancera survey. FV (2010) Usefulness of cell free plasma DNA,
Biochim Biophys Acta 1775:181232 procalcitonin and C-reactive protein as markers
50. The American College of Obstetricians and of infection in febrile patients. Ann Clin
Gynecologists Committee Opinion Number Biochem 47:253258
640 September 2015. Cell-free DNA screening 55. Slowey PD, Giese U, Hofner M, Kegler U,
for fetal aneuploidy, http://www.acog.org/ Weber M, Buck RL, Laughlin MJ (2014)
Resources-And-Publications/Committee- Comparison of RNAProSAL saliva collection
Opinions/Committee-on-Genetics/ versus centrifugation for cell-free DNA isolation
Cell-fr ee-DNA-Scr eening-for-Fetal- from saliva specimens. Paper presented at the
Aneuploidy Molecular Medicine Tri- Conference, San
Francisco CA, February 2014
Chapter 2
RNA Sequencing Analysis ofSalivary Extracellular RNA

BlancaMajem, FengLi, JieSun, andDavidT.W.Wong
Abstract
Salivary biomarkers for disease detection, diagnostic and prognostic assessments have become increasingly
well established in recent years. In this chapter we explain the current leading technology that has been
used to characterize salivary non-coding RNAs (ncRNAs) from the extracellular RNA (exRNA) fraction:
HiSeq from Illumina platform for RNA sequencing. Therefore, the chapter is divided into two main sec-
tions regarding the type of the library constructed (small and long ncRNA libraries), from saliva collection,
RNA extraction and quantification to cDNA library generation and corresponding QCs. Using these
invaluable technical tools, one can identify thousands of ncRNA species in saliva. These methods indicate
that salivary exRNA provides an efficient medium for biomarker discovery of oral and systemic diseases.
Key words Saliva, exRNAs, Small and long ncRNA profiling, Biomarkers, RNA sequencing
1 Introduction
Extracellular RNA (exRNA) in human saliva is an emerging field for

noninvasive diagnostic applications. The discovery of saliva-derived
mRNA in normal and oral cancer patients [13] and other forensic
applications [4, 5] opened up a new field for noninvasive molecular
diagnosis. Our laboratory has extensively studied microarray-based
gene profiling followed by real-time quantitative-PCR (RT-qPCR)
for saliva mRNA detection. We have identified certain macromole-
cules associated with salivary mRNA that were protecting against
ribonucleases [6]. Salivary RNA was found in complexes with lipids,
proteins, lipoproteins, and phospholipids as well [7, 8]. Apoptotic
bodies [9] or other vesicular structures in saliva also play a protec-
tion role. Therefore, RNA in the saliva may not be as fragile as it was
previously assumed to be. Despite the numerous studies based on
characterizing and finding mRNA diagnostic biomarkers in saliva,
the introduction of deep sequencing technologies [10, 11] has
revealed a new landscape of salivary exRNA [12]: micro-RNAs
(miRNAs), piwi-interacting-RNAs (piRNAs), circular-RNAs (cir-
cRNAs), and other noncoding RNAs (ncRNAs). To date, only a few
17
18 Blanca Majem et al.
studies characterizing ncRNAs in saliva have used RNA-sequencing

(RNA-Seq) technologies [13]. In this chapter, we present the
detailed methodology for RNA extraction, cDNA library construc-
tion and quality controls (QCs), and data analysis of sequencing
data. Although a variety of platforms are available for RNA-seq, the
Illumina platform is the most used nowadays. Increasing knowl-
edge on salivary composition thanks to this platform will make a
difference in understanding the biology of the diagnostic biomark-
ers found in saliva for local and systemic diseases.
The purpose of this chapter is to provide robust and reliable
methods for isolating and profiling of salivary exRNA, dividing it in
two main sections regarding the type of the library (small and long
ncRNA libraries) constructed. We also describe a protocol for RNA
extraction after saliva collection, including detailed explanations of
DNase treatment, RNA precipitation for sample concentration and
specific QCs for the extracted RNA.The commercial kits for RNA
extraction allow high RNA yield, but the eluted RNA usually is
contained in big volumes and therefore low concentrations, which
is not recommended for subsequent steps, since library preparation
starts with little volumes and requires a high concentration sample.
Either way, the lower limit of detection of the QCs makes the RNA
precipitation crucial for sample concentration, resulting in high
reproducibility among samples and accurate RNA and cDNA quan-
tification, which at the same time is translated into good quality of
raw read data after sequencing. Thus, our protocols are a guide for
RNA-seq of salivary exRNA, but some concepts and methodology
may also be applied to other types of body fluids.
2 Materials
2.1 Saliva Collection 1. 50mL sterile tube.

andProcessing 2. Laboratory vortex.
3. Refrigerated benchtop centrifuge with 50mL tube adapters.
4. SUPERase-In RNase inhibitor, Cat# AM2694 Ambion.
2.2 RNA Sequencing 1. Qiazol.

ofSalivaryncRNAs 2. Chloroform.
2.2.1 RNA Isolation 3. miRNeasy micro kit.
4. Absolute ethanol.
2.2.2 DNase Treatment 1. DNase.

andRNA Precipitation 2. NaOAc 3M.
3. Glycogen.
Salivary RNA-Seq 19
4. Absolute ethanol.
5. Nuclease-free water.
2.2.3 RNA Quantification 1. QuanTi RiboGreen RNA assay kit.

andQCs 2. 96-Well half area microplate (black solid plate), Cat# 3694
Corning.
3. Agilent RNA 6000 Pico kit.
2.2.4 cDNA Library 1. NEBNext Multiplex Small RNA library Prep Set for Illumina.
Preparation 2. Exiqon Spike-in miRNA kit v2, Cat# 208041 Exiqon.
Small ncRNA Library 3. 8-Tube PCR strip.
4. Thermal Cycler PCR machine.
5. 6% Novex TBE PAGE gel, 1.0mM 10-well.
6. SYBR Gold Nucleic Acid Gel Stain (Life Technologies, Inc.
#S-11494).
7. Gel broker tubes, Cat#, 3388-100 SeqMatic.
8. Corning, Costar, Spin-X Centrifuge Tube Filters (Cellulose
Acetate Filters).
9. 3M Sodium Acetate, pH 5.5.
10. 100 and 80% ethanol (freshly prepared).
11. QIAquick PCR purification kit.
Long ncRNA Library 1. NEBNext Ultra Directional RNA Library Prep kit for
Illumina.
2. ERCC spike-in, Cat# 4456740 Ambion.
3. NEBNext Singleplex or NEBNext Multiplex Oligos for
Illumina.
4. Actinomycin D (Sigma# A1410, dissolved in dimethylsulfox-
ide [DMSO] to 5g/L).
5. 8-tube PCR strip.
6. 80% thanol (freshly prepared).
7. Thermal Cycler PCR machine.
8. DynaMag-2 Magnet.
9. Agencourt AMPure XP Beads (Beckman Coulter, Inc.
#A63881).
2.2.5 cDNA Library 1. Qubit dsDNA BR assay kit.

Quantification andQCs 2. 96-well half area microplate (black solid plate), Cat# 3694
Corning.
3. Agilent High Sensitivity DNA Kit.
2.2.6 RNA Sequencing 1. EB buffer.

byIllumina 2. Tween 20 Surfact-Amps detergent solution.
3. HiSeq2000 Illumina system.
2.2.7 De-Multiplexing 1. Cutadapt.

andData Processing 2. Bowtie mapping 16s rRNA/Microbiome.
3. Bowtie mapping Human Genome.
3 Methods
3.1 Saliva Collection Saliva collection from human subjects has to be approved by the
andProcessing Institutional Review Board.
We have used the following inclusion criteria for normal sub-
ject selection: age 30 years, and no history of malignancy, immu-
nodeficiency, autoimmune disorders, hepatitis, HIV infection, or
smoking.
1. Ask subjects to refrain from eating, drinking, smoking, or oral
hygiene procedures for at least 1h prior to collection.
2. Instruct subjects to rinse the mouth thoroughly with water
and to void the mouth of saliva. The subject should be seated
comfortably with eyes open and head tilted slightly forward.
For unstimulated saliva collection subjects should rest for
5min and minimize orofacial movements.
3. To collect un-stimulated saliva (see Note 1) allow, saliva to
accumulate in the floor of the mouth and ask the subject to spit
into a preweighed or graduated test tube every 60s. Collection
for 5min usually yields sufficient saliva (~5mL) for analysis.
4. Following collection, centrifuge saliva samples at 2600g for
15min at 4C.Saliva supernatant will then be separated from
the cellular phase.
5. Add SUPERase-In RNase inhibitor (at a ratio of 1L/mL) to
1mL of cell-free saliva (CFS) supernatant for preserving
exRNA degradation.
6. Store aliquots of 1mL CFS at 80C for further analysis.
3.2 RNA Sequencing 1. Thaw 4 aliquots of 1mL of saliva, resting the tubes on ice and
ofSalivary exRNA not for more than half an hour (see Note 2).
3.2.1 RNA Isolation 2. Split the sample in 500L of CFS and centrifuge for 5min at
10,000g (see Note 3). Collect the supernatant to proceed
with step 2, and discard the pellet fraction.
3. Split 0.5mL of cell free saliva (CFS) in two tubes250L in
each.
4. Add 750L of Qiazol to 250L of CFS.Vortex for 30s and
incubate 5min at RT.
Salivary RNA-Seq 21
5. Add 200L chloroform and mix by vortex for 30s, and then
incubate 5min at RT.
6. Centrifuge the sample at 12,000g for 15min at 4C.
7. Carefully collect 600L (at least) of upper aqueous phase and
transfer to the new tubes.
8. Add 900L (1.5 Vol) of 100% ethanol and mix thoroughly by
pipetting up and down several times. Do not centrifuge.
Continue without delay to the next step.
9. Pipette 700L of the sample into an RNeasy MinElute spin
column. Centrifuge at 9300g for 30s at RT.Discard the
flow-through. Repeat this step using the remaining sample.
10. Pipette 700 L buffer RWT into the RNeasy MinElute spin
column and centrifuge at 9300g for 30s to wash. Discard
the tube with flow-through and place the column in a new
2mL collection tube.
11. Pipette 700 L Buffer RPE onto the RNeasy MinElute spin
column. Close the lid gently and centrifuge at 9300g for 30s
to wash the column. Discard the flow-through.
12. Pipette 300 L Buffer RPE onto the RNeasy MinElute spin
column. Close the lid gently and centrifuge at 9300g for 30s
to wash the column. Discard the flow-through.
13. Pipette 500 L of 80% ethanol onto the RNeasy MinElute
spin column (see Note 4). Close the lid and centrifuge at
9300g for 2min to wash membrane. Discard the collection
tube with the flow-through.
14. Place the RNeasy MinElute spin column into a new 2mL col-
lection tube. Centrifuge at full speed for 5min to dry the
membrane. Discard the collection tube.
15. Place the column in a new 1.5-mL tube. Add 30L preheated
water (~50C) directly to the center of the membrane. Close
the lid and incubate for 12min at RT, and then centrifuge for
1min at full speed.
16. Maintain the column in the same tube. Add 30L more (see
Note 5) of preheated water directly to the center of the mem-
brane. Close the lid and incubate for 12min at RT (see Note
6), and then centrifuge for 1min at full speed. Proceed directly
to DNase treatment and RNA precipitation step (see Note 7).
3.2.2 DNase Treatment 1. Mix the next components to perform off-column DNase treat-
andRNA Precipitation ment to the eluted RNA of 8 samples at the same time:
2L TURBO DNase18L (for 8 samples).
11L Buffer99L (for 8 samples).
27L H2O (nuclease-free)243L (for 8 samples).
2. Add 40L DNase Mix/sample (100L final volume=60 L

RNA+40 L DNase Mix).
3. Leave it for 15min at RT. Continue with step 4 for RNA
precipitation.
4. Add 10L (0.1 Vol) of sodium acetate 3M pH5.5.
5. Add 1 L (5g) of glycogen (Glycogen is at 5g/L
concentration).
6. Vortex briefly.
7. Add 250L (2.5Vol) of 100% ethanol (see Note 8).
8. Vortex briefly.
9. Incubate at 80C overnight (O/N) or for at least 1.5h at
80C.
3.2.3 RNA Quantification 1. Prepare serial dilution of rRNA standards (12.5200ng/mL).

andQCs 2. Make 70L aliquots of each standard and stock at 80C for
QuanTi RiboGreen future use.
RNAAssay 3. Make 5 L aliquots of fluorescent dye (Component A in
RiboGreen kit) and stock them at 80C (see Note 9).
4. Take one set of standards and one Fluorescent Dye aliquot
from freezer and thaw them at room temperature (RT) in the
dark (important for the Dye) (see Note 10).
5. Prepare RNA sample dilutions at 1/30in 1 TE buffer: Mix
1 L of RNA/sample and 29L of 1 TE buffer for each
sample.
6. Prepare enough working solution (WS) for all the experiment
at a ratio of 1:200 dilution of Fluorescent Dye: 1 TE into a
15mL tube (in the darkness).
7. Plate 15L of the standards (multichannel micropipette is rec-
ommended for reproducibility) in triplicate, and 15L of
diluted samples in duplicate.
8. Add 15L of the WS into each well (standard and samples)
and incubate the plate for 15min at RT in the dark (with lid).
9. Read the plate at 480520nm in a spectrophotometer (see
Note 11).
Agilent Bioanalyzer, 1. Take out the reagents 30min prior to running the Chip and
Eukaryotic RNA Pico Chip allow them to reach RT in the dark.
2. Follow the manufacturer instructions for preparing the gel-
dye-matrix properly, and running the chip (45min in total).
Criteria of QCs for extracted RNA:
Quant-iT Ribogreen RNA assay: salivary RNA concentra-
tion normally ranges from 50 to 80ng/mL saliva. If total
RNA amount is <5ng it is not recommended to proceed
with the library construction.
Salivary RNA-Seq 23
RNA 6000 Pico Chip, Bioanalyzer: detection of intact

ribosomal RNA peak indicates residual cell contamination
(eukaryotic: 18S (1869nt), 28S rRNA (5070nt); pro-
karyotic: 16S (1542nt), 23S rRNA (2906nt)) and
excludes the sample for further analysis.
3.2.4 cDNA Library Prepare the Exiqon Spike-in miRNA kit v2: Dissolve the miR-
Preparation CURY LNA Array Spike-in microRNA Kit v2in 30L/vial of
nuclease-free water (supplied) upon receipt. Vortex to thoroughly
Small ncRNA Library
dissolve the lyophilized RNA, pulse briefly in a microfuge, and
leave the suspension on ice for 30min to dissolve. Aliquot the
dissolved spike-in miRNAs and store at 80C until use and avoid
repeated cycles of freeze/thawing.
Ligate the3 SR Adaptor 1. Mix the following components in a sterile nuclease-free PCR
tube:
Saliva RNA 5.5L

Exiqon Spike-in 0.5L
(Green) 3 SR adaptor for Illumina 1L
Total volume 7L
2. Incubate in a preheated thermal cycler for 2min at

70C.Transfer tube to ice.
3. Add the following components:
RNA+ 3 SR adaptor mix 7L

(Green) 3 ligation reaction buffer (2) 10L
(Green) 3 ligation enzyme mix 3L
Total volume 20L
4. Incubate for 1h at 25C in a thermal cycler.
Hybridize theReverse 1. Add the following components to the ligation mixture from
Transcription Primer step 4 and mix well:
3 Ligation reaction mix from step 1 20L

Nuclease-free water 4.5L
(Pink) SR RT primer for Illumina 1L
Total volume now should be 25.5L
2. Heat samples for 5min at 75C.Transfer to 37C for 15min,

followed by 15min at 25C.
Ligate the5 SR Adaptor 1. With 5min remaining, resuspend the (yellow) 5 SR adaptor in
120L of nuclease-free water and store at 80C (This step is
only necessary when the kit is first opened).
2. Aliquot 1.1LN of the (yellow) 5 SR Adaptor into a sepa-
rate, 200L nuclease-free PCR tube, with N equal to the
number of samples being processed for the current
experiment.
3. Incubate the adaptor in the thermal cycler at 70C for 2min
and then immediately place the tube on ice. Keep the tube on
ice and use the denatured adaptor within 30min of
denaturation.
4. Add the following components to the ligation mixture from
step 6 and mix well:
Reaction mix from step 2 25.5L

(Yellow) 5 SR Adaptor for Illumina 1L
(denatured)
(Yellow) 5 ligation reaction buffer (10) 1L
(Yellow) 5 ligation enzyme mix 2.5L
Total volume 30L
5. Incubate for 1h at 25C in a thermal cycler.
Perform Reverse Mix the following components in a sterile, nuclease-free tube:

Transcription
Adaptor ligated RNA from step 3 30L
(Red) first strand synthesis reaction buffer 8L
(Red) murine RNase inhibitor 1L
(Red) ProtoScript II reverse transcriptase 1L
Total volume 40L
Incubate for 60min at 50C.

Immediately proceed to PCR amplification.
Safe Stopping Point: If you do not plan to proceed immediately to
PCR amplification, then heat inactivate the RT reaction at 70C for
15min. Samples can be safely stored at 15 to 25C.
Perform PCR Amplification Add the following components to the RT reaction mix from step
4 and mix well:
Salivary RNA-Seq 25
RT reaction mix from step 4 40L

(Blue) LongAmp Taq 2 Master Mix 50L
(Blue) SR Primer for Illumina 2.5L
(Blue) Index Primer 2.5L
Nuclease-free water 5L
Total volume now should be 100L
PCR Cycling Conditions
Initial denaturation 94C 30s 1 Cycle

Denaturation 94C 15s
Annealing 62C 30s 15 Cycles
Extension 70C 15s
Final extension 70C 5min 1 Cycle
Hold 4C
QIAQuick PCR Purification Purify the PCR amplified cDNA construct (100L) using a
QIAQuick PCR Purification Kit.
1. Add 500L Buffer PB to the PCR reaction and mix.
2. Apply the sample to the QIAquick column and centrifuge at
15,600g for 3060s.
3. Add 750L Buffer PE to the QIAquick column and centri-
fuge at 15,600g for 3060s.
4. Centrifuge the column with the lid of the spin column open
for 5min at 15,600g (see Note 12).
5. Place each QIAquick column in a clean 1.5mL microcentri-
fuge tube.
6. To elute amplified DNA add 26L Nuclease-free Water. Let
the column stand for 1min, and then centrifuge at 15,600g
for 1min.
Size Selection Using 6% Prepare 500mL Running Buffer (100mL 5 Running

Polyacrylamide Gel Buffer+400mL Water), leave 100mL Running Buffer with 10L
andPurification ofSize SYBR Gold stain for later use.
Selected Library
1. Mix the purified PCR product (25L) with 10L of Gel
Loading Dye, Blue (6).
2. Load 5L of Quick-Load pBR322 DNA-MspI Digest in a
well on the 6%PAGE 10-well gel.
3. Load two wells with 17L each of mixed amplified cDNA and
loading dye on the 6% PAGE 10-well gel.
4. Run the gel for ~1.5h at 90V.Do not let the blue dye exit the
gel.
5. Remove the gel from the apparatus and stain the gel with SYBR
Gold nucleic acid gel stain in a clean container for 23min and
view the gel on a UV transilluminator. The 140 and 150nt
bands correspond to adapter-ligated constructs derived from
the 21 and 30nt RNA fragments, respectively. For miRNAs,
isolate the bands corresponding to ~140bp. For piRNAs, iso-
late the band corresponding to ~150bp (see Fig.1).
6. Place the two gel slices from the same sample in a Gel Broker tube
(SeqMatic) with a 2mL tube, then centrifuge at 14,000g for
1min, and then soak in 400L DNA Gel Elution buffer (1).
7. Rotate in eppendorf shaker for at least 2h at RT.
8. Transfer the eluate and the gel debris to SpinX column with
1cm diameter Whatman filter.
9. Centrifuge the filter for 2min at >15,600g.
10. Recover eluate and add 1L linear acrylamide, 40L 3M
sodium acetate pH 5.5, 500L of 100% ethanol, and 500L
of isopropanol. Vortex well.
11. Precipitate at 20C for at least 4h or 80C at least 1.5h.
12. Spin >15,600g for 30min at 4C.
13. Remove the supernatant, taking care not to disturb the pellet.
14. Wash the pellet with 500L 80% ethanol.
Fig. 1 Transilluminator view of miRNA and piRNA bands. The lanes S1 to S4 cor-
respond to 4 different small ncRNA libraries. Each library has been run per dupli-
cate and bands were cut below 140bp and above 300bp. miRNAs isolated
bands correspond to ~140bp. piRNAs isolated bands correspond to ~150bp
Salivary RNA-Seq 27
15. Spin >15,600g for 10min at 4C.

16. Air-dry pellet for up to 10min at RT to remove residual
ethanol.
17. Resuspend pellet in 12LEB Buffer (2 of 12L will be used
for cDNA library quantification in Subheading3.2.5).
Long ncRNA Library Prepare the ERCC spike-in: Dissolve in nuclease-free water the
lyophilized product making 1:100 dilution stocks. Vortex to thor-
oughly dissolve the lyophilized RNA, pulse briefly in a microfuge,
and leave the suspension on ice for 30min to dissolve. Aliquot
(12L) the dissolved spike-in RNAs and store at 80C until use
and avoid repeated cycles of freeze/thawing.
Preparation ofFirst Strand

Total saliva RNA+ERCC spike-in (4.5L of 5L
Reaction Buffer
RNA+0.5 L Spike-in)
andRandom Primer Mix
(Pink) NEBNext first strand synthesis reaction buffer 4L
(5)
(Pink) NEBNext random primers 1L
Total volume 10L
RNA Fragmentation 1. Incubate the samples at 94C for 2min.

2. Transfer the tube on ice.
3. Proceed to First Strand cDNA Synthesis.
First Strand cDNA Dilute Actinomycin D stock solution (5g/L) to 0.1g/L

Synthesis in nuclease-free water for immediate use.
The fragmented and primed mRNA 10L

(Pink) murine RNase inhibitor 0.5L
Actinomycin D (0.1g/L) 5L
(Pink) ProtoScript II reverse transcriptase 1L
Final volume 20L
Incubate the sample in a preheated thermal cycler as follows:

10min at 25C.
15min at 42C.
15min at 70C.
Hold at 4C.
Second Strand cDNA The First Strand Synthesis reaction mixes 20L
Synthesis
Nuclease-free water 48L
(Orange) second strand synthesis reaction 8L
buffer (10)
(Orange) second strand synthesis enzyme mix 4L
Total volume 80L
1. Mix thoroughly by gentle pipetting.

2. Incubate in thermal cycler for 1h at 16C, with heated lid set
at 40C.
Purify theDouble-Stranded 1. Vortex AMPure XP beads to resuspend.

cDNA Using 2.5 2. Add 200L (2.5) of resuspended AMPure XP beads to the
Agencourt AMPure second strand synthesis reaction (80L). Mix well on a vor-
XPBeads tex mixer or by pipetting up and down at least 10 times.
3. Incubate for 5min at RT.
4. Quickly spin the tube in a microcentrifuge to collect any sam-
ple on the sides of a tube. Place the tube on an appropriate
magnetic rack (DynaMag-2 Magnet) to separate beads from
supernatant. After the solution is clear (about 5min), carefully
remove and discard the supernatant. Be careful not to disturb
the beads that contain DNA targets.
5. Add 200L of freshly prepared 80% ethanol to the tube while
in the magnetic rack. Incubate at RT for 30s, and then care-
fully remove and discard the supernatant.
6. Repeat step 5 once for a total of 2 washing steps.
7. Air-dry the beads for 10min while the tube is on the magnetic
rack with lid open (recommend hood).
8. Elute the DNA target from the beads into 60L nuclease-free
water. Mix well on a vortex mixer or by pipetting up and down.
Quickly spin the tube and then place it in the magnetic rack
until the solution is clear.
9. Remove 55.5L of the supernatant and transfer to a clean
nuclease-free PCR tube.
Perform End Repair/dA-Tail

The purified double-stranded cDNA 55.5L
ofcDNA Library
(Green) NEBNext end repair reaction buffer (10) 6.5L
(Green) NEBNext end prep enzyme mix 3L
Total volume 65L
Incubate the sample in a thermal cycler as follows:

30min at 20C.
Salivary RNA-Seq 29
30min at 65C.
Hold at 4C.
Proceed immediately to Adaptor Ligation.
Perform Adaptor Ligation Dilute the NEBNext Adaptor for Illumina (15M) to 1.5M
with a 10-fold dilution (1:9) with nuclease-free water for immedi-
ate use.
The dA-Tailed cDNA 65L

(Red) Blunt/TA Ligase Master Mix 15L
(Red) Diluted NEBNext adaptor 1L
Total volume 83.5L
Incubate 15min at 20C in a thermal cycler.

The adaptor is provided in NEBNext Singleplex or NEBNext
Multiplex Oligos for Illumina
Purify theLigation 1. To the ligation reaction (83.5L), add 16.5L nuclease-free

Reaction Using AMPure water to bring the reaction volume to 100L.
XPBeads 2. Add 100L (1.0) resuspended AMPure XP beads and mix well
on a vortex mixer or by pipetting up and down at least 10 times.
4. Quickly spin the tube in a microcentrifuge and place the tube
on an appropriate magnetic rack to separate beads from super-
natant. After the solution is clear (about 5min), discard the
supernatant that contains unwanted fragments (Caution: do
not disturb the beads).
6. Repeat step 5 once for a total of two washing steps.
7. Briefly spin the tube, and put the tube back in the magnetic
rack.
8. Completely remove the residual ethanol, and air-dry beads for
10min while the tube is on the magnetic rack with the lid
open (recommend hood).
9. Elute DNA target from the beads with 50L nuclease-free
water. Mix well on a vortex mixer or by pipetting up and down,
and put the tube in the magnetic rack until the solution is clear.
10. Transfer the 50L supernatant to a clean PCR tube. Discard
the beads.
11. To the 50L supernatant, add 50L (1.0) of the resus-

pended AMPure XP beads and mix well on a vortex or by
pipetting up and down at least 10 times.
on an appropriate magnetic rack to separate beads from the
supernatant. After the solution is clear (about 5min), discard
the supernatant that contains unwanted fragments (Caution:
do not discard the beads).
16. Briefly spin the tube, and put the tube back in the magnetic rack.
17. Completely remove the residual ethanol, and air-dry beads for
10min while the tube is on the magnetic rack with the lid
open (recommend hood).
18. Elute DNA target from the bead with 25L nuclease-free
and put the tube in the magnetic rack until the solution is clear.
19. Without disturbing the bead pellet, transfer 20L of the
supernatant to a clean PCR tube and proceed to PCR enrich-
ment (see Note 13).
Optional stopping point: at this point cDNA library can be
stored at 20C.
Perform USER Excision The Universal PCR primer and Index (X) Primer are contained in
andPCR Library the NEBNext SinglePlex or NEBNext Multiplex Oligos for
Enrichment Illumina.
The size selected cDNA 20L

(Blue) NEBNext USER enzyme 3L
(Blue) NEBNext High-Fidelity PCR Master Mix, 2 25L
(Blue) Universal PCR Primer (25M) 1L
(Blue) Index (X) Primer (25M) 1L
Total volume 50L
PCR cycling conditions
User digestion 37C 15min 1 Cycle

Initial denaturation 98C 30s 1 Cycle
Salivary RNA-Seq 31
Denaturation 98C 10s

Annealing 65C 30s 15 Cycle
Extension 72C 30s
Final extension 72C 5min 1 Cycle
Hold 4C
Purify thePCR Reaction 1. Vortex AMPure XP beads to resuspend.

Using AMPure XP Beads 2. Add 50L (1.0) of resuspended AMPure XP beads to the
PCR reaction (50 L). Mix well on a vortex mixer or by
pipetting up and down at least 10 times.
on an appropriate magnetic rack to separate beads from super-
natant. After the solution is clear (about 5min), carefully
remove and discard the supernatant. Be careful not to disturb
the beads that contain DNA targets.
7. Air-dry the beads for 5min while the tube is on the magnetic
rack with the lid open.
8. Elute the DNA target from the beads into 23L nuclease-free
quickly spin the tube in a microcentrifuge and place it in the
magnetic rack until the solution is clear.
9. Transfer 20L of the supernatant to a clean PCR tube, pro-
ceed with the QCs and quantification step and/or store at
20C.
3.2.5 cDNA Library 1. Prepare serial dilution of DNA standards (120ng/L).

Quantification andQCs 2. Make 500L aliquots of each standard and stock at 4C for
Qubit dsDNA BR Assay future use (use within a month).
3. Make 10L aliquots of fluorescent dye in Qubit DNA kit and
stock at 80C (see Note 14).
4. Remove the high concentrated STD curve from 4C and one
Dye aliquot from 80C, and bring them to RT in the dark
(important for the Dye) (see Note 10).
5. Prepare enough WS for all the experiment at a ratio of 1:200
(means of 1L of Dye/Reagent per 200L of Buffer (see
Note 15).
6. Dilute the DNA library samples with 1:60 against WS.
7. Prepare the low concentrated STD curve (1/10 ratio of high

concentrated STD, see Table1).
8. Plate 30L of the low concentration standard curve (multi-
channel micropipette is recommended for reproducibility) in
triplicate, and 30L of sample in duplicate.
9. Incubate the plate for 15min at RT in the dark (cover the plate
well to avoid DMSO evaporation and therefore, variation in
the lecture).
10. Read the plate at 485530nm in a spectrophotometer.
Agilent Bioanalyzer, High 1. Take out the reagents 30min prior to running the Chip and
Sensitivity DNA Chip allow them to reach RT in the dark.
2. Dilute the DNA samples 1/10 with nuclease-free water (Only
for the Long RNA libraries).
3. Follow the manufacturer instructions for preparing the gel-
dye-matrix properly, and running the chip (45min in total)
(see Fig.2).
Criteria of QCs of constructed library:
Qubit dsDNA BR assay: concentration >10nM.
High Sensitivity DNA Chip, Bioanalyzer: Small RNA
library should have a major peak of 140200bp; Long
RNA library should have a major peak of 300400bp.
3.2.6 RNA Sequencing 1. Sample pooling: 8 libraries of small ncRNA are pooled at 10
byIllumina nM in total, per lane; 4 libraries of long ncRNA are pooled at
10nM in total, per lane (see Note 16). Samples are pooled
with QIAGEN EB Buffer 0.1% Tween 20in a total volume of
at least 20L (preferable).
2. Sequencing: samples are sequenced by HiSeq Illumina system,
and stranded and Single-End 50 base paired (SE50) used for
the procedure (5 days long).
Table 1
Qubit DNA standards
Initial high concentration Volume of diluted DNA Volume of WS Final low concentration
(ng/L) standard (L) (ng/L)
20 10L of 20ng/L 90 2
10 10L of 10ng/L 90 1
5 10L of 5ng/L 90 0.5
2 10L of 2ng/L 90 0.2
1 10L of 1ng/L 90 0.1
0 10L of 0ng/L 90 0
Salivary RNA-Seq 33
Fig. 2 Agilent bioanalyzer: eukaryotic RNA pico chip and high sensitivity DNA chip. (a) RNA profile of salivary
exRNA after RNA precipitation. The average length of salivary exRNA is 25200 nucleotides. (b) cDNA library
of small ncRNAs (left) and long ncRNAs (right) after size selection and cDNA library purification
3.2.7 Barcode 1. Raw data: one FastQ file is obtained per lane (8 lanes/flow
De-Multiplexing andData cell).
Processing 2. De-multiplexing: Each lane needs to be de-multiplexed follow-
ing the indexing codes of the samples (148 index in one set).
3. Adaptor Trimmed reads: raw data of each individual sample is
submitted to Cutadapt software to remove the adaptor
sequences from RNA-Seq raw data.
4. Quality control: QC on adaptor-trimmed reads follows the
next aspects: (a) Per Base Sequence quality, (b) Per Read qual-
ity, (c) per Base N content, and (d) Adaptor content.
5. Mapping:
Small RNA libraries: Bowtie mapping and Human Genome
are used to align the reads to the human genome. Then,
RNA read counts are measured using mapping results and
RNA annotation (see Table2).
Long RNA libraries: Bowtie mapping to 16S rRNA/
microbial genome is used before mapping to the Human
Genome, and RNA read counts are measured using map-
ping results and RNA annotation (see Table2).
4 Notes
1. Unstimulated saliva is collected between 9a.m. and 10a.m.

following published protocols [14].
Table 2
Output data of small and long ncRNA libraries
Small ncRNA libraries Long ncRNA libraries

Average number of RNA detected Average number of genes detected
miRNA 386 RefSeq genes 3050
piRNA 99 lncRNA genes 1419
Other ncRNAs a
145
Includes snoRNAs, tRNAs, snRNA, and others
a
2. Each aliquot allows two RNA extractionsstarting volume of

500L CFS.
3. This step gets rid of residual bacteria and cell debris.
4. Adding a washing step with ethanol 80% step to the commer-
cial protocol improves washing salts and concentrates the RNA
in the silica-gel membrane.
5. Two times of elution may reduce the concentration of RNA
but will translate into higher yield after RNA precipitation
since more volume for eluting the RNA allows better recovery
of the RNA trapped in the silica-gel membrane.
6. Waiting time (12min) for the membrane to get well soaked
as well as using preheated water facilitates the elution of all
RNA content.
7. RNA precipitation after DNase treatment will clean the pro-
tein content in the sample regarding the DNase enzyme and
will result in a high RNA concentration, suitable for starting
cDNA library construction.
8. The use of ethanol instead of isopropanol is because the pre-
cipitated pellet is firmer and adheres more strongly to the tube
wall with ethanol than isopropanol. Ethanol is more volatile
which facilitates removal and less salt will co-precipitate with
ethanol than with isopropanol.
9. Avoiding freezing and thawing of the rRNA standard and fluo-
rescent dye improves notably the reproducibility of the
Ribogreen quantification assay.
10. Little changes in temperature will affect the fluorescence lec-
ture, so it is important to avoid heating the tubes containing
Dye before plate lecture.
11. Both standard curve and samples will end up in dilutions
when mixed with the DyeSTD curve points: 100, 50, 25,
12.5, 6.25, and 0ng/mL, RNA samples: 1/60 dilution.
12. Centrifugation with the lid open ensures that no ethanol
remains during DNA elution. Residual ethanol may interfere
with the correct loading of the sample on the PAGE gel.
Salivary RNA-Seq 35
13. Be sure not to transfer any beads. Trace amounts of bead carry
over may affect the optimal performance of the polymerase
used in the NEBNext High-Fidelity 2 PCR Master Mix in the
subsequent PCR step.
14. Avoiding freezing and thawing of the fluorescent dye notably
improves the reproducibility of the Qubit DNA quantification
assay.
15. WS already contains the dye so, needs to be prepared freshly
and protected from light.
16. Each pooled library contributes to 1.25 and 2.5nM of the
total lane, for small and long ncRNA libraries respectively.
Acknowledgments
This work was supported by National Institutes of Health grant

(UH2/UH3 TR000923). We thank Kai Kao, Leo lee, and Hui
Zhou for technical suggestions.
Conflict of InterestsDavid Wong is cofounder of RNAmeTRIX
Inc., a molecular diagnostic company. He holds equity in
RNAmeTRIX, and serves as Scientific Advisor. The University of
California also holds equity in RNAmeTRIX.Intellectual property
that David Wong invented and which was patented by the University
of California has been licensed to RNAmeTRIX.
References
1. Hu Z, Zimmermann BG, Zhou H, Wang J, 7. Rosi A, Guidoni L, Luciani AM, Mariutti G,

Henson BS, Yu W, Elashoff D, Krupp G, Wong Viti V (1988) RNA-lipid complexes released
DTW (2008) Exon-level expression profiling: a from the plasma membrane of human colon
comprehensive transcriptome analysis of oral carcinoma cells. Cancer Lett 39:153160
fluids. Clin Chem 54:824832 8. Whitelegge JP, Zabrouskov V, Halgand F, Souda
2. Li Y, Zhou X, St John MAR, Wong DTW P, Bassilian S, Yan W, Wolinsky L, Loo JA, Wong
(2004) RNA profiling of cell-free saliva using DTW, Faull KF (2007) Protein-sequence poly-
microarray technology. JDent Res 83:199203 morphisms and post-translational modifications in
3. Li Y, St John MA, Zhou X, Kim Y, Sinha U, proteins from human saliva using top-down
Jordan RC, Eisele D, Abemayor E, Elashoff D, Fourier-transform ion cyclotron resonance mass
Park NH, Wong DTW (2004) Salivary tran- spectrometry. Int JMass Spectrom 268:190197
scriptome diagnostics for oral cancer detection. 9. Halicka HD, Bedner E, Darzynkiewicz Z
Clin Cancer Res 10:84428450 (2000) Segregation of RNA and separate pack-
4. Juusola J, Ballantyne J(2005) Multiplex aging of DNA and RNA in apoptotic bodies
mRNA profiling for the identification of body during apoptosis. Exp Cell Res 260:248256
fluids. Forensic Sci Int 152:112 10. Park NJ, Zhou X, Yu T, Brinkman BM,
5. Juusola J, Ballantyne J(2007) mRNA profiling Zimmermann BG, Palanisamy V, Wong DTW
for body fluid identification by multiplex quan- (2007) Characterization of salivary RNA by
titative RT-PCR.J Forensic Sci 52:12521262 cDNA library analysis. Arch Oral Biol 52:3035
6. Park NJ, Li Y, Yu T, Brinkman BMN, Wong 11. Spielmann N, Ilsley D, Gu J, Lea K, Brockman
DTW (2006) Characterization of RNA in J, Heater S, Setterquist R, Wong DTW (2012)
saliva. Clin Chem 52:988994 The human salivary RNA transcriptome
revealed by massively parallel sequencing. Clin 13. Majem B, Rigau M, Revents J, Wong DTW
Chem 58:13141321 (2015) Non-coding RNAs in saliva: emerging
12. Bahn JH, Zhang Q, Li F, Chan TM, Lin X, Kim biomarkers for molecular diagnostics. Int
Y, Wong DTW, Xiao X (2014) The landscape of JMol Sci 16:86768698
microRNA, Piwi-interacting RNA, and circular 14. Navazesh M (1993) Methods for collecting
RNA in human saliva. Clin Chem 61:221230 saliva. Ann NY Acad Sci 694:7277
Chapter 3
Qualitative andQuantitative Proteome Analysis ofOral

Fluids inHealth andPeriodontal Disease by Mass
Spectrometry
ErdjanSalih
Abstract
The significance of protein identification and characterization by classical protein chemistry approaches is
clearly highlighted by our detailed understanding of the biological systems assembled over a time period
of almost a century. The advent of state-of-the-art mass spectrometry (MS) with sensitivity, speed, and
global protein analysis capacity without individual protein purification has transformed the classical protein
chemistry with premise to accelerate discovery. These combined with the ability of the oral fluids such as
whole saliva (WS) and gingival crevicular fluid (GCF) to reflect both systemic and locally derived proteins
have generated significant interest to characterize these fluids more extensively by MS technology. This
chapter deals with the experimental details of preanalytical steps using multidimensional protein separation
combined with MS analysis of WS and GCF to achieve detailed protein composition at qualitative and
quantitative levels. These approaches are interfaced with gold standard stable-isotope labeling technologies
for large-scale quantitative MS analysis which is a prerequisite to determine accurate alterations in protein
levels as a function of disease progression. The latter incorporates two stable-isotope chemistries one
specific for cysteine containing proteins and the other universal amine-specific reagent in conjunction with
oral fluids in health and periodontal disease to perform quantitative MS analysis. In addition, specific
preanalytical steps demanded by the oral fluids such as GCF and WS for sample preparations to overcome
limitations and uncertainties are elaborated for reliable large-scale quantitative MS analysis.
Key words Mass spectrometry, Protein analysis, Oral fluids, Saliva, Gingival crevicular fluid,
Stable-isotope labeling chemistries, Qualitative and quantitative proteomics
1 Introduction
Protein identification and characterization by classical protein

chemistry approaches for almost a century have been one of the
major contributors and an indispensable field in the study of
biomedical/biological systems. However, the extensive insights
gained within the cellular molecular mechanisms and development
of diagnostic biomarkers to a variety of human diseases have been
at the expense of time and laborious series of experimentation with
very incremental progress. The advances made in mass spectrometry
37
38 Erdjan Salih
(MS) have overcome major limitations of classical protein chemistry

and provided rapid, sensitive, and more global determination of
protein composition of biological samples. MS technology permits
the identification of a large-scale proteome at a qualitative and
quantitative level with relative ease in complex biological samples
without purification of the individual proteins. One area of the
biological fields that promises to yield important advances is the
oral fluids such as gingival crevicular fluid (GCF) and whole saliva
(WS) both of which can be collected noninvasively and may serve
as source of diagnostic biomarkers for general systemic and oral
diseases including periodontal disease. The protein composition of
WS from healthy individuals [13] and periodontal disease patients
[4, 5] have been studied by MS technology. Whereas identification
of protein composition of GCF by both classical biochemical meth-
ods and highly sensitive MS technology have been very limited due
to very small sample size availability of ~0.30.5L per site and
0.551.4L per site from healthy and diseased sites, respectively.
GCF contains serum and locally generated extracellular proteins,
inflammatory mediators, microbial plaque, and antibodies directed
against bacteria [610]. Due to these properties, GCF is consid-
ered a valuable body fluid that may serve as an important source of
biomarkers for both systemic and periodontal diseases [1015].
Despite major advantages provided by contemporary MS technol-
ogy, there still remain limitations such as dynamic protein range
and the presence of highly abundant proteins that need to be over-
come. These relate to the abundant serum-derived proteins in
GCF and further complications introduced by changes in contri-
butions of serum from 30 to 70% in GCF from periodontally
healthy versus inflamed periodontal disease sites [1619]. Unlike
GCF, the serum-derived abundant proteins such as albumin do not
pose limitations in WS analysis by MS.However, there are other
very abundant proteins such as amylase, acidic proline-rich pro-
teins, mucins, and cystatins that require preanalytical simplification
prior to MS analysis.
While MS-based proteome studies of GCF from healthy and
periodontal disease sites have been carried out, these were pre-
dominantly at a qualitative level [2024] or at a quantitative
level but using GCF from experimentally induced gingivitis model
[25, 26]. A comparative quantitative analysis of GCF from healthy
and periodontal patients by large-scale MS-based technology using
label-free approach has also been carried out [27] by relating the
separate MS runs and quantifying identified proteins found in both
healthy and periodontal disease samples. Such approaches have
limitations and uncertainties associated with the accuracy of the
quantifications due to inherent variations in the flight of the peptides
from one MS run to next. To overcome many of the limitations
outlined above and to establish quantitative proteome analysis of
oral fluids from periodontally healthy individuals versus those with
Protein Analysis of Oral Fluids using Mass Spectrometry 39
periodontal disease by large-scale MS technology gold standard

analytical methods were utilized [19]. These included: (1) SDS-
PAGE to accomplish, (a) complete elution of the GCF proteins
from the PerioPaper collection strips, and (b) separation of GCF
and WS proteins to reduce complexity and limit the adverse impact
of abundant proteins to enhance the number of proteins identified
and quantified; and (2) introduction of two stable-isotope reagents
with distinct chemical reactivity for achieving reliable relative quan-
titation. One of these is the mTRAQ reagent that is an amine-
specific stable-isotope labeling agent universal for all proteins and
their peptides generated by trypsin digestion. The commercially
available forms consist of three different isotopic variants, light
[12C] mTRAQ reagents 0 (140Da mass addition), heavy [13C]
mTRAQ reagents with 4 (144Da mass addition), or heavy
[13C] mTRAQ reagents with 8 (148Da mass addition). In rela-
tive quantitative MS analytical approaches using mTRAQ stable
isotope reagents, one sample is labeled with light [12C] mTRAQ
reagents 0 (140Da mass addition) and the other with either of
the heavy mTRAQ reagent which is optional (Fig.1). A second
stable-isotope agent is the cleavable-ICAT reagent with specific
chemical reactivity toward free sulphdryl (-SH) containing pro-
teins/peptides for relative quantitation of cysteine-containing
proteins/peptides in control (e.g., healthy) versus experimental
(e.g., periodontal disease) samples. The cleavable-ICAT reagent
was designed to have either nine [12C] (light reagent) or nine [13C]
(heavy reagent) atoms to provide a mass difference of 9Da
between the control versus the corresponding experimental sample
of the same cysteine-containing tryptic peptide observed within
the same MS spectra for relative quantification. In addition, the
ICAT Reagents
O
Cysteine peptide
C TFA cleavage site
HN NH
S-
HC CH O

H2C CH CH2 CH2 CH2 CH2 C CH2 CH2 O CH2 CH2 O CH2 CH2 O CH2 CH2 CH2 NH C CH I
S
O
Biotin Linker (heavy or light reagent) -haloketone

Heavy Reagent: 9 C13
reactive moiety
Light Reagent: 9 C12
= C12 or C13
Fig. 1 Physical and chemical properties of ICAT reagents showing the free sulphydryl (-SH) cysteine containing
peptide reactive -haloketone moiety, the stable 12C-light and 13C-heavy isotope carrying backbone and the
cleavable biotin-side chain for avidin-solid support affinity enrichment
40 Erdjan Salih
mTRAQ Reagents
O
CH2 CH2 C CH2

H3C N NH CH2 C O N
CH2 CH2 O C CH2 + NH peptide
N
N
* Peptide reactive site Free amine group of peptide
0 4 8
=13CH3 , = 15N Chemical
Coupling step
O
CH2 CH2 C CH2
H3C N NH CH2 C NH peptide + HO N
CH2 CH2 O C CH2
O
N-Hydroxysuccinimide
Labeled Peptide Leaving group
Fig. 2 Physical and chemical properties of mTRAQ reagents showing the free amine reactive
N-Hydroxysuccinimides ester bond, and the stable 12C/14N-light and 13C/15N-heavy isotope carrying sites
alkyl linker in the ICAT reagent is coupled to a (cleavable) biotin

moiety which allows for rapid avidin-based solid-phase affinity
enrichment of cysteine-containing tryptic peptides (Fig.2). The
detailed practical utility of these technologies for oral fluids is the
focus of this chapter.
2 Materials
2.1 Whole Saliva 1. 50mL sterile Falcon tubes (USA Scientific, Inc.) for WS
(WS) andGingival collection, chilled on ice, until 1015mL of whole saliva were
Crevicular Fluid (GCF) collected from each subject.
Collection 2. PerioPaper strips (Oraflow, Plainview, NY) for GCF collections,
and sterile Eppendorf tubes.
3. Dental-cotton rolls -water jet and -air jet.
4. Periotron model 8000 (Proflow, Inc., Amityville, NY) for GCF
volume determination.
5. 80C freezer for oral fluid sample storage.
2.2 Electroelution 1. NuPage 12% BisTris, Mini- Gels with 1.010mm wells
Using SDS (Bio-Rad, Richmond, CA).
PolyacrylamideGels 2. Coomassie blue R-250 (Bio-Rad, Richmond, CA) protein
staining solution in 50% methanol+30% acetic acid+20%
H2O.
3. A destaining solution of 40% methanol+10% acetic acid+50%
H2O.In all cases whenever used H2O should be ultra-pure
deionized water.
4. Sharp straight edge razor.
2.3 In Gel Digestion, 1. Flat bottom 2-mL Eppendorf tubes.

Peptide Extraction 2. Buffer 1: 50mM ammonium bicarbonate (NH4HCO3), pH
~8.0; prepared by dissolving 3.95g of solid NH4HCO3 in 1L
of H2O (see Note 1).
3. Buffer 2: 50mM NH4HCO3 pH ~8.0+50% acetonitrile
(CH3CN) HPLC grade; prepared using 1L of 50% H2O+50%
pure CH3CN liquid solvent and dissolving 3.95g of solid
NH4HCO3.
4. 0.5g trypsin/25L of 50mM NH4HCO3, prepared by
dissolving 0.5mg of crystalline solid Trypsin TPCK-treated
from bovine pancrease (Sigma-Aldrich) in 25mL of 50mM
NH4HCO3, pH ~8.0.
2.4 Removal 1. Buffer A: 0.1% trifluoroacetic acid (HPLC grade); prepared by

of Sodium-Dodecyl- adding 0.1mL of pure TFA to 100mL of H2O.
Sulphate (SDS) 2. C-18 reverse-phase MicroSpin columns (The Nest Group,
andSalts inGeneral Inc., Southborough, MA) for removal of SDS/Salts.
3. Speed-Vac freeze dry apparatus.
2.5 Total Protein/ 1. Bicinchoninic acid (BCA) micro-protein assay reagents,

Peptide Determination Protein Assay Kit (PIERCE, Rockford, IL), a modified Lowrys
protein assay [28].
2. Bovine serum albumin standard (1mg/mL; Sigma-Aldrich).
3. 96-Well ELISA plates and plate reader.
2.6 Quantitative 1. Light [12C] and heavy [13C] mTRAQ reagents (Applied
LC-ESI-MS/MS Biosystems, Foster City, CA).
Analysis Using 2. 50mM Tris+0.1% SDS denaturing buffer, prepared by dis-
Differential Stable- solving 0.84g of Trizma base in 100mL H2O and pH adjusted
Isotope Labeling to 9.0.
with Amine-Specific 3. 50mM Tris-(2-carboxyethyl) phosphine (TCEP), non-thiol-
Tag forRelative based reducing agent.
andAbsolute
4. 50mM NH4HCO3, pH ~8.0.
Quantification
(mTRAQ) 5. Isopropanol (2-propanol).
42 Erdjan Salih
6. Cation-exchange cartridge (POROS 50 HS, 50m particle

size; Applied Biosystems).
7. Cation-exchange buffer A: 10mM potassium phosphate with
25% acetonitrile, pH 3.0 (Applied Biosystems).
8. Cation-exchange buffer B: 10mM potassium phosphate with
25
% acetonitrile 350mM potassium chloride, pH 3.0
+
(Applied Biosystems).
9. C18 MicroSpin Column (The Nest Group).
2.7 Quantitative 1. Light [12C] and heavy [13C] ICAT reagents (Applied
LC-ESI-MS/MS Biosystems).
Analysis Using 2. 50mM Tris+0.1% SDS denaturing buffer, prepared by dis-
Stable-Isotope solving 0.84g of Trizma base in 100mL H2O and pH adjusted
Labeling to 9.0.
withCleavable- 3. 50mM Tris-(2-carboxyethyl) phosphine, a volatile reducing
Isotope-Coded- agent.
Affinity-Tag (ICAT)
4. 50mM NH4HCO3, pH ~8.0.
andAffinity Peptide
Enrichment 5. Acetonitrile (CH3CN), HPLC grade (Sigma-Aldrich).
6. Cation-exchange cartridge (POROS 50 HS, 50m particle
size; Applied Biosystems).
7. Cation-exchange buffer A: 10mM potassium phosphate with
25% acetonitrile, pH 3.0 (Applied Biosystems).
8. Cation-exchange buffer B: 10mM potassium phosphate with
% acetonitrile
25 +
350mM potassium chloride, pH 3.0
9. Avidin-affinity cartridge (Applied Biosystems).
10. Affinity buffer A, 2 phosphate buffer solution pH 7.2.
11. Affinity buffer B, 30
% acetonitrile
+ % TFA (Applied
0.4
Biosystems).
3 Methods
For collection of human oral fluids such as whole saliva (WS) and
gingival crevicular fluid (GCF) in a clinical setting need: (a) an
institutional review board approval of the procedures and patient
privacy/consent documentation for human study, (b) access to a
dental clinic and assistance from an experienced periodontist for
selection and assessment of the clinical parameters for healthy and
periodontal disease periodontium at specific sites and generally in
the oral cavity, and (c) the patient evaluation and GCF collections
should be all performed by the same trained and calibrated clinical
dentist.
Strategies for Multi-Dimensional Separation Mass Spectrometric Analysis
Eluted from the periopaper

with 50 mM
ammonium Bicarbonate
Eluted from the Proteins Electroeluted from

After albumin Proteins
periopaper with remaining the periopaper &
removal with remaining on
ammonium on the blue fractionation by SDS-
Blue Gel periopaper
bicarbonate 50mM gel PAGE
Trypsin digestion
LC-ESI-MS/MS Analysis and identification
Fig. 3 Schematic representation of the strategies for multiple fractionation/separation techniques and different
processing of the GCF PerioPaper samples for multidimensional quantitative mass spectrometric proteome
analysis (reproduced from [22] with permission from John Wiley & Sons)
3.1 Gingival 1. First clean teeth with water jet, gently air-dry and isolate with
Crevicular Fluid (GCF) cotton rolls placed in the buccal/labial sulcus.
Collection 2. Insert commercially available collection PerioPaper strips into
andProcessing the entrance of the sulcus/periodontal pocket of periodontally
healthy subjects or patients with periodontal disease for 30s
(see Note 2 and Fig.3).
3. The GCF volume should be determined immediately after col-
lection using a calibrated Periotron 8000 or any other instru-
ment available with similar performance that can be used to
measure very small fluid volumes bound to paper (see Note 3).
4. Convert Periotron reading to actual volume (L) by reference
to the standard curve that was accessed during calibration of
the Periotron with different volumes of water (see Note 4).
5. Place PerioPaper strips in Eppendorf tubes on ice and either
process immediately for MS analysis or store at 80C until
needed.
3.2 Processing 1. Process each PerioPaper GCF collection from different individ-
ofPerioPaper GCF uals separately. Add 100L of 50mM NH4HCO3, pH 8.0, to a
Collections andElution 1.5-mL Eppendorf tube containing a single PerioPaper strip
ofGCF Proteins with the collection tip immersed in the buffer. Agitate at 1min
intervals for 5min and then raise the PerioPaper collection tip
44 Erdjan Salih
above the buffer level and clip the top edge of the paper under
the Eppendorf tube lid edge and centrifuge at 10,600 g using
bench top Eppendorf centrifuge (see Note 5).
2. Remove the 100L of the NH4HCO3 containing the extracted
GCF proteins and place in a new Eppendorf tube.
3. Repeat step 1 above 3 and combine the extracts providing a
total of 300L of GCF sample from each PerioPaper strip which
may represent GCF samples derived from the same or different
individuals and from healthy or periodontal disease sites.
4. Determine total protein concentrations of the GCF eluate
samples using the micro-protein/peptide modified Lowrys
protein assay [28]. Prepare a dilution series of bovine serum
albumin (BSA) standard range from 0 to 0.4mg/mL (0, 0.03,
0.06, 0.1, 0.2, 0.3, 0.4mg/mL) and use 25L of each stan-
dard or unknown GCF sample in duplicates for microplate
assay (see Note 6). The working reagent was obtained by mix-
ing 50 parts of BCA Reagent A (sodium carbonate, sodium
bicarbonate, bicinchoninic acid, and sodium tartrate in 0.1M
sodium hydroxide) with one part of BCA Reagent B (4%
cupric sulfate). For analysis, use 200L of the working reagent
in each well plus the 25L of each standard or unknown sam-
ple, incubate at 37C for 30min followed by absorbance read-
ing at 562nm using an ELISA plate reader. The protein
concentrations in the unknown samples are calculated from
the linear regression plots of the BSA standard curve.
5. Keep these individual PerioPaper strip GCF eluates separately
to be used for MS analysis individually (~50100L at a time)
or use equal protein aliquots (510g) from these extracts to
generate a variety of pooled samples for qualitative and quanti-
tative MS analysis. Process aliquots of such samples for MS
analysis immediately or store at 80C until needed.
3.3 Qualitative 1. Take 50100L (~815g protein) aliquots of GCF samples

Large-Scale Protein prepared above (Subheading3.2, step 4) and freeze dry sepa-
Analysis ofGCF rately. For generating multiple pooled samples use 10g from
Eluates Directly each individual GCF eluate pool together and process similarly.
2. Resuspend in 20L of dissolution buffer (0.5M triethylammo-
nium bicarbonate, TEAB) followed by the addition of 1L of
2% SDS for protein denaturation, 2L of 50mM TCEP, pH
8.5, for disulphide reduction, and incubate for 1h at 60C.
3. Dilute to 200mL using 50mM NH4HCO3, pH 8.0, and add
0.30g trypsin (15L of the stock trypsin 0.5mg/25mL)
and incubate at 37C overnight (~20h).
4. Freeze dry the trypsin digested GCF samples and resuspend in
15L of MS buffer of 97.4% H2O: 2.5% CH3CN:0.1% formic
acid and perform LC-ESI-MS/MS analysis as detailed above
(Subheading3.2, step 5).
3.4 Removal 1. Remove the highly abundant serum albumin by SwellGel

ofAbundant Serum Blue (Pierce, Co. Rockford, IL 61105) in GCF (see Note 7).
Albumin fromGCF Use ten PerioPaper strips of GCF collection and process as
Samples bySwellGel detailed in Subheading3.2 above.
Blue forLC-ESI-MS/ 2. Combine all 10 PerioPaper GCF extracts ~3mL and concen-
MS Analysis trate to ~1mL using Speed-Vac freeze dry evaporation.
3. Use the SwellGel Blue Albumin Removal Kit (PIERCE,
Rockford, IL) to rapidly remove abundant albumin from sam-
ples. Place three separate SwellGel Blue Albumin Removal
discs into 3 Mini-spin columns and hydrate each disc with
400L of ultrapure H2O.
4. Place the mini-spin columns with their respective SwellGel
blue discs into separate 2.0mL collection tubes and centrifuge
at 12,000g for 1min in a microcentrifuge to remove excess
liquid.
5. Load the GCF extracts ~350mL per Mini-spin column, incubate
for 2min, and centrifuge at 12,000g for 1min.
6. Reload the flow-through/recovered fluid back onto the
respective Mini-spin columns and recentrifuge to collect the
GCF samples (see Note 8).
7. Place the mini-spin columns in new 2-mL collection tubes and
wash the resin with 350mL binding/wash buffer to remove
unbound proteins by centrifugation as detailed above. Repeat
this step 3 times and combine all the wash eluates with the
original first eluate from step 9.
8. Subject the albumin depleted GCF samples from 11 above
after albumin removal to 2% w/w trypsin digestion by incuba-
tion at 37C overnight (~20h) (see Note 9).
9. Freeze dry the SwellGel Blue discs after albumin removal and
rehydrate with 350L of 0.5g/25L of 50mM NH4HCO3,
pH ~8.0, and incubate at 37C overnight (~20h).
10. Subject the PerioPaper strips used (10 PerioPapers) to trypsin
digest using 0.5g of trypsin/25L of 50mM NH4HCO3,
pH ~8.0, at 37C overnight (~20h).
11. Freeze dry each of the trypsin digests from steps 12 to 14 and
perform LC-ESI-MS/MS analysis.
3.5 SDS-PAGE 1. Place one PerioPaper collection strip per SDS-PAGE well using
andIn-Gel Digest NuPage 12% BisTris mini-gel 1.0mm10mm wells and per-
ofGCF Proteins form electrophoresis at constant 120V until the dye front
forLC-ESI-MS/MS reaches the bottom of the gel and the molecular weight stan-
dards can be seen to be separated (Fig.4).
2. Remove the gel from the glass plates and stain with Coomassie
blue (50% methanol+30% acetic acid+20% H2O) and destain
with a solution of 40% methanol+10% acetic acid+50%H2O.The
destained gel with GCF protein bands visible can be sectioned
46 Erdjan Salih
Fig. 4 SDS-PAGE eletroelution of the GCF proteins from PerioPaper strips show-
ing the actual gel electrophoresis run with the separated prestained protein stan-
dards, PerioPaper within the wells, the dy-front at the bottom gel, and GCF
protein not visible yet as the gel is still between the glass plates and not stained
with Coomassie blue
into a number of different molecular weight ranges by excising

these regions with a sharp straight edge razor. For example, the
sections comprised of Cut 1 (above 75kDa), Cut 2 (75
50kDa), Cut 3 (below 5035kDa), Cut 4 (below 25 to above
10kDa), and Cut 5 (10kDa and below), as shown in Fig.5
(see Note 10).
3. Cut each molecular weight gel section (Cut 1 to Cut 5) into
smaller pieces (12mm) and place in their respective separate
Eppendorf tubes, generating five samples to be processed and
each group analyzed by MS/MS separately.
4. Remove SDS and Coomassie blue stain by treatment with buf-
fer 1 (50mM NH4HCO3, pH 8.0) and buffer 2 (25mM
NH4HCO3 pH 8.0+50% CH3CN) alternatively (see Note 11).
Repeat this step 3 times.
5. After the last buffer 2 treatment, suspend the gel pieces in
buffer 1 containing trypsin (0.5g of trypsin per 25L of
buffer) and incubate in Eppendorf tubes for 20h at 37C.
6. Following in-gel digestion, remove and save the buffer con-
taining trypsin and the released peptides and place in a new
Eppendorf tube. Further extract the gel pieces by the addition
of buffer 1, agitate for ~5min, remove buffer 1, and pool with
the original peptide extract that contained the trypsin.
7. Extract the gel pieces with the addition of buffer 2 and repeat
the process as described above, pooling the extracts together.
Fig. 5 SDS-PAGE eletroelution of the GCF proteins from PerioPaper strips. A picture
of SDS-PAGE, coomassie stained gel, showing an example of sectioning of the
gel according to different molecular weight regions. All GCF samples were from
healthy sites. Lane 1, standard molecular weight proteins; Lanes 210, nine indi-
vidual PerioPapers run separately. On the right-hand side under CUT are the
sections of different molecular weight regions excised across all five lanes and
processed for MS analysis, cut 1: Mr range ~100kDa and above, cut 2: Mr range
~4080kDa, cut 3: Mr range ~2538kDa, cut 411 to 24kDa, and cut 5: Mr range
~210kDa, (reproduced from [22] with permission from John Wiley & Sons)
Repeat the alternative treatment/extraction with buffer 1

followed by buffer 2 three times.
8. Freeze dry the pooled specific molecular weight range trypsin
peptides using Speed-Vac and clean using C-18 reverse-phase
Micro-Spin column for the removal of SDS and salts from sam-
ples prior to LC-ESI-MS/MS analysis, as outlined below
(Subheading3.6).
3.6 Removal ofSDS 1. Condition the MicroSpin Vydac C18 column with 200L of
andSalts buffer B, 0.1% Trifluoroacetic Acid+80% Acetonitrile+20%
withMicroSpin Vydac and H2O) and centrifuge for 3min at 500 g using an Eppendorf
Silica C18 bench top centrifuge.
Reverse- 2. Equilibrate the mini-column with 200L of buffer A, 0.1%
PhaseColumn Trifluoroacetic acid in H2O and centrifuge for 3min at 3000g
using an Eppendorf bench top centrifuge.
3. Repeat steps 1 and 2 twice.
4. Suspend the freeze dried samples from (step 8, Subheading3.5
above) in a maximum of 300L of buffer A and apply onto the
48 Erdjan Salih
top of the spin column and centrifuge for 3min at 3000g,

(Eppendorf bench top centrifuge) discard the flow through.
5. Wash the unbound material from the column by the addition
of 200L of buffer A followed by centrifugation for 3min at
500g (Eppendorf bench top centrifuge). Repeat this step
one more time. Discard the flow through and wash eluates.
6. Elute the bound peptides by the addition of 200L of buffer
B to the column and centrifuge for 3min at 3000g. Freeze
dry this final eluate containing the trypsin peptides and save for
MS analysis.
The combined approaches and steps under Subheadings3.5
and 3.6 provide a multidimensional separation and LC-MS/MS
analysis of GCF, as summarized in Fig.3.
3.7 Whole Saliva 1. Depending on the extent of experimentation and sample pro-
(WS) Collection cessing 1015mL of WS may be collected in 50mL sterile
andProcessing Falcon tubes chilled on ice from each individual. Immediately
after collection the WS should be centrifuged at 12,000g for
30min at 4C to separate the whole saliva supernatant (WSS)
from the pellet (see Note 12).
2. Determine total protein concentrations of the WSS samples by
micro-protein/peptide assay as detailed in Subheading3.4,
step 4 for GCF except prepare a dilution series of bovine serum
albumin (BSA) standard range from 0 to 1.0mg/mL (0, 0.1,
0.2, 0.3, 0.4, 0.5, 0.7, and 1.0mg/mL).
3. Process and prepare the WSS samples either immediately for
MS analysis or store frozen at 80C until needed.
3.8 Qualitative 1. Freeze dry separately equal aliquots of WSS from different
Nano-Flow Liquid- individuals each equivalent to 100g of protein in 1.5-mL
Chromatography Eppendorf tubes (see Note 13).
andElectrospray- 2. Resuspend each individual sample in 90L of dissolution buf-
Ionization-Tandem fer (0.5M triethylammonium bicarbonate, pH 8.5) followed
Mass Spectrometric by the addition of 4L of 2% SDS for protein denaturation,
(LC-ESI-MS/MS) 8 L of 50mM [tris(2-carboxyethyl) phosphine] for disul-
Analysis ofWSS phide reduction, and incubate for 1h at 60C.
3. Dilute to 1.0mL using 50mM NH4HCO3, pH 8.0, and tryp-
sin digest using 3g of trypsin (3% w/w) at 37C overnight.
Repeat trypsin digestion one more time with additional 3g of
trypsin overnight (see Note 14).
4. Freeze dry the samples and resuspend in 30L in 97.4% H2O:
2.5% CH3CN:0.1% formic acid for MS analysis.
3.9 SDS-PAGE 1. Prepare at least five (200g each, ~200L of WSS) separate
ofWSS Proteins WSS samples either separately from each subject or as pooled
forLC-ESI-MS/MS sample from different individuals for example 40g (~40L)
from each of five individuals (see Note 15).
2. Freeze dry such separate samples to be applied into multiple

lanes of the SDS-PAGE for separation.
3. Resuspend the samples in gel electrophoresis buffer and per-
form protein separation as described in Subheading3.7.
4. Follow the steps and processing of the gels as detailed in
Subheading3.7 before LC-ESI-MS/MS analysis.
3.10 Qualitative Analyze the samples by nano-electrospray using an online autos-

LC-MS/MS Analysis ampler (Micro AS, ThermoFinnigan, CA) with autoinjections of
ofGCF andWSS 3L of sample onto an inline fused silica microcapillary column,
Samples (75m10cm), packed with C18 resin (Micron Bioresource, Inc.
Auburn, CA) at a flow rate of 250nL/min (see Note 16). Separate
the peptides by a 90min elution comprising of multi-step-linear
gradient using solvent A, H2O/2.5% CH3CN/0.1% formic acid,
and solvent B, CH3CN/0.1% formic acid. The gradient steps to be
used from 100% solvent A to 8% solvent B in 10min, to 15% sol-
vent B in 15min, to 25% solvent B in 20min, to 50% solvent B in
30min, and to 95% solvent B in 15min using a MS micro-pump
(ThermoFinnigan, CA). Generate the MS/MS data using data-
dependent acquisition with an MS survey scan range between 390
and 2000m/z to generate a total ion chromatogram. For protein
identification each survey scan (MS) to be followed by automated
sequential selection of five most abundant peptides for CID, at
35% normalized collision energy, with dynamic exclusion time of
20s of the previously selected ions.
3.11 Database For identification of the proteins search the MS/MS de novo pep-
Search andProtein/ tide fragmentation spectra from LC-ESI-MS/MS against the
Peptide Identification human database, Uniprot (Universal Protein Resource, Version
9.0), which combines the data from Swiss-Prot (Version 51),
TreMBL (Version 36) and PIR using Bioworks 3.3.1 software and
SEQUEST search engine or alternative search software as deemed
necessary or available. To determine the false positive rate, the
data should be searched against a concatenated human sequence
database containing both the forward and the reverse sequence
version. Calculate the false-positive rate by using the number of
matches to the reverse database multiplied by 2 and dividing by the
total number of matches (forward plus reverse) [19, 29]. Search
the data using partial trypsin specification and 2 miscleavages (see
Note 17). During the database search use filtered criteria:
Cn0.1; probability 0.1; and for fully tryptic peptides,
XCorr1.6, 1.8, 3.5 (Z=+1, +2, +3 respectively); and for partial
tryptic peptides, XCorr 1.8, 2.1, 3.75 (Z=+1, +2, +3
respectively).
50 Erdjan Salih
3.12 Relative The individual steps in the overall sample preparation and chemical
Quantitative LC-MS/ labeling with the mTRAQ reagent are summarized in Fig.6.
MS Analysis
1. Place one PerioPaper collection strip per SDS-PAGE well using
ofGingival Crevicular NuPage 12% BisTris mini-Gel 1.0mm10mm wells using
Fluid (GCF) inHealth one section of the gel, i.e., four wells for four PerioPaper strips
andPeriodontal derived from healthy subjects separated by one empty lane
Disease Using mTRAQ from the remaining four wells to be used for four PerioPaper
Stable Isotope- strips derived from periodontal disease patients (Fig.7). For
Labeling Chemistries multiple healthy individuals and periodontal disease patients it
may be required to run multiple such SDS-PAGE to generate
representative samples of each and provide sufficient protein
samples.
2. Run gel electrophoresis at constant 120V until (a) dye front
is only half way down the gel, a short run to achieve just com-
plete elution of the GCF proteins into the gel (Fig.7a, b) the
dye front reaches the bottom of the gel and visually the sepa-
rated molecular weight (Mr) standards can be observed
(Fig.7b).
ICAT-labeling of GCF & WSS mTRAQ-labeling of GCF & WSS

Healthy Periodontitis Healthy Periodontitis
Subjects Patients Subjects Patients
(A) (B)
SDS-PAGE SDS-PAGE SDS-PAGE SDS-PAGE
In gel digestion and In gel digestion and In gel digestion and In gel digestion and
protein measurament protein measurament protein measurament protein measurament
Labeling with Light Labeling with Heavy Labeling with Light Labeling
with Heavy
ICAT Reagent (227 Da) ICAT Reagent (236 Da) Reagent (140 Da) Reagent (148 Da)
Combining Combining
Healthy and Healthy and
Periodontitis Periodontitis
Removal of excess reagents Removal of excess reagents

using Cation-Exchange using Cation-Exchange
Affinity purification on Avidin-column Removal of salts using reverse-

followed by Biotin-Cleavage phase-C-18 micro-column
Analysis by LC/MS/MS for Protein Analysis by LC/MS/MS for Protein

Identification and Quantification Identification and Quantification
Fig. 6 Sequential steps used for stable-isotope labeling with ICAT or mTRAQ reagents of GCF and WSS samples
from healthy subjects and periodontal disease patients and processing for LC-MS/MS analysis, (a), ICAT and
(b), mTRAQ (reproduced from [19] with permission from John Wiley & Sons)
Fig. 7 SDS-PAGE electroelution of GCF proteins from PerioPaper strips from

healthy and periodontal subjects for stable-isotope labeling and quantitative pro-
teome analysis by MS. (a) Short time SDS-PAGE run: a single PerioPaper contain-
ing ~0.20.7L GCF was placed in each well containing sample buffer. The
SDS-PAGE was run until the dyefront was approximately half-way the gel length.
This was sufficient to electroelute proteins from the PerioPaper into the gel visu-
alized by Coomassie blue staining. Molecular weight standards can be seen on
the left side of the gel. The electroeluted GCF proteins are not visible at this point
since the gel has not been yet stained with coomassie blue. (b) Long time SDS-
PAGE run: four PerioPaper strips derived from healthy individuals and four
PerioPapers derived from periodontal patients, respectively, were run in a single
SDS-PAGE separated by an empty lane. After comassie staining the gel was
sectioned into different molecular weight regions as indicated. Lane 1, standard
molecular weight proteins; Lanes 25, 4 individual PerioPapers from healthy and
lanes 710 individual PerioPapers from periodontal patients run separately. On
the right-hand side under CUT are the sections of different molecular weight
regions excised across and processed for MS analysis, cut 1: Mr range ~100kDa
and above, cut 2: Mr range ~4080kDa, cut 3: Mr range ~2538kDa, cuts 411
to 24kDa, and cut 5: Mr range ~210kDa, (reproduced from [19] with permis-
sion from John Wiley & Sons)
3. Process the gel(s) as in steps 28 of Subheading3.7, with the

exception that in the case of short runs the whole lane is excised
and processed, such that all of the gel lanes for healthy or peri-
odontal disease samples are used and processed separately,
whereas for the long runs both healthy and periodontal disease
samples must be run on the same gel to ensure Cut sections
will have appropriate counterparts for accuracy of quantitation,
i.e., as shown in Fig.7b. In the latter case, four lanes of healthy
and four lanes of periodontal disease samples are kept separate,
generating five different Mr Cut sections (Cuts 15) for the
52 Erdjan Salih
healthy GCF samples and the corresponding counterparts for

the periodontal disease GCF samples to be used as pairs for
chemical labeling.
4. Resuspend the freeze dried trypsin peptides in 200L of dis-
solution buffer (0.5M triethylammonium bicarbonate, pH 8.5)
and determine total protein concentration by micro-protein/
peptide assay as detailed in step 4 in Subheading3.4.
5. For differential labeling generate two 100g proteins/peptides
in 100L dissolution buffer (see step 4 above), one represent-
ing the healthy individuals and the other periodontal disease
patients. This may be derived from a single individual or from
multiple healthy individuals and multiple periodontal disease
patients. For instance, for each multiple pooled samples use
10g from 10 healthy individuals and similarly use 10g from
10 periodontal disease patients, as outlined in Fig.8.
6. Repeat step 5 above for each of the five different Mr Cut
sections or for those GCF samples from short SDS-PAGE runs
composed of unseparated total GCF proteins.
7. For chemical labeling, add 50L of isopropanol to each
reagent vial (light [12C] and heavy [13C]).
Relative Quantitative Mass Spectrometric Analysis
GCF samples from 40 healthy (160 GCF periopapers) and 40

periodontal pateints (160 GCF periopapers) were analyzed
Randomly selected different pools of 10 patients from each group were prepared
10 10 10 10 10 10 10 10
Healthy Healthy Healthy Healthy Periodontal Periodontal Periodontal Periodontal
Patients Patients Patients Patients Patients Patients Patients Patients
Comparisons of pools from different groups were conducted according to quantitative

technique utilized followed by
LC-ESI-MS/MS analysis
Fig. 8 Schematic representation of sample pools to generate GCF samples from ten random patients from
each group, healthy subjects and periodontal disease patients, for stable-isotope labeling with ICAT or mTRAQ
reagents for relative quantitative LC-MS/MS analysis (reproduced from [19] with permission from John Wiley
& Sons)
8. Add the healthy sample to the Light [12C] mTRAQ reagent

vial and the periodontal disease sample to the Heavy [13C]
mTRAQ reagent vial followed by incubation for 1h at room
temperature.
9. Combine the differentially labeled healthy and periodontal dis-
ease samples in an Eppendorf tube and freeze dry.
10. Resuspend the sample with 4mL of cation exchange loading
buffer A (10mM potassium phosphate in 25% acetonitrile at
pH 3.0) and adjust the pH if necessary to be between 2.5 and
3.3 (see Note 18).
11. Pre-equilibrate the cation-exchange cartridges (POROS 50 HS,
50-m particle size) with 2mL of cation exchange buffer A.
12. To remove the excess reagents load the sample onto the cation-
exchange micro-column (~12 drop/s) followed by injection
of 1mL of cation exchange buffer A to remove unbound
material.
13. Elute the bound peptides from the cartridge using 500L of
cation exchange buffer B (10mM potassium phosphate in
25% acetonitrile/350mM potassium chloride at pH 3.0) and
save the eluate containing the peptides.
14. Remove the cation-exchange buffer salts from the sample
using C18 MicroSpin Column as outlined in Subheading3.8
before LC-ESI-MS/MS analysis.
15. Carry out LC-MS/MS analysis for protein identification and
quantification.
16. Perform database searches for the identification and relative
quantification of proteins with light (from healthy) and heavy
(from periodontal disease) using a static mass addition of
140Da (light label) on lysine and N-terminal residues and
144Da or 148Da (heavy labels) for dynamic mass modifica-
tion on lysine and N-terminal residue and the mass differences
between light and heavy labeled peptides (either 4 or 8Da) as
specified modifications to calculate the relative [12C]/[13C]
ratios of peptide pairs identified, further information may be
found in [19].
3.13 Relative The WSS proteins are already in solution and can be processed and
Quantitative used directly for relative quantitative LC-MS/MS analysis, as out-
LC-ESI-MS/MS lined in steps 17 below. However, for penetrating into the
Analysis ofWhole detailed proteome of WSS there is a need to perform preanalytical
Saliva Supernatant separation such as SDS-PAGE to simplify the sample complexity.
(WSS) inHealth 1. Freeze dry separately equal aliquots of WSS from different
andPeriodontal individuals within the healthy group and those of periodontal
Disease Using mTRAQ disease group, each equivalent to 500g (~0.5mL WSS; see
Stable Isotope- Note 13) of protein in 1.5-mL Eppendorf tubes or generate
Labeling Chemistries pooled samples separately of equal protein amount as above for
54 Erdjan Salih
healthy and periodontal disease patients, as illustrated in Fig.8

(see Note 19).
2. Resuspend in 180L of dissolution buffer (0.5M triethylam-
monium bicarbonate, pH 8.5) followed by the addition of
8L of 2% SDS for protein denaturation, 16L of 50mM
[tris(2-carboxyethyl) phosphine] for disulphide bond reduction,
and incubate for 1h at 60C.
3. Dilute to 1.8mL using 50mM NH4HCO3, pH 8.0, and tryp-
sin digest using 3g of trypsin (3% w/w) at 37C overnight.
Repeat trypsin digestion one more time with additional 3g of
trypsin overnight (see Note 14).
4. Freeze dry trypsin digests and suspended in 100L dissolution
buffer.
5. For chemical labeling, add 50L of isopropanol to each
reagent vial (light [12C] and heavy [13C]).
6. Add the healthy sample to the Light [12C] mTRAQ reagent
vial and the periodontal disease sample to the Heavy [13C]
mTRAQ reagent vial followed by incubation for 1h at room
temperature.
7. Combine the differentially labeled healthy and periodontal dis-
ease samples in an Eppendorf tube and freeze dry.
8. Follow the steps 716 in Subheading3.12 for GCF which is
common from here onward.
9. For preanalytical separation of WSS using SDS-PAGE follow
the steps 116 in Subheading3.12, except in this case (a) WSS
samples are first denatured and reduced and added to the wells
instead of the PerioPaper strips, and (b) the SDS-PAGE is
performed until dye-front reaches the bottom of the gel (as in
Fig.7b).
3.14 Relative 1. Aliquots of GCF samples derived from healthy subjects and
Quantitative periodontal disease patients as described in Subheading3.12,
LC-ESI-MS/MS steps 16 can be used to prepare samples for ICAT labeling
Analysis ofGingival and LC-ESI-MS/MS analysis. Prepare 100g total protein
Crevicular Fluid (GCF) per group each, either as samples representing individual sub-
inHealth jects (see Note 20) or as pooled samples using 10g protein
andPeriodontal from each of the 10 different individuals.
Disease Using Stable 2. Treat 100g of GCF proteins from healthy and 100g from
Isotope-Coded- periodontal disease patients separately suspended in 100L
Affinity-Tag (ICAT) dissolution buffer (0.5M triethylammonium bicarbonate,
Labeling pH 8.5) with 8L of 50mM Tris-(2-carboxyethyl)-
phosphine reducing agent and incubate for 1h at 60C
(see Note 21).
3. Add 20L of pure acetonitrile to each of the labeling reagent

vials with cleavable ICAT (light [12C] and heavy [13C]).
4. Transfer the 100g of GCF proteins from healthy subjects to
the light [12C-ICAT] vial and the 100g GCF proteins from
periodontal disease patients to the heavy [13C-ICAT] vial and
incubate for 2h at 37C in a water bath and then cool.
5. Combine the entire contents of the differentially labeled sam-
ples of GCF and perform the excess reagent removal steps out-
line in Subheading3.12, steps 810 using the
micro-cation-exchange cartridge.
6. Affinity purify the eluted peptides from the cation-exchange
cartridge using the avidin affinity cartridge to isolate the
containing tryptic peptides. Neutralize the cation-
cysteine-
exchange eluted samples with 500L of affinity buffer A (2
phosphate buffer solution pH 7.2) (see Note 22).
7. Pre-equilibrate the avidin-affinity cartridge using 2mL of
affinity buffer B (30% acetonitrile+0.4% TFA) followed by
2mL of affinity buffer A.
8. Inject the labeled sample slowly onto the cartridge followed by
injection of 500L affinity buffer A.Remove the unbound/
unlabeled peptides by sequential injection of 1mL of affinity
buffer wash 1 (1 phosphate buffer solution pH 7.2), 1mL of
affinity buffer wash 2 (50mM ammonium bicarbonate/ 20%
methanol, pH 8.3), and 1mL ultra-pure deionized water.
9. Elute the cysteine-containing peptides using 800L of affinity
buffer B and freeze dry.
10. Resuspend the ICAT labeled peptides in 90L of a solution
containing ratio of 95:5, cleaving Reagent A (100% TFA) and
Reagent B (stabilizer) and incubate for 2h at 37C in a water
bath to cleave/remove the biotin group from the ICAT
moiety.
11. Cool the sample, freeze dry, and subject to LC MS/MS analy-
sis to identify ICAT peptide pairs and to quantify the relative
[12C]/[13C] ratios.
12. Perform database searches for identification and relative quan-
tification of proteins with light (from healthy) and heavy
(from periodontal disease) using a static mass addition of
227Da (light ICAT label) and 236Da (heavy labels) for
dynamic mass modification on the cysteine residues. A mass
difference between light and heavy labeled peptides of 9Da as
specified modifications to calculate the relative [12C]/[13C]
ratios of peptide pairs identified, further information may be
found in [23].
56 Erdjan Salih
3.15 Relative 1. Prepare 100g total protein per group each using WSS sam-
Quantitative ples from healthy and periodontal disease patients separately,
LC-ESI-MS/MS either as samples representing individual subjects or as pooled
Analysis ofWhole samples using 10g protein from each of the ten different
Saliva Supernatant individuals and freeze dry, as illustrated in Fig.8.
(WSS) inHealth 2. Resuspend each group separately in 80L protein denaturing
andPeriodontal buffer, 50mM Tris, pH 8.0, and 0.1% SDS, and add 4L of
Disease Using Stable 50mM Tris-(2-carboxyethyl)-phosphine reducing agent and
Isotope-Coded- incubate for 1h at 60C (see Note 23).
Affinity-Tag (ICAT) 3. Follow the same procedures as for GCF in Subheading3.14,
Labeling steps 312 for labeling and processing of WSS samples fol-
lowed by LC-MS/MS analysis and data search approaches for
relative quantification.
4 Notes
1. Dissolving 3.95g of solid NH4HCO3 in 1L of H2O to prepare

50mM buffer, pH ~8.0; requires no adjustment of the pH is
necessary as it is automatically pH ~7.88.0.
2. During GCF sample collection with PerioPaper strips avoid
mechanical irritation to reduce contamination with blood if
this occurs those PerioPaper strips should be discarded.
3. Calibrate the Periotron every time by using a range of volumes
of water 0.12.0L and construct a standard curve.
4. Normally, GCF collected by PerioPaper strips from periodontally
healthy individuals would contain 0.350.6L of fluid and a total
of ~3658g protein, whereas those collected from patients with
periodontal disease would contain 0.551.4L of fluid and a total
of ~5070g protein per site per paper strip [19].
5. The soluble proteins from the PerioPaper strips can mostly dif-
fuse into the surrounding buffer within 5min with frequent
agitation, however, to ensure maximum extraction/elution of
the GCF proteins raising the PerioPaper collection strip above
the buffer level and securing under Eppendorf cap edge with
the collection tip in suspension during centrifugation elutes
the adsorbed fluid completely.
6. The standard BSA series of dilutions for protein determination
need to be covering the lower range since the individual GCF
eluates from a single PerioPaper strip will require a standard
protein linear curve to cover a protein range closely related to
a single GCF PerioPaper strip with a total protein of ~3050g
in 300L extraction buffer.
7. Removal of serum albumin that is an abundant protein in GCF
and makes up ~50% of the total protein in serum, improves
significantly the number of proteins identified and maximizes

the identification of the low level proteins in GCF.
8. Reloading of the first flow through after centrifugation multiple
times ensures maximum removal of the abundant albumin.
9. Determine the protein concentrations of the SwellGel Blue
albumin depleted GCF samples and use 2% w/w trypsin
digestion.
10. The gel pieces in SDS-PAGE sections with specific molecular
weight range such as Cut 1 representing multiple and different
PerioPaper strips used for elution/separation of GCF proteins
can be processed and analyzed by LC-MS/MS separately or as
pooled samples.
11. The treatment of the gel pieces that are cut into 12mm sizes
requires an amount of each of buffers 1 and 2 which is depen-
dent on the total gel volume in the Eppendorf tube such that
the buffer level is above the gel level. At each treatment each
buffer should be left with the gel for 23min with agitation at
intervals. Buffer 1 leads to swelling of the gel pieces while buf-
fer 2 shrinks them.
12. Human WS after collection requires removal of particulate
matter such as food and oral cavity related cell debris and cells
(both bacterial and host) by centrifugation to obtain WSS for
MS analysis. This ensures protein identification by MS to be
strictly related to proteins derived from systemic and local
glandular origin. If WSS is not to be used immediately for MS
analysis storage at 80C is necessary to reduce/eliminate
protein modification/degradation.
13. The amount of WSS to be used from multiple healthy or peri-
odontal disease patients is defined by the aim of the study,
i.e., qualitative or quantitative and the number of distinct
samples obtained, keeping in mind that on average human
whole saliva has 1.11.5g protein/L (or 1.11.5mg protein/
mL of saliva).
14. Trypsin digestion of WSS proteins requires frequently repeat
digestion with higher concentrations (3% w/w) of trypsin
(total 6% w/w equivalence) than the normal 2% w/w to
ensure complete digestion of the proteins.
15. Each of the trypsin digested individual samples can be sub-
jected to MS analysis separately or by combining aliquots of
equal protein amounts from different individuals to generate a
series of pooled samples for MS analysis.
16. In addition to the different protein separation approaches used
pre-LC-MS/MS analysis, the C-18 reverse-phase nano-column
provides one more final separation of the peptides on the basis
of hydrophobicity.
58 Erdjan Salih
17. The use of partial trypsin searches avoids the exclusion of any
peptides generated by the oral fluid proteolytic activity in the
WSS and GCF.
18. After resuspension of the sample with cation exchange buffer
the pH of the solution can be checked by using 1 drop of the
solution on pH paper and, if above pH 3 acidify with drop-
wise addition of 1.0M HCL and checking the pH after each
addition.
19. To increase the number of identified WSS proteins and their
relative quantitation, the same type of approach as for GCF
using SDS-PAGE long runs to separate the proteins and carry
out protein identification and quantification on the different
Mr Cut sections is required.
20. In order to use 100g GCF protein from a single individual
one needs to collect at least 34 PerioPaper strips from the
same person and combine the protein extracts from all the
PerioPaper strips.
21. The GCF samples are eluted and generated by SDS-PAGE in
gel trypsin digested after disulfide bond reduction; this gener-
ates peptides with free thiol (-SH) groups of cysteine residues.
Despite the number of steps to clean the samples, storage at
80C has the potential to generate oxidized states of the
thiol groups both by molecular oxygen and by reoxidation to
form disulphide bonds that need to be reduced. This is because
the quantitative approach with ICAT relies on the presence of
free thiols for reaction. Use of Tris-(2-carboxyethyl)-phosphine
as a non-thiol-based reducing agent is essential instead of the
commonly used thiol containing reducing agents such as
-mercaptoethanol or dithiothreitol (DTT) which will seri-
ously impact the ICAT quantitation approaches.
22. After neutralization, check the pH using pH paper. The pH
should be around 7. If not, adjust with 2M NaOH.
23. If WSS samples are to be analyzed after SDS-PAGE using dif-
ferent Mr Cuts, then the 0.1% SDS may be excluded as at this
point since the proteins are already trypsin digested. The non-
reducing agent should still be included to ensure complete
retention of the free thiol groups for reaction with ICAT.
References
1. Xie H, Rhodus NL, Griffin RJ, Carlis JV, Griffin vary proteome by liquid chromatography/
TJ (2005) A catalogue of human saliva proteins mass spectrometry and two-dimensional gel
identified by free flow electrophoresis- based electrophoresis-mass spectrometry. Proteomics
peptide separation and tandem mass spectrom- 5:17141728
etry. Mol Cell Proteomics 4:18261830 Salih E, Siqueira WL, Helmerhorst EJ,
3.
2. Hu S, Xie Y, Ramachandran P, Ogorzalek Loo Oppenheim FG (2010) Large-scale phospho-
RR, Li Y, Loo JA, Wong DT (2005) Large- proteome of human whole saliva using
scale identification of proteins in human sali- disulfide-
thiol interchange covalent chroma-
tography and mass spectrometry. Anal Biochem 15. Offenbacher S, Barros S, Mendoza L, Mauriello
407:1933 S, Preisser J, Moss K, de Jager M, Aspiras M
4. Salazar MG, Jehmlich N, Murr A, Dhople VM, (2010) Changes in gingival crevicular fluid
Holtfreter B, Hammer E, Vlker U, Kocher T inflammatory mediator levels during the induc-
(2013) Identification of periodontitis associated tion and resolution of experimental gingivitis
changes in the proteome of whole saliva by mass in humans. JClin Periodontol 37:324333
spectrometric analysis. JClin Periodontol 16. Hattingh J, Ho E (1980) The concentration of
40:825832 proteins in human gingival crevicular fluid.
5. Wu Y, Shu R, Luo LJ, Ge LH, Xie YF (2009) JPeriodontal Res 15:9095
Initial comparison of proteomic profiles of 17. Alfano MC (1974) The origin of gingival fluid.
whole unstimulated saliva obtained from gen- JTheor Biol 47:127136
eralized aggressive periodontitis patients and 18. Bickel M, Cimasoni G, Andersen E (1985)
healthy control subjects. JPeriodontal Res Flow and albumin content of early (pre-
44:636644 inflammatory) gingival crevicular fluid from
6. Curtis MA, Gillett IR, Griffiths GS, Maiden human subjects. Arch Oral Biol 30:599602
MF, Sterne JA, Wilson DT, Wilton JMA, 19. Carneiro LG, Nouh H, Salih E (2014)
Johnson NW (1989) Detection of high-risk Quantitative gingival crevicular fluid proteome
groups and individuals for periodontal- in health and periodontal disease using stable
diseaseslaboratory markers from analysis of isotope chemistries and mass spectrometry.
gingival crevicular fluid. JClin Periodontol JClin Periodontol 41:733747
16:111 20. Ngo LH, Veith PD, Chen YY, Chen D, Darby IB,
7. Champagne CM, Buchanan W, Reddy MS, Reynolds EC (2010) Mass spectrometric analyses
Preisser JS, Beck JD, Offenbacher S (2003) of peptides and proteins in human gingival cre-
Potential for gingival crevice fluid measures as vicular fluid. JProteome Res 9:16831693
predictors of risk for periodontal diseases. 21. Kido J, Bando M, Hiroshima Y, Iwasaka H,
Periodontol 2000 31:167180 Yamada K, Ohgami N, Nambu T, Kataoka M,
8. Delima AJ, Van Dyke TE (2003) Origin and Yamamoto T, Shinohara Y, Sagawa I, Nagata T
function of the cellular components in gingival (2012) Analysis of proteins in human gingival
crevice fluid. Periodontol 2000 31:5576 crevicular fluid by mass spectrometry.
9. Armitage GC (2004) Analysis of gingival crev- JPeriodontal Res 47:488499
ice fluid and risk of progression of periodonti- 22. Carneiro LG, Venuleo C, Oppenheim FG,
tis. Periodontol 2000 34:109119 Salih E (2012) Proteome data set of human
10. Lamster IB, Ahlo JK (2007) Analysis of gingi- gingival crevicular fluid from healthy periodon-
val crevicular fluid as applied to the diagnosis of tium sites by multidimensional protein separa-
oral and systemic diseases. Ann NY Acad Sci tion and mass spectrometry. JPeriodont Res
1098:216229 47:248262
11. Kojima T, Andersen E, Sanchez JC, Wilkins 23. Baliban RC, Sakellari D, Li Z, DiMaggio PA,
MR, Hochstrasser DF, Pralong WF, Cimasoni Garcia BA, Floudas CA (2012) Novel protein
G (2000) Human gingival crevicular fluid con- identification methods for biomarker discovery
tains MRP8 (S100A8) and MRP14 (S100A9), via a proteomic analysis of periodontally healthy
two calcium-binding proteins of the S100 fam- and diseased gingival crevicular fluid samples.
ily. JDent Res 79:740747 JClin Periodontol 39:203212
12. Loos BG, Tjoa S (2005) Host-derived diag- 24. Tsuchida S, Satoh M, Umemura H, Sogawa K,
nostic markers for periodontitis: do they exist Kawashima Y, Kado S, Sawai S, Nishimura M,
in gingival crevice fluid? Periodontol 2000 Kodera Y, Matsushita K, Nomura F (2012)
39:5372 Proteomic analysis of gingival crevicular fluid
13. Mntyl P, Stenman M, Kinane D, Salo T, for discovery of novel periodontal disease
Suomalainen K, Tikanoja S, Sorsa T (2006) markers. Proteomics 12:21902202
Monitoring periodontal disease status in smok- 25. Grant MM, Creese AJ, Barr G, Ling MR, Scott
ers and nonsmokers using a gingival crevicular AE, Matthews JB, Griffiths HR, Cooper HJ,
fluid matrix metalloproteinase-8-specific chair- Chapple IL (2010) Proteomic analysis of a non-
side test. JPeriodontal Res 41:503512 invasive human model of acute inflammation
14. Fitzsimmons TR, Sanders AE, Bartold PM, and its resolution: the twenty-one day gingivitis
Slade GD (2010) Local and systemic biomark- model. JProteome Res 9:47324744
ers in gingival crevicular fluid increase odds of 26. Bostanci N, Ramberg P, Wahlander ,
periodontitis. JClin Periodontol 37:3036 Grossman J, Jnsson D, Barnes VM, Papapanou
60 Erdjan Salih
PN (2013) Label-free quantitative proteomics 28. Lowry OH, Rosebrough NJ, Farr AL, Randall
revealed differentially regulated proteins in RJ (1951) Protein measurement with folin
experimental gingivatis. JProteome Res phenol reagent. JBiol Chem 193:265275
12:657678 29. Peng J, Elias JE, Thoreen CC, Licklider LJ,
27. Bostanci N, Heywood W, Mills K, Parkar M, Gygi SP (2003) Evaluation of multidimen-
Nibali L, Donos N (2010) Application of label- sional chromatography coupled with tandem
free absolute quantitative proteomics in human mass spectrometry (LC/LC-MS/MS) for
gingival crevicular fluid by LC/MS E (gingival large-scale protein analysis: the yeast proteome.
exudatome). JProteome Res 9:21912199 JProteome Res 2:4350
Chapter 4
Antioxidant Micronutrients andOxidative Stress

Biomarkers
IainL.C.Chapple, HelenR.Griffiths, MikeR.Milward, MartinR.Ling,
andMelissaM.Grant
Abstract
Chronic inflammatory diseases are the major causes of mortality in humans and recent research has improved
our understanding of the major impact of life-style factors upon inflammatory diseases and conditions.
One of the most influential of these is nutrition, which may drive both pro-inflammatory as well as anti-
inflammatory cascades at molecular and cellular levels. There are a variety of model systems that may be
employed to investigate the impact of micronutrients and macronutrients upon inflammatory pathways,
many of which operate through oxidative stress, either at the level of controlling the redox state of the cell
and downstream redox-regulated gene transcription factors, and other acting as free radical generating or
scavenging agents. This chapter focuses upon biological sample preparation prior to assay and details methods
for analyzing certain antioxidant micronutrients and biomarkers of oxidative stress.
Key words Antioxidant, Micronutrient, Oxidative stress, Ascorbic acid, Protein carbonyl, Comet
assay, 8-Oxo-2-deoxyguanosine, Gingival crevicular fluid, Plasma, Saliva
1 Introduction
Diet and nutrition are generally categorized as modifiable

lifestyle risk factors for the development of disease [1]. In Latin
diet means way of living, but the term is also derived from the
Greek word dieta which meant a room or a place of meeting or
assembly. In ancient German culture, political assemblies were
linked to feasting and hence diet is literally derived from words
that historically define lifestyles. The study of nutrition is broad
and complex and has only recently started to receive attention within
scientific literature in relation to chronic diseases of humans.
Nutrients may be categorized as macronutrients (proteins,
fats, carbohydrates, poly-unsaturated fatty acids/PUFAs), required
by the body in relatively large quantities from the diet, or micro-
nutrients (minerals, vitamins), required by the body in much
smaller amounts [2]. It has long been recognized that diet has
61
62 Iain L.C. Chapple et al.
powerful effects upon inflammatory status, arguably as strong as or

stronger than bacterial deposits, and until recently the mechanisms
underpinning those effects were poorly understood. This chapter
will therefore include nutritional mechanisms underpinning pro-
inflammatory, anti-inflammatory, and pro-resolving inflammatory
mechanisms relevant to the study of disease pathogenesis.
Ascorbic acid remains one of the dietary antioxidants that
attracts considerable interest, with intracellular ascorbic acid pro-
tecting important biomolecules from oxidation and contributing
to the cellular redox potential. It is oxidized to dehydroascorbic
acid (DHA) by the actions of oxygen radicals and also during the
regeneration of other antioxidant molecules such as vitamin E and
glutathione. This chapter will therefore begin by outlining the pro-
cess of determining the concentration of ascorbic acid and dehy-
droascorbic acid within human plasma and will then continue by
describing methods of plasma carotenoid analyses. Carotenoids are
predominantly hydrophobic species, comprising all-trans carbon
chains of C40 polyenes with some cyclization possible at one or
both ends of the carbon chain. In the order of 600 carotenoids
have been identified in nature, but only a few of them are ingested
in sufficient quantity to be detected in human plasma [3]. Of the
34 carotenoids identified in biological samples, the most abundant
are beta-carotene, lutein, lycopene, alpha-carotene, beta-
cryptoxanthin, and zeaxanthin, their common cis-isomers and
some oxidation products.
Furthermore, in the absence of effective antioxidant scaveng-
ing or enzymatic removal of oxidizing species, free radicals and
reactive oxygen/nitrogen species will combine with all manner of
biomolecules due to their inherent reactivity. Proteins are the
major component of the cell and as such are likely targets for post-
translational modifications [4]. Methods to measure protein bio-
markers of oxidation have been reviewed extensively [5], and rather
than covering a wide range of methods, this chapter will focus on
the measurement of protein carbonyls by ELISA [6] and western
blot. Protein carbonyls are generic markers of protein oxidation,
and occur ubiquitously on many amino acids [7] (Fig.1).
In addition to protein oxidation, various methods have been
developed to measure oxidative DNA damage, although they
remain complicated techniques to establish [8] (Fig.2). An alter-
native approach that is of increasing interest is the enzymatic
comet assay described within this chapter, because it depends on
the bacterial DNA repair enzyme formamidopyrimidine DNA gly-
cosylase (FPG) to convert oxidized purines (principally 8-oxoGua)
to DNA breaks. Many labs have adopted an enzyme-dependent
Comet assay for ease of operation and reduced the risk of artefac-
tual oxidation that may result from DNA extraction [8]. Moreover,
there are quantitative imaging tools for assessing distribution of
DNA in the comet tails and reducing inter-laboratory variation.
Antioxidant Micronutrients andOxidative Stress Biomarkers 63
R R
C + X C + O2
Protein core ( R ) Carbon-centred

+ side chain side-chain radical
R R
O decomposition COO
R +
C
C= O
Protein Stable Side-chain Side-chain

radical carbonyl alkoxyl radical peroxyl radical
group
Conformational changes Changes in

immunogenicity
Inactivation Increased
(enzyme activity) susceptibility to Carbonyls by ELISA or
protease Spectrophotometry
Fig. 1 Biomarkers of oxidative damage to proteins
Fig. 2 Biomarkers of oxidative damage to DNA
Pioneering work in the 1980s led to the concept of total

antioxidant capacity (TAC) within biochemistry [9], and the
development of a number of assays that claimed to measure the
combined, global effects of all antioxidants within biological samples.
The two best characterized TAC assays are the TRAP (total per-
oxyl radical trapping antioxidant parameter) [9] and the enhanced
chemiluminescence (ECL) assay [1012]. However, characteriza-
tion and validation of TAC assays indicate that they are sensitive
only to specific antioxidants and therefore data requires very care-
ful interpretation and it may not accurately represent invivo mea-
sures in patients. For this reason, invivo antioxidant analyses may
be best performed using assays for individual species, including
vitamin C as described in this chapter.
To afford protection against oxidative damage to DNA, lipids,
and proteins, small molecule and enzymatic antioxidants must be
active within specific subcellular compartments or within the
plasma. When analyzing biological fluid samples for antioxidant
molecules or capacity, it is advisable to ask patients to fast over-
night and not to drink anything other than water. This controls, to
some degree, confounding by dietary antioxidant intake at the
time of sample collection. The protocols for the processing of
blood samples for serum and plasma for antioxidant analyses are
covered within this chapter. Buffy coat preparations of white blood
cells are also included for 8-Oxo-2-deoxyguanosine (8-OHdG)
analysis, whereas the specific preparation of neutrophilic polymor-
phonuclear leukocytes is covered in Chapter 29.
In addition, gingival crevicular fluid (GCF) and saliva samples
may be collected for analyses regarding dietary aspects of disease
pathogenesis, and may benefit by reflecting local intra-oral
inflammatory-immune processes. It is also worth noting that saliva is
largely a surrogate fluid for GCF, since GCF changes are more likely
to reflect the periodontal environment than saliva, but the latter is
frequently used due to the technical challenges of utilizing GCF
(small sample volumes and limited assay sensitivities). This chapter
therefore concludes by outlining methods of GCF and saliva collec-
tion for subsequent analyses.
2 Materials
2.1 Determination 1. 5% Metaphosphoric acid.

ofAscorbic Acid 2. Centrifuge.
andDehydroascorbic
3. 1mM 1,2 phenylene diamine.
Acid
4. High-performance liquid chromatography (HPLC) with UV,
fluorescence and electrochemical detectors.
2.2 Measurement 1. 10mg/mL albumin in PBS (5mL).

ofProtein Carbonyls 2. 500mM ferrous sulphate (2mL) (make fresh on day).
3. 500mM sodium citrate (6mL).
4. Mix ferrous sulphate with sodium citrate and 2mL bicarbonate
(ELISA coating) buffer immediately before use.
5. 50mM deferoxamine (10mL).

6. 2mg/mL sodium borohydride (2mL).
7. 2PD10 columns.
8. PBS.
9. 0.05% PBS Tween.
10. 10mM DNPH in 2N HCl.
11. 2N HCl.
12. Microcentrifuge tubes.
13. 20% Trichloroacetic acid.
14. 6M guanidine HCl in 20mM potassium phosphate adjusted
to pH 2.3 with trifluoroacetic acid.
15. 0.05M carbonate buffer, pH 9.6 (made from 1.59g/L sodium
carbonate to 2.93g/L sodium hydrogen carbonate).
16. Ethyl acetate/ethanol wash.
17. Color development substrate: 0.15M citrate phosphate buffer
pH 5.0, 20mg tablet O-Phenylene diamine (OPD).
18. 8.8M hydrogen peroxide.
19. Nunc-maxisorp 96-well plates.
20. 4% (w/v) marvel milk powder in dH2O.
21. 1:500 anti-DNP mouse IgE in 4% (w/v) marvel.
22. 1:2000 anti-mouse IgE conjugate in 4% (w/v) marvel.
23. Microplate reader.
2.3 Preparation 1. Vacutainers containing lithium-heparin.

ofWhite Cells 2. Ice/ice box.
for8-OHdG Analyses
3. PBS.
4. Universal containers.
5. 3% Gelatin (see below).
6. Microbiological incubator.
7. 50mL centrifuge tube.
8. Centrifuge.
9. 0.83% NH4Cl solution (lysis buffer).
10. Pasteur pipette.
11. 1.5mL cryotube.
12. Liquid nitrogen.
13. Chlorine release tablet.
2.3.1 3% Gelatin inPBS 1. PBS.

2. Gelatin (150 bloom from pig skin).
3. Bunsen burner.
4. 50100mL glass bottles.

5. Boiling water bath.
6. Microbiological refrigerator.
2.4 Comet Assay 1. Single forested glass microscope slides.

2. 22mm22mm coverslips.
3. Plastic universal container.
4. Incubator.
5. Centrifuge.
6. Agarose, normal melting point.
7. Agarose, low melting point (LMP).
8. Lysis buffer: 100mM disodium EDTA, 2.5M NaCl, 10mM
TrisHCl, pH to 10.0 with sodium hydroxide, 1% Triton
X-100 to be added immediately before use.
9. Electrophoresis buffer: 300mM NaOH, 1mM disodium
EDTA, pH 13.
10. Neutralization buffer: 0.4M TrisHCl, pH 7.5.
11. 30cm25cm flat-bed electrophoresis tank.
12. 1mg/mL propidium iodide stock solution; for work solution
dilute to 2.5g/mL.
2.5 Serum 1. Plain vacutainers with no preservative or added anticlotting

Preparation factors.
2. Ice/ice box.
3. Microbiological fridge.
4. Pasteur pipette.
5. Centrifuge.
6. Liquid nitrogen.
7. Microbiological freezer.
2.6 Plasma 1. Vacutainers containing lithium-heparin.

Preparation 2. Ice/ice box.
3. Centrifuge.
4. Pasteur pipette.
5. Cryogenic vials.
6. Liquid nitrogen.
2.7 Meta- 1. 33.5% Meta-phosphoric acid.

PhosphoricAcid 2. Volumetric tube.
3. Weighing scales.
4. Distilled water.
5. Microbiological fridge.
2.8 Collection 1. Paper sampling strip, e.g., PerioPaper (Oralflow Inc, USA).
ofGingival Crevicular 2. Triple air syringe/cotton wool rolls.
Fluid (GCF)
3. Tweezers.
4. Timer.
5. Precalibrated Periotron 6000 or 8000 (Oraflow Inc, USA).
6. Cryogenic vials containing an appropriate buffer (e.g., PBS or
PBS with 5mg/L BSA or ammonium bicarbonate, depending
upon subsequent analyses).
7. Liquid nitrogen/80C microbiological freezer.
2.9 Collection 1. Saliva sampling marble.

ofSaliva 2. Graduated Falcon tube.
3. Funnel.
4. Timer.
5. 80C microbiological freezer.
3 Methods
3.1 Determination 1. Immediately after sample collection add an equal volume of

ofAscorbic Acid 5% metaphosphoric acid (see Notes 1 and 2).
andDehydroascorbic 2. Centrifuge (1000g for 30min at 4C).
Acid inHuman Plasma
3. Collect stabilized ascorbate containing supernatant and store
at 70C until analyzed.
4. Add 1mM 1,2 phenylene diamine under nitrogen for 60min
at 4C to yield quinoxaline that is detected by its fluorescence
at 425nm when excited at 350nm (derivatization increases
the sensitivity of detecting DHA).
5. Determination of ascorbic acid and DHA is then performed
simultaneously by HPLC using an amino column, elution
with water/methanol and simultaneous detection by UV
(AA at 265nm) and fluorescence (DHA) [13, 14] (see Notes 3
and 4).
3.2 Carotenoid 1. To detect carotenoids and their metabolites within biological

Analyses (See Notes samples, high-performance liquid chromatography (HPLC)-
57) photodiode array (PDA) detection-mass spectrometry (MS)
has been used [15], although such resources are not available
to all laboratories.
2. A reduced number of major carotenoids can be more easily

detected using conventional HPLC and UV-based detection
systems as described [16].
3.3 Measurement 1. Mix 2.5mL of 10mg/mL BSA with 0.25mL of 500mM fer-
ofProtein Carbonyls rous sulfate, mix and vortex.
toDetect Oxidative 2. Incubate at room temperature for 30min (oxidized sample).
Protein Biomarkers
3. Add 0.25mL of 50mM deferoxamine to oxidized sample.
3.3.1 Preparation 4. For reduced sample, mix 2.5mL of 10mg/mL BSA with
ofELISA Standards[6] 0.25mL of 2mg/mL sodium borohydride (reduced sample;
see Note 8).
5. Prepare PD10 columns by removing cap, breaking off tip, and
eluting storage buffer.
6. Wash through with PBS and allow to run through completely.
7. Add 3mL of oxidized/reduced sample to independent PD10
columns.
8. Add further 3mL of PBS to elute proteins. Collect eluant into
separate bijoux tubes for each column labeled as OX (oxidized)
and RED (reduced).
9. Take an aliquot of protein eluate from each tube (2L) and
dilute into 20L PBS.
10. Measure protein concentration in OX and RED using bicin-
choninic acid assay.
11. Adjust protein concentration of OX and RED independently
to 2mg/mL in PBS.
4 Method
1. In triplicate, take 500L of standard (OX or RED) in duplicate

and mix with 500L of DNPH in HCl and with HCl alone in
separate tubes at room temperature for 1h with vortexing
every 15min.
2. Discard supernatant.
3. Wash samples by adding 1mL ethanol-ethyl acetate (1:1v/v),
mixing on vortex and centrifuging as before (see Note 9).
Discard supernatant.
4. Repeat wash twice more.
5. Redissolve precipitate in 1mL of 6M guanidine HCl (above).
6. Leave for 30min at 37C to redissolve.
7. Remove any remaining insoluble material by centrifugation
(13,000g for 1min).
8. Measure the absorbance at 360nm and calculate the carbonyl
content using the Beer-Lambert Law (see Note 10).
4.1 Carbonyl 1. Dilute samples (or cell lysates) and standards to 0.02mg/mL
ELISAMethod in 0.05M carbonate buffer pH 9.6.
2. Aliquot 50L standard per well into Nunc-maxisorp 96-well
plates using standards of known carbonyl content within the
range of 010nmol/mg.
3. Samples (50L) were aliquoted into independent wells in trip-
licate and incubated for 1618h at 4C.
4. Wash the plate three times with 0.05 % PBS Tween.
Subsequently block the plate by incubation with 4% (w/v)
Marvel milk powder in distilled water, for 1h to inhibit any
nonspecific binding.
5. Following a further three washes using PBS Tween, incubate
samples and standards for 2h with primary antibody (anti-
DNP mouse IgE, at 1:500in 4% (w/v) marvel).
6. Wash the plate three times using PBS Tween and incubate for
1h with secondary antibody (anti-mouse IgE HRP conjugate,
at 1:2000in 4% (w/v) marvel).
7. Remove excess unbound antibody by washing three times
using PBS Tween.
8. Add color development substrate and 10L of hydrogen per-
oxide (total 50L) to each well and leave for 15min in the
dark for color to develop and then stop using 2N H2SO4.
9. Measure absorbance spectrophotometrically at 490nm, using
a microplate reader (see Note 10).
10. Calculate carbonyl content from standard curve and express as
nmol (carbonyl) per mg of protein (see Note 11).
4.2 8-OHdG Analyses 1. Add 2mL of PBS (at room temperature) to each tube contain-
or Comet Assay ing cells (red and white blood cells (see Notes 12 and 13) and
toDetermine Oxidative gently invert three times to mix.
DNA Damage (See 2. Place tubes at 37C for 10min.
Notes 12 and13) 3. Transfer cells from all tubes into a plastic Universal container.
4.2.1 Preparation 4. Add an equal volume of 3% gelatin (Subheading3.5.1) to the
ofWhite Cells (Buffy Coats) cells and gently invert container five times to mix.
5. Place container in 37C incubator for 30min.
6. Carefully remove container from the incubator and transfer as
much of the white cell-rich layer (see Subheading4) to a 50mL
centrifuge tube.
7. Add an equal volume of PBS (room temperature) and gently
invert tube three times.
8. Centrifuge at 1000g for 20min at 4C.
9. Tip off supernatant into a flask.
10. Flick base of tube several times to loosen cell pellet.
11. Add 10mL lysis buffer, carefully invert tube five times to mix
and stand at 4C for 15min to lyse erythrocytes.
15. Add 25mL of PBS and carefully invert tube five times to mix.
19. Using a Pasteur pipette, transfer as much material as possible
to a 1.5-mL cryotube.
20. Add 0.5mL of PBS to centrifuge tube, gently shake and then
transfer to cryotube.
21. Snap freeze and store in liquid nitrogen.
22. Add a chlorine release tablet to the flask containing washings
and place in fume hood overnight (not necessary to switch on;
mark do not touch). Dispose of washings down a sink using
plenty of water.
4.2.2 3% Gelatin inPBS 1. Add 200mL PBS to a conical flask.

(150 Bloom fromPig Skin) 2. Weigh out 6g gelatin and sprinkle on the surface of the
forPreparation ofBuffy PBS.
Coats
3. Gently heat gelatin suspension over a Bunsen burner with con-
stant agitation (see Note 14).
4. Aliquot into glass bottles (50100mL).
5. Sterilize by placing bottles (with loose caps) in a boiling water
bath for 30min on 3 consecutive days (see Note 15).
6. Store at 4C.
4.3 Comet Assay 1. Pre-coat slides with 100L of hot 1% normal melting point
toDetermine Oxidative agarose by dropping the agarose at one end of the slide and
DNA Damage smearing it in the other direction using another slide. Leave
(See Note 16) overnight at room temperature.
4.3.1 Day 1
4.3.2 Day 2 1. Centrifuge 10,000 cells at 400g for 4min at 4C.Discard

(See Note 17) supernatant and keep samples on ice.
2. In parallel, warm 0.6% low melting point agarose in a water
bath at 37C.
3. Add 200L of low melting pointing point agarose to the cell
pellet and quickly pour 80L of the gel onto two pre-coated
slides, distributing gel with the aid of a coverslip.
4. Allow 510min on ice for agarose to set then remove the

coverslip.
5. Place slides in ice-cold lysis buffer 1% Triton X-100 overnight
in the dark.
4.3.3 Day 3 1. Carefully remove lysis buffer without disturbing gels.

2. Wash the slides twice for 10min using ice-cold, DNase-free
water and keep in the dark to prevent any DNA damage due to
light.
3. For enzyme-specific cleavage, incubate slides in specific enzyme
reaction buffer twice each for 510min (see Note 18).
4. Add the DNA glycosylase enzyme (FPG: see Note 19) to the
gels and cover them with cover slip and put them at 37C in
5% CO2 humidified incubator.
5. Transfer the slides into the electrophoresis tank so that all the
slides are laid lengthwise in the same direction.
6. Cover the slides with ~875mL ice-cold alkali buffer and incu-
bate for 20min in the dark.
7. Electrophorese in the dark for 20min at 27V, 300mA.
8. Carefully remove the slides from the tank, flood each slide with
1mL neutralization buffer, and leave for 20min.
9. Rinse slides twice with 1mL water and after 10min drain off
excess liquid and allow slides to dry in the oven at 37C or at
room temperature overnight.
10. Incubate with 2.5pg/mL propidium iodide for 15min prior
to visualization under epifluorescence microscopy using green
light excitation. The percentage of DNA in the tails in the
presence or absence of specific enzymes is used as a measure of
specific oxidative lesions.
4.4 Serum 1. Collect blood into plain tubes with no preservative or anti-
Preparation clotting factor and immediately place on ice at the
toDetermine chairside.
Antioxidant Capacity 2. Transfer to the laboratory and take out of ice and stand at
(See Note 20) room temperature for 30min.
3. Place in fridge for a further 30min to allow clot to contract
(see Note 21).
4. Centrifuge at 3000g for 10min (sealed buckets; 4C).
5. Carefully open the tubes and, using a Pasteur pipette, carefully
aliquot clear serum into 1.5-mL cryotubes (1.5mL/tube).
6. Store tubes in liquid nitrogen or at 80C.
4.5 Determination 1. Collect blood into lithium-heparin containing tubes and place
ofAntioxidant on ice at the chairside.
Capacity (See Note 22) 2. Transfer to laboratory and take heparinized blood tubes out of
4.5.1 Plasma Preparation the ice and transfer to sealed buckets in a centrifuge.
4. Carefully open tubes and remove plasma using a Pasteur
pipette, being careful not to include any cells.
5. After removal of plasma, keep tubes containing cells on ice for
preparation of white cells (buffy coats).
6. Aliquot plasma into 1.5mL cryogenic vials.
7. Store in liquid nitrogen or at 80C.
8. For vitamin C analysesadd 0.75mL of 100g/L metaphos-
phoric acid (see below) to each tube containing 0.75mL
plasma. Cap tubes and invert ten times. Store at 80C.
4.5.2 33.5% Meta- 1. Allow bottle to come to room temperature.

Phosphoric Acid; 2. Weigh out 3g of crystals and transfer into a 10mL volumetric
ForPlasma Vitamin C tube.
Analyses (See Notes 23
3. Add distilled water until level in tube is approximately 9mL.
and24)
4. Stopper tube and gently mix by inversion until all crystals have
dissolved (approximately 10min).
5. Carefully remove stopper and drain excess liquid back into
the tube.
6. Make up volume to 10mL by addition of distilled water.
7. Replace stopper, gently invert to mix, and then store at 4C.
4.6 Collection 1. Use a paper sampling strip such as a PerioPaper (Oralflow

ofGingival Crevicular Inc, USA).
Fluid (GCF) toMeasure 2. Having air dried the designated sample site thoroughly to
Antioxidant Activity remove saliva film/contamination, place the strip gently into
(See Note 25) the gingival crevice (or periodontal pocket) until light resis-
tance is felt.
3. Leave for 30s (shorter induces a sampling time error, and lon-
ger irritates the crevice and may induce excess GCF flow) and
remove [17].
4. Immediately measure the volume, ideally using a precalibrated
[18] Periotron 6000 or 8000 (Oraflow Inc, USA), since this
is faster than weighing.
5. Place the strip into a cryogenic vial containing an appropriate
buffer (e.g., PBS or PBS with 5mg/L BSA or ammonium
bicarbonate, depending upon subsequent analyses), such that
the paper strip is fully submersed in elution buffer.
6. Hold the tube carefully with tweezers into a portable liquid

nitrogen containing flask/dewar (as appropriate to the room
size for health and safety measures) and allow the nitrogen to
flow into the vial and snap freeze the sample.
7. Allow excess nitrogen to evaporate before securing the cryo-
genic vial lid and dropping the tube into the liquid nitrogen
until time to transfer to definitive storage chamber.
8. Ideally store under liquid nitrogen, otherwise store within a
80C freezer.
9. Prior to assay, thaw the sample and allow to elute for 30min
prior to immediate assay.
10. Express resulting data as total amount of antioxidant per 30s
sample time and also as a concentration to account for the
confounding influence of differences in GCF volume or flow
rate [19].
4.7 Collection 1. Ask volunteer if they have followed the pre-sampling instruc-
ofSaliva Samples tions with respect to eating, drinking, and brushing teeth prior
toMeasure to sampling appointment and record this.
Antioxidant Activity 2. Give volunteer the saliva sampling marble.
(See Notes 2628) 3. Remove correctly labeled lid from the saliva sample tube (grad-
uated Falcon tube).
4. Place saliva sampling funnel into the saliva sample tube (Falcon
tube).
5. Give volunteer the combination of saliva sampling funnel and
the saliva sample tube (Falcon tube) to hold.
6. Instruct volunteer to place marble in their mouth and continu-
ally roll it around their mouth for 5min.
7. Instruct the volunteer to retain the marble in their mouth
while expectorating the resulting saliva.
8. Time the volunteer for 5min and ensure that a minimum of
1mL of saliva has been collected.
9. If 1mL of saliva has not been collected in 5min then have the
volunteer continue until 1mL has been collected.
10. Record the total amount of time it took to reach 1mL.
11. If the volunteer spits marble into funnel by accident, they can
retrieve with their fingers and replace in their mouths.
12. Take the apparatus from the volunteer and place lid on the
saliva sample tube.
13. The saliva sample tube is to be taken to the laboratory and
stored in a 80C freezer.
5 Notes
1. Owing to its highly oxidizable nature, there are a number of

important considerations for effective analysis of ascorbic acid
in biological samples. Immediately after collection, samples
must be preserved using metaphosphoric acid to precipitate
proteins, reduce the pH and therefore the likelihood of oxida-
tion reactions [20].
2. Some sample loss can be expected following storage, and it is
recommended that any laboratory assess the rate of decay
under storage during method validation. Samples should ide-
ally be analyzed within 1 month.
3. Greater sensitivity and specificity can be achieved by employing
electrochemical coulometric detection as described [21].
4. The majority of AA/DHA analyses are undertaken in plasma,
although the measurement of the AA/DHA ratio within single
cells using coulometric approaches applied to chip technology
is also possible [22].
5. Throughout processing and analysis, samples are kept in the
dark to avoid photodegradation.
6. Small synthetic lipophilic antioxidant molecules such as butyl-
ated hydroxyanisole are introduced to minimize artefactual
oxidation invitro.
7. The high lipid solubility of the molecules requires the use of
stringent organic solvents to elute samples from the Suplex
PKB 100 column which is suited to separation of basic organic
molecules such as carotenoids, prior to UV detection at
460nm.
8. For reduced sample, mix 2.5mL of BSA with 0.25mL sodium
borohydride (may fizz =release of O2). After fizzing
stops=reduced sample.
9. Wash samples by adding 1mL ethanol-ethyl acetate (1:1v/v),
mixing on vortex, and centrifuging as beforeit is really
important that the pellet is disrupted as much as possible
although this may appear trickyit needs plenty of vortexing.
10. The carbonyl content is calculated using the molar absorption
coefficient of 22,000M1cm1 for aliphatic hydrazones.
11. Carbonyl content of samples are typically recorded as nmol
carbonyl /mg protein.
12. If buffy coats containing white cells are required for 8-OHdG
analyses, the blood is collected in the same manner as for
plasma.
13. The white cell-layer may be located immediately above the
erythrocyte cell layer.
14. The heated gelatin suspension must not boil, a temperature of

approximately 50C is sufficient to allow gelatin to dissolve.
15. Do not sterilize gelatin suspension by autoclaving.
16. This methodological approach has been shown to be robust in a
number of laboratories and is sensitive to change under oxida-
tive stress. However, while comparative measures can be made,
pre- and posttreatment, it is not directly quantitative.
17. Day 2undertake all steps on ice.
18. The conditions for each enzyme should be optimized in
advance.
19. The Comet assay depends on the bacterial DNA repair enzyme
formamidopyrimidine DNA glycosylase (FRG) to convert oxi-
dized purines (principally 8-oxoGua) to DNA breaks.
20. When analyzing biological fluid samples for antioxidant mole-
cules or capacity, it is advisable to ask patients to fast overnight
and not to drink anything other than water. This controls to
time of sample collection [23].
21. Halfway through this time (i.e., after ~15min) carefully remove
the cap of the tube and, using a Pasteur pipette, gently encircle
the clot to stop adherence to the tube wall. Replace cap.
22. When analyzing biological fluid samples for antioxidant mole-
cules or capacity, it is advisable to ask patients to fast overnight
and not to drink anything other than water. This controls to
time of sample collection.
23. For vitamin C analyses using plasma, it is important to precipi-
tate proteins prior to sample storage that can be achieved using
meta-phosphoric acid.
24. Once prepared, remove from fridge and allow to reach room
temperature prior to use.
25. Please note that many key antioxidant species are labile and eas-
ily oxidized by environmental exposure prior to freezing and/or
assay or oxidation during centrifugation [12]. The protocol
described is therefore advised for the collection and storage of
GCF samples where antioxidant activity is to be analyzed.
26. It is important to decide whether to collect unstimulated or
stimulated samples. The former requires the volunteer to be
subject to minimal sensory stimulation, whether it be ocular,
olfactory, auditory, gustatory, etc. Therefore, the volunteer
must be sat in a dark, quiet room, with no atmospheric smells
and they must simply lean forward and drool saliva into the
collection pot. Any attempt to expectorate saliva utilises muscles
and causes stimulation.
27. Stimulated saliva can be collected in different ways, using

paraffin wax to stimulate flow, or a sterile marble rolled around
the mouth [12] and expectorating saliva into a narrow funnel
(in case the marble is lost) that leads to a collection tube
(e.g., graduated Falcon tubes).
28. The saliva should be collected for a fixed period of time (e.g.,
5min) such that in due course the saliva flow rate may be
calculated. Saliva composition has been shown to vary accord-
ing to the flow rate, because those components actively secreted
into saliva via the acinus may not change or may increase with
increases in flow rate, whereas those components within saliva
but derived from, for example GCF, will be diluted by increases
in saliva flow rate. As for GCF, it is therefore best to express
saliva components as amount per mL (concentration) as well as
amount per minute (rate of delivery of saliva component).
These values differ for antioxidant species [12, 13].
References
1. Chapple IL (2009) Potential mechanisms peroxyl radical-trapping antioxidant capability
underpinning the nutritional modulation of of human blood plasma by controlled peroxi-
periodontal inflammation. JAm Dent Assoc dation. The important contribution made by
140:178184 plasma proteins. FEBS Lett 187:3337
2. Van der Velden U, Kuzmanova D, Chapple IL 10. Cao G, Alessio HM, Cutler RG (1993)
(2011) Micronutritional approaches to peri- Oxygen-radical absorbance capacity assay for
odontal therapy. JClin Periodontol 38(Suppl antioxidants. Free Rad Biol Med 14:303311
11):142158 11. Whitehead TP, Thorpe GH, Maxwell SR
3. Faulks RM, Southon S (2005) Challenges to (1992) Enhanced chemiluminescent assay for
understanding and measuring carotenoid bio- antioxidant capacity in biological fluids. Anal
availability. Biochim Biophys Acta Chim Acta 266:265277
1740:95100 12. Chapple IL, Mason GI, Garner I, Matthews
4. Griffiths HR, Grant MM (2006) The use of JB, Thorpe GH, Maxwell SR, Whitehead TP
proteomic techniques to explore the holistic (1997) Enhanced chemiluminescent assay for
effects of nutrients invivo. Nutr Res Rev measuring the total antioxidant capacity of
19:284293 serum, saliva and crevicular fluid. Ann Clin
5. Hawkins CL, Morgan PE, Davies MJ (2009) Biochem 34:412421
Quantification of protein modification by oxi- 13. Cooke MS, Mistry N, Ahmad J, Waller H,
dants. Free Rad Biol Med 46:965988 Langford L, Bevan RJ, Evans MD, Jones GD,
6. Carty JL, Bevan R, Waller H, Mistry N, Cooke Herbert KE, Griffiths HR, Lunec J(2003)
M, Lunec J, Griffiths HR (2000) The effects of Deoxycytidine glyoxal: lesion induction and
vitamin C supplementation on protein oxida- evidence of repair following vitamin C supple-
tion in healthy volunteers. Biochem Biophys mentation invivo. Free Rad Biol Med
Res Comm 273:729735 34:218225
7. Stadtman ER, Levine RL (2000) Protein oxi- 14. Bevan RJ, Durand MF, Hickenbotham PT, Kitas
dation. Ann NY Acad Sci 899:191208 GD, Patel PR, Podmore ID, Griffiths HR, Waller
8. European Standards Committee on Oxidative HL, Lunec J(2003) Validation of a novel ELISA
DNA Damage (ESCODD) (2003) for measurement of MDA-LDL in human
Measurement of DNA oxidation in human plasma. Free Rad Biol Med 35:517527
cells by chromatographic and enzymic meth- 15. Khachik F, Bernstein PS, Garland DL (1997)
ods. Free Rad Biol Med 34:10891099 Identification of lutein and zeaxanthin oxida-
9. Wayner DD, Burton GW, Ingold KU, Locke S tion products in human and monkey retinas.
(1985) Quantitative measurement of the total, Invest Ophthal Vis Sci 38:18021811
16. Stahl W, Sundquist AR, Hanusch M, Schwarz 20. Lyman CM, Schultze MO, King CG (1937)
W, Sies H (1993) Separation of beta-carotene The effect of metaphosphoric acid and some
and lycopene geometrical isomers in biological other inorganic acids on the catalytic oxidation
samples. Clin Chem 39:810814 of ascorbic acid. JBiol Chem 118:757764
17. Chapple IL, Garner I, Saxby MS, Moscrop H, 21. Lykkesfeldt J, Loft S, Poulsen HE (1995)
Matthews JB (1999) Prediction and diagnosis Determination of ascorbic acid and dehydro-
of attachment loss by enhanced chemilumines- ascorbic acid in plasma by high-performance
cent assay of crevicular fluid alkaline phospha- liquid chromatography with coulometric
tase levels. JClin Periodontol 26:190198 detectionare they reliable biomarkers of oxi-
18. Chapple IL, Landini G, Griffiths GS, Patel NC, dative stress? Anal Biochem 229:329335
Ward RS (1999) Calibration of the Periotron 22. Zhao S, Huang Y, Shi M, Liu YM (2009)
8000 and 6000 by polynomial regression. Quantification of biogenic amines by micro-
JPeriodontal Res 34:7986 chip electrophoresis with chemiluminescence
19. Lamster IB, Oshrain RL, Gordon JM (1986) detection. JChromatog A 1216:51555159
Enzyme activity in human gingival crevicular 23. Brock GR, Butterworth CJ, Matthews JB,
fluid: considerations in data reporting based on Chapple IL (2004) Local and systemic total
analysis of individual crevicular sites. JClin antioxidant capacity in periodontitis and
Periodontol 13:799804 health. JClin Periodontol 31:515521
Chapter 5
NMR-Based Metabolomics ofOral Biofluids

HorstJoachimSchirra andPaulineJ.Ford
Abstract
NMR-based metabolomics is an established technique for characterizing the metabolite profile of biological
fluids and investigating how metabolite profiles change in response to biological and/or clinical stimuli.
Thus, NMR-based metabolomics has the potential to discover biomarkers for diagnosis, prognosis, and/or
therapy of clinical conditions, as well as to unravel the physiology underlying clinical conditions. Here, we
describe a detailed protocol for NMR-based metabolomics of oral biofluids, including sample collection,
sample handling, NMR data acquisition, and processing. In addition, we give a general overview of the
statistical analysis of the resulting metabolomic data.
Key words Metabolomics, Systems biology, NMR spectroscopy, Saliva, Gingival crevicular fluid
1 Introduction
Metabolomics is a systems biology technique that either aims to

characterize all metabolites in a biological sample (metabolo-
mics) [1] or to investigate how metabolite levels change in
response to external stimuli (metabonomics) [2]. Despite this
historical distinction, recently both terms are generally used inter-
changeable, with metabolomics being more common and used
throughout this chapter.
Metabolomics combines methods of analytical chemistry with
high-level methods of multivariate statistical analysis (MVSA) and
recently even with computational modeling to characterize metabolic
changes inherent to biological processes and disease [3]. The two
most common analytical methods are mass spectrometry (MS) and
nuclear magnetic resonance (NMR) spectroscopy. Both methods are
complementary to each other, having their own distinct advantages
and disadvantages [47]. MS is more sensitive than NMR (nM-pM
range vs. mM-M range), but also more selective in the metabolite
classes that can be observed with a specific technique, and less quan-
titative. In contrast, NMR spectroscopy provides information on core
metabolites in an inherently quantitative manner and with minimal
79
80 HorstJoachimSchirra andPaulineJ.Ford
sample preparation. Thus, NMR spectroscopy is an ideal technique

for initial metabolomics investigation, and in this chapter we will
focus on NMR-based metabolomics.
The applications of metabolomics can generally be divided into
two subclasses that are defined by the motivation: (1) Metabolomics
can be used to discover, validate, and use biomarkers, which are either
individual metabolites or metabolite patterns that are characteristic
for specific diseases [8, 9]. Subsequently, the validated biomarkers
can then be used for disease diagnosis, prognosis, and to monitor
therapy (theranosis). In this context, metabolomics spans all three
areas of personalized medicine. However, when used for biomarker
discovery, metabolomics is only establishing that metabolites are
associated with disease, but not why. (2) In contrast, by identifying
the metabolite changes associated with biological processes and
developing disease, metabolomics can characterize the metabolic
pathways that are affected (and unaffected) in disease. Thus, metab-
olomics can be used to gain a mechanistic understanding of disease
development/progression and biological processes.
Since demonstrating the usefulness of saliva as biofluid in
NMR-based metabolomics [10, 11], metabolic profiling of oral
biofluids with NMR spectroscopy has used saliva and gingival cre-
vicular fluid (GCF) in the following areas and applications:
(1) Dentistry and oral health: metabolomics has been used to
characterize metabolite signatures for generalized chronic periodontitis
[12], gingivitis [13], caries lesions in children [14] before and after
caries treatment [15], and effect of oral rinsing on saliva composition
[16, 17]. (2) Food science and nutrition: Saliva is a useful biofluid for
characterizing nutritional changes [18, 19]. It has been shown that
the metabolite profiles of stimulated and unstimulated saliva are dif-
ferent [20], and that the saliva response depends on taste perception
[21, 22]. (3) In more general applications, salivary metabolomics has
been used to evaluate sports performance [23], and correlating
exhaled breath condensate and saliva profiles [24].
In this chapter, we will be outlining the experimental protocol
for sampling saliva as well as GCF, and carrying out metabolomic
analysis with NMR spectroscopy. In our description, we will draw
equally on our own experience, description of NMR-based metab-
olomics methods in established oral biology papers, and on stan-
dard protocols recommended for NMR-based metabolomics of
clinical samples [25, 26]. We will also provide a short overview and
introduction into the downstream analysis of metabolomics data
by multivariate statistical analysis (MVSA), but provide no detailed
protocols for this analysis, as these are beyond the scope of this
chapter, owing to the multitude of different methods available to
perform MVSA.Instead, we refer the reader to detailed overviews
[2729] and hope that our description provides an entry point to
the more advanced statistical literature.
NMR-Based Metabolomics ofOral Biofluids 81
2 Materials
2.1 Sample 1. Graduated polypropylene tubes, e.g., Falcon tubes.

Collection 2. Sodium fluoride.
andPreparation
3. Pad of paraffin wax.
4. GCF collection and NMR buffer: 15mM sodium fluoride,
0.05% (w/v) sodium azide, 10mM NaH2PO4, and 150mM
NaCl, pH 7.2.
5. Ice.
6. Vortex mixer.
7. Benchtop centrifuge.
8. Glass beaker.
9. Magnetic stirrer.
10. Eppendorf tubes, 1.5mL.
11. 80C freezer for sample storage.
2.2 Sample 1. Centrifugal ultrafiltration devices with a molecular weight cut-

Preparation off of 3kDa, e.g., Amicon Ultra or Sartorius Vivaspin.
2. Milli-Q (deionized) water.
3. NMR sample buffer for saliva samples: 75mM sodium phosphate
buffer, pH 7.4, containing 2mM 2,2-dimethyl-2-silapentane-5-
sulfonate-d6 sodium salt (DSS-d6), 2mM difluorotrimethylsi-
lanylphosphonic acid (DFTMP), 0.05% (w/v) sodium azide
(NaN3), and 20% (v/v) D2O.
4. NMR lock and standards solution for GCF samples: 5mM
2,2-dimethyl-2-silapentane-5-sulfonate-d6 sodium salt (DSS-
d6), 5mM difluorotrimethylsilanylphosphonic acid (DFTMP),
0.05% (w/v) sodium azide (NaN3) in 100% D2O.
5. Optional: Gilson sample handling robot.
6. 5-mm or 3-mm diameter NMR tubes, 103.5mm length, with
number code, with hole, in a 96-tube rack (5mm: Bruker
Biospin part Number Z112273, 3mm: Bruker Biospin part
Number Z112272).
7. POM balls for closing NMR tubes (Bruker Biospin part
Number Z72497).
2.3 NMR 1. 600MHz Avance III NMR spectrometer (Bruker Biospin)

Spectroscopy and equipped with (1) BBI 5mm Z-gradient probe and automatic
MVSA tuning and matching (ATMA) unit, (2) SampleJet sample changer
system with cooling of samples while stored and temperature
conditioning system for preheating samples before insertion into
magnet (see Note 1).
2. Bruker Biospin TopSpin 3.2 software.

3. Bruker Biospin Amix software (optional).
4. Desktop computer with programs such as:
(a) Bruker Biospin Amix software: statistical preprocessing,
MVSA, metabolite identification.
(b) Chenomx NMR Suite (Chenomx, Edmonton, Canada):
statistical preprocessing, metabolite identification.
(c) e.g., Simca P+ (Umetrics, Sweden): MVSA.
Matlab (MathWorks, USA): statistical preprocessing, MVSA.
(d)
(e) R (https://www.r-project.org): statistical preprocessing,
MVSA.
3 Methods
This protocol describes the process of using NMR spectroscopy to

collect metabolomic data from the oral biofluids saliva and GCF
(see Fig.1). Standard procedures for the preparation and analysis of
clinical/biological samples with NMR-based metabolomics have
been outlined in two seminal publications [25, 26]. Thus, we will
not repeat these papers but refer to them where appropriate. In our
description of equipment, experiments, and experimental parame-
ters, we will follow the nomenclature of Bruker Biospin, as Bruker
spectrometers are the most common type of NMR spectrometers.
Other manufacturers have equipment and experiments that are
analogous to the procedures described here, and conversion of the
parameters should be straightforward.
3.1 Sample Saliva is of two types: stimulated (characterized by serous secre-

Collection tions, watery consistency, and high flow rate) or unstimulated
(characterized by mucous secretions, thicker consistency, and low
flow rate). Since the constituents and their concentration will vary
according to the type of saliva collected, it is important to carefully
control the collection method. In both cases, participants are
requested to refrain from any oral activities (eating, drinking, rins-
ing, smoking, brushing teeth) for at least 2h before saliva collec-
tion. Saliva is usually collected at defined times to avoid differences
in circadian rhythm between different samples. Often collection is
done directly after waking up, as this means subjects are usually in
a consistent metabolic state (fasted overnight) and have not yet
undertaken any oral activities.
GCF is an inflammatory exudate of the gingival tissues which
flows continuously into the gingival sulcus (space between the
tooth and gum). Due to its proximity to the biofilm of dental
Sample Collection Saliva GCF
Vortex
Transport to laboratory on ice
Sample Preparation Centrifuge Pooled QC Sample*
Aliquot
Store at -80C
Randomise sample order
Thaw
Pooled QC Sample*
Ultrafiltration
NMR sample preparation
Setup/optimisation of NMR parameters with QC sample

NMR
Spectroscopy
Setup automation run
Run NMR data acquisition in automation
Process NMR spectra
Quality Check
Repeat failed samples
NMR spectra
MVSA Statistical Pre-processing
Unsupervised MVSA (e.g. PCA)
Supervised MVSA (e.g. PLS, O2PLS)
Metabolite Identification
Biological/Clinical Interpretation
Fig. 1 Overview of the process pipeline for NMR-based metabolomics of oral biofluids. The individual steps
involved in NMR-based metabolomics of saliva and GCF are outlined as described in this chapter, starting from
sample collection via sample preparation, and NMR spectroscopy to multivariate statistical analysis (MVSA).
The ultrafiltration step is optional. *The generation of pooled quality control (QC) samples can be performed
either immediately after sample collection or during preparation of NMR samples. GCF gingival crevicular
fluid, PCA principal components analysis, PLS partial least squares projections to latent structures, O2PLS
bidirectional orthogonal PLS
plaque located on the tooth surface, GCF also contains bacterial

products. There are different methods available for collection [30].
Here, we will focus on the filter strip collection method, as it is easy
to use and is a commonly employed collection method.
Detailed instructions for the collection of saliva and GCF are

also provided in [30, 31], and the instructions below should be
read in conjunction with these references.
3.1.1 Unstimulated 1. Provide subjects with a graduated polypropylene tube containing

Saliva Collection 15mol NaF.
2. Have the participant drool in the polypropylene tube for
510min. The duration should be consistent across all samples
and depends on how much saliva needs to be collected.
3. Close the tube.
4. Transport the collected sample(s) to the laboratory on ice.
Processing of the sample should occur within 1h of
collection.
3.1.2 Stimulated Saliva 1. Have the participant chew on a piece of paraffin wax, and let
Collection the stimulated saliva dribble into a graduated polypropylene
tube (see Note 2) for 510min. The duration should be con-
sistent across all samples and depends on how much saliva
needs to be collected.
2. Transport the collected sample(s) to the laboratory on ice.
Processing of the sample should occur within 1h of
collection.
3.1.3 Gingival Crevicular 1. GCF collection is ideally performed in a clinic, however can be
Fluid Collection done elsewhere if appropriate lighting and participant posi-
tioning is available.
2. Select two sites representative of the periodontal status for the
participant, generally the buccal sulcus of the maxillary molars
or premolars. Isolate the site from saliva contamination using
cotton rolls.
3. Insert Millipore filter paper strips (1mm wide and pre-
weighed) into the gingival sulcus and leave for 30s (or replace
when almost saturated, up to a total sample time of 30s).
Strips visibly contaminated by blood should be discarded.
4. Weigh strip to determine the volume of GCF collected (see
Note 3).
5. Place strips in Eppendorf tubes containing 220L 15mM
sodium fluoride, sodium azide, 10mM NaH2PO4, and
150mM NaCl at pH 7.2 for NMR analysis (see Note 4).
6. Vortex the Eppendorf tubes to allow the collected GCF to be
released from the filter paper.
7. Discard the filter paper.
8. Transport to the laboratory on ice. Processing of the sample
should occur within 1h of collection.
3.2 Sample 1. Centrifuge the saliva or GCF samples at the following condi-
Processing tions to remove particulate matter:
andStorage
(a) Saliva: 2600g for 15min at 4C.
(b) GCF: 800g for 10min at 4C.
2. If samples from all subjects enrolled in the study are collected
at once, then create at this time several pooled quality control
(QC) samples: Take an aliquot (e.g., 50 or 100L for saliva,
20L for GCF) of biofluid from each sample and combine in
a glass beaker chilled on ice. Stir the resulting biofluid mixture
for 10min. Aliquot this sample pool into multiple pooled QC
samples of the same volume as regular biofluid samples. Then
treat these pooled QC samples as any other sample in the study.
If samples are taken from volunteers at different dates, then
perform this step instead at step 3 in Subheading3.4.
3. Aliquot the supernatants and store at 80C.
3.3 Optional: This section can optionally be inserted at step 4 of Subheading3.4,

Ultrafiltration if removal of high molecular weight compounds from saliva is
toRemove High desired.
Molecular Weight
Saliva Constituents 1. Follow the manufacturers instructions for handling centrifu-
gal ultrafiltration devices with a molecular weight cutoff of
3kDa in the following steps.
2. The membranes of most centrifugal ultrafiltration devices are
preserved with glycerol as humectant. It is imperative to
remove this glycerol before sample handling to avoid contami-
nating the samples with glycerol: First rinse ultrafiltration filters
by manual rinsing with water (squirt bottle).
3. Then rinse centrifugal ultrafiltration devices by centrifuging 510
volumes of water through the device, according to the manufac-
turers instructions. The exact number of rinse cycles depends on
the brand of the device, and should be tested beforehand by
repeating rinse cycles until the filtrate is free of glycerol.
4. Now filter each saliva sample in an individual prerinsed
centrifugal ultrafiltration device.
5. Collect the filtrate as saliva sample and follow the steps in the
following sections.
6. The remaining retentate in the top of the centrifugal ultrafil-
tration device is concentrated in saliva proteins, peptides, and
other high-molecular weight components, and could be used
for characterizing these constituents.
3.4 NMR Sample 1. Randomize the order of sample preparation and NMR data
Preparation acquisition by randomizing the order in which samples are
prepared and inserted into each 96-rack of NMR tubes.
Biological fluids are unstable, and changes to their composition

can begin immediately after thawing (see Note 5). For this
reason, it is imperative to randomize sample order to prevent
that s ystematic errors accumulate preferentially into any of the
specific study groups along the pipeline of sample preparation,
sample changer queuing, and NMR data acquisition. Insert
pooled QC samples at regular intervals into the sample run
order, typically 2 pooled QC samples plus one long-term QC
sample per 96-tube rack, leaving 93 regular saliva samples
available per 96-tube rack.
2. Thaw the stored frozen saliva or GCF samples on ice (see Note 5).
3. If samples from subjects enrolled in the study were collected at
different dates, then create now several pooled QC samples:
Take an aliquot (e.g., 50 or 100L for saliva, 20L for GCF)
of biofluid from each sample and combine in a glass beaker
chilled on ice. Stir the resulting biofluid mixture for 10min.
Aliquot this sample pool into multiple pooled QC samples of
the same volume as regular biofluid samples. Then treat these
pooled QC samples as any other sample in the study. If samples
were taken from volunteers at the same date and time, then
perform this step instead at step 2 in Subheading3.2.
4. If removal of high molecular weight compounds by centrifugal
filtration is desired insert the steps of Subheading3.3 here.
3.4.1 Manual Saliva 1. In an Eppendorf tube mix 300L saliva sample with 300L
Sample Preparation NMR sample buffer. These volumes are for samples in 5mm
NMR tubes. When using 3mm NMR tubes, mix 150L saliva
sample with 150L NMR sample buffer.
2. Centrifuge at 12,000g and 4C for 5min to remove any
potential sediment.
3. Transfer 550L of the supernatant to a 5mm NMR tube and
close the tube with a POM ball or a closed tube cap with a
number code. (When using 3mm NMR tubes transfer 200L
of the supernatant.)
4. Note the tube number and position of the sample in the
96-well rack. It is good practice to document the tube posi-
tions in a rack by taking a photo of the 96-well rack with all
samples before inserting the rack into the sample changer.
3.4.2 Automated Saliva 1. Centrifuge the thawed saliva samples at 12,000g and 4C for
Sample Preparation 5min to remove any potential sediment.
2. Use the sample preparation robot to mix in a deep-well 96-well
plate 300L of each saliva sample with 300L NMR sample
buffer. (When using 3mm NMR tubes, mix 150L saliva
sample with 150L NMR sample buffer.)
3. Transfer 550L of the centrifuged sample mixes to 5mm

NMR tubes in a 96-tube rack. (When using 3mm NMR tubes
transfer 200L of the mixed samples.)
4. Close the tubes with POM balls.
5. Note the tube numbers and positions of the sample in the
96-well rack. It is good practice to document the tube positions
in a rack by taking a photo of the 96-well rack with all samples
before inserting the rack into the automated sample changer.
3.4.3 Manual GCF 1. In an Eppendorf tube mix 200L GCF sample (already in buffer,
Sample Preparation see Note 4) with 22L NMR lock and standards solution. These
volumes are optimized for samples in 3mm NMR tubes which
are recommended for GCF samples due to the low volume of
CGF collected per sample.
2. Centrifuge at 12,000g and 4C for 5min to remove any
potential sediment.
3. Transfer 200L of the supernatant to a 3mm NMR tube and
close the tube with a POM ball or a closed tube cap with a
number code.
4. Note the tube number and position of the sample in the
before inserting the rack into the sample changer.
3.4.4 Automated GCF 1. Use the sample preparation robot to add 22L NMR lock and
Sample Preparation standard solution to each GCF sample and mix.
2. Centrifuge the thawed GCF samples (already in buffer, see

Note 4) at 12,000g and 4C for 5min to remove any poten-
tial sediment.
3. Transfer 200L of the centrifuged sample mixes to 3mm
NMR tubes in a 96-tube rack.
4. Close the tubes with POM balls.
5. Note the tube numbers and positions of the sample in the
before inserting the rack into the automated sample changer.
3.5 General Setup The general setup and maintenance of an NMR spectrometer
ofNMR Spectrometer involves steps such as temperature calibration, calibration of water
suppression, and calibration of external references such as a
synthetic ERETIC signal [32, 33]. These steps are common to all
metabolomics studies, and standard recommendations for carrying
out these procedures have recently been described in detail in [26].
For this reason, we refer to that publication rather than repeating

the recommendations here. Instead, we will describe the experimental
setup specific for saliva and GCF samples. In the following
instructions commands used within the Bruker Biospin software
TopSpin are written in fixed font, and placeholders for filenames
or parameters are enclosed in angular brackets (<>).
1. Set the temperature of the NMR spectrometer probe head to
the desired measurement temperature, usually 298K, and
allow the spectrometer to equilibrate.
2. Set the automatic sample changer to the correct automation
mode: 5mm shuttle when using 5mm NMR tubes; and 3mm
shuttle when using 3mm tubes.
3. Ensure the preheater settings of the sample changer are cor-
rect: The temperature of the preheating positions within the
SampleJet temperature conditioning system should be the
same as the measurement temperature of the probe head (usu-
ally 298K). The sample changer should preheat the next sam-
ple, and the minimum preheating time is 3min. These settings
are available from the SampleJet web interface, which can be
accessed with the command ha (see Note 6).
4. Ensure the sample carousel of the sample changer is chilled to
the appropriate temperature (approximately 6C).
5. Insert the sample rack containing the NMR tubes into the
sample changer.
6. Insert one of the pooled QC samples into the magnet. The
following steps of experimental setup and optimization of
experimental parameters that are specific to each metabolomics
study should be carried out on this pooled QC samples, as it is
likely to be a representative average of all samples encountered
in a particular study.
7. Create a dataset for experimental setup of the QC sample
(edc) and read in a parameter set for the desired NMR experi-
ment to be run (rpar <parametersetname>).
8. Lock the NMR spectrometer to the appropriate solvent system
(lock). Usually, this is 90% H2O+10% D2O.
9. Tune and match the probe head using the automatic tuning
and matching routines (atma or atmm).
10. Perform automatic shimming with the command topshim
tunea. If shimming problems occur, a (more time-consum-
ing) fully three-dimensional shimming procedure via topshim
3d might be required.
11. Save the optimized shim parameters to be available as starting
shim for each sample during the automation run (wsh
<shimfilename>).
12. Determine the optimal 90 hard-power pulse and correspond-

ing power level using the automated routine (pulsecal
man). This will also automatically determine the optimal power
level for water presaturation (corresponding to a 50Hz field).
13. Now optimize the carrier frequency (O1) of the spectrometer.
This is typically done in two steps (step 14 and step 15 in this
section):
14. First use the gs mode in an NMR experiment with presatura-
tion, with a long relaxation delay and the number of scans
(NS) set to 1, and adjust the O1 frequency until the intensity
of the acquired free induction decay (FID) signal is at a mini-
mum (see Note 7). In this experiment, use the pulse length
and power level parameters determined so far.
15. Then perform the following three-step iteration cycle:
(a) Acquire a spectrum initially with the optimized settings
used in Step 14 (NS=1, long relaxation delay, optimized
90 pulse length, optimized pulse poser levels).
(b) Phase the resulting spectrum on the DSS signal.
(c) Now change the O1 frequency and repeat this three-step
iterative cycle until the whole spectrum can be phased by
zero-order phase correction alone, including the residual
water signal.
16. The experimental parameters optimized so far are common to all
NMR experiments described in Subheading3.6. Thus, the
datasets described below should be updated with the
experimental parameters optimized in steps 1215 of this
Subheading3.5 (also see Note 8).
3.6 Detailed Setup There are a variety of different NMR experiments that are recom-
ofNMR Experiments mended for NMR-based metabolomics studies, including (1) a
one-dimensional (1D) Nuclear Overhauser Spectroscopy (NOESY)
with presaturation (noesypr1d); (2) a 1D Carr-Purcell-Meiboom-
Gill (CPMG) experiment with presaturation (cpmgpr1d); and (3) a
two-dimensional (2D) J-resolved experiment with presaturation
(jresgpprqf). Their specific setup is described in this section.
3.6.1 1D NOESY The general workhorse experiment of NMR-based metabolomics

Experiment studies is a one-dimensional NOESY experiment (pulse sequence
noesypr1d). It is quantitative, has excellent water suppression
properties, and usually leads to flat baselines, but it can be
affected by high-molecular weight compounds in a biological
sample (see Subheading3.6.2).
The noesypr1d experiment has the pulse sequence RD90t90
m90ACQ, with RD being the relaxation delay (4s), t a short
delay (usually 4s), 90 the radiofrequency pulses, m the NOESY
mixing time (100ms), and ACQ the acquisition time (the length of
which is automatically determined by the acquisition parameters TD

and SW as well as the base frequency of the spectrometer).
Experimental Setup
1. As described in step 16 of Subheading3.5 update the

experimental parameters of the NOESY dataset (see Table1).
2. Confirm that the maximum receiver gain (RG) achievable with
the experiment is larger than the recommended standard RG
of 128 (see Note 9), by either using the command rga (see
Note 10), or by running the experiment with the standard RG
of 128 and observing whether overload of the analog-digital
converter occurs.
3. Ensure the parameter set contains an automation routine
(defined in acquisition parameter AUNM) that reads in the
correct parameter set (defined in acquisition parameter
EXP), shims on the sample, and automatically optimizes
pulses (see Note 11).
4. Save the updated parameterset (wpar <parametersetname>
all).
3.6.2 1D CPMG Saliva samples contain significant amounts of high-molecular weight

Experiment constituents such as proteins, peptides, lipids, and/or lipoproteins
that cause broad signals in the NMR spectrum that interfere with
the interpretation and quantification of the sharp signals from
Table 1
Experimental parameters noesypr1d
Acquisition parameters:
Pulse program (PULPROG) noesypr1d
Time domain (TD) 65,536
Number of scans (NS) 128
Dummy scans (DS) 8
Sweep width (SW) 14ppm
Receiver gain (RG) 128
Relaxation delay (D1) 4.0s
Mixing time (D8) 100ms
Temperature (TE) 298K
Experiment time (expt) 18:07min at 600MHz
Processing parameters:
Window function (WDW) EM
Line broadening (LB) 0.3Hz
low-molecular weight metabolites. These broad signals can either

be avoided by physically removing high-molecular weight constituents
from the saliva samples (see Subheading3.3) or by running a CPMG
experiment (cpmgpr1d) that filters out the signals from high-
molecular weight compounds from the NMR spectra by exploiting
their relaxation properties.
The cpmgpr1d experiment has the pulse sequence RD90
(t180t)nACQ, with RD being the relaxation delay (4s), 90
the radiofrequency pulses, t the spin-echo delay (usually 500s),
n the number of loops (a minimum of 128), and ACQ the
acquisition time (the length of which is automatically determined
by the acquisition parameters TD and SW as well as the base
frequency of the spectrometer).
Experimental Setup
1. As described in step 16 of Subheading3.5, update the experi-

mental parameters of the CPMG dataset (see Table2).
2. Confirm that the maximum RG achievable with the experiment
is larger than the recommended standard RG of 128 (see Note 9),
by either using the command rga (see Note 10), or by running
the experiment with the standard RG of 128 and observing
whether overload of the analog-digital converter occurs.
Table 2
Experimental parameters cpmgpr1d
Acquisition parameters:
Pulse program (PULPROG) cpmgpr1d
Time domain (TD) 65536
Dummy scans (DS) 16
Sweep width (SW) 20ppm
Spin-eco delay (D20) 500s
Number of loops (L4) 128
Window function (WDW) EM
Line broadening (LB) 0.3Hz
3. Ensure the parameter set contains an automation routine (defined

in acquisition parameter AUNM) that reads in the correct param-
eter set (defined in acquisition parameter EXP), shims on the
sample, and automatically optimizes pulses (see Note 11).
all).
3.6.3 2D J-Resolved The 2D J-resolved experiment (jresgpprqf) is a good complement

Experiment to both 1D experiments above. In its 1D projection, it collapses
NMR signals from multiplets to singulets, thus simplifying the
NMR spectrum and alleviating spectral overlap. This can be help-
ful in quantifying metabolites in regions affected by spectral over-
lap. The full 2D spectrum enables better identification of
metabolites as it allows in regions of spectral overlap to identify
which lines of a signal cluster belong to which multiplet.
The jresgpprqf experiment has the pulse sequence RD90t1
180t1ACQ, with RD being the relaxation delay (4s), t1 the
indirect evolution time, 90 and 180 the radiofrequency pulses with
respective flip angles, and ACQ the acquisition time (the length of
which is automatically determined by the acquisition parameters TD
and SW as well as the base frequency of the spectrometer).
Experimental Setup
1. As described in step 16 of Subheading3.5, update the

experimental parameters of the J-resolved dataset (see Table3).
2. Confirm that the maximum RG achievable with the experiment
is larger than the recommended standard RG of 128 (see Note 9),
by either using the command rga (see Note 10), or running the
experiment with the standard receiver gain of 128 and observing
whether overload of the analog-digital converter occurs.
3. Ensure the parameter set contains an automation routine
(defined in acquisition parameter AUNM) that reads in the
correct parameter set (defined in acquisition parameter
EXP), shims on the sample, and automatically optimizes
pulses (see Note 11).
all).
3.7 Automated NMR Automated NMR data acquisition is usually performed with the
Data Acquisition ICON-NMR interface of Bruker Biospin, but other automation pro-
cedures are also available [34]. Our descriptions refer to ICON-NMR,
but can be translated to analogous other automation routines.
1. Transfer the updated parameter sets that were optimized in

Subheading 3.6 (and saved in step 4 of each experiments
individual subsection) to the experiment list in ICON-NMRs
Configuration module.
Table 3
Experimental parameters jresgpprqf
Acquisition parameters: f2 f1
Pulse program (PULPROG) jresgpprqf
Time domain (TD) 8192 40
2D acquisition mode (FnMODE) QF
Dummy scans (DS) 16
Sweep width (SW) 16.6ppm 78Hz
Window function (WDW) EM EM
Line broadening (LB) 0.3Hz 0.3Hz
2. Update any critical run parameters for automation in ICON-

NMRs Configuration module. These typically might include:
(a)
The temperature conditioning system of the SampleJet
sample changer: The temperature conditioning system
should be enabled, with one subsequent sample in the
heater, and the minimum sample conditioning time set to
3min (see Note 12).
(b) The temperature handling system should set and check the
temperature after sample insertion, and allow additional
equilibration time in the magnet if the temperature equili-
bration check fails (Post-Insertion Temperature Set/
Check Routine=TEREADY 60 0.2).
(c) Automatic tuning and matching of each sample should be
enabled.
(d) Samples should be automatically locked (Lock Program
LOCK) and shimmed (Shim Program topshim tunea). It
is advisable to define a shim parameter file that is automati-
cally read in to provide a good starting shim. This shimfile
should contain the optimized shim parameters from step
11 in Subheading3.5.
(e) Steps 2.12.4 of this section can alternatively also be
incorporated into an acquisition automation program
(analogous to the one provided in Note 11) if automation
is not done via the ICON-NMR interface.
(f)
In the Priority section of the Configuration module it
should be decided whether the samples should be run in the
order in which they are positioned in the tube rack(s) (only
Enable Priority ticked)which is usually the order in
which they are submitted to the measurement queue in step
3 of this Subheading3.7 belowor whether they should be
run in random order (only Randomize Measurement
Order ticked). It is preferable that samples are already in
random order in the tube racks and then are run sequentially
(see step 1 in Subheading3.4), as this method is faster. If
samples are arranged systematically in the tube racks and run
randomly, then the total runtime of the sample queue will be
considerably longer because (a) the sample pre-equilibration
time in the Temperature Conditioning System is now inserted
between each sample instead of occurring during acquisition
of the previous sample, and (b) individual samples might be
inserted multiple times into the magnet if multiple experi-
ments are acquired per sample.
3. Set up a queue of all samples and experiments to run in
automation mode in the Automation section of ICON-NMR.
For each saliva sample this typically includes the 1D-CPMG
experiment and the 2D J-resolved experiment, but the
1D-CPMG experiment might be replaced by the 1D-NOESY
experiment if suppression of high-molecular weight constitu-
ents is not required (i.e., if the steps in Subheading3.3 were
performed). For GCF samples the 1D-CPMG experiment and
the 2D J-resolved experiment are essential. If for each sample
all three experiments are acquired (as advocated by some [25,
26]) with the parameters as in Subheading3.6 then the handover
time between individual samples is about 45min. If shorter
runtimes per sample and thus a higher sample throughput are
desired then the parameters in Subheading3.6 (especially
number of scans NS) should be adjusted accordingly. E.g., if
NS is adjusted to 32 for both 1D spectra and to NS=2 for the
J-resolved experiment, the handover time between samples
shortens to approximately 19min.
3.8 NMR Data 1. Fourier transform the data using the following apodization
Processing of1D NMR parameters: Zero-filling by a factor of 2 (SI=2*TD), multiplica-
Spectra tion of the FID by a exponential window function with a line-
broadening factor of 0.3 (WDW=EM, LB=0.3) (see Note 13).
2. Correct the phase of the NMR spectrum. If the setup of the
spectrometer and experiment was correct then no first-order
phase correction should be necessary (set phc1=0), and only a
small phase zero-order phase correction should be needed.
3. Calibrate the chemical shift of the NMR spectrum to the DSS
signal at 0ppm.
4. Correct the baseline of the NMR spectrum to achieve a base-

line that is flat across the whole spectrum and centered at 0
intensity.
Automated processing routines are available that can speed up
and simplify the processing steps for studies with a large number of
samples.
3.9 Processing of2D 1. The following apodization parameters apply for the J-resolved
NMR Spectra spectrum: Zero-filling by a factor of 2 (SI=2*TD) in f2,
(J-Resolved Spectra) increasing the number of processed data points in f1 to 256
(SI=256in f1), multiplication of the FID by a exponential
window function with a line-broadening factor of 0.3
(WDW=EM, LB=0.3) in both dimensions (see Note 13). The
phasing mode in f1 should be set to magnitude calculation
(PH_mod=mc).
2. Now perform a two-dimensional Fourier transformation of the
data (xfb).
3. Phase the spectrum in the acquisition dimension f2.
4. Tilt the spectrum (tilt).
5. Symmetrize the spectrum around the 0Hz middle line (symj).
6. Correct the baseline in f2 (abs2). Make sure the left and right
limits for baseline correction (absf1 and absf2) as well as the
desired polynomial grade of baseline correction (absg) are set
correctly.
7. Calibrate the spectrum to the DSS signal at 0ppm in f2 and
0Hz in f1.
8. If desired produce a 1D f2 projection of the spectrum (f2sum
1 256 <destination processing number> n). This
1D projection is essentially a simplified 1D spectrum in which
all multiplet signals have been collapsed to central singulets
(effectively decoupling all proton signals). Thus signal overlap
is greatly reduced, and it might be possible to distinguish
metabolites from each other that have close chemical shifts and
complex signal splitting.
Automated processing routines are available that can speed up
and simplify the processing steps for studies with a large number of
samples.
3.10 Quality After processing, it should be checked that the acquired spectra
Assurance ofNMR meet quality standards with respect to line width/line shape (shim
Data Acquisition quality during acquisition), baseline flatness, magnitude of the
residual water signal, absence of phase errors, and absence of a
receiver gain overload. These quality criteria are outlined in detail
in [26]. If a particular spectrum does not meet these criteria, it
should either be rerun, or the respective sample should be excluded
from further analysis. It is advisable to periodically check the quality
of spectra during an automation run to detect (and rerun) any

failed samples as soon as possible. Even with optimized parameters,
one can expect about 5% of all samples to fail on first attempt.
3.11 Multivariate After acquisition and processing, the NMR spectra from metabo-
Statistical Analysis: lomics experiments are typically analyzed with Multivariate
General Comments Statistical Analysis (MVSA) to identify spectral features in the
NMR spectra that are correlated with the biological factors tested
in the metabolomics study.
The typical individual steps in the MVSA pipeline are:
(a) Statistical preprocessing: Collating and treating the NMR
data to bring them into a form suitable for MVSA.This
involves collating the 1D data to a data matrix.
(b) Bucketing (=data reducing the matrix), normalizing and

scaling the matrix [28, 29, 35].
(c) Performing the actual MVSA: Multiple methods both

unsupervised and supervised are available.
(d) Interpreting the MVSA: This step involves identifying the
spectral features that are correlated with the biological
factors studied.
(e) Identifying the metabolites represented by the spectral

features in the previous step.
(f) Biological interpretation: This step can involve a pathway
analysis of the identified metabolites, correlation of the
metabolomics data with other data (e.g., metadata, other
omics platforms), or using the data in the context of
genome-scale metabolic modeling.
There are a wide range of different MVSA methods and
multiple program platforms available to perform each step of this
analysis pipeline [2729]. Thus, a detailed description of MVSA
and the steps included in the analysis go beyond the scope of this
chapter. Instead, we will, in the following sections, provide general
advice on the general pipeline and the individual steps involved
rather than detailed guidelines.
3.12 Statistical Before MVSA, metabolomics data are usually arranged in a data
Preprocessing matrix, called X matrix, in which the intensity data from the
individual NMR spectra are arranged in rows (rows=samples).
Thus, the columns represent the intensities of the spectra at one
frequency point or range (columns=variables). Often this X matrix
is not used at full data resolution (as, for example, a data matrix of
100 1D-NOESY spectra at full resolution is 25MB data), but is
data reduced by dividing each NMR spectrum into segments of
equal width, called buckets.
Historically, bucket widths of 0.04ppm have been used and

are still useful for an initial exploratory data analysis, but as com-
puting power is not a limiting factor anymore, higher resolutions
(smaller bucket widths) or even analysis of spectra at full resolution
is preferable. The position of some signals in the NMR spectrum is
not constant, but depends on factors such as pH and ionic strength
of each sample. Thus, there will be signals whose position is vari-
able between spectra. There are two possible methods of avoiding
these shifting signals affecting the subsequent MVSA: (1) Bucketing
with buckets of sufficient width (or with variable bucket widths
[36, 37]) so that signals only shift position within one bucket [35,
38]. (2) Using spectral alignment tools [3943] that align shifting
signals with each other, thus enabling MVSA of spectra at full
resolution.
Two further operations need to be performed on the X
matrix before MVSA: First each row of data needs to be normal-
ized (= normalization of each spectrum) to cancel out artifacts
from sample dilution or unbalanced regulation. Secondly, each
data column (= each variable) needs to be appropriately scaled.
This statistical preprocessing of NMR data can be performed
with the Amix program from within the Bruker Biospin suite of
programs. Alternatively, Chenomx NMR Suite (Chenomx Inc,
Edmonton, Canada) can import Bruker NMR data and bucket
them, as well as aid metabolite identification (see Subheading3.14).
Moreover, the command convbin2asc can be used to export
NMR spectra from within Bruker Biospins TopSpin program to
ASCII files, which can then be collated and imported into, e.g.,
Matlab or R for subsequent MVSA.
3.13 Multivariate As its name suggests, MVSA analyzes the statistical trends of
Statistical Analysis multiple variables in a data set at once, and is thus highly appropriate
for the analysis of NMR-based metabolomics data, which at full
spectral resolution could contain 64k different variables. Generally,
MVSA methods are also data reduction methods that try to present
the complexity of a highly multidimensional data set by projecting
them to a lower dimensional space of latent variables that are easier
to understand and often align with the biological factors present in
the study. There are two different types of MVSA methods:
1. Unsupervised methods, such as Principal Components Analysis
(PCA) [44], which only analyze the X data matrix itself and thus
provide an unbiased overview of the data. Because unsupervised
methods are unbiased, they are an ideal first entry point of data
exploration and analysis in metabolomics studies. However for
that same reason, the latent variables (= the principal components)
might not align with biological factors but rather with confounding
factors in the study, especially if the confounding factors are larger
in magnitude than the biological effects (see Note 14).
2. Supervised MVSA methods, such as Partial Least Squares

Projections to Latent Structures (PLS) [45], orthogonal PLS
[46], or bidirectional orthogonal PLS (O2PLS) [47, 48], aim
to correlate the X data matrix with a Y matrix, containing the
metadata available for a study, such as group identity of each
sample, etc. Thus, supervised MVSA methods are highly useful
for uncovering correlations between variables in the X data
matrix and biological factors (see Note 15). Essentially, super-
vised methods enable the discovery of potential biomarkers in
the X data. However, supervised MVSA methods are vulnerable
to bias (= to finding spurious correlations between variables and
biological factors), and thus need to be rigorously validated. For
both supervised and unsupervised methods it is highly encour-
aged to report all figures of merit characterizing the statistical
analysis, such as number of samples (n), number of variables (k),
number of latent variables in the fitted statistical model (A), the
% X and Y variance explained by the model (R2X and R2Y,
respectively), and the cross-validated predictability of the model
(Q2) [49, 50].
MVSA methods such as PCA, PLS, or O2PLS yield graphical
representations of the data that are identical and comprise usually a
scores plot and a loadings plot. The scores plot shows how similar/
different samples are to each other, and thus allows both the identi-
fication of sample clustering and/or the identification of individual
sample outliers [28, 29]. The accompanying loadings plot then
shows which variables (=spectral regions) are responsible for any
sample clustering observed in the scores plot. Identification of the
metabolites associated with the respective spectral regions forms the
bridge between MVSA and biological interpretation.
Several programs are available to carry out MVSA, both
commercially and open source, such as Simca P+ (Umetrics,
Sweden), Amix (Bruker Biospin), Matlab (MathWorks, USA), or
R (https://www.r-project.org). Further lists of available programs
and analysis strategies are documented in [2729]. These lists are
by no means exhaustive, and researchers are encouraged to work
with the analysis pipeline that is most adequate in their individual
circumstances.
3.14 Metabolite Once spectral features that correlate with biological factors have
Identification been identified in the MVSA, it becomes important to identify the
metabolites that are associated with these spectral features, as that
enables biological/clinical interpretation of a metabolomic study.
In the most simple cases, metabolite identification can be made
from the 1D NMR spectra alone, by recognizing the characteristic
position and signal splitting of the NMR signals of individual
metabolites. As an example, annotated 1D 1H-NMR spectra of
saliva and GCF are shown in Fig.2. Programs such as Amix (Bruker
Biospin) or Chenomx NMR Suite (Chenomx Inc, Edmonton,
A Formate AcOH
Prop
Glu,
Glu,
Pro, PG
EtOH Pro
Isoval
His His Pyr EtOH
Im MeOH
Succ Prop
Tyr Glu Gluc Fuc
4-OHPhAc Urea Gal
Choline Lact
PhAc Fuc, Valine
Gly Putr Putr
Phe Gal, Lys Sarc Ile
Im Gluc But
Isoval
TMA DMA
Lact Ala But
MA
5.5
B Formate AcOH
Prop, EtOH
But,
Val,
Capryl, Val,
Capr Lact
Capryl,
EtOH Capr Prop
AcAc But
Glycerol MeOH But,
Acetone
Val, Val,
Tau Lev
CreP Capryl, Capryl,
Lact Crn Capr Capr
Cre Cit
5.5
Fig. 2 1D 1H-NMR spectra of saliva and GCF at 900MHz. (a) Saliva. The intensity in the region between 9 and
5.4ppm has been scaled up by a factor of 10 compared to the rest of the spectrum. (b) GCF.The major metabo-
lites have been annotated in both panels, and the following abbreviations have been used: 4-OHPhAc
4-hydroxyphenylacetate, AcAc acetoacetate, AcOH acetate, Ala alanine, But butyrate, Capr caprate, Capyl
caprylate, Cit citrate, Cre creatine, CreP creatine phosphate, Crn creatinine, DMA dimethylamine, EtOH
ethanol, Fuc fucose, Gal galactose, Gly glycine, Glu glutamate, Gluc glucose, His histidine, Ile isoleucine, Im
imidazole, Isovlr isovalerate, Lact lactate, Lev levulinate, Lys lysine, MA methylamine, MeOH methanol, PG
propylene glycol, Phe phenylalanine, PhAc phenyl acetate, Pro proline, Prop propionate, Putr putrescine, Pyr
pyruvate, Sarc sarcosine, Succ succinate, Tau taurine, TMA trimethylamine, Tyr tyrosine, Vlr valerate
Canada) have inbuilt and validated data bases of the signal patterns
of a wide range of metabolites. Alternatively, public available
databases, such as the BioMagRes DataBank (http://www.bmrb.
wisc.edu) [51] or the Human Metabolome Data Bank (http://
www.hmdb.ca) [52, 53], contain both 1D and 2D NMR data of at
present more than 1000 metabolites. In addition, the HMDB has
been recently updated with specialist data for the human saliva
metabolome [54]. The J-resolved spectra recoded of each sample
can help with metabolite identification in regions of the NMR
spectrum that are affected by spectral overlap, by clarifying the
splitting patterns of individual metabolite signals and/or simplifying
the NMR spectrum (see Subheading3.6.3). In addition, several
homo- and hetero-nuclear 2D NMR experiments, which are also in
use in organic chemistry and/or natural products discovery, are
useful in identifying metabolites (for example TOCSY, 13C-HSQC,
13
C-HSQC-TOCSY, 13C-HMBC) [55]. However, the acquisition
times of these experiments are in the order of several hours and are
thus too long to be routinely employed on every sample. Instead,
these spectra are usually measured manually on a single,
representative sample of the study, most suitably on the pooled QC
sample, which should represent an average of all samples present in
the study.
In extreme cases, or where metabolites are unknown, NMR
methods for metabolite identification might have to be comple-
mented by hyphenated methods (fractionation, HPLC-NMR) and
other analytical methods, such as mass spectrometry.
3.15 Biological/ Identification of the metabolites that correlate with specific

Clinical Interpretation biological/clinical factors allows identification of the metabolic
pathways that are altered (as well as the metabolic pathways that
are not changed) under specific conditions. Several tools of pathway
analysis are available for this purpose [5659]. This opens the door
for a biological interpretation of the metabolomics results.
Alternatively, the identification of specific metabolites that are
associated with specific disease conditions opens the door to the
discovery of biomarkers for disease diagnosis, prognosis, and/or
theranosis.
4 Notes
1. These specifications are suggested minimum requirements.

Operation at 600MHz proton frequency is recommended,
but not mandatory. Clinical phenome centers are likely to
operate at that 1H frequency. Spectrometers operating at
higher field strength offer higher resolution and higher sensi-
tivity that is beneficial in research applications, and it is likely
that 800MHz 1H frequency will become a second standard
in the future. Similarly, the presence of a BBI probe is a mini-
mum equipment requirement, and probes such as triple reso-
nance (TXI) probes are as suitable. Even more advantageous
are cryoprobes, which offer significantly higher signal-to-noise
ratios that lead to shorter acquisition times, more intense
NMR spectra, and higher sample throughput.
2. There are other collection devices for collecting stimulated
saliva (e.g., Salivette), but it should be noted that different
techniques for collecting saliva can have different recovery
rates for metabolites, and specifically that Salivettes have sig-
nificantly lower metabolite recovery rates as well as an altered
profile compared to the paraffin collection technique [60].
3. Alternatively, an electronic measuring device, the Periotron,
can be used to determine the volume of GCF collected [30].
4. Saliva samples are collected as pure saliva and then have to be
diluted with a buffer during NMR sample preparation. In contrast,
GCF samples are already collected in essentially the final NMR

sample buffer, without the field-lock D2O and the chemical shift/
pH standards. These are the only substances that need to be
added to CGF samples during NMR sample preparation
(Subheading3.4.3 or Subheading3.4.4). Different sample buf-
fers are used because the volume of collected saliva is usually in the
mL range, whereas the volume of collected GCF is in the L range,
thus any dilution needs to be minimized during sample prepara-
tion. For that reason, it is also advised to prepare GCF samples in
3mm NMR tubes with a total sample volume of 200L rather
than 5mm NMR tubes, which require a sample volume of at
least 550L.
5. Sample degradation usually starts directly after thawing a
sample. In addition, biological samples are usually affected by
repeated freeze-thaw cycles. Thus, once samples have been
thawed, it is important to work through the full pipeline of
sample preparation and NMR data acquisition as quickly as
expedient, to keep samples cooled (at 4C or on ice) at all
times, to add bactericidal preservatives, such as sodium azide,
and to avoid freeze-thaw cycles during sample preparation.
6. With this setting of the temperature conditioning system the
next sample is effectively preheated during the total NMR
acquisition time of all experiments of the preceding sample,
which is usually longer the minimum preheating time of 3min.
7. This slightly counterintuitive instruction arises, because the
proton concentration of water is 111M, whereas the concen-
tration of most metabolites is in the millimolar or sub-
millimolar range. Thus, in a biofluid spectrum without water
suppression less than a 1/100,000th of the observable FID
signal is not from water. Water suppression is achieved by
continuous-wave irradiation on the water frequency. Thus, the
better the match between the carrier frequency of the proton
channel (O1) and the water frequency, the better the water
suppression, and the less total signal is observed.
8. For quality control of the experimental parameters optimized
so far the reader is referred to [26].
9. On modern Bruker Biospin NMR spectrometers RG values
larger than 203 do not achieve any higher sensitivity due to the
digital oversampling of the analog-digital converter. Conversely,
signal and dynamic range is lost at RG values of less than 128.
Thus, after optimization of parameters one should achieve RG
values of at least 128. Maximum achievable RG values larger
than 203 can safely be reset to RG=203.
10. If the command rga has been used to check the receiver gain,
then afterward the receiver gain has to be manually changed
back from the receiver gain achieved by rga to the desired
receiver gain of 128.
11. An example for an acquisition AU program that reads in the

correct parameter set, shims on the sample, and optimizes
pulse lengths and power levels is:
/* AU program au_zg_auto */
GETCURDATA
XCMD(topshim tunea)/*automated shim-
ming on the sample.*/
/* ^^^This line might be omitted if shim-
ming is done within ICON-NMR*/
XCMD(pulsecal same)/*automated deter-
mination of pulse lengths*/
ZG
QUIT
12. The temperature conditioning system settings in ICON-NMR
will for the duration of the automation run override any settings
made in the SampleJet web interface during step 3 in
Subheading3.5. Note though that the temperature of the tem-
perature conditioning system cannot be set in ICON-NMR,
but has to be set in the SampleJet web interface during step 3
in Subheading3.5. The temperature of the temperature condi-
tioning system should be set to the same temperature as the
measurement temperature of the experiment (298K) to shorten
temperature equilibration times for each sample.
13. Some spectra will after processing with these parameters show
artifacts around intense signals (so-called sinc wiggles). These
sinc wiggles have two potential origins: (1) The receiver gain
of the experiment was too high (likely caused by a sample with
an abnormally high concentration), resulting in overloading of
the analog-digital converter. In this case, the only solution is to
remeasure the corresponding experiment with a lower RG. (2)
Sinc wiggles can also occur if the signal of certain intense
metabolites has not decayed to zero by the end of the free
induction decay (or in other words AQ was too small and the
free induction decay was truncated). This can be solved at the
processing stage by using an alternative window function to
exponential multiplication prior to Fourier transformation,
such as a sine function shifted by /2 (WDW=sine and SSB=2,
then using sinm and ft for processing).
14. For this reason both stringent experimental study design and
stringent quality control procedures during sample handling and
data acquisition are essential for metabolomic studies. Otherwise,
the biological effects to be characterized/studied can easily be
overwhelmed by larger (and avoidable) confounding effects.
15. Instead of using metadata as content, the Y matrix can also
contain other omics data that were derived from the same
samples as the X matrix data, such as mass spectrometry-
derived metabolomics data, proteomics data, or transcriptomics
data. In this case, supervised MVSA methods can be used to
make inroads into multiomics data analysis [6163].
Acknowledgments
We gratefully acknowledge Dr Shaneen Leishman for assistance in

refining the collection methods. We are grateful to Dr Emma
Broughton and Dr Rachel Dunn for preparing and analyzing the
saliva and GCF samples used for Fig.2. NMR spectra for Fig.2 were
measured at the University of Queenslands 900MHz spectrometer,
which is part of the Queensland NMR Network (QNN), and the
authors acknowledge financial support provided by the Queensland
State Government to the Queensland NMR Network facilities at
The University of Queensland. We wish to thank Dr Gregory Pierens
for critical reading of the manuscript and helpful advice.
References
1. Fiehn O (2002) Metabolomicsthe link between (2009) Understanding the human salivary
genotypes and phenotypes. Plant Mol Biol metabolome. NMR Biomed 22:577584
48:155171 12. Aimetti M, Cacciatore S, Graziano A, Tenori L
2. Nicholson JK, Lindon JC, Holmes E (1999) (2012) Metabonomic analysis of saliva reveals
Metabonomics: understanding the meta- generalized chronic periodontitis signature.
bolic responses of living systems to pathophysi- Metabolomics 8:465474
ological stimuli via multivariate statistical 13. Klukowska M, Goyal CR, Khambe D, Cannon
analysis of biological NMR spectroscopic data. M, Miner M, Gurich N, Circello B, Huggins T,
Xenobiotica 29:11811189 Barker ML, Furnish C, Conde E, Hoke P,
3. Aurich MK, Thiele I (2016)Computational mod- Haught C, Xie SC, White DJ (2015) Response of
eling of human metabolism and its application to chronic gingivitis to hygiene therapy and experi-
systems biomedicine. In: Schmitz U, Wolkenhauer mental gingivitis. Clinical, microbiological and
O (eds) Systems medicine, pp253-281 metabonomic changes. Am JDent 28:273284
4. Issaq HJ, Van QN, Waybright TJ, Muschik 14. Fidalgo TKS, Freitas-Fernandes LB, Angeli R,
GM, Veenstra TD (2009) Analytical and statis- Muniz AMS, Gonsalves E, Santos R, Nadal J,
tical approaches to metabolomics research. Almeida FCL, Valente AP, Souza IPR (2013)
JSep Sci 32:21832199 Salivary metabolite signatures of children with
5. Shepherd LVT, Fraser P, Stewart D (2011) and without dental caries lesions. Metabolomics
Metabolomics: a second-generation platform for 9:657666
crop and food analysis. Bioanalysis 3:11431159 15. Fidalgo TKS, Freitas-Fernandes LB, Almeida
6. Shulaev V (2006) Metabolomics technology FCL, Valente AP, Souza IPR (2015)
and bioinformatics. Brief Bioinform 7:128139 Longitudinal evaluation of salivary profile from
7. Zhang AH, Sun H, Wang P, Han Y, Wang XJ children with dental caries before and after
(2012) Modern analytical techniques in metab- treatment. Metabolomics 11:583593
olomics analysis. Analyst 137:293300 16. Lemanska A, Grootveld M, Silwood CJL,
8. Gebregiworgis T, Powers R (2012) Application Brereton RG (2012) Chemometric variance
of NMR metabolomics to search for human analysis of 1H NMR metabolomics data on the
disease biomarkers. Comb Chem High effects of oral rinse on saliva. Metabolomics
Throughput Screen 15:595610 8:S64S80
9. Emwas AHM, Salek RM, Griffin JL, Merzaban 17. Lloyd GR, Wongravee K, Silwood CJL,
J(2013) NMR-based metabolomics in human Grootveld M, Brereton RG (2009) Self
disease diagnosis: applications, limitations, and Organising Maps for variable selection: applica-
recommendations. Metabolomics 9:10481072 tion to human saliva analysed by nuclear mag-
netic resonance spectroscopy to investigate the
10. Bertram HC, Eggers N, Eller N (2009) effect of an oral healthcare product. Chemometr
Potential of human saliva for nuclear magnetic Intell Lab Syst 98:149161
resonance-based metabolomics and for health-
related biomarker identification. Anal Chem 18. Misawa T, Date Y, Kikuchi J(2015) Human
81:91889193 metabolic, mineral, and microbiota fluctua-
tions across daily nutritional intake visualized
11. Takeda I, Stretch C, Barnaby P, Bhatnager K, by a data-driven approach. JProteome Res
Rankin K, Fu H, Weljie A, Jha N, Slupsky C 14:15261534
19. Walsh MC, Brennan L, Malthouse JPG, Roche RA, Willmitzer L (eds) Plant metabolomics.
HM, Gibney MJ (2006) Effect of acute dietary pp117128
standardization on the urinary, plasma, and 30. Griffiths GS (2003) Formation, collection

salivary metabolomic profiles of healthy and significance of gingival crevice fluid.
humans. Am JClin Nutr 84:531539 Periodontology 2000(31):3242
20. Neyraud E, Tremblay-Franco M, Gregoire S, 31. Henson BS, Wong DT (2010) Collection,
Berdeaux O, Canlet C (2013) Relationships storage, and processing of saliva samples for
between the metabolome and the fatty acid downstream molecular applications. In:
composition of human saliva; effects of stimu- Seymour GJ, Cullinan MP, Heng NCK (eds)
lation. Metabolomics 9:213222 Oral biology: molecular techniques and appli-
21. Mounayar R, Morzel M, Brignot H, Tremblay- cations, pp2130
Franco M, Canlet C, Lucchi G, Ducoroy P, 32. Akoka S, Barantin L, Trierweiler M (1999)
Feron G, Neyraud E (2014) Nutri-metabolomics Concentration measurement by proton NMR
applied to taste perception phenotype: human using the ERETIC method. Anal Chem
subjects with high and low sensitivity to taste of 71:25542557
fat differ in salivary response to oleic acid. 33. Silvestre V, Goupry S, Trierweiler M, Robins
OMICS J Integrat Biol 18:666672 R, Akoka S (2001) Determination of substrate
22. Mounayar R, Morzel M, Brignot H, Tremblay- and product concentrations in lactic acid bacte-
Franco M, Canlet C, Lucchi G, Ducoroy P, rial fermentations by proton NMR using the
Feron G, Neyraud E (2014) Salivary markers ERETIC method. Anal Chem 73:18621868
of taste sensitivity to oleic acid: a combined 34. Clos LJ, Jofre MF, Ellinger JJ, Westler WM,
proteomics and metabolomics approach. Markley JL (2013) NMRbot: Python scripts
Metabolomics 10:688696 enable high-throughput data collection on cur-
23. Santone C, Dinallo V, Paci M, D'Ottavio S, rent Bruker BioSpin NMR spectrometers.
Barbato G, Bernardini S (2014) Saliva metabo- Metabolomics 9:558563
lomics by NMR for the evaluation of sport per- 35. Euceda LR, Giskedegrd GF, Bathen TF
formance. JPharm Biomed Anal 88:441446 (2015) Preprocessing of NMR metabolomics
24. de Laurentiis G, Paris D, Melck D, Maniscalco data. Scand JClin Lab Investig 75:193203
M, Marsico S, Corso G, Motta A, Sofia M 36. Davis RA, Charlton AJ, Godward J, Jones SA,
(2008) Metabonomic analysis of exhaled Harrison M, Wilson JC (2007) Adaptive binning:
breath condensate in adults by nuclear mag- an improved binning method for metabolomics
netic resonance spectroscopy. Eur Respir data using the undecimated wavelet transform.
J32:11751183 Chemometr Intell Lab Syst 85:144154
25. Beckonert O, Keun HC, Ebbels TMD, Bundy 37. Sousa SAA, Magalhaes A, Ferreira MMC
JG, Holmes E, Lindon JC, Nicholson JK (2013) Optimized bucketing for NMR spectra:
(2007) Metabolic profiling, metabolomic and three case studies. Chemometr Intell Lab Syst
metabonomic procedures for NMR spectros- 122:93102
copy of urine, plasma, serum and tissue extracts.
Nat Protoc 2:26922703 38. Wishart DS (2008) Quantitative metabolomics
using NMR.TrAC Trends Anal Chem
26. Dona AC, Jimnez B, Schfer H, Humpfer E, 27:228237
Spraul M, Lewis MR, Pearce JTM, Holmes E,
Lindon JC, Nicholson JK (2014) Precision high- 39. Kim SB, Wang Z, Hiremath B (2010) A
throughput proton NMR spectroscopy of human Bayesian approach for the alignment of high-
urine, serum, and plasma for large-scale meta- resolution NMR spectra. Ann Oper Res
bolic phenotyping. Anal Chem 86:98879894 174:1932
27. Eliasson M, Rannar S, Trygg J(2011) From 40. MacKinnon N, Ge W, Khan AP, Somashekar
data processing to multivariate validation BS, Tripathi P, Siddiqui J, Wei JT, Chinnaiyan
essential steps in extracting interpretable infor- AM, Rajendiran TM, Ramamoorthy A (2012)
mation from metabolomics data. Curr Pharm Variable reference alignment: an improved
Biotechnol 12:9961004 peak alignment protocol for NMR spectral data
with large intersample variation. Anal Chem
28. Roberts MJ, Schirra HJ, Lavin MF, Gardiner 84:53725379
RA (2014) NMR-based metabolomics: global
analysis of metabolites to address problems in 41. Tomasi G, Savorani F, Engelsen SB (2011) ico-
prostate cancer. In: iConcept Press (ed) Cervical, shift: an effective tool for the alignment of
breast and prostate cancer. iConcept Press, chromatographic data. JChromatogr A
Tokwawan, Kowloon, Hong Kong, pp143 1218:78327840
29. Trygg J, Gullberg J, Johansson AI, Jonsson 42. Veselkov KA, Lindon JC, Ebbels TMD,

P, Moritz T (2006) Chemometrics in metab- Crockford D, Volynkin VV, Holmes E,
olomicsan introduction. In: Saito K, Dixon Davies DB, Nicholson JK (2009) Recursive
segment- wise peak alignment of biological
1
H-1 NMR spectra for improved metabolic 53. Wishart DS, Tzur D, Knox C, Eisner R, Guo
biomarker recovery. Anal Chem 81:5666 AC, Young N, Cheng D, Jewell K, Arndt D,
43. Wang K, Barding GA, Larive CK (2015) Peak Sawhney S, Fung C, Nikolai L, Lewis M,
alignment of one-dimensional NMR spectra by Coutouly MA, Forsythe I, Tang P, Shrivastava
means of an intensity fluctuation frequency dif- S, Jeroncic K, Stothard P, Amegbey G, Block
ference (IFFD) segment-wise algorithm. Anal D, Hau DD, Wagner J, Miniaci J, Clements
Meth 7:96739682 M, Gebremedhin M, Guo N, Zhang Y,
44. Pearson K (1901) On lines and planes of clos- Duggan GE, MacInnis GD, Weljie AM,
est fit to systems of points in space. Phil Mag Dowlatabadi R, Bamforth F, Clive D, Greiner
2:559572 R, Li L, Marrie T, Sykes BD, Vogel HJ,
Querengesser L (2007) HMDB: the human
45. Wold S, Ruhe A, Wold H, Dunn WJ (1984) The metabolome database. Nucleic Acids Res
collinearity problem in linear-regressionthe par- 35:D521D526
tial least-squares (PLS) approach to generalized
inverses. SIAM JSci Stat Comput 5:735743 54. Dame ZT, Aziat F, Mandal R, Krishnamurthy
R, Bouatra S, Borzouie S, Guo AC, Sajed T,
46. Trygg J, Wold S (2002) Orthogonal projec- Deng L, Lin H, Liu P, Dong E, Wishart DS
tions to latent structures (O-PLS). JChemometr (2015) The human saliva metabolome.
16:119128 Metabolomics 11:18641883
47. Trygg J(2002) O2-PLS for qualitative and 55. Cavanagh J, Fairbrother WJ, Palmer AG,
quantitative analysis in multivariate calibration. Rance M, Skelton NJ (2007) Protein NMR
JChemometr 16:283293 spectroscopy: principles and practice, 2nd edn.
48. Trygg J, Wold S (2003) O2-PLS, a two-block Protein NMR Spectroscopy: Principles and
(X-Y) latent variable regression (LVR) method Practice, pp1888
with an integral OSC filter. JChemometr 56. Frolkis A, Knox C, Lim E, Jewison T, Law V,
17:5364 Hau DD, Liu P, Gautam B, Ly S, Guo AC, Xia
49. Goodacre R, Broadhurst D, Smilde AK, Kristal JG, Liang YJ, Shrivastava S, Wishart DS (2010)
BS, Baker JD, Beger R, Bessant C, Connor S, SMPDB: the small molecule pathway database.
Calmani G, Craig A, Ebbels T, Kell DB, Manetti Nucleic Acids Res 38:D480D487
C, Newton J, Paternostro G, Somorjai R, 57. Aggio R, Ruggiero K, Villas-Bas S (2009)
Sjstrm M, Trygg J, Wulfert F (2007) Proposed Pathway activity profiling (papi): an integration
minimum reporting standards for data analysis system for metabolomics data. N Biotechnol
in metabolomics. Metabolomics 3:231241 25:S334S335
50. Lindon JC, Nicholson JK, Holmes E, Keun HC, 58. Aggio RBM, Ruggiero K, Villas-Bas SG
Craig A, Pearce JTM, Bruce SJ, Hardy N, (2010) Pathway activity profiling (PAPi): from
Sansone SA, Antti H, Jonsson P, Daykin C, the metabolite profile to the metabolic path-
Navarange M, Beger RD, Verheij ER, Amberg A, way activity. Bioinformatics 26:29692976
Baunsgaard D, Cantor GH, Lehman-McKeeman
L, Earll M, Wold S, Johansson E, Haselden JN, 59. Xia JG, Sinelnikov IV, Han B, Wishart DS
Kramer K, Thomas C, Lindberg J, Schuppe- (2015) MetaboAnalyst 3.0-making metabolo-
Koistinen I, Wilson ID, Reily MD, Robertson mics more meaningful. Nucleic Acids Res
DG, Senn H, Krotzky A, Kochhar S, Powell J, 43:W251W257
van der Ouderaa F, Plumb R, Schaefer H, Spraul 60. Lenander-Lumikari M, Johansson I, Vilja P,
M, Standard Metabolic Reporting Structures Samaranayake LP (1995) Newer saliva collec-
Working Group (2005) Summary recommendation methods and saliva composition: a study
tions for standardization and reporting of meta- of two Salivette kits. Oral Dis 1:8691
bolic analyses. Nat Biotechnol 23:833838 61. Bylesj M, Eriksson D, Kusano M, Moritz T,
51. Ulrich EL, Akutsu H, Doreleijers JF, Harano Trygg J(2007) Data integration in plant biology:
Y, Ioannidis YE, Lin J, Livny M, Mading S, the O2PLS method for combined modeling of
Maziuk D, Miller Z, Nakatani E, Schulte CF, transcript and metabolite data. Plant
Tolmie DE, Wenger RK, Yao HY, Markley JL J52:11811191
(2008) BioMagResBank. Nucleic Acids Res 62. Bylesj M, Nilsson R, Srivastava V, Grnlund
36:D402D408 A, Johansson AI, Jansson S, Karlsson J, Moritz
52. Wishart DS, Jewison T, Guo AC, Wilson M, T, Wingsle G, Trygg J(2009) Integrated anal-
Knox C, Liu YF, Djoumbou Y, Mandal R, ysis of transcript, protein and metabolite data
Aziat F, Dong E, Bouatra S, Sinelnikov I, to study lignin biosynthesis in hybrid aspen.
Arndt D, Xia JG, Liu P, Yallou F, Bjorndahl T, JProteome Res 8:199210
Perez-Pineiro R, Eisner R, Allen F, Neveu V, 63. Kirwan GM, Johansson E, Kleemann R, Verheij
Greiner R, Scalbert A (2013) HMDB 3.0-the ER, Wheelock AM, Goto S, Trygg J, Wheelock
human metabolome database in 2013. Nucleic CE (2012) Building multivariate systems biol-
Acids Res 41:D801D807 ogy models. Anal Chem 84:70647071
Chapter 6
Gene Therapy ofSalivary Diseases

BruceJ.Baum, SandraAfione, JohnA.Chiorini, AnaP.Cotrim,
CorinneM.Goldsmith, andChangyuZheng
Abstract
For many years, our research group worked to develop gene transfer approaches for salivary gland disorders
that lacked effective conventional therapy. The purpose of this chapter is to describe and update key meth-
ods used in this process. As described in our original chapter from the 2010 volume, we focus on one clini-
cal condition, irradiation-induced salivary hypofunction, and address the choice of transgene and vector to
be used, the construction of recombinant viral vectors, how vector delivery is accomplished, and methods
for assessing vector function invitro and in an appropriate animal model.
Key words Gene therapy, Salivary glands, Adenovirus, Adeno-associated virus, Radiation damage,
Salivary hypofunction
1 Introduction
There are two major disorders that lead to the irreversible loss of
salivary gland function: (1) irradiation damage that occurs during
the course of treatment for a head and neck cancer and (2) the
autoimmune exocrinopathy Sjgrens syndrome. Both disorders
are fairly common. For 2015, the estimated number of new cases
of oro-pharyngeal cancers diagnosed in the US was 45,780,
accounting for 2.76% of all estimated new malignancies [1]. The
treatment for most such patients, in developed societies, includes
surgery and irradiationchemotherapy. Sjgrens syndrome has a
prevalence of ~0.51%, making it the second most common rheu-
matic disease after rheumatoid arthritis [2]. Although the etiolo-
gies of these two disorders are dramatically different, both
conditions result in the loss of salivary acinar cells, the only cell
type that normally secretes the fluid component of saliva. With
both conditions, the predominant remaining epithelial cells are of
duct origin, and incapable of fluid secretion. Patients lacking saliva
suffer considerable morbidity, including dysphagia, oral infections,
delayed mucosal healing, and considerable pain and discomfort.
107
108 Bruce J. Baum et al.
If patients in either disease group have a reasonable mass of

acinar cells remaining, treatment with sialogogues (salivary stimu-
lants, e.g., pilocarpine, civemaline) can be beneficial. For patients
who lack most or all salivary acinar cells there currently is no suit-
able treatment available, a situation that provided the impetus for
our beginning to explore the use of invivo gene transfer (gene
therapy). It also is important to recognize a key difference in these
two conditions: irradiation-induced salivary hypofunction is a
localized gland problem. While this condition certainly leads to
some systemic concerns (e.g., dysphagia, infections), a primary
treatment for it needs only to be targeted to the damaged gland.
Furthermore, the pathologic etiology, i.e., the radiation treatment,
is time-limited. A patient presenting with irradiation-induced sali-
vary hypofunction was treated with radiotherapy in the past and is
without any ongoing active disease process. Conversely, a patient
with Sjgrens syndrome experiences a systemic autoimmune dis-
ease, albeit one commonly having salivary glands as a major target
organ. Localized salivary gland gene therapy for a patient with
Sjgrens syndrome can address their salivary hypofunction [3],
but in all likelihood, at least for the present, will have no beneficial
effects on the systemic disease process. Additionally, it is important
to recognize that Sjgrens syndrome patients show continuous
disease activity, e.g., the presence of serum autoimmune markers.
Based on its more localized nature, and the absence of an active
disease process, irradiation-induced salivary hypofunction is a dis-
order more readily treatable by salivary gland gene therapy. It also
lends itself to a more useful presentation for a chapter such as this.
Hence, the focus of this chapter is only on the repair of
irradiation-induced salivary hypofunction.
As with the development of a therapy for any disease condi-
tion, an essential initial element is to have a good understanding of
the physiology of the normal target tissue, in addition to a good
understanding of the pathophysiological situation. Fortunately,
our laboratory has such a background, as a result of working many
years in the salivary gland field [4]. Based on that understanding,
we decided that surviving duct cells in the irradiated gland were
capable of generating an osmotic gradient into the gland lumen,
but needed water permeability pathways to allow fluid to follow
this gradient [5]. Accordingly, the gene of choice for our approach
was aquaporin-1 (AQP1), the first described water channel [6].
Given that salivary epithelial cells are slowly dividing post-mitotic
cells, it was appropriate to employ nonintegrating vectors for the
actual gene transfer. Accordingly, we have used recombinant sero-
type 5 adenoviral (rAd5) and serotype 2 adeno-associated viral
(rAAV2) vectors [7]. Each vector type has distinct advantages. For
example, rAd5 vectors are relatively easy to produce, lead to high
transgene expression, but induce a potent immune response that in
animal models renders the expression transient (<24 weeks).
Gene Therapy ofSalivary Diseases 109
rAAV2 vectors elicit a modest immune response and generally lead

to low, but stable, transgene expression; however, they are more
difficult to produce. We initially used rAd5 vectors for proof of
concept studies in both animal models and humans, and now are
working on developing a long-lived therapeutic correction of
irradiation-induced salivary hypofunction by using a rAAV2 vector.
Herein, we will describe the production of both vector types.
1.1 Recombinant Adenoviruses are non-enveloped, double stranded DNA viruses.

Serotype 5 Adenoviral There are more than 50 human adenoviral serotypes. A common
Vector Production adenoviral gene transfer vector is based on adenovirus serotype 5
and we typically use an E1 gene-deleted, replication-deficient rAd5
vector in our studies. Importantly, these vectors can efficiently
transduce salivary acinar and duct cells [8]. To construct, propa-
gate and purify an E1 deleted rAd5 vector, various products and
services from different companies can be used, including ABL, Inc
(Rockville, MD), Agilent Technologies (Santa Clara, CA), Cell
Biolabs, Inc (San Diego, CA), Clontech (Mountain View, CA),
Creative Biogene (Shirley, NY), Life Technologies (Grand Island,
NY), Microbix Biosystems (Toronto, Canada), SignaGen
Laboratories (Rockville, MD), ViGene Biosciences (Rockville,
MD), Vector Biolabs (Malvern, PA), and Viral Vector Core Facility
(Iowa City, IA). In the first generation (simplest) rAd5 vectors, the
E1 gene is deleted. This gene encodes the E1a and E1b proteins,
which are required for the replication of a wild-type Ad5. This
gene deletion creates space in the vector genome to insert a foreign
gene or cDNA (the transgene).
Currently, there are three principal ways to generate E1 deleted
rAd5 vectors; using a two-plasmid co-transfection method with
eukaryotic (293) cells, a one-plasmid transfection method
(ViralPower Adenoviral Expression System from Life Technologies)
and an E. coli-based system (from Agilent Technologies). Herein,
we will describe the two-plasmid co-transfection method to gener-
ate a rAd5 vector, as this is the method we routinely use. In the
two-plasmid system, one plasmid is termed a shuttle vector (the
one used in our laboratory is pACCMV-pLpA) and can express in
both prokaryotic and eukaryotic cells. This plasmid has a human
cytomegalovirus (CMV) promoter, a multiple cloning site, and a
SV40 polyadenylation signal that can be placed in the E1 region
between map units 1.3 and 9.1 of the adenoviral genome. This
plasmid also contains 455 base pairs of Ad5 sequence (0.01.3
map units) upstream of the CMV promoter and 3039 base pairs of
Ad5 sequence (9.317.0 map units) downstream of the SV40
polyadenylation signal. The second plasmid is pJM17, and it pro-
vides most of the Ad5 sequence except for the E1 gene and a small
part of the E3 gene region. The pJM17 plasmid (Fig.1) contains
overlapping sequences with the Ad5 sequences found in pACCMV-
pLpA, which permit homologous exchange when they are
A 0 B bp M pJM17
Hind III (39080)
Hind III (3380)
Hind III (36143) 8114
Tet 7126
6108
AMPr 5090
Ori
4072
Hind III (7175)
pJM17 3054
Hind III (30697) 40.07 kb
Hind III (10612)
2036
Ad5
1636
Hind III (15934)

Hind III (18015)
Hind III (22687) 1018
Hind III (18090)
Fig. 1 (a) Map of pJM17 plasmid showing HindIII restriction sites. (b) Gel showing HindIII digestion results.
There are 10 HindIII restriction endonuclease sites in the pJM17 plasmid. After digestion, it yields 10 fragments
(8010, 5446, 5322, 4597, 4370, 3795, 3437, 2937, 2081, and 75bp). Eight of these fragments from the
HindIII-digested pJM17 are readily visualized as bands on 1% agarose gels after electrophoresis
co-transfected into 293 cells, a human embryonic kidney cell line

[9]. The 293 cells stably express the E1 proteins necessary to allow
replication of the constructed E1-deleted rAd5 vector [10].
1.2 Recombinant The production of rAAV2 can be generally accomplished with

AAV Serotype 2 Vector either of two methods: the use of stable packaging cell lines, or
Production transient transfection of production cell lines. Both methods
require three genetic elements: (1) the recombinant AAV2
genome, consisting of the wild-type AAV2 inverted terminal
repeats (ITRs) framing the expression cassette. The ITRs contain
the cis sequence information needed for replication and encapsida-
tion. (2) the genes coding for the replication and structural pro-
teins (rep and cap, respectively), expressed in trans. (3) the required
helper functions naturally encoded by a helper virus such as adeno-
virus, also expressed in trans.
To generate rAAV2 using stable packaging cell lines, the
expression cassette is stably transfected together with rep and cap
genes into a cell line (e.g., HeLa cells). These cells can be stored
and expanded for rAAV2 production, which then only requires a
wild-type Ad5 infection. This method is easily scalable and thus
very suitable for large-scale production in vector core facilities.
However, this process is quite laborious if used for only a few prep-
arations. Also, stable transfection of the rep and cap genes, encod-
ing toxic proteins, can be technically challenging. Contamination
with wild-type Ad5 can be a problem due to the infection of stably
transfected cell lines with Ad5 to start the production process, but
new purification methods to overcome this are being developed.
In contrast, the relatively undemanding transient transfection

method is ideally suited for mid-scale production, and is what we
use in our laboratory. This classical rAAV2 production method is
based on a trans-complementary co-transfection of plasmids
containing the required genetic elements into an appropriate tar-
get cell (we use 293T cells). It allows flexible and quick rAAV2
production using the straightforward protocol described below.
The expression cassette is present in the pAAV plasmid flanked by
ITRs [11]. For optimal yield and convenience, we use a packag-
ing/helper plasmid, pmmtvRC, which contains the rep and cap
sequences, and an Ad5 helper plasmid, pRS449B.With this
method, the typical yield, per 108 cells, after CsCl centrifugation
and subsequent dialysis is between 11012 and 61012 total par-
ticles, as determined by QPCR.
The significant interest in the use of AAV vectors clinically has
led to recent advances in scalable production and purification
methods. Transfection of cells with a set of plasmids, such as
described above, has been used for many years to produce a cell
lysate containing large amounts of AAV vector particles that are
free of contaminating helper virus. The AAV particles are then
purified from this crude lysate by a combination of methods such
density gradient centrifugation (with CsCl or iodixanol), affinity or
ion exchange chromatography, and gel filtration.
Although effective, these methods lack easy scalability, which is
required for clinical applications. The use of suspension cells [12],
roller bottles [13], or cell factories in combination with cell lines
[14], which stably maintain the rAAV vector and some of the AAV
helper functions required to produce rAAV, has addressed some of
the scalability issues. These cells can then be induced to produce
rAAV vector by infection with replication defective lytic viruses
such as derived from Ad5 or herpes simplex virus [15, 16]. Similarly,
the use of Baculovirus-based systems [17], in which the AAV vec-
tor and AAV helper functions are delivered to suspensions of insect
cells by infection, has increased both the scale and the yield of
particles per cell to the levels sufficient for large-scale manufactur-
ing as required for either preclinical studies in large animal models
or clinical studies for localized gene therapy applications, such as in
the salivary glands.
1.3 Cannulation Among the most common target organs for invivo gene therapy
ofSalivary Glands are the liver, lung, and muscle. The liver and lung have considerable
metabolic activity and bulk protein production, but are not easy to
access. Muscle, while easy to access for vector delivery, is not a tissue
known for protein production and secretion. Salivary glands pres-
ent an attractive target with some particular advantages for invivo
gene transfer. First, they are easy to access, as the ductal orifices of
the glands open into the mouth and typically can be visualized and
cannulated without significant complications. Indeed, contrast
imaging of cannulated glands (sialography) is a routine clinical

diagnostic procedure. Secondly, although salivary glands are densely
packed, almost all acinar and ductal cells have their apical mem-
branes directly in contact with the oral cavity along the lumen.
Third, the majority of salivary parenchymal cells (acinar) are capable
of producing high amounts of protein for both exocrine and endo-
crine secretion. Therefore, a transgene can be targeted to a large
number of protein-producing cells with little difficulty.
1.4 Assessing After delivery of the transgene of interest into salivary glands using
Functional Response a viral vector, it is essential to assess the function of the transgene.
(Saliva Collection) For the purpose outlined in this chapter, the key biological func-
tion to measure is salivary flow rate. We typically collect whole
saliva from the floor of the mouth in rodent experiments (Fig.2),
although saliva can be collected from individual glands as well.
Whole saliva is much easier and is better for animal survival, which
is particularly important if you are conducting long-term studies
with multiple sample collections.
1.5 Animal Models An animal model provides an invivo, nonhuman experimental tool
that mimics a disease or injury that is similar to a human condition.
The use of animal models allows the investigation of pathology and
pathogenesis in ways that are impossible in a human patient. Thus,
animal models provide scientists with a rich experimental resource,
but also require high ethical behavior and great care when used. All
animal studies should be conducted using a detailed, written
experimental protocol that is reviewed by an appropriate and inde-
pendent oversight body to ensure that minimal animal pain or dis-
comfort occurs, and that the study is scientifically valid.
Fig. 2 Figure of mouse during the whole saliva collection procedure. Note the
positioning of the capillary tube draining the saliva from the floor of the mouth
into a pre-weighed Eppendorf tube
1.5.1 Irradiation Irradiation in animal models has long been used to mimic salivary
ofSalivary Glands hypofunction in patients who receive irradiation to treat head and
neck cancer. Many different animals can be used and our group has
employed rhesus monkeys, miniature pigs, rats, and mice at various
times and for various purposes as preclinical models of salivary
gland irradiation damage. Mice are advantageous, as they are small,
easy to handle, and inexpensive. Thus, for the purpose of this chap-
ter, we will provide details on irradiating mice and the particular
procedures we use. Anyone who is contemplating studying irradi-
ated animals, of whatever species, is urged to consult radiation
biology experts at your institution.
For our studies, the mice are 710 weeks old (young adults),
and typically weigh 2030g. We irradiate salivary glands by placing
each animal into a specially built Lucite jig in such a way that the
animal can be immobilized without the use of anesthesia. The use
of anesthesia can change the sensitivity to irradiation, so if anesthe-
sia must be used, it must be used in all experimental and control
groups for proper comparison. Additionally, the jig is fitted with a
Lucite cone that surrounds the animals head and prevents head
movement during the radiation exposure (see Fig.3). For single
dose irradiation experiments in mice, we use a 15Gy dose [18].
Single dose studies are very convenient, but they do not fully
mimic clinical irradiation protocols. Clinically, patients receive frac-
tionated doses, commonly ~1.52.5Gy/day 5 days/week, for up
to 6 weeks. For our studies in mice we utilize an adapted fraction-
ated dose scheme, five daily fractions of 6Gy. This is more conve-
nient than the human regimen, yet one that leads to significant
salivary hypofunction [19]. Radiation is delivered to the animals
Fig. 3 Figure of three mice in the specially made Lucite jig prior to irradiation. The
animals are immobilized without the use of anesthesia. Note that the jig is fitted
with a Lucite cone that surrounds the animals head and prevents head move-
ment during the radiation exposure
head with a Therapax DXT300 X-ray irradiator (Precision X-ray

Inc., North Branford, CT) using 2.0mm Al filtration (300 KVp)
at a dose rate of 1.9Gy/min. The single dose and fractionated
dose schemes described achieve comparable salivary hypofunction,
resulting in ~5060% reduction when measured 8 weeks after irra-
diation was completed [18, 19]. Immediately after radiation, ani-
mals are removed from the Lucite jig and housed (4 animals/cage)
in a climate and light controlled environment, and allowed free
access to food and water.
2 Materials
2.1 Generating 1. 293 Cell line (Microbix Biosystems, Toronto, Canada).

arAd5 2. IMEM medium supplemented with 10% fetal bovine serum,
100U/mL penicillin G, and 100g/mL streptomycin for
293 cell line.
3. pJM17 plasmid (Microbix Biosystems).
4. Calcium Phosphate Transfection Kit.
5. Ultra-clear centrifuge tubes (Beckman Coulter).
6. Cesium Chloride (CsCl) gradients (1.25mg/mL, 1.33mg/
mL, and 1.4mg/mL) for rAd5 purification. CsCl is dissolved
in TD buffer: 140mM NaCl, 5mM KCl, 0.7mM Na2HPO4,
25mM TrisHCl, pH 7.4.
7. SW41 rotor (Beckman Coulter) or equivalent.
8.
PIERCE Slide-A-Lyzer dialysis cassette (ThermoFisher
Scientific, Rockford, IL).
9. Virus dialysis buffer: 100mM TrisHCl, pH 7.4, 10mM
MgCl2, and 10%v/v glycerol.
10. Virus dilution buffer: 5mM MgCl2, 20%v/v glycerol, 10mM
TrisHCl, pH 7.4.
11. SYBR Green PCR Master Mix (Applied Biosystems, Foster
City, CA).
12. ABI StepOne Plus Sequence Detector (Applied Biosystems).
2.2 Generating 1. 293T cells (American Type Culture Collection, Manassas, VA).
arAAV2 Vector 2. 150mm plates.
3. DMEM low glucose (4.51g/L d-glucose, Gibco) media:
100U/mL penicillin G, 100g/mL streptomycin, 2mM glu-
tamine, 10% fetal bovine serum.
4. pAAV2 plasmid containing expression cassette with the trans-
gene and ITRs (6g/plate).
5. pmmtv2RC plasmid ([20]; termed p2RepCap in the citation;
6g/plate).
6. pRS449B Ad5 helper plasmid ([21]; 12g/plate).
7. Solution A (stock for ten 150mm plates): 2 Hepes buffered

Saline pH 7.0302, 10mL.
8. Solution B (for ten 150mm plates): 9.0mL double distilled
water, followed by adding plasmid DNA at 60g pmmtv2RC,
120g p449B, 60g pAAV2-transgene, followed by adding
1000L of 2M CaCl2 and mixing by pipetting up and down.
9. TD buffer: As described in Subheading2.1, item 6.
10. Benzonase (also called endonuclease from Sigma, 25,000 units).
11. 50mL conical tubes.
12. 250mL conical tubes.
13. CsCl (as above for rAd5 production).
14. Beckman Ultra Clear Tube.
15. Ultra centrifuge and Beckman SW41 Ti rotor and buckets.
16. Refractometer.
17. Butterfly needle (21G Vacutainer; Becton Dickinson).
18. Several Eppendorf micro-centrifuge tubes.
19. Dialysis cassette (Pierce, Slide-A-Lyzer 10K).
20. Saline.
2.3 Delivery 1. Mice.

ofVector toMurine 2. Rack made by us for oral cavity display (Fig.4a).
SubmandibularGlands
3. Spring for spreading cheeks (made from paperclip).
4. Cotton to hold back tongue.
5. Microfine IV needle BD insulin syringe 0.3cc, 28-gauge
needle.
6. Preformed PE10 tubing (Intramedic PE10 polyethylene tub-
ing, Clay Adams Brand, Becton Dickinson), heated and pulled
to make cannulae (see Note 2).
7. Scalpel blade to bevel cannulae.
8. Binocular dissecting microscope.
9. Krazy glue (ethyl cyanoacrylate; Elmers Products, Inc;
Columbus, OH).
10. Curved forceps with very fine tip.
11. Ketamine (100mg/mL).
12. Xylazine (20mg/mL).
13. Atropine (Sigma) made by dissolving powder in water at
0.5mg/mL (can be stored in aliquots at 20C).
2.4 Mouse Whole 1. Heparinized Micro-Hematocrit capillary tubes.

Saliva Collections 2. Pilocarpine.
3. Ketamine (100mg/mL) and xylazine (20mg/mL).
Fig. 4 (a) Mouse with cannulas inserted into the orifices of Whartons duct (sub-
mandibular gland) prior to vector delivery (note needle and syringe attached to
each cannula are out of the plane of this picture). Place the upper jaw of the
mouse on a wire, i.e., teeth over the wire, and pull the lower jaw down with rub-
ber band onto rack. Finally, expand cheeks with wire spring made from a bent
paper clip. (b) Close-up view of right (1 arrow) and left (2 arrow) Whartons
ducts in a mouse
3 Methods
3.1 Generating The following protocol is what we use routinely to produce an

arAd5 E1-deleted rAd5 vector.
1. Insert transgene of interest into multiple cloning site of
pACCMV-pLpA using conventional molecular cloning meth-
ods to create the shuttle vector.
2. Make a high quality preparation of shuttle vector.
3. Make a high quality preparation of the plasmid pJM17 (see

Fig.1; Note 1).
4. Grow 293 cells in IMEM.
5. Co-transfect pJM17 and the shuttle vector into 293 cells by
calcium phosphate co-precipitation using the Calcium
Phosphate Transfection Kit. Add 15g of pJM17, 5g of
shuttle vector, 36L of 2M CaCl2 and use double-distilled
H2O to bring volume to 300L in a 14mL polypropylene
tube, and then mix well.
6. Quickly add 300L 2 Hepes buffered saline and bubble with
an automatic pipet for 1min.
7. Incubate at room temperature for 30min until a fine precipi-
tate is formed.
8. Add the entire mixture to a 100mm plate of 293 cells, and
gently rotate the plate to mix well.
9. Put the plate back into the tissue culture incubator, and replace the
growth medium every 2 days. To increase the chances of successful
vector formation, usually 45 plates of cells are co-transfected.
10. After ~1012 days, plaques (holes in the 293 monolayer) can
be observed, which indicate the replication of the recombinant
adenovirus. Continually, and carefully, replace the medium
until ~50% of 293 cells are lysed.
11. Harvest the entire plate (293 cells and growth medium) by
pipetting the growth medium present to dislodge the cell
monolayer, i.e., do not use trypsin, and collect into a 50mL
conical test tube.
12. Completely lyse the 293 cells by freezing (dry ice) and thawing
(37C water bath) five times, vortexing for 1min after each
time, to release rAd5 vector from the cells. Centrifuge at
~700g for 5min.
13. Transfer supernatant [crude viral lysate (CVL)] to a new 50mL
tube. Use 50L of the CVL to infect one well of 293 cells (105
cells/well in a 12-well plate). Measure the recombinant protein
production by an appropriate method (e.g., by Western blot,
ELISA, enzyme assay) in the lysate or medium of these cells
after 24h to confirm that the CVL contained functional virus
(See Note 3).
14. To make a large-scale preparation of the recombinant rAd5
vector, use the above CVL to transduce three fresh 100mm
plates of 293 cells.
15. Harvest the plates of cells at day 3. Freeze and thaw five times as
above, combine and again centrifuge to collect supernatant.
16. Use this CVL to transduce 20 150mm plates of 293 cells.
Harvest the cells at day 3, centrifuge the harvested cells and
aspirate all but ~6mL of the media. Use the remaining 6mL
to resuspend the cell pellet. Freeze and thaw the cell suspen-
sion five times, then centrifuge at ~1600g for 10min.
17. Make the first CsCl gradient by placing 2.5mL of density 1.25
CsCl in sterile ultra-clear centrifuge tubes, then slowly under-
lay with 2.5mL of density 1.40 CsCl. Transfer the CVL super-
natant to the top of the CsCl gradient.
18. Spin the tubes in a SW41 rotor at ~151,000g, 22C for 1h
(balance carefully with culture medium or phosphate-buffered
saline).
19. Following centrifugation, clean the outside of the tube with
70% alcohol, then use a 21-gauge needle and syringe to poke
through the side of the tube. Collect the lower opalescent
band, which contains the rAd5 vector.
20. Place 8mL of density 1.33 CsCl into sterile ultra-clear centri-
fuge tubes and transfer the vector band to this second CsCl
gradient.
21. Spin the tubes in a SW41 rotor at ~151,000g, 22C for 18h
(again balancing carefully). As above, clean the outside of the
tube with 70% alcohol, and again use a 21-gauge needle and
syringe to collect the opalescent band and add glycerol to 10%.
22. Transfer the vector into a Slide-A-Lyzer dialysis cassette and
dialyze two times in 500mL dialysis buffer for 30min, and
then in 1-L fresh dialysis buffer for 1h, three times.
23. Remove the vector suspension from the dialysis cassette and
aliquot in sterile Eppendorf tubes at ~100L/tube and store
at 80C.
24. Use real-time PCR (QPCR) to titer the rAd5 vector with prim-
ers from the E2 region of adenovius; E2q1 (5-GCAGAACCA
CCAGCACAGTGT-3) and E2q2 (5-TCCACGCATTTCC
TTCTAAGCTA-3). Titers are expressed as particles/mL.The
plasmid pACCMV-pLpA can be used as a standard for QPCR,
with 1g of the plasmid (~8570 base pairs) being equivalent
to 9.01010 molecules. Standard curves are established from
101 molecules to 108 molecules of shuttle vector, and rAd5
vectors are tested at three dilutions over a 100-fold range.
These QPCR assays are typically carried out with the SYBR
Green PCR Master Mix using a StepOne Plus Sequence
Detector, with the following conditions: stage 1, 95C for
2min; stage 2, 95C for 10min; stage 3, 95C for 15s, 60C
for 1min, repeated 40 times.
3.2 Generating 1. Culture ten 150mm plates of 293T cells in DMEM media at
arAAV2 Vector ~1.5106 cells/plate.
2. Incubate cells in DMEM with 10% FBS overnight.
3. For transfection of plasmids, allow 10mL 2 Hepes (Solution

A) to come to room temperature in 50mL tube.
4. Place a 2mL pipette in solution A and continuously bubble air
while adding solution B drop by drop from a pipet containing
10mL of solution B.
5. Incubate until a white precipitate is seen (at least 30min), mix
up and down with pipet.
6. Add 2.0mL of this mixture per plate, drop by drop while gen-
tly swirling the plate.
7. 24h after transfection, remove medium and add 20mL fresh
DMEM with 10% FBS.
8. 48h after transfection, harvest the cells by scraping them in
their own media and transferring to 250m>conical tube.
9. Centrifuge cells at ~500g for 15min.
10. Discard media and while washing the cell pellet with 50mL 1
PBS transfer to 50mL conical tube and centrifuge again.
11. Discard 1 PBS and resuspend cell pellet with 10mL 1 TD
buffer (this can be safely stored at 20C).
12. Freeze cell pellets on dry ice, then thaw pellets (37C), and
vortex cell lysate for 2min. Repeat twice.
13. Add Benzonase to a final concentration of 20U/mL of cell
lysate.
14. Sodium deoxycholate is then added to a final concentration of
0.5%, and incubate for 60min at 37C.
15. Centrifuge lysate at 340g for 15min.
16. Collect supernatant and add 0.55g of CsCl per mL of super-
natant and mix until dissolved, about 10min at room
temperature.
17. Adjust Refractive Index (RI) of supernatant to 1.372 with
refractometer. If the RI is too high, add small amounts of TD
buffer. If RI is too low, add small amounts of CsCl.
18. Add supernatant to one Ultra Clear Tube and fill tube to the
top (approximately 11mL) with TD buffer adjusted with CsCl
to an RI of 1.372.
19. Centrifuge tubes for 72h at ~182,300g using low accelera-
tion and low deceleration settings.
20. As above with rAd5 vector preparation, clean the outside of
the tube with 70% alcohol.
21. With a butterfly needle, puncture tube 1-cm above the bottom
of the tube. Be sure to twist needle while pushing so that the
plastic plug from the tube wall does not get stuck in the needle
preventing the fractions from going through.
22. Using a clamp to control the flow, collect ~700L fractions of

purified virus in several 1.5mL microcentrifuge tubes. Before
collecting the next fraction, measure the RI of the previous
fraction, keeping only those fractions with an RI between
1.371 and 1.372. These fractions typically contain the highest
titers of the rAAV2 vector. The fraction in which the vector is
found depends on the size of the transgene (i.e., larger trans-
gene would mean fraction with a higher RI).
23. Perform QPCR to measure vector titer from the different frac-
tions, on 1L aliquots from 1:1000L dilution in double
distilled water, as described in Subheading3.1, step 24. The
sequences for the forward primer, reverse primer are based on
the promoter of the expression cassette. The sequences used
are as follows for CMV promoter: Forward
CATCTACGTATTAGTCATCGCTATTACCAT; Reverse
TGGAAATCCCCGTGAGTCA.The reaction mixture is
incubated at 95C for 2min (stage 1), 95C for 8min (stage
2), then denaturation at 95C for 15s (step 3), and annealing
and extension at 60C for 1min, repeated 40 times (stage 4).
After completion, store selected fractions at 4C.
24. This stock virus can be stored in aliquots as desired and kept at
80C.Once an aliquot is thawed on ice, it can be stored at
4C for up to 2 weeks. Note that repeat thawing and refreez-
ing will result in loss of vector activity.
25. Before using invivo dialyze fractions against 0.9% Sodium
Chloride, putting fraction in dialysis cassette using a 21-gauge
needle with a 1.5mL syringe (see Pierce manual) and dialyzing
by floating the cassette in 400mL of saline for 30min, twice,
while stirring at room temperature.
26. Change saline and repeat twice (total of 1.2L of saline).
3.3 Methods 1. Weigh animal.

forDelivery ofVector 2. Inject 1L/g ketamine: xylazine (3:2) intramuscularly into
toSubmandibular animals hind legthe animal is usually anesthetized within
Glands 5min.
3. Place animal on a specially constructed rack (Fig.4a) with the
upper jaw on wireteeth over wire, the lower jaw pulled down
with rubber band on the rack, the cheeks expanded with a wire
spring, and a ball of cotton (~0.8cm) positioning the tongue
back toward the throat.
4. Adjust binocular dissecting microscope to focus on both sub-
mandibular gland duct orifices, which are located slightly lat-
eral to the midline of the floor of the mouth, about 45mm
posterior from the lower incisors. You should see two whitish
papillae. The duct orifice is located on the ventral aspect, about
half way down the papillae (Fig.4b).
5. Pick up preformed cannula made from PE10 tubing for inser-

tion. Using the scalpel blade, cut a bevel (~45) on the narrow
end in the middle of the thinnest part of the tubing. Note that
younger animals tend to require thinner tubing, while older
animals require thicker tubing.
6. Using very delicate forceps, pick up the beveled (narrow) end
of the tubing (cannula), approximately 2mm from the end of
the tubing, and push it gently into the duct orifice. The angle
of the cannula should be 45 to the floor of the mouth. Push
the cannula approximately 34mm into the duct. Be sure that
the cannula is well sealed with the duct orifice by visualizing
the fitness of the orifice rim to the cannula.
7. After successful cannula insertion, place a drop of Krazy glue at
the site of insertion to secure the cannula.
8. Inject 0.5mg/kg atropine intramuscularly, wait 10min. While
waiting, fill a syringe with sample or control solution for deliv-
ery. For mice the optimal volume to infuse is 50L.Dilute
vector in saline to the desired concentration.
9. Place distal end of cannula around the needle of a 0.3cc syringe
and inject sample slowly through the cannula into the gland.
10. Wait 10min before gently removing the syringe to prevent
backflow.
11. Gently remove cannulae and Krazy glue by pulling on tubing.
3.4 Methods 1. Pre-weigh 1.5mL Eppendorf tube (one/mouse) and record.

forMurine Saliva 2. Anesthetize mouse with ketamine (60mg/kg)
+xylazine
Collections (8mg/kg).
3. Inject pilocarpine (0.25g/g; this dose may need to be
adjusted depending on the mouse strain used) subcutaneously
in back of neck. Pilocarpine will stimulate the salivary glands to
secrete saliva. Normally, the saliva will start to flow about
35min after pilocarpine is injected.
4. Place the mouse on a box about 8cm high, with the head of
the mouse over the edge of the box (see Fig.2).
5. Put one end of a Micro-Hematocrit capillary tube under the
mouses tongue and the other end into the pre-weighed
Eppendorf tube. Check capillary tube position and saliva flow
constantly.
6. Monitor mouse carefully for signs of difficulty breathing, and
supply with oxygen if necessary.
7. Stop saliva collection after 20min, making sure that all of the
saliva drains from mouses mouth and capillary tube into the
Eppendorf tube.
8. Weigh the tube, then subtract the pre-weighed value of the

tube. The amount of saliva will be based on a specific gravity of
1 (i.e., 1g/mL) and typically will be expressed as L/20min.
You can also convert the saliva output to L/g-body weight or
milligram-gland weight.
9. The saliva can be directly used as output (end point measure)
or be used to assess transgenic protein production or activity, if
appropriate. Additionally, salivary composition can be deter-
mined, e.g., calcium, sodium, amylase, total protein, etc. Saliva
should be stored at 80C until assayed.
4 Notes
1. Difficulty with the preparation of the pJM17 plasmid (see gel

picture of high quality pJM17 preparation in Fig.1). The
pJM17 plasmid is a large plasmid, 40.07kb. When preparing
this plasmid it is necessary to be careful about two things. First,
remember when making this plasmid it results in a low yield. It
is best to use 100g/mL of ampicillin for cultures of bacteria
containing pJM17, and to culture the bacteria for less than
16h. Second, during the plasmid extraction procedure,
because of its size, this plasmid breaks easily. One needs to be
very careful at each step to prevent artificial shearing of the
plasmid. A high quality preparation of pJM17 will show eight
bands following 1% agarose gel electrophoresis after HindIII
digestion (Fig.1).
2. Making a cannula.
(a) Cut a segment of tubing about 5cm long.
(b) Hold both ends of tubing between index finger and thumb
(keep tubing a little bit loose).
(c) Hold tubing about 15cm above a delicate flame for a very
short timewhen the tubing begins to SOFTENremove
from the flame and pull gently at both ends (not too much!).
(d) Wait approximately 20s for tubing to cool and solidify,
store at room temperature.
3. Test for viral activity of produced rAd5 and rAAV2.
Seed a 12-well dish of 293 or 293T cells. At 90% confluency,
transduce individual wells with equal amounts of viral particles
(as determined by QPCR) from fractions being tested from the
CsCl gradients. For rAd5 vectors use 10100 viral particles/
cell and for rAAV2 vectors use 1000100,000 viral particles/
cell. Perform an ELISA or Western immunoblot on media or
cell extracts later, as appropriate, from each transduced well to
determine the viral fractions that best express transgenic protein
and use those fractions for invivo experiments.
Acknowledgment
The authors research is supported by the intramural research pro-

gram of the National Institute of Dental and Craniofacial Research.
References
1. Siegel RL, Miller KD, Jemal A (2015) Cancer virus type 2 vectors via suspension transfection.
statistics 2015 CA.Cancer JClin 65:529 Biotechnol Bioeng 94:416430
2. Mitsias DI, Kapsogeorgou EK, Moutsopoulos 13. Qu G, McClelland A, Wright JF (2000)
HM (2006) Sjgrens syndrome: why autoim- Scaling-up production of recombinant AAV
mune epithelitis. Oral Dis 12:523532 vectors for clinical applications. Curr Opin
3. Kok MR, Yamano S, Lodde BM, Wang J, Drug Discov Devel 3:750755
Couwenhoven RI, Yakar S, Voutetakis A, 14. Clark KR, Voulgaropoulou F, Johnson PR
Leroith D, Schmidt M, Afione S, Pillemer (1996) A stable cell line carrying adenovirus-
SR, Tsutsui MT, Tak PP, Chiorini JA, Baum inducible rep and cap genes allows for infectiv-
BJ (2003) Local adeno-associated virus- ity titration of adeno-associated virus vectors.
mediated interleukin 10 gene transfer has Gene Ther 3:11241132
disease-modifying effects in a murine model 15. Conway JE, Rhys CM, Zolotukhin I, Zolotukhin
of Sjgrens syndrome. Hum Gene Ther S, Muzyczka N, Hayward GS, Byrne BJ (1999)
14:16051618 High-titer recombinant adeno-associated virus
4. Baum BJ (1993) Principles of saliva secretion. production utilizing a recombinant herpes sim-
Ann NY Acad Sci 694:1723 plex virus type I vector expressing AAV-2 Rep
5. Vitolo JM, Baum BJ (2002) The use of gene and Cap. Gene Ther 6:986993
transfer for the protection and repair of salivary 16. Gao GP, Lu F, Sanmiguel JC, Tran PT, Abbas
glands. Oral Dis 8:183191 Z, Lynd KS, Marsh J, Spinner NB, Wilson JM
6. Preston GM, Agre P (1991) Isolation of the (2002) Rep/Cap gene amplification and high-
cDNA for erythrocyte integral membrane pro- yield production of AAV in an A549 cell line
tein of 28 kilodaltons: member of an ancient expressing Rep/Cap. Mol Ther 5:644649
channel family. Proc Natl Acad Sci U S A 17. Urabe M, Ding C, Kotin RM (2002) Insect
88:1111011114 cells as a factory to produce adeno-associated
7. Baum BJ, Wellner RB, Zheng C (2002) Gene virus type 2 vectors. Hum Sene Ther
transfer to salivary glands. Int Rev Cytol 13:19351943
213:93146 18. Vitolo JM, Cotrim AP, Sowers AL, Russo A,
8. Mastrangeli A, O'Connell B, Aladib W, Wellner RB, Pillemer SR, Mitchell JB, Baum
Fox PC, Baum BJ, Crystal RG (1994) BJ (2004) The stable nitroxide Tempol facili-
Direct invivo adenovirus-mediated gene tates salivary gland protection during head and
transfer to salivary glands. Am JPhysiol neck irradiation in a mouse model. Clin Cancer
266:G1146G1155 Res 10:18071812
9. McGrory WJ, Bautista DS, Graham FL (1988) 19. Cotrim AP, Hyodo F, Matsumoto K, Sowers
A simple technique for the rescue of early AL, Cook JA, Baum BJ, Krishna MC, Mitchell
region I mutations into infectious human ade- JB (2007) Differential radiation protection of
novirus type 5. Virology 163:614617 salivary glands versus tumor by Tempol with
10. Graham FL, Smiley J, Russell WC, Nairn R accompanying tissue assessment of Tempol by
(1977) Characteristics of a human cell line magnetic resonance imaging. Clin Cancer Res
transformed by DNA from human adenovirus 13:49284933
type 5. JGen Virol 36:5974 20. Walters RW, Yi SM, Keshavjee S, Brown KE,
11. Chiorini JA, Wendtner CM, Urcelay E, Welsh MJ, Chiorini JA, Zabner J(2001)
Safer B, Hallek M, Kotin RM (1995) High- Binding of adeno-associated virus type 5 to
efficiency transfer of the T cell co-stimulatory 2,3-linked sialic acid is required for gene trans-
molecule B7-2 to lymphoid cells using high- fer. JBiol Chem 273:2061020616
titer recombinant adeno-associated virus vec- 21. Smith RH, Afione S, Kotin RM (2002)
tors. Hum Gene Ther 6:15311541 Transposase-mediated construction of an inte-
12. Park JY, Lim BP, Lee K, Kim YG, Jo EC grated adeno associated virus type 5 helper
(2006) Scalable production of adeno- associated plasmid. Biotechniques 208:210211
Part II
Molecular Biosciences
Chapter 7
The Oral Microbiota in Health and Disease:

An Overview of Molecular Findings
Jos F. Siqueira Jr. and Isabela N. Ras
Abstract
Culture-independent nucleic acid technologies have been extensively applied to the analysis of oral bacterial
communities associated with healthy and diseased conditions. These methods have confirmed and substan-
tially expanded the findings from culture studies to reveal the oral microbial inhabitants and candidate
pathogens associated with the major oral diseases. Over 1000 bacterial distinct species-level taxa have been
identified in the oral cavity and studies using next-generation DNA sequencing approaches indicate that
the breadth of bacterial diversity may be even much larger. Nucleic acid technologies have also been helpful
in profiling bacterial communities and identifying disease-related patterns. This chapter provides an
overview of the diversity and taxonomy of oral bacteria associated with health and disease.
Key words Oral microbiology, Molecular biology methods, Taxonomy, Oral diseases
1 Introduction
The term microbiota consists of the symbiotic microbial cells

colonizing the host surfaces and cavities, while the term microbi-
ome is commonly used to define the genes these cells harbor [1,
2]. The microbiota colonizing the oral cavity is composed of
diverse groups of microbial species, each one possessing its specific
nutritional and physico-chemical requirements. Although fungi,
viruses, protozoa, and archaea have been found, bacteria are the
most predominant microorganisms present in the oral cavity.
They colonize different oral sites, including the teeth and mucosal
surfaces (lips, cheek, palate, and tongue); each one of these distinct
oral habitats possesses their idiosyncrasies that will influence the
composition of the microbiota. Saliva bathes most of these areas
and may contain approximately 100 million bacterial cells per ml
[3]. For the most part, the oral microbiota lives in harmony with
the host, usually in a commensal state; however, under certain
circumstances, this relationship can break down, convert to parasit-
ism and disease may ensue.
DOI 10.1007/978-1-4939-6685-1_7, Springer Science+Business Media LLC 2017
127
128 Jos F. Siqueira Jr. and Isabela N. Ras
Culture has been traditionally used to study the diversity of the

oral microbiota. For oral bacteria to be successfully grown in the
laboratory, culturing conditions have to be adjusted to suit their
varied requirements [46]. A high diversity of bacterial species has
been cultivated from the oral cavity, but early microscopy studies
had already suggested that roughly one-half of the oral microbiota
cannot be cultivated in vitro [7]. Introduction of culture-
independent nucleic acid methods to the analysis of oral bacterial
diversity has not only confirmed this picture suggested by micro-
scopic studies, but also demonstrated a still broader and more
diverse spectrum of extant oral bacteria.
This chapter provides an overview of bacterial diversity and
taxonomy in the oral cavity under healthy and diseased conditions,
with the main focus on the findings from nucleic acid technologies.
2 Nucleic Acid Technologies
Nucleic acid (or molecular biology) techniques have revolution-

ized the field of medical microbiology given their numerous appli-
cations and advantages over other commonly used methods
(Table 1). As with any other technology, molecular methods also
have their own limitations (Table 1). A large selection of molecular
methods for the study of microorganisms is currently available and
the choice of a particular approach depends on the questions being
addressed. Molecular methods for microbial identification can be
directly used in clinical samples to detect the unexpected (open-
ended analysis) or to target specific taxa (closed-ended analysis).
Broad-range polymerase chain reaction (PCR) amplification of the
16S rRNA gene followed by cloning and Sanger-based sequencing
(clone library analysis) can be used to disclose the microbial diver-
sity in a given environment. Microbial community structures can
be analyzed and components be identified via community profiling
techniques, such as denaturing gradient gel electrophoresis
(DGGE) and terminal restriction fragment length polymorphism
(T-RFLP). Among other applications, DNA-DNA hybridization
arrays (checkerboard and microarrays), specific single-primer PCR,
nested PCR, multiplex PCR, and quantitative real-time PCR can
be used to survey large numbers of clinical samples for the presence
of target species. Fluorescence in situ hybridization (FISH) can
identify, measure abundance of target species, and provide infor-
mation on their spatial distribution in tissues. Next-generation
DNA sequencing (NGS) technologies have recently emerged as
valuable tools for microbial identification and community profil-
ing. These technologies permit massive DNA sequencing with a
much higher throughput than the conventional Sanger sequencing
approach [8]. The most currently used NGS technologies include
the 454 pyrosequencing (Roche Applied Science), HiSeq or MiSeq
(Illumina), and SOLiD (Applied Biosystems).
The Oral Microbiota: An Overview 129
Table 1
Advantages and limitations of nucleic acid technologies
Advantages Limitations
1. Detect both cultivable and as-yet- 1. Most assays are qualitative or semi-quantitative
uncultivated species or strains (exception: real-time PCR)
2. High specificity and accurate 2. Most assays only detect one species or a few
identification of strains with ambiguous different species at a time (exceptions: broad-range
phenotypic behavior PCR, DGGE, T-RFLP, checkerboard, DNA
3. Detect species directly in clinical samples microarrays, next-generation DNA sequencing)
4. High sensitivity 3. Most assays detect only the target species and fail
5. Rapididentification can be achieved in to detect unexpected species (exceptions: broad-
no more than minutes to a few hours range PCR, DGGE, T-RFLP, next-generation
6. Do not require carefully controlled DNA sequencing)
anaerobic conditions during sampling, 4. Some assays can be laborious and costly (e.g.,
transportation and handling broad-range PCR, next-generation DNA
7. Can be used during antimicrobial sequencing)
treatment 5. Biases in broad-range PCR introduced by
8. Samples can be stored frozen for later homogenization procedures, preferential DNA
analysis amplification, and differential DNA extraction
9. DNA can be transported easily between 6. Detect dead microorganismsa
laboratories
10. Detect dead microorganismsa
a
Detection of dead cells can be an advantage and a limitation. On the plus side, this ability allows detection of hitherto
uncultivated or fastidious bacteria that can die during sampling, transportation or isolation procedures. On the down
side, detection of dead bacteria may give rise to misinterpretations as to their role in the habitat
3 Diversity and Taxonomy of Oral Bacteria
Data from culture and molecular studies have collectively revealed

that almost 1000 distinct bacterial species-level taxa may be able to
live in the human oral cavity [9, 10]. Not all of them are present in
the same individual at any one time, and a particular individual can
harbor about 100200 taxa in his/her mouth. Whereas some spe-
cies are common to different oral sites, the majority of species are
selective for a particular site [11]. Over a third of the oral bacterial
taxa remain to be cultivated and fully characterized. This raises the
interesting possibility that as-yet-uncultivated and uncharacterized
species that have passed unnoticed by culturing studies may
actually play important ecological, beneficial or pathogenic roles in
the oral cavity.
Molecular studies have shown that the most prevalent and abun-
dant oral bacterial species-level taxa fall into 8 phyla: Firmicutes,
Fusobacteria, Bacteroidetes, Actinobacteria, Proteobacteria,
Spirochaetes, Synergistetes, and TM7 [10, 1217] (Table 2). In fact,
NGS studies have reported the occurrence of representatives of
more than 20 phyla in different oral sites [18], most of them occur-
ring as low-abundance members of bacterial communities [18, 19].
Table 2
Bacterial phyla and respective genera commonly found in the oral cavity
Phyla and genera Species-level representatives

Firmicutes
Anaerococcus A. prevotii
Catonella C. morbi
Centipeda C. periodontii
Dialister D. invisus, D. pneumosintes, uncultivated phylotypes
Eggerthella Eg. lenta
Enterococcus E. faecalis
Eubacterium E. sulci, E. infirmum, E. saphenum, E. nodatum, E. brachy, E. minutum,
uncultivated phylotypes
Lachnoanaerobaculum Lc. saburreum
Filifactor Fl. alocis
Finegoldia Fn. magna
Gemella Ge. morbillorum
Granulicatella Gr. adiacens
Lactobacillus L. salivarius, L. acidophilus, L. fermentum, L. paracasei, L. catenaformis
Megasphaera Uncultivated phylotypes
Mogibacterium Mo. timidum, Mo. pumilum,
Mo. neglectum, Mo. vescum
Parvimonas Pa. micra
Peptoniphilus Pn. asaccharolyticus, Pn. lacrimalis
Peptostreptococcus Pep. anaerobius, uncultivated phylotypes
Pseudoramibacter Ps. alactolyticus
Selenomonas Se. sputigena, Se. noxia, uncultivated phylotypes
Solobacterium So. moorei, uncultivated phylotypes
Streptococcus S. mutans, S. sobrinus, S. mitis, S. sanguinis, S. gordonii, S. oralis, S.
anginosus, S. constellatus,
S. intermedius, uncultivated phylotypes
Veillonella V. parvula, uncultivated phylotypes
Bacteroidetes
Capnocytophaga Ca. gingivalis, Ca. ochracea
Porphyromonas P. endodontalis, P. gingivalis
Prevotella Pr. intermedia, Pr. nigrescens, Pr. multissacharivorax, Pr. baroniae, Pr.
denticola, uncultivated phylotypes
Alloprevotella A. tannerae, A. rava
Tannerella T. forsythia
Actinobacteria
Actinomyces A. israelii, A. gerencseriae, A. naeslundii, A. meyeri, A. odontolyticus,
uncultivated phylotypes
Atopobium At. parvulum, At. minutum, At. rimae, uncultivated phylotypes
Bifidobacterium B. dentium, B. adolescentis, B. bifidum
Corynebacterium Co. matruchotii
Olsenella O. uli, O. profusa, uncultivated phylotypes
Propionibacterium Pp. acnes, Pp. propionicum
Rothia R. dentocariosa
Slackia Sl. exigua
(continued)
Table 2
(continued)
Phyla and genera Species-level representatives

Proteobacteria
Aggregatibacter Ag. actinomycetemcomitans, Ag. aphrophilus
Campylobacter Cm. rectus, Cm. gracilis, Cm. curvus
Cm. showae, Cm. concisus
Eikenella Ei. corrodens
Neisseria N. mucosa, N. sicca
Fusobacteria
Fusobacterium F. nucleatum, F. periodonticum, uncultivated phylotypes
Leptotrichia Lp. buccalis
Spirochaetes
Treponema Tr. denticola, Tr. socranskii
Tr. parvum, Tr. maltophilum
Tr. lecithinolyticum, uncultivated phylotypes
Synergistetes
Fretibacterium Fr. fastidiosum
Pyramidobacter Py. piscolens
The dominant members of the microbiota have been shown to

change from deciduous to permanent teeth dentition; Proteobacteria
predominate in the former condition, while Bacteroidetes,
Veillonellaceae family, Spirochaetes and TM7 increase with the per-
manent dentition [20]. However, with adulthood, it seems that
the oral microbiota becomes more stable [21]. This stability along
with the high inter-individual variation in the oral bacterial com-
munities open perspectives for a potential application to forensic
identification of individuals.
The composition of the oral microbiota in health and disease
can be affected by some general conditions, such as diabetes [22].
In addition, Human Immunodeficiency Virus (HIV)-positive
adults with chronic periodontitis have a higher bacterial diversity in
the subgingival biofilm as compared to HIV-negative controls and
the bacterial community profiles significantly differ between these
two conditions [23].
The oral fungal diversity has also been examined by using a
NGS approach: 74 cultivatable and 11 as-yet-uncultivated fungal
genera were detected [24]. Fifteen genera occurred in 20 % or
more of the samples; Candida species were the most prevalent,
detected in 75 % of the individuals, followed by Cladosporium
(65 %), Aureobasidium (50 %), and Saccharomycetales (50 %).
4 Refined Bacterial Taxonomy Associated with Oral Diseases
4.1 Dental Caries Dental caries (tooth decay) is one of the most common biofilm-
related infectious diseases that affects humans. Culturable species
of Streptococcus, Lactobacillus, and Actinomyces are closely associ-
ated with the etiopathogenesis of different forms and stages of car-
ies [25, 26]. However, nucleic acid approaches have demonstrated
that the diversity of the microbiota associated with caries is much
greater than anticipated by culture studies [2729]. Overall, over
40 % of the microbiota occurring in caries lesions is made up of as-
yet uncultivated species [3034]. As-yet-uncultivated phylotypes
or uncharacterized strains of Bifidobacterium, Propionibacterium,
and Atopobium have been added to the list of candidate pathogens
associated with this disease [30, 31, 3436].
Several recent studies have used NGS approaches to evaluate
the caries microbiota and substantial new information has been
generated. A pyrosequencing study of shifting bacterial profiles in
different stages of caries progression revealed representatives of 18
phyla and 145 genera [29]. The abundance of several genera,
including Lactobacillus, Megasphaera, Olsenella, Scardovia,
Shuttleworthia, and Streptococcus, was significantly increased in
dentinal caries; Actinomyces and Corynebacterium dominated in
white spot lesions; and Flavobacterium, Neisseria, Bergeyella, and
Derxia were more abundant in the intact surfaces of caries indi-
viduals. Dentinal caries exhibited reduced bacterial diversity in
comparison to the other sites evaluated. The latter finding was
confirmed by another study that evaluated the composition of the
microbiota of dentinal caries layers with pH values ranging from
4.5 to 7.8 and found that acidic conditions were associated with
lower bacterial diversity and dominance of Lactobacillus species
[37]. In general, studies of the microbiota of advanced dentinal
caries reveal a predominance of lactobacilli and/or species/phylo-
types of the genera Prevotella, Atopobium, Selenomonas, Dialister,
Fusobacterium, Olsenella, Pseudoramibacter, Bifidobacterium,
Veillonella, Streptococcus, Granulicatella, and members of the
Lachnospiraceae family [28, 3641].
If untreated, caries can advance to expose the dental pulp and
cause irreversible inflammation. Several bacterial taxa have been
found in the advanced front of dentinal caries associated with pulp
exposure and irreversible pulpitis; the most frequently include
Atopobium genomospecies C1, Pseudoramibacter alactolyticus,
Streptococcus species, Parvimonas micra, Fusobacterium nuclea-
tum, and Veillonella species [42]. Some taxa, such as Streptococcus
species, Parvimonas micra, and Dialister invisus, were significantly
associated with symptoms.
Other forms of caries have also been investigated by NGS tech-
niques. Pyrosequencing analysis of root caries identified
Propionibacterium acidifaciens, Streptococcus mutans, Olsenella

profusa, Prevotella multisaccharivorax, and Lactobacillus crispatus
in association with these lesions, whereas Delftia acidovorans,
Bacteroidetes [G-2] sp., Lachnospiraceae [G-3] sp., and Prevotella
intermedia were associated with health [43]. The genera
Streptococcus, Granulicatella, and Actinomyces were significantly
increased in children with severe early childhood caries [44].
4.2 Halitosis Colonization of the tongue dorsum by bacteria producing volatile

sulfur compounds and other metabolites has been implicated as a
major source of oral malodor in subjects with halitosis [45]. A
molecular study revealed that about 60 % of the bacteria detected
on the tongue dorsum remain uncultivated and species-level taxa
most associated with halitosis include Atopobium parvulum,
Eubacterium sulci, Solobacterium moorei, and some as-yet-
uncultivated phylotypes from the genera Dialister and Streptococcus,
and the division TM7 [12].
4.3 Periodontal
Disease Periodontal disease results from the subgingival presence of complex
bacterial biofilms. It has been recently postulated that periodontitis
is a disease caused by dysbiotic bacterial communities [46, 47].
Important advances in understanding the infectious agents of peri-
odontal diseases have occurred after introduction of nucleic acid
identification approaches. The bacterial taxa that have been strongly
associated with periodontal diseases by culture and molecular studies
include Porphyromonas gingivalis, Tannerella forsythia, Treponema
denticola, Aggregatibacter actinomycetemcomitans, Fusobacterium
nucleatum, Filifactor alocis, Prevotella intermedia, and many as-yet-
uncultivated phylotypes [14, 4854].
NGS studies of the periodontal microbiota have expanded the
coverage of bacterial diversity and provided substantial informa-
tion about community profile differences. Analyses of subgingival
bacterial biofilm communities in individuals with different peri-
odontal conditions (health, gingivitis, and periodontitis) revealed
representatives of 26 phyla, 433 genera, and over 1000 species-
level taxa [55]. Bacteroidetes, Fusobacteria, Synergistetes, and
Spirochaetes are the most abundant phyla in periodontitis subjects,
whereas Firmicutes and Proteobacteria dominate the microbiota in
gingivitis and healthy subjects, respectively [55, 56]. Proportions
of Actinobacteria have also been shown to be higher in health
[57]. High levels of Porphyromonas, Fusobacterium, Fretibacterium,
Filifactor, and Treponema genera occur in periodontitis subjects,
while Streptococcus, Capnocytophaga, Leptotrichia, and Haemophilus
genera prevail in gingivitis subjects [55]. Approximately 20 % of
the sequences found in subgingival biofilms are related to as-yet-
uncultivated and uncharacterized phylotypes [56].
Different community profiles have been identified in health
and disease, with diversity being higher in the latter [55 57 ].
A study showed higher bacterial diversity in periodontitis patients

and less diverse communities in smoking-associated periodontitis
[58]. Some genera, such as Fusobacterium, Fretibacterium,
Streptococcus, Veillonella, and Filifactor, are more abundant in
smokers, while Prevotella, Campylobacter, Aggregatibacter, and
Haemophilus are less abundant in these individuals [59].
Microorganisms other than bacteria, namely archaea [60] and
herpesviruses [61, 62], have also been found in association with
periodontal diseases. Their role in the disease pathogenesis remains
to be elucidated.
4.4 Apical Apical periodontitis is an inflammatory disease that affects both the
Periodontitis periodontal ligament and alveolar bone, usually around the apex of
the root, caused by bacterial biofilms formed in the necrotic dental
root canal. Root canal infection is generally a sequel to caries, but
can also occur after trauma, iatrogenic procedures or in teeth with
advanced periodontal disease. No specific etiologic agents have
been unequivocally identified for endodontic infections, but a
group of 1020 species have been more frequently found in most
studies [63, 64]. The species commonly occurring in the canals of
teeth with primary apical periodontitis belong to the genera
Fusobacterium, Dialister, Filifactor, Streptococcus, Porphyromonas,
Prevotella, Tannerella, Treponema, and Parvimonas [64].
Enterococcus faecalis and streptococci are the most common taxa
detected in posttreatment apical periodontitis [65, 66].
NGS studies of endodontic infections have revealed that even in
the relatively isolated root canal environment, significant bacterial
diversity exists and representatives of over 20 phyla have been identi-
fied [6771]. The most represented, abundant, and prevalent phyla
are Firmicutes, Bacteroidetes, Proteobacteria, Fusobacteria, and
Actinobacteria. A great inter-individual variation in the composition
of the endodontic microbiota has been quite evident [70, 72].
Separate analyses of the microbiota in the apical part of the root
canal have also shown large bacterial diversity [68, 69]. Bacteria in
this area are expected to be in direct contact with the host defenses
and be directly involved in disease pathogenesis.
The acute apical abscess is a severe and symptomatic form of
the disease and is caused by advance of the infection to the periapi-
cal tissues. A significant difference in bacterial community profiles
has been observed when comparing these abscesses with asymp-
tomatic infections [70]. Acute infections are significantly more
diverse than chronic infections, and exhibit higher abundance and
prevalence of members of Fusobacteria and Parvimonas [70].
Persistent infections are the main cause of posttreatment
apical periodontitis. The most abundant phyla in association with
these cases are Firmicutes, Proteobacteria, Actinobacteria, and
Bacteroidetes [73]. In comparison with primary infections, persis-
tent infections have been shown to be significantly enriched with
members of the phyla Proteobacteria and Tenericutes and the genera

Lactobacillus, Streptococcus, and Sphingomonas [74]. When analyz-
ing different types of endodontic infections, a study detected
Enterococcus faecalis only in patients with persistent infections
[72]. No significant differences in bacterial diversity have been
observed for primary and persistent infections [71], but studies
showed a higher phylogenetic diversity for persistent infections
associated with substandard endodontic treatments [72, 74].
5 Concluding Remarks
Traditionally, the oral microbiota in health and disease has been

studied by means of culture approaches. Such studies have resulted
in the establishment of a set of species thought to play an impor-
tant role in the pathogenesis of several oral diseases. Over the last
two decades, not only have findings from culture-based methods
been confirmed but they have also been significantly expanded by
those from culture-independent nucleic acid techniques. Molecular
methods have confirmed and strengthened the association of
several bacterial species with oral diseases and have also revealed
new suspected pathogens. The list of oral inhabitants, including
candidate pathogens, has increased to include culture-difficult spe-
cies and as-yet-uncultivated bacteria. As a consequence of the
resolution and high throughput of many nucleic acid approaches,
knowledge of the oral microbiota has been comprehensively refined
and some previously established concepts of disease etiology may
need to be readdressed.
References
1. Ursell LK, Metcalf JL, Parfrey LW, Knight R 6. Sizova MV, Hohmann T, Hazen A, Paster BJ,
(2012) Defining the human microbiome. Nutr Halem SR, Murphy CM, Panikov NS, Epstein
Rev 70(Suppl 1):S38S44 SS (2012) New approaches for isolation of pre-
2. Turnbaugh PJ, Ley RE, Hamady M, Fraser- viously uncultivated oral bacteria. Appl Environ
Liggett CM, Knight R, Gordon JI (2007) The Microbiol 78:194203
human microbiome project. Nature 7. Socransky SS, Gibbons RJ, Dale AC, Bortnick
449:804810 L, Rosenthal E, MacDonald JB (1963) The
3. Wade WG (2013) Characterisation of the human microbiota of the gingival crevice in man. 1.
oral microbiome. J Oral Biosci 55:143148 Total microscopic and viable counts and counts
4. Leys EJ, Griffen AL, Kumar PS, Maiden MF of specific organisms. Arch Oral Biol 8:
(2006) Isolation, classification, and identifica- 275280
tion of oral microorganisms. In: Lamont RJ, 8. Voelkerding KV, Dames SA, Durtschi JD
Burne RA, Lantz MS, Leblanc DJ (eds) Oral (2009) Next-generation sequencing: from
microbiology and immunology. ASM Press, basic research to diagnostics. Clin Chem
Washington, DC, pp 7388 55:641658
5. Siqueira JF Jr, Ras IN (2013) As-yet- 9. Paster BJ, Olsen I, Aas JA, Dewhirst FE (2006)
uncultivated oral bacteria: breadth and association The breadth of bacterial diversity in the human
with oral and extra-oral diseases. J Oral Microbiol periodontal pocket and other oral sites.
5:doi: 10.3402/jom.v3405i3400.21077 Periodontol 2000 42:8087
10. Dewhirst FE, Chen T, Izard J, Paster BJ, 22. Zhou M, Rong R, Munro D, Zhu C, Gao X,
Tanner AC, Yu WH, Lakshmanan A, Wade Zhang Q, Dong Q (2013) Investigation of the
WG (2010) The human oral microbiome. effect of type 2 diabetes mellitus on subgingival
J Bacteriol 192:50025017 plaque microbiota by high-throughput 16S
11. Aas JA, Paster BJ, Stokes LN, Olsen I, Dewhirst rDNA pyrosequencing. PLoS One 8, e61516
FE (2005) Defining the normal bacterial flora of 23. Ferreira DC, Gonalves LS, Siqueira JF Jr,
the oral cavity. J Clin Microbiol 43:57215732 Carmo FL, Santos HF, Feres M, Figueiredo
12. Kazor CE, Mitchell PM, Lee AM, Stokes LN, LC, Soares GM, Rosado AS, dos Santos KR,
Loesche WJ, Dewhirst FE, Paster BJ (2003) Colombo AP (2015) Subgingival bacterial
Diversity of bacterial populations on the community profiles in HIV-infected Brazilian
tongue dorsa of patients with halitosis and adults with chronic periodontitis. J Periodontal
healthy patients. J Clin Microbiol 41:558563 Res 51:95102
13. Paster BJ, Falkler Jr WA Jr, Enwonwu CO, 24. Ghannoum MA, Jurevic RJ, Mukherjee PK,
Idigbe EO, Savage KO, Levanos VA, Tamer Cui F, Sikaroodi M, Naqvi A, Gillevet PM
MA, Ericson RL, Lau CN, Dewhirst FE (2002) (2010) Characterization of the oral fungal
Prevalent bacterial species and novel phylo- microbiome (mycobiome) in healthy individu-
types in advanced noma lesions. J Clin als. PLoS Pathog 6:e1000713
Microbiol 40:21872191 25. Marsh P, Martin MV (1999) Oral microbiology,
14. Paster BJ, Boches SK, Galvin JL, Ericson RE, 4th edn. Wright, Oxford
Lau CN, Levanos VA, Sahasrabudhe A, 26. Bowden GH (2000) The microbial ecology of
Dewhirst FE (2001) Bacterial diversity in dental caries. Microb Ecol Health Dis
human subgingival plaque. J Bacteriol 12:138148
183:37703783 27. Belda-Ferre P, Alcaraz LD, Cabrera-Rubio R,
15. de Lillo A, Ashley FP, Palmer RM, Munson Romero H, Simn-Soro A, Pignatelli M, Mira
MA, Kyriacou L, Weightman AJ, Wade WG A (2012) The oral metagenome in health and
(2006) Novel subgingival bacterial phylotypes disease. ISME J 6:4656
detected using multiple universal polymerase 28. Schulze-Schweifing K, Banerjee A, Wade WG
chain reaction primer sets. Oral Microbiol (2014) Comparison of bacterial culture and
Immunol 21:6168 16S rRNA community profiling by clonal anal-
16. Aas JA, Barbuto SM, Alpagot T, Olsen I, ysis and pyrosequencing for the characteriza-
Dewhirst FE, Paster BJ (2007) Subgingival tion of the dentine caries-associated
plaque microbiota in HIV positive patients. microbiome. Front Cell Infect Microbiol 4:164
J Clin Periodontol 34:189195 29. Jiang W, Ling Z, Lin X, Chen Y, Zhang J, Yu J,
17. Huyghe A, Francois P, Charbonnier Y, Xiang C, Chen H (2014) Pyrosequencing anal-
Tangomo-Bento M, Bonetti EJ, Paster BJ, ysis of oral microbiota shifting in various caries
Bolivar I, Baratti-Mayer D, Pittet D, Schrenzel states in childhood. Microb Ecol 67:962969
J, Geneva Study Group on Noma (GESNOMA) 30. Prez D, Olsen I, Aas JA, Willumsen T, Grinde
(2008) Novel microarray design strategy to B, Paster BJ (2008) Bacterial profiles of root
study complex bacterial communities. Appl caries in elderly patients. J Clin Microbiol
Environ Microbiol 74:18761885 46:20152021
18. Keijser BJ, Zaura E, Huse SM, van der Vossen 31. Aas JA, Dardis SR, Griffen AL, Stokes LN, Lee
JM, Schuren FH, Montijn RC, ten Cate JM, AM, Olsen I, Dewhirst FE, Leys EJ, Paster BJ
Crielaard W (2008) Pyrosequencing analysis of (2003) Molecular analysis of bacteria associ-
the oral microflora of healthy adults. J Dent ated with caries in permanent teeth. J Dent Res
Res 87:10161020 82 (Spec Iss A):IADR Abstract No. 1025.
19. Zaura E, Keijser BJ, Huse SM, Crielaard W http://www.dentalresearch.org
(2009) Defining the healthy core microbi- 32. Munson MA, Banerjee A, Watson TF, Wade
ome of oral microbial communities. BMC WG (2004) Molecular analysis of the micro-
Microbiol 9:259 flora associated with dental caries. J Clin
20. Crielaard W, Zaura E, Schuller AA, Huse SM, Microbiol 42:30233029
Montijn RC, Keijser BJ (2011) Exploring the 33. Aas JA, Dardis SR, Griffen AL, Stokes LN, Lee
oral microbiota of children at various develop- AM, Olsen I, Dewhirs FE, Leys EJ, Paster BJ
mental stages of their dentition in the relation (2005) Most of the microbiota in caries has not
to their oral health. BMC Med Genomics 4:22 yet been cultivated. J Dent Res 84 (Spec Iss A),
21. Lazarevic V, Whiteson K, Hernandez D, IADR Abstract No 2805. http:// www.dental-
Francois P, Schrenzel J (2010) Study of inter- research.org
and intra-individual variations in the salivary 34. Aas JA, Griffen AL, Dardis SR, Lee AM, Olsen
microbiota. BMC Genomics 11:523 I, Dewhirst FE, Leys EJ, Paster BJ (2008)
Bacteria of dental caries in primary and perma- 47. Hajishengallis G, Lamont RJ (2014) Breaking
nent teeth in children and young adults. J Clin bad: manipulation of the host response by
Microbiol 46:14071417 Porphyromonas gingivalis. Eur J Immunol
35. Becker MR, Paster BJ, Leys EJ, Moeschberger 44:328338
ML, Kenyon SG, Galvin JL, Boches SK, 48. Socransky SS, Haffajee AD, Cugini MA, Smith
Dewhirst FE, Griffen AL (2002) Molecular C, Kent RL Jr (1998) Microbial complexes in
analysis of bacterial species associated with child- subgingival plaque. J Clin Periodontol
hood caries. J Clin Microbiol 40:10011009 25:134144
36. Lima KC, Coelho LT, Pinheiro IV, Ras IN, 49. Sakamoto M, Huang Y, Umeda M, Ishikawa I,
Siqueira JF Jr (2011) Microbiota of dentinal Benno Y (2002) Detection of novel oral phylo-
caries as assessed by reverse-capture checker- types associated with periodontitis. FEMS
board analysis. Caries Res 45:2130 Microbiol Lett 217:6569
37. Kianoush N, Adler CJ, Nguyen KA, Browne 50. Griffen AL, Kumar PS, Leys EJ (2003) A
GV, Simonian M, Hunter N (2014) Bacterial quantitative, molecular view of oral biofilm
profile of dentine caries and the impact of pH communities in health and disease suggests a
on bacterial population diversity. PLoS One 9, role for uncultivated species, polymicrobial dis-
e92940 eases. American Society for Microbiology
38. Nadkarni MA, Caldon CE, Chhour KL, Fisher Conferences, Lake Tahoe, Nevada, p 13
IP, Martin FE, Jacques NA, Hunter N (2004) 51. Kumar PS, Griffen AL, Barton JA, Paster BJ,
Carious dentine provides a habitat for a com- Moeschberger ML, Leys EJ (2003) New bac-
plex array of novel Prevotella-like bacteria. terial species associated with chronic periodon-
J Clin Microbiol 42:52385244 titis. J Dent Res 82:338344
39. Martin FE, Nadkarni MA, Jacques NA, Hunter 52. Kumar PS, Griffen AL, Moeschberger ML,
N (2002) Quantitative microbiological study Leys EJ (2005) Identification of candidate
of human carious dentine by culture and real- periodontal pathogens and beneficial species by
time PCR: association of anaerobes with histo- quantitative 16S clonal analysis. J Clin
pathological changes in chronic pulpitis. J Clin Microbiol 43:39443955
Microbiol 40:16981704 53. Brinig MM, Lepp PW, Ouverney CC, Armitage
40. Chhour KL, Nadkarni MA, Byun R, Martin GC, Relman DA (2003) Prevalence of bacteria
FE, Jacques NA, Hunter N (2005) Molecular of division TM7 in human subgingival plaque
analysis of microbial diversity in advanced car- and their association with disease. Appl Environ
ies. J Clin Microbiol 43:843849 Microbiol 69:16871694
41. Ling Z, Kong J, Jia P, Wei C, Wang Y, Pan Z, 54. Harper-Owen R, Dymock D, Booth V,
Huang W, Li L, Chen H, Xiang C (2010) Weightman AJ, Wade WG (1999) Detection of
Analysis of oral microbiota in children with unculturable bacteria in periodontal health and
dental caries by PCR-DGGE and barcoded disease by PCR. J Clin Microbiol
pyrosequencing. Microb Ecol 60:677690 37:14691473
42. Ras IN, Lima KC, Assuncao IV, Gomes PN, 55. Park OJ, Yi H, Jeon JH, Kang SS, Koo KT,
Bracks IV, Siqueira JF Jr (2015) Advanced car- Kum KY, Chun J, Yun CH, Han SH (2015)
ies microbiota in teeth with irreversible pulpi- Pyrosequencing analysis of subgingival micro-
tis. J Endod 41:14501455 biota in distinct periodontal conditions. J Dent
43. Chen L, Qin B, Du M, Zhong H, Xu Q, Li Y, Res 94:921927
Zhang P, Fan M (2015) Extensive description 56. Griffen AL, Beall CJ, Campbell JH, Firestone
and comparison of human supra-gingival ND, Kumar PS, Yang ZK, Podar M, Leys EJ
microbiome in root caries and health. PLoS (2012) Distinct and complex bacterial profiles
One 10, e0117064 in human periodontitis and health revealed by
44. Jiang W, Zhang J, Chen H (2013) 16S pyrosequencing. ISME J 6:11761185
Pyrosequencing analysis of oral microbiota in 57. Abusleme L, Dupuy AK, Dutzan N, Silva N,
children with severe early childhood dental Burleson JA, Strausbaugh LD, Gamonal J,
caries. Curr Microbiol 67:537542 Diaz PI (2013) The subgingival microbiome in
45. Loesche WJ, Kazor C (2002) Microbiology health and periodontitis and its relationship
and treatment of halitosis. Periodontol 2000 with community biomass and inflammation.
28:256279 ISME J 7:10161025
46. Lamont RJ, Hajishengallis G (2015) 58. Camelo-Castillo AJ, Mira A, Pico A, Nibali L,
Polymicrobial synergy and dysbiosis in inflam- Henderson B, Donos N, Tomas I (2015)
matory disease. Trends Mol Med 21:172183 Subgingival microbiota in health compared to
periodontitis and the influence of smoking. 68. Siqueira JF Jr, Alves FR, Ras IN (2011)
Front Microbiol 6:119 Pyrosequencing analysis of the apical root canal
59. Moon JH, Lee JH, Lee JY (2015) Subgingival microbiota. J Endod 37:14991503
microbiome in smokers and non-smokers in 69. Ozok AR, Persoon IF, Huse SM, Keijser BJ,
Korean chronic periodontitis patients. Mol Wesselink PR, Crielaard W, Zaura E (2012)
Oral Microbiol 30:227241 Ecology of the microbiome of the infected
60. Lepp PW, Brinig MM, Ouverney CC, Palm K, root canal system: a comparison between api-
Armitage GC, Relman DA (2004) Methanogenic cal and coronal root segments. Int Endod
Archaea and human periodontal disease. Proc J 45:530541
Natl Acad Sci U S A 101:61766181 70. Santos AL, Siqueira JF Jr, Ras IN, Jesus EC,
61. Slots J (2005) Herpesviruses in periodontal Rosado AS, Tiedje JM (2011) Comparing the
diseases. Periodontol 2000 38:3362 bacterial diversity of acute and chronic dental
62. Slots J (2010) Human viruses in periodontitis. root canal infections. PLoS One 6, e28088
Periodontol 2000 53:89110 71. Hong BY, Lee TK, Lim SM, Chang SW, Park
63. Siqueira JF Jr (2002) Endodontic infections: J, Han SH, Zhu Q, Safavi KE, Fouad AF, Kum
concepts, paradigms, and perspectives. Oral KY (2013) Microbial analysis in primary and
Surg Oral Med Oral Pathol Oral Radiol Endod persistent endodontic infections by using pyro-
94:281293 sequencing. J Endod 39:11361140
64. Siqueira JF Jr, Ras IN (2009) Diversity of 72. Vengerfeldt V, pilka K, Saag M, Preem JK,
endodontic microbiota revisited. J Dent Res Oopkaup K, Truu J, Mndar R (2014) Highly
88:969981 diverse microbiota in dental root canals in cases
of apical periodontitis (data of illumina
65. Sundqvist G, Figdor D, Persson S, Sjogren U sequencing). J Endod 40:17781783
(1998) Microbiologic analysis of teeth with
failed endodontic treatment and the outcome of 73. Anderson AC, Al-Ahmad A, Elamin F, Jonas
conservative re-treatment. Oral Surg Oral Med D, Mirghani Y, Schilhabel M, Karygianni L,
Oral Pathol Oral Radiol Endod 85:8693 Hellwig E, Rehman A (2013) Comparison of
the bacterial composition and structure in
66. Ras IN, Siqueira JF Jr (2012) Characterization symptomatic and asymptomatic endodontic
of microbiota of root canal-treated teeth with infections associated with root-filled teeth
posttreatment disease. J Clin Microbiol using pyrosequencing. PLoS One 8, e84960
50:17211724
74. Tzanetakis GN, Azcarate-Peril MA, Zachaki S,
67. Li L, Hsiao WW, Nandakumar R, Barbuto SM, Panopoulos P, Kontakiotis EG, Madianos PN,
Mongodin EF, Paster BJ, Fraser-Liggett CM, Divaris K (2015) Comparison of bacterial com-
Fouad AF (2010) Analyzing endodontic infec- munity composition of primary and persistent
tions by deep coverage pyrosequencing. J Dent endodontic infections using pyrosequencing.
Res 89:980984 J Endod 41:12261233
Chapter 8
Microbial Community Profiling Using Terminal Restriction

Fragment Length Polymorphism (T-RFLP) and Denaturing
Gradient Gel Electrophoresis (DGGE)
Jos F. Siqueira Jr., Mitsuo Sakamoto, and Alexandre S. Rosado
Abstract
In their natural environments, microorganisms usually live in organized communities. Profiling analysis of
microbial communities has recently assumed special relevance as it allows a thorough understanding of the
diversity of the microbiota, its behavior over time, and the establishment of patterns associated with health
and disease. The application of molecular biology approaches holds the advantage of including culture-
difficult and as-yet-uncultivated phylotypes in the profiles, providing a more comprehensive picture of the
microbial community. This chapter focuses on two particular techniques, namely, terminal restriction
fragment length polymorphism (T-RFLP) and denaturing gradient gel electrophoresis (DGGE), both of
which have been widely used in environmental studies and have been successfully used by the authors in
the study of the oral microbial communities associated with conditions of health and disease.
Key words Human oral microbiota, 16S rRNA gene, Terminal restriction fragment length polymor-
phism (T-RFLP), Denaturing gradient gel electrophoresis (DGGE)
1 Introduction
Microbial community profiling techniques are genetic fingerprinting

approaches that can be used to determine the structure and diver-
sity of microbial communities living in a given environment and to
monitor changes in the community over time, including after anti-
microbial treatment. Species identification can also be obtained
with these techniques. There are several molecular methods for
community profiling, but the terminal restriction fragment length
polymorphism (T-RFLP) and the denaturing gradient gel electro-
phoresis (DGGE) have been frequently used in the study of oral
communities in health and disease [112].
T-RFLP allows the assessment of the diversity of complex
bacterial communities and rapid comparison of the community
structure from different ecosystems [13]. T-RFLP analysis measures
the size polymorphism of terminal restriction fragments from a
139
140 Jos F. Siqueira et al.
PCR amplified marker. When T-RFLP is used to analyze bacterial

communities, PCR is first carried out to amplify the 16S rRNA
gene from different species in the sample. One of the PCR primers
is labeled with a fluorescent dye [14]. PCR amplicons are then
digested with restriction enzymes, generating fluorescently labeled
fragments of different lengths (the terminal fragments). These
fragments are separated on high-resolution sequencing gels in an
automated DNA sequencer, which is used to read both the size
and the intensity of terminally labeled restriction fragments
(T-RFs), creating a typical profile. In such a profile, size is repre-
sented on the horizontal axis, and intensity (relative to the abun-
dance of a given fragment size) is represented on the vertical axis
[15]. In theory, each T-RF represents a single species. Extensive
databases exist for 16S rRNA gene sequences and can be used to
identify all T-RFs predicted from known sequences, considering a
given set of primers and restriction enzymes [16]. T-RF lengths are
predicted by finding the restriction site closest to the site where the
labeled primer will anneal and counting the number of nucleotides
in between. Multiple restriction enzymes (usually 4 or 5) are nec-
essary to provide reliable identification since distinct species may
generate the same T-RF when only one enzyme is used [17].
The DGGE technique is based on electrophoretic separation
of PCR-amplified 16S rRNA gene (or other genes) fragments in
polyacrylamide gels containing a linearly increasing gradient of
DNA denaturants (a mixture of urea and formamide). As the PCR
product migrates in the gel, it encounters increasing concentra-
tions of denaturants, and, at some position in the gel, it will become
partially or fully denatured. Partial denaturation causes a significant
decrease in the electrophoretic mobility of the DNA molecule.
Molecules with different sequences may have a different melting
behavior and will therefore stop migrating at different positions in
the gel. The position in the gel at which the DNA melts is deter-
mined by its nucleotide sequence and composition [18]. Therefore,
in DGGE, PCR products of the same length but with different
sequences can be separated [19, 20]. A GC-rich sequence (or
GC-clamp) is added to the 5-end of one of the primers used in the
PCR reaction and makes the DNA unable to denature completely
in the gel [21]. DNA bands in DGGE can be visualized using
ethidium bromide, SYBR Green, or silver staining. If species
identification is desired, specific bands can be excised from the
gels, re-amplified by PCR, and sequenced [22].
2 Materials
2.1 DNA Extraction 1. Buffer A: 10 mM TrisHCl (pH 8.0), 50 mM ethylenediami-

netetraacetic acid (EDTA).
2. Lysis buffer: 0.5 % (w/v) lysozyme and 0.1 % (w/v) N -
acetylmuramidase in buffer A. Store in aliquots at 20 C.
Microbial Community Profiling with T-RFLP and DGGE 141
3. TE buffer: 10 mM TrisHCl, pH 8.0, 1 mM EDTA.

4. Alternatively, other techniques may be used for DNA extrac-
tion (see Note 1).
2.2 Terminal 1. Forward primer 8F: 5-AGA GTT TGA TCC TGG CTC
Restriction Fragment AG-3. This primer is labeled at the 5-end with 6-carboxy-
Length Polymorphism fluorescein (6-FAM), which is synthesized by Applied
Biosystems Japan (see Note 2).
2.2.1 PCR Amplification
of the 16S rRNA Gene 2. Reverse primer 1492R: 5-GGT TAC CTT GTT ACG ACT
T-3.
3. Tris-acetate-EDTA (TAE) buffer (50): 2 M Tris (do not
adjust pH), 2 M glacial acetic acid, 0.05 M EDTA, pH 8.0.
4. Polyethyleneglycol (PEG) solution: 40 % (w/v) PEG 6000,
10 mM MgCl2 (see Note 3).
2.2.2 T-RFLP Analysis 1. Capillaries: 310 Capillary 47 cm, 3130xl & 3100 Capillary
Array 36 cm, 3130xl & 3100 Capillary Array 50 cm (Applied
Biosystems, Foster City, CA, USA) (see Note 4).
2. Polymers: POP-4 (for the ABI Genetic Analyzer 310 and ABI
PRISM 3100 instruments); POP-7 (for the ABI 3130xl
Genetic Analyzer) (Applied Biosystems).
3. Running buffer: buffer (10) with EDTA (Applied Biosystems).
4. Size standards: GeneScan 500 ROX Size Standard, GeneScan
1000 ROX Size Standard, GeneScan 1200 LIZ Size Standard
(all supplied by Applied Biosystems).
5. Template preparation reagent: Hi-Di Formamide (Applied
Biosystems).
2.3 Denaturing 1. Forward primer 968F: 5-AAC GCG AAG AAC CTT AC-3,
Gradient Gel containing a 40-base GC clamp (5-CGC CCG CCG CGC
Electrophoresis GCG GCG GGC GGG GCG GGG GCA CGG GGG G-3)
added to its 5-end, which makes it suitable for DGGE.
of 16S rRNA Gene 2. Reverse primer 1401R: 5-GCG TGT GTA CAA GAC CC-3.
3. Deionized formamide (see below).
4. 1 % (w/v) bovine serum albumin (BSA). Store in aliquots of
50 L at 20 C.
5. DNA polymerase kit (including buffers) for PCR (e.g., Ex Taq
Hot Start DNA polymerase manufactured by Takara Japan).
2.3.2 DGGE Analysis 1. TAE buffer: 20 mM Tris-acetate, pH 7.4, 10 mM sodium ace-

tate, 0.5 mM disodium EDTA. Store at room temperature.
2. Deionized formamide: add 12.5 g of AG 501-X8 resin (Bio-Rad)
to 250 mL formamide 100 %. Stir for 1 h at room temperature.
Remove beads by passing the solution through folded filter
paper in a funnel. Store in the dark at 4 C.
3. 10 % (w/v) ammonium persulfate (APS) in deionized water.

Store in 800 L aliquots at 20 C.
4. N,N,N,N-tetramethylethylenediamine (TEMED).
5. Loading buffer 6: 1.5 mL glycerol and 12.5 mg bromophe-
nol blue (BPB) in 5 mL deionized water. Store at 4 C.
6. Gel dye: 0.05 g bromophenol blue in 10 mL 1 TAE.
7. Acrylamide/bis-acrylamide, 40 % solution for electrophoresis,
37.5:1 (Sigma-Aldrich).
8. Zero percentage urea/formamide (UF) in 6 % acrylamide/bis-
acrylamide: 15 % (v/v) acrylamide/bis-acrylamide, 40 % solu-
tion for electrophoresis (37.5: 1), 2 % (v/v) TAE buffer 50.
Store at 4 C in a dark bottle (storable up to 6 months) (see
Note 5).
9. 100 % UF in 6 % acrylamide/bis-acrylamide: 42 % (w/v) urea
P.A., 40 % (v/v) deionized formamide, 15 % (v/v) acrylamide/
bis-acrylamide, 40 % solution for electrophoresis, 2 % (v/v)
TAE buffer 50x. The final volume must be completed to
100 mL after dissolving the urea (see Note 6).
10. Staining solution for DGGE: SYBR Green in deionized
water in the proportion of 1:10,000 (this solution should be
prepared fresh and kept in the dark or in an amber vial).
3 Methods
3.1 DNA Extraction 1. An aliquot of 0.5 mL of clinical sample (saliva, pus, and plaque
or root canal contents suspended in Tris-EDTA buffer) is
diluted with buffer A in a 1:2 ratio (v/v) and washed with the
same buffer (see Note 1).
2. The bacterial cell pellet obtained is resuspended in 0.5 mL of
the lysis buffer. After incubation at 37 C for 1 h, proteinase K
and sodium dodecyl sulfate (SDS) are added to a final concen-
tration of 2 mg/mL and 1 % (w/v), respectively. The mixture
is incubated at 50 C for 2 h.
3. Nucleic acid is released by three cycles of freezing in a 80 C
freezer followed by thawing in a 65 C water bath.
4. The nucleic acid is then extracted with equal volumes of phenol
(saturated with 10 mM TrisHCl, pH 8.0) and phenol/
chloroform/isoamyl alcohol (25:24:1).
5. Bulk nucleic acids are precipitated from solution with 0.1 volume
of 3 M sodium acetate and 0.8 volume of isopropyl alcohol
followed by centrifugation (16,000 g for 15 min).
6. The DNA precipitate is washed with 70 % ethanol and resus-
pended in 100 L TE.
7. RNase is added to a final concentration of 10 g/mL and the

mixture is incubated at 37 C for 1 h.
8. The mixture is then treated with equal volumes of phenol and
phenol/chloroform/isoamyl alcohol (25:24:1).
9. The DNA is precipitated again with 0.1 volume of 3 M sodium
acetate and 0.8 volume of isopropyl alcohol.
10. The DNA is pelleted by centrifugation (16,000 g for 15 min),
washed with 70 % ethanol, dried in vacuum for 10 min, and
dissolved in 100 L TE.
3.2 Terminal 1. Amplification reactions are performed in a total volume of

Restriction Fragment 50 L containing 5 L of DNA extract (100 ng), 1.25 U
Length Polymorphism Takara Ex Taq (Takara Bio, Japan), 5 L of 10 Ex Taq buffer,
4 L of dNTP mixture (2.5 mM each), and 10 pmol of each
primer.
of 16S rRNA Gene
2. 16S rRNA genes are amplified in a Biometra Tgradient
Thermocycler using the following program: 95 C for 3 min,
followed by 30 cycles of 95 C for 30 s, 50 C for 30 s, and
72 C for 1.5 min, with a final extension at 72 C for 10 min.
3. Amplified DNA is verified by electrophoresis of aliquots of
PCR mixture (2 L) in 1.5 % agarose in 1 TAE buffer.
4. A 50 L aliquot of the 16S rRNA gene solution is mixed with
30 L of a PEG solution and 12 L of 3 M sodium acetate,
gently shaken for 10 min at room temperature, and centri-
fuged at >16,000 g for 15 min.
5. The supernatant is removed carefully by pipetting, and then
precipitated DNA is washed twice with 70 % ethanol (see Note 7)
and redissolved in 20 L of sterile distilled water. Purified 16S
rRNA genes are stored at 20 C until analysis.
3.2.2 T-RFLP Analysis The following protocol can be used in the ABI PRISM 310 Genetic
for ABI PRISM 310 Genetic Analyzer, ABI PRISM 3100 Genetic Analyzer, and ABI 3130xl
Analyzer Genetic Analyzer instruments. Any modification specific for each
instrument is also noted.
1. Purified PCR product (2 L) is digested with 20 U of restric-
tion enzyme HhaI, MspI, AluI, HaeIII, or RsaI in a total
volume of 10 L at 37 C for 3 h.
2. The restriction digest product (1 L) is mixed with 12 L of
Hi-Di Formamide and 1 L of DNA fragment length stan-
dard. The standard size marker is a 1:1 mixture of GS 500
ROX and GS 1000 ROX. In the case of ABI 3130xl Genetic
Analyzer, GS 1200 LIZ is used as a standard size marker.
3. Each sample is denatured at 95 C for 2 min and then imme-
diately placed on ice.
4. The length of T-RF is determined on an ABI PRISM 310

Genetic Analyzer (Applied Biosystems) in GeneScan mode
(15 kV, 8 A, and 60 C for 48 min for each sample). 310
Capillary 47 cm and 310 POP-4 are used (see Note 8).
5. Fragment sizes are estimated by using the Local Southern method
in GeneScan 3.1 software (Applied Biosystems) (see Note 9).
6. T-RFs with a peak area of less than 25 fluorescence units are
excluded from the analysis. In the case of ABI 3100 and 3130xl
Genetic Analyzers, T-RFs with a peak area of <2 % of the total
area are excluded.
7. Fragments are resolved to one base pair by manual alignment
of the size standard peaks from different electropherograms.
8. Predicted T-RFLP patterns of the 16S rRNA genes of known
bacterial species are obtained using the GENETYX-MAC pro-
gram (Software Developing Co., Tokyo) (see Note 10). An
example is given in Fig. 1.
3.3 Denaturing The electrophoresis gel is made of polyacrylamide (acrylamide and

Gradient Gel bis-acrylamide) containing a linear gradient of the denaturants
Electrophoresis urea and formamide, increasing from the top of the gel to the bot-
tom. The experiments are carried out in a Dcode universal muta-
tion detection system (Bio-Rad, Richmond, VA, USA). The
gradient of denaturants can be adjusted according to the fragment
of DNA amplified by PCR. The example we give here is of a poly-
acrylamide gel with a linear gradient of denaturant ranging from
30 to 70 %, formed from mixing stock solutions of polyacrylamide
(6 %): (a) a solution containing 0 % UF and another containing
100 % of denaturants UF (100 % corresponds to 7 M urea and 40 %
of deionized formamide).
3.3.1 PCR Amplification 1. Amplification reactions are performed in a total volume of

of 16S rRNA Gene 50 L containing 5 L of DNA extract (100 ng), 0.2 M of
each primer, 5 L of 10 PCR buffer, 2.5 mM MgCl2, 1.5
units Taq DNA polymerase, and 0.2 mM of each dNTP and
deionized water to make up the volume.
2. 16S rRNA genes are amplified in an Eppendorf thermocycler
(Hamburg, Germany) using the following programs: initial
denaturation step at 94 C for 3 min, 36 cycles of denaturation
at 94 C for 1 min, primer annealing at 55 C for 1 min, and
extension at 72 C for 2 min, followed by a final extension step
of 72 C for 10 min.
Prior to DGGE analysis, the presence of PCR products is con-
firmed by electrophoresis in a 1.5 % agarose gel conducted at 4 V/
cm in Tris-borate-EDTA buffer. The gel is stained for 15 min with
0.5 g/mL ethidium bromide and viewed under ultraviolet transil-
lumination. A 100-bp DNA ladder (New England Biolabs, Beverly,
MA, USA) served as the molecular size standard.
a Terminal restriction fragment (bp)

0 100 200 300 400 500 600 700 800 900 1000 1100
E. saphenum, Peptostreptococcus sp.

Before
Po. gingivalis
S. cristatus, (clone AJ062, AP24, BS044 & FG014)
S. mitis,
S. salivarius,
Pr. intermedia S. sanguinis
Fu. nucleatum
S. anginosus,
Fi. alocis S. intermedius,
V. dispar
After
S. cristatus, S. anginosus,
S. mitis, S. gordonii,
N. pharyngis S. salivarius, S. intermedius,
Fu. nucleatum V. atypica,
S. mutans S. sanguinis
V. dispar,
V. parvula
T. aromaticivorans
b Terminal restriction fragment (bp)

0 100 200 300 400 500 600 700
Pr. intermedia
Po. gingivalis
Eubacterium sp.
E. nodatum, Before
Peptostreptococcus sp.
(clone BP1-27 & PUS9.170) (clone AJ062, AP24, BS044 & FG014) S. cristatus,
S. mitis,
E. saphenum Fi. alocis S. salivarius,
S. sanguinis
Eubacterium sp.
(clone AP54)
S. anginosus,
V. dispar
Fu. nucleatum S. intermedius
V. atypica, S. cristatus,
S. mitis,
After
V. dispar,
S. salivarius,
V. parvula S. anginosus,
S. sanguinis
S. gordonii,
S. intermedius,
Fu. nucleatum N. pharyngis S. mutans
T. aromaticivorans
Fig. 1 Terminal restriction fragment length polymorphism patterns of 16S rRNA genes from subgingival plaque
samples of a patient with periodontitis taken before treatment and after treatment generated after digestion
with HhaI (a) and MspI (b). 16S rRNA genes were amplified with universal primers 27F and 1492R. Almost all
the terminal restriction fragments were presumed to be species or phylotypes detected by the 16S rRNA gene
clone library analysis. E., Eubacterium; Fi., Filifactor; Fu., Fusobacterium; N., Neisseria; Po., Porphyromonas;
Pr., Prevotella; S., Streptococcus; T., Terrahaemophilus; V., Veillonella. Reproduced with permission from
Sakamoto et al. [3]. (2004) Society for General Microbiology
3.3.2 DGGE Analysis 1. The following instructions assume the use of a Dcode uni-
versal mutation detection system. They can be easily adapted
to other DGGE instruments.
2. Pouring the gel: Two glass slabs are used in the assembly of the
gel. Assemble the slabs using the spacers and clamps as depicted
Fig. 2 General alignment of the DGGE plates
in Fig. 2. It is highly recommended to clean the trays with

gauze and alcohol. Do not use cotton or paper, because their
residue can fluoresce under exposure to UV light.
3. Insert the alignment card to keep the spacers parallel and push
the slabs and spacers down and to each other at the same time
you squeeze both clamps (not too tight).
4. Remove the card from the rack, and check if the alignment was
correct, moving a finger under the borders to see the
alignment.
5. Place the sandwich on the gray sponge into the casting stand,
and turn the handles down to lock the assembly in place. Find
the comb you want to use and place it near the stand.
6. Preparation of the denaturant gradient (prepare these solu-
tions with the 0 % UF and 100 % UF solutions (see Note 11):
Prepare 12 mL of the lower and upper gradient solutions in
separate Falcon tubes (marked L and H), being UF low (for
example 30 %), UF high (for example 70 %). Also prepare a
tube with 5 mL of 0 % solution for the stacking gel.
7. Add 30 L of gel dye (without glycerol) in 12 mL of the solu-
tion with the highest concentration of denaturants (the gel
coloration will present a gradient of blue color, corresponding
to the gradient concentration).
8. After preparation of the denaturing solutions, add the APS and
TEMED quickly to prevent polymerization of the gel in the
syringe. Add 4.2 L of 10 % APS and 0.83 L of TEMED per
mL of solution. Thus, each 12 mL of solution will take 50 L
of 10 % APS and 10 L of TEMED. These reagents are respon-
sible for polymerization of acrylamide and bis-acrylamide
(polymerization begins immediately after addition of TEMED,
so work quickly).
9. Draw each solution into the syringes. For our gels, the low
denaturant syringe is on the front side of the gradient maker
and the high denaturant is on the backside. Make sure air is

removed from syringes (avoid bubble formation). The needles
are then coupled to the device that will form the gradient
(gradient wheel from Bio-Rad) along the denaturant gel.
10. Set up the gradient former by inserting 30 mL syringes into a
closed gradient wheel. Secure the syringes and back out
the syringe plungers by reversing the gradient wheel. Note the
resulting volume measured on each syringe (between 12 mL),
which will be the amount of solution that the system will
deliver to create the new gel. Immediately wash the syringes
with distilled water after the use, to avoid polymerization of
the remaining solutions.
11. Place the delivery tube in between the two plates near the center
of the top edge of the plate assembly. Slowly and consistently
turn the wheel until the gel is poured (see Fig. 3a).
12. Finally, a solution without denaturing agents completes the gel
(stacking gel) and where the comb is coupled (after polymer-
ization, the comb is removed and the slots will be formed)
(see Fig. 3b).
13. Allow the gel to polymerize for about 2 h. Then, it must be
mounted in the DGGE device, which permits two gels to run
simultaneously (if you are running only one gel, use glass slabs
Fig. 3 Preparation of denaturant gradient. (a) The lower and the upper gradient solutions are loaded into separate
syringes. (b) Denaturing gel with different UF concentrationsUF low (e.g., 30 %) and UF high (e.g., 70 %)
without spacers to close the system and fit it in the support).

After polymerization, remove the comb carefully. Rinse the
slots to remove non-polymerized gel with 0.5 TAE buffer.
14. Running the gel: add 7 L of fresh 1 TAE buffer to the buffer
tank up to the Fill mark. Switch on the DCode universal
mutation detection system (Bio-Rad) at least 60 min before
electrophoresis so that the buffer can heat up to 60 C (it is not
needed to activate the stirrer underneath the tank and the
pump of the Dcode system).
15. Load about 20 L of PCR products (see Note 12) and 20 L
of loading buffer (1:1). The application of the samples should
occur with the slots previously immersed in 1 TAE (run-
ning buffer).
16. Load a marker sample along with the test samples (for determi-
nation of band positions and comparisons between different
gels using software (e.g., Gel Compar II), we use 3 marker
lanes: 1 in the middle slot and 1 at each slot at the extremities
of the gel).
17. Run the gel at 60 C for either 16 h at 75 V or 4 h at 200 V.
18. After running, turn off the power supply, remove the control
unit, and take out the core. Remove the sandwich and the
clamps and remove one glass plate.
19. Gels are stained with SYBR Green I nucleic acid gel stain
(Molecular Probes) for 40 min and then scanned using a Storm
Phosphorimager (Amersham Biosciences, Uppsala, Sweden).
20. The digitized images of DGGE gels can be analyzed using the
Gel Compar II software (Applied Maths, Kortrijk, Belgium) or
other approaches [6] (Fig. 4).
4 Notes
1. We have also had good results using the QIAamp DNA Mini
Kit, following the recommendations of the manufacturer
(Qiagen, Valencia, CA, USA).
2. 6-FAM-labeled forward primer 8f is light sensitive. It must be
stored in the dark at 20 C.
3. PEG solution is very viscous.
4. The condition of the capillary will significantly affect the
T-RFLP patterns.
5. During preparation and development of DGGE gels, highly
toxic and carcinogenic materials are used. Always wear a lab
coat, gloves, and a mask during all steps, and change them even
when the slightest contamination is suspected. Use blue nitrile
gloves when handling acrylamide/bis-acrylamide solutions.
Fig. 4 Dendrograms obtained by the UPGMA method for clustering of DGGE banding patterns of oral samples
using the Gel Compar II (version 5.10) software package
6. Store the bottles in the dark at 4 C. For good-quality gels, do

not use a 100 % denaturant solution which is older than 1
month.
7. The precipitated DNA may be easily washed out from micro-
centrifuge tube. It is important for a good result to ensure
recovering of the precipitated DNA.
8. The length of T-RF is influenced by the type of genetic ana-
lyzer, capillary, polymer, and size standard. In the case of ABI
3100 and 3130xl Genetic Analyzer instruments, the length of
T-RF is determined in GeneScan mode (15 kV, 100 A, and
60 C for 40 min for each sample) and GeneMapper mode
(8.5 kV, 100 A, and 60 C for 112 min for each sample),
respectively. 3130xl & 3100 Capillary Array 36 cm and 3100
POP-4 (for 3100 Genetic Analyzer) and 3130xl & 3100
Capillary Array 50 cm and 3100 POP-7 (for 3130xl Genetic
Analyzer) are used.
9. In the case of ABI 3100 and 3130xl Genetic Analyzers, frag-
ment sizes are estimated by using the Local Southern method
in GeneScan 3.7 and GeneMapper 4.0 software packages
(Applied Biosystems), respectively.
10. There are size discrepancies between predicted and observed
T-RF lengths [1].
11. Example of calculations to prepare a gel with gradient denaturant
(gradient 30 % UF to 70 % UF): prepare the 30 % UF solution of
denaturant (final volume of 12 mL) using the formula:
Ci Vi = Cf Vf (1)
where Ci is the initial concentration of 100 % denaturant solu-

tion, Vi is the volume necessary to reach the 30 % UF concentra-
tion, Cf is the final concentration of the solution (in this example,
30 % UF denaturant), and Vf is the final volume of the solution
(here we are using 12 mL for one syringe). We have:
100 Vi = 30 12 (2)
In the above example, 3.6 mL from the 100 % UF solution

will be added to 8.4 mL of 0 % denaturant solution (to com-
plete the total volume of 12 mL of denaturant solution).
Prepare the 70 % UF solution of denaturant (final volume
of 12 mL). Here, just replace the value of Cf in Eq. (1) by 70.
Thus, we have:
100 Vi = 70 12 (3)
The value of Vi, therefore, will be 8.4 mL. To complete

the volume of 12 mL, add 3.6 mL of 0 % denaturant solution.
Note that these calculations are valid for the construction of a
gel with gradient denaturant from 30 to 70 %. These values

may be adjusted in Eq. (1) to make gels with a range of values
of denaturation.
12. The quantity of PCR products applied to the gel may vary
depending on the quality of amplicons and the sensitivity of
the method used for gel staining.
Acknowledgments
This work was supported in part by grants from the Brazilian

National Research Council (CNPq) and FAPERJ (Jos Siqueira
and Alexandre Rosado) and by a Grant-in-Aid for Scientific
Research (No. 13672202) from the Japan Society for the
Promotion of Science (Mitsuo Sakamoto). We are grateful to
Natalia Franco for drawing the DGGE schemes.
References
1. Sakamoto M, Takeuchi Y, Umeda M, Ishikawa 7. Li Y, Ku CY, Xu J, Saxena D, Caufield PW
I, Benno Y (2003) Application of terminal (2005) Survey of oral microbial diversity using
RFLP analysis to characterize oral bacterial PCR-based denaturing gradient gel electro-
flora in saliva of healthy subjects and patients phoresis. J Dent Res 84:559564
with periodontitis. J Med Microbiol 52:7989 8. Sakamoto M, Ras IN, Siqueira JF Jr, Benno
2. Zijnge V, Harmsen HJ, Kleinfelder JW, van der Y (2006) Molecular analysis of bacteria in
Rest ME, Degener JE, Welling GW (2003) asymptomatic and symptomatic endodontic
Denaturing gradient gel electrophoresis analy- infections. Oral Microbiol Immunol
sis to study bacterial community structure in 21:112122
pockets of periodontitis patients. Oral 9. Zijnge V, Welling GW, Degener JE, van
Microbiol Immunol 18:5965 Winkelhoff AJ, Abbas F, Harmsen HJ (2006)
3. Sakamoto M, Huang Y, Ohnishi M, Umeda Denaturing gradient gel electrophoresis as a
M, Ishikawa I, Benno Y (2004) Changes in diagnostic tool in periodontal microbiology.
oral microbial profiles after periodontal treat- J Clin Microbiol 44:36283633
ment as determined by molecular analysis of 10. Machado de Oliveira JC, Siqueira JF Jr, Ras
16S rRNA genes. J Med Microbiol IN, Baumgartner JC, Xia T, Peixoto RS,
53:563571 Rosado AS (2007) Bacterial community pro-
4. Siqueira JF Jr, Ras IN, Rosado AS (2004) files of endodontic abscesses from Brazilian and
Investigation of bacterial communities associ- USA subjects as compared by denaturing gra-
ated with asymptomatic and symptomatic end- dient gel electrophoresis analysis. Oral
odontic infections by denaturing gradient gel Microbiol Immunol 22:1418
electrophoresis fingerprinting approach. Oral 11. Siqueira JF Jr, Ras IN, Debelian GJ, Carmo
Microbiol Immunol 19:363370 FL, Paiva SS, Alves FR, Rosado AS (2008)
5. Ras IN, Siqueira JF Jr, Aboim MC, Rosado Profiling of root canal bacterial communities
AS (2004) Denaturing gradient gel electro- associated with chronic apical periodontitis
phoresis analysis of bacterial communities asso- from Brazilian and Norwegian subjects.
ciated with failed endodontic treatment. Oral J Endod 34:14571461
Surg Oral Med Oral Pathol Oral Radiol Endod 12. Alves FR, Siqueira JF Jr, Carmo FL, Santos AL,
98:741749 Peixoto RS, Ras IN, Rosado AS (2009)
6. Siqueira JF Jr, Ras IN, Rosado AS (2005) Bacterial community profiling of cryogenically
Application of denaturing gradient gel electro- ground samples from the apical and coronal
phoresis (DGGE) to the analysis of endodontic root segments of teeth with apical periodonti-
infections. J Endod 31:775782 tis. J Endod 35:486492
13. Liu WT, Marsh TL, Cheng H, Forney LJ 18. Gasser RB (1998) Whats in that band? Int
(1997) Characterization of microbial diversity J Parasitol 28:989996
by determining terminal restriction fragment 19. Myers RM, Fischer SG, Lerman LS, Maniatis T
length polymorphisms of genes encoding 16S (1985) Nearly all single base substitutions in
rRNA. Appl Environ Microbiol 63: DNA fragments joined to a GC-clamp can be
45164522 detected by denaturing gradient gel electro-
14. Clement BG, Kehl LE, De Bord KL, Kitts CL phoresis. Nucleic Acids Res 13:31313145
(1998) Terminal restriction fragment patterns 20. Muyzer G, de Waal EC, Uitterlinden AG
(TRFPs), a rapid, PCR-based method for the (1993) Profiling of complex microbial popula-
comparison of complex bacterial communities. tions by denaturing gradient gel electrophoresis
J Microbiol Methods 31:135142 analysis of polymerase chain reaction-amplified
15. Spiegelman D, Whissell G, Greer CW (2005) A genes coding for 16S rRNA. Appl Environ
survey of the methods for the characterization Microbiol 59:695700
of microbial consortia and communities. Can 21. Muyzer G, Smalla K (1998) Application of
J Microbiol 51:355386 denaturing gradient gel electrophoresis
16. Matsumoto M, Sakamoto M, Hayashi H, (DGGE) and temperature gradient gel electro-
Benno Y (2005) Novel phylogenetic assign- phoresis (TGGE) in microbial ecology. Antonie
ment database for terminal-restriction frag- Van Leeuwenhoek 73:127141
ment length polymorphism analysis of human 22. Machado de Oliveira JC, Gama TG, Siqueira
colonic microbiota. J Microbiol Methods JF Jr, Ras IN, Peixoto RS, Rosado AS (2007)
61:305319 On the use of denaturing gradient gel electro-
17. Sakamoto M, Umeda M, Benno Y (2005) phoresis approach for bacterial identification in
Molecular analysis of human oral microbiota. endodontic infections. Clin Oral Investig 11:
J Periodontal Res 40:277285 127132
Chapter 9
Analysis of 16S rRNA Gene Amplicon Sequences Using

the QIIME Software Package
Blair Lawley and Gerald W. Tannock
Abstract
The study of microbial ecology has undergone a paradigm shift in recent years, with rapid advances in
molecular and bioinformatic tools allowing researchers with wide-ranging interests and backgrounds
access to community profiling methods. While these advances have undoubtedly led to exciting new
understanding of many systems, the array of protocols available and the idiosyncrasies of particular
approaches can lead to confusion or, at worst, erroneous interpretation of results. Here, we describe a
workflow from raw 16S rRNA gene amplicon sequence data, generated on an Illumina MiSeq instrument,
to microbial community taxonomy profiles and basic diversity measures. The workflow can be adapted to
input from major sequence platforms and uses freely available open source software that can be imple-
mented on a range of operating systems.
Key words High-throughput sequencing, 16S rRNA gene, QIIME, Microbial ecology, Bioinformatics,
Sequence analysis, Operational taxonomic unit (OTU)
1 Introduction
The rapid development of high-throughput sequencing (HTS)

instrumentation, driven initially by the Human Genome Project,
has changed the way many researchers in the biological sciences
work. HTS-associated methods have inevitably had a profound
impact on the study of microbial ecology and the advent of national
and international projects such as the Human Microbiome Project
(HMP) [1] and MetaHIT [2], and, more recently, the Earth
Microbiome Project (EMP) [3] have brought microbial ecology to
the public eye. The start point for many projects is community
profiling through sequencing of highly conserved marker genes,
predominantly the universal 16S rRNA gene, when studying pro-
karyote or archaeal populations. As methods have improved and
sequencing costs have decreased so has the demand for simple and
powerful analysis software packages increased. Development of a
universal sequencing and analysis pipeline that suits all projects
153
154 Blair Lawley and Gerald W. Tannock
appears to be unrealistic due to the complexity and variety of

research being undertaken. Analysis pipelines such as QIIME [4]
and Mothur [5] have evolved and expanded as the field has grown.
However, as advanced and powerful as these packages are, they still
present scientists with a steep learning curve. Our aim here is to
provide an overview of steps required to move from raw 16S rRNA
amplicon sequence data to taxa summaries and diversity measures.
Included are details of one possible analysis walk-through we have
used in our laboratory that can be modified to suit various sequence
platforms and research aims.
2 Methods
When undertaking a microbial diversity project, there are various

issues that researchers have to consider in order for the project to be
successful. These include selection of an appropriate hypervariable
region of the 16S rRNA gene to sequence, the type of HTS platform,
and the choice of post-sequencing bioinformatic software. Other
equally important aspects include the technique(s) employed to
sample the environment of interest (e.g., gut, oral, or hot springs),
the subsequent processing of those samples, and the genomic DNA
extraction procedures. The latter aspect is paramount as poor-quality
DNA will compromise severely all downstream procedures, especially
the HTS phase. As genomic DNA purification techniques are often
environment specific and HTS services are location dependent, such
decisions are left to the readers discretion.
2.1 Which Target Current HTS approaches do not allow full-length sequencing of
Sequence? the 16S rRNA gene; thus, researchers are restricted to sequencing
across a limited number of hypervariable regions (nine exist in the
16S rRNA gene). Each region has its advantages and disadvantages
when it comes to identifying microbes of interest. The more com-
monly sequenced variable regions are V1, V2, V3, V4, V6, and V7
(or combinations of these regions [6, 7]). Looking for consensus
in related literature may be a valuable approach, as it will allow
comparison with previously published work from the same envi-
ronment. If there is no consensus, V4 is a good choice as it is the
standard sequence region employed by the Human Microbiome
Project and the Earth Microbiome Project (see Note 1).
2.2 Which Sequence A range of HTS platforms, each using different chemistries, are
Chemistry? available to researchers for amplicon sequencing. This technical
space is currently dominated by Illumina (predominantly through
the MiSeq platform) and Thermo Fisher (via the Ion Torrent
platform). At time of writing, the previously popular Roche/454
system (via the GS-FLX platform) is no longer available as an HTS
option. Many researchers will be restricted to platforms available
in-house or through preferred third-party sequence providers.
Analysis of 16S rRNA Gene Amplicon Sequences Using the QIIME Software Package 155
The 454 and Ion Torrent chemistries offer longer reads but higher
error rates than Illumina, possibly providing an advantage by allow-
ing discrimination of closely related bacteria through sequencing
of more than one variable region. The Illumina system offers large
numbers of reads with very high accuracy. If using the V4 region,
the Illumina paired-end 250 base read system is a good choice as it
generates millions of reads with very low error rates and allows
multiplexing of hundreds of samples on a single MiSeq run.
2.3 Sequence Many software packages are available for analysis of amplicon HTS
Analysis data (see Note 2). There is rarely a best choice. Here we will
describe a typical analysis pipeline (Fig. 1) using QIIME and its
associated packages. Specific command line inputs will vary
Fig. 1 Flow chart showing the main steps in a typical sequence analysis pipeline. Commonly used programs
and QIIME scripts are shown in parentheses. The point where OTU clustering workflows would be included is
shown as a gray box
depending on the format of the input data. Thus, in the interests

of simplicity, we will focus on sequence input files commonly gen-
erated via amplicon sequencing on an Illumina MiSeq instrument.
QIIME (current version is 1.9.1 at the time of writing) can be
obtained through the official web site (http://qiime.org) where a
range of install options is listed for various operating systems (see
Note 3).
2.4 Quality Control Obtain FastQC software from the Babraham Institute web site
(QC) of Raw Data http://www.bioinformatics.babraham.ac.uk/projects/fastqc/
and install on your system. Run the software, then use FileOpen
and navigate to your raw data file (.fastq, BAM or SAM; see Note 4).
When an appropriate file is opened, the software will run automati-
cally and produce graphical output. The output window will show
a list of test modules in the left-hand pane. These will be preceded
by a green icon if the raw data passes QC, an orange icon if the data
shows some abnormalities, and a red icon if the data is particularly
unusual. Orange and/or red icons do not automatically mean that
the data is of poor quality but suggest closer analysis of the raw
data would be valuable (see Note 5).
2.5 Generating
a Mapping File Many QIIME scripts utilize a mapping file to associate sample bar-
codes or metadata with sequences. The simplest way to generate a
mapping file is to use Microsoft Excel and save in tab-delimited
format. The first column must be headed #SampleID, the second
column must be headed BarcodeSequence, the third column
must be headed LinkerPrimerSequence, and the final column
must be headed Description (all without quote marks). Sample
metadata columns can be inserted between the
LinkerPrimerSequence and Description columns and can
include any data relevant to the samples (age, site, treatment,
batch, etc.). An example is shown in Fig. 2. Once a mapping file
has been generated and saved (as a tab-delimited text file), it can be
validated with the QIIME script validate_mapping_file.py. This
will report any errors in the mapping file and will create a corrected
version of the file if required.
2.6 Combining When using the Illumina MiSeq platform, sequencing from both
Paired-End Reads ends of an amplicon (paired-end reads), and combining both reads,
can improve the overall quality score of a sequence (see Note 6).
QIIME offers two sequence pairing algorithms, and inputs are the
forward read, reverse read, and, optionally, the index read (the
index read includes the sample barcodes). To join paired-end reads
and update the index read file to include only successfully paired
reads, employing default parameters, use the following command:
join_paired_ends.py f forward_read.fastq
r reverse_read.fastq b index_read.fastq o
output_directory_path
#SampleID BarcodeSequence LinkerPrimerSequence Diet Sex Birth Description
#Example mapping file from Tannock et al 2013
8884 ACGAGTGCGT GTATTACCGCGGCTGCTGGCAC Cow Male V 01_C
8899 ACGCTCGACA GTATTACCGCGGCTGCTGGCAC Goat Male V 02_G
18127 AGACGCACTC GTATTACCGCGGCTGCTGGCAC Breast Female V 03_B
18155 AGCACTGTAG GTATTACCGCGGCTGCTGGCAC Cow Female C 04_C
18292 ATCAGACACG GTATTACCGCGGCTGCTGGCAC Goat Female V 05_G
18468 ATATCGCGAG GTATTACCGCGGCTGCTGGCAC Cow Female V 06_C
18563 CGTGTCTCTA GTATTACCGCGGCTGCTGGCAC Breast Female C 07_B
18649 CTCGCGTGTC GTATTACCGCGGCTGCTGGCAC Goat Female V 08_G
Fig. 2 Example mapping file to be used within the QIIME pipeline. The first three columns and the final column
must be included with the specified headers. Other columns contain metadata that describe the experiment
and may be used to identify groups of samples during diversity analyses. The file is saved as tab-delimited text
2.7 Sequence Raw sequence data may contain reads from hundreds of samples.
Demultiplexing These reads need to be associated with their sample of origin and
checked for quality. The QIIME scripts split_libraries.py (for data
provided as .fasta and .qual files), split_libraries_fastq.py (for data
provided as .fastq), and multiple_split_libraries_fastq.py (for data
that are demultiplexed prior to return from the sequence provider)
are used to achieve this. Using the example of a MiSeq run, where
reads have been paired, demultiplexing can be achieved with the
following command:
split_libraries_fastq.py -i paired_reads.
fastq -b index_reads.fastq --rev_comp_mapping_
barcodes -m mapping_file.txt -o output_direc-
tory_path -q 19
This command includes a Phred quality threshold set at 19
(-q 19) that ensures retained sequences have high-quality base calls
(in other words, the probability of an incorrect base call is 0.01).
It also allows reverse complementing of barcodes in the mapping
file (--rev_comp_mapping_barcodes) as Illumina index (barcode)
reads are often the reverse complement of those included in the
mapping file (check this with your sequence provider).
2.8 Sequence Clustering of sequences into operational taxonomic units (OTUs)

Clustering is a critical step in the analysis pipeline. A plethora of algorithms
can be employed, and a wide variety of parameters can be adjusted
depending on the biological questions being asked and the quality
of the data set. QIIME offers three main OTU-picking strategies.
Closed-reference OTU clustering compares sequences against a
curated marker gene database (see Note 7) with a defined similarity
threshold. In an environment where the bulk of the microbes likely

to be present have been described previously (such as the human
gut or oral cavity), closed-reference OTU clustering is an appropri-
ate approach [8]. This approach will, however, discard sequences
that do not match the reference database at the stipulated identity
threshold. In contrast, an open-reference approach first utilizes
closed-reference clustering but takes sequences that fail to match
with the reference database and performs de novo clustering with
this sequence subset, thus ensuring all sequences are utilized and
potential rare community members are captured [9] (see Note 8).
Finally, a purely de novo approach can be taken in which reads are
compared with each other and no reference database is used. This
is a preferred method when the environment under study is likely
to have microbes that are poorly defined or absent from current
reference databases (e.g., soil or sediment).
A command for clustering sequences using an open-reference
approach on demultiplexed sequence data is shown below, with
script defaults employed (see Note 9):
pick_open_reference_otus.py -i paired_fil-
tered_reads.fna -r PATH_TO_/refseqs.fna -o
output_folder/ -s 0.1
2.9 Chimera Chimeric sequences include regions from two different organisms
Removal and are commonly generated during amplicon preparation. These
artifactual sequences will not cluster with either parent sequence
and thus can inflate OTU numbers. If a closed-reference clustering
approach is taken, chimeras will automatically be removed, as they
will not match the reference database. If an open-reference cluster-
ing approach is used, chimera removal can reduce OTU numbers
to more biologically relevant levels [10]. Chimera removal is not
included as a default setting in the open-reference OTU-picking
workflow (described above) but can be employed retrospectively.
To do this, use the aligned representative sequences generated dur-
ing open-reference OTU picking (found in the output folder called
pynast_aligned_seqs) and the aligned version of the database used
during open-reference clustering:
identify_chimeric_seqs.py -i rep_set_
aligned.fasta -a reference_set_aligned.fasta
-o chimeric_seqs.txt
Chimeras can be removed from the list of representative
sequences as follows:
filter_fasta.py -f rep_set.fna -o rep_set_
no_chimeras.fna -s chimeric_seqs.txt -n
The OTU table must now be regenerated, removing sequences
identified as chimeric, using the following command:
make_otu_table.py -i final_otu_map.txt -o
otu_table_chimeras_removed.biom -e chimeric_
seqs.txt -t rep_set_tax_assignments.txt
The final_otu_map.txt and rep_set_tax_assignments.txt files

can be found in the output directories generated during open-
reference OTU picking.
2.10 Taxonomy Assignment of taxonomy to OTU representative sequences is

Assignment included as a default step in the open-reference OTU-picking
workflow. When using closed-reference OTU-picking approach,
taxonomy is taken directly from the curated database. In most
cases, the workflow-assigned taxonomy will be sufficient; however,
the behavior of taxonomy assignment can be modified with a
parameters file if required, or alternative taxonomy assignment
methods can be explored using the OTU representative sequence
file and the QIIME script assign_taxonomy.py.
2.11 Generating Taxa The open-reference OTU-picking workflow will generate a file
Summary Tables named otu_table_mc2_w_tax.biom (see Note 10). This file is a
starting point for downstream sequence analyses. A rapid way to
gain insights into the taxonomic structure of each sample is to
generate taxa summary tables. These collapse the original OTU
table at available taxonomic levels (dependent on the database used
for clustering and taxonomy assignment). The following script will
generate tables and plots showing the relative abundance of taxo-
nomic groups for each sample in the mapping file:
summarize_taxa_through_plots.py -i otu_ta-
ble_mc2_w_tax.biom -m Mapping_file.txt -o taxa_
summary
2.12 Measuring Alpha diversity is a measure of diversity within a community [11]

Alpha Diversity and, at its simplest level, provides a quantitative description of the
number of species (or OTUs) in the data set (species richness).
Alpha diversity metrics are often used to generate collectors curves
or rarefaction plots, providing information relating to how exten-
sively a community has been sampled. A wide range of diversity
metrics can be implemented via the script alpha_diversity.py (use
alpha_diversity.py -s to show the available metrics). A more com-
mon approach is to use the alpha_rarefaction.py script. This work-
flow script generates multiple rarefied OTU tables, measures alpha
diversity metrics for each table, collates the results, and returns
collated tables and rarefaction plots (see Note 11). Generation of
rarefaction plots and tables, using default metrics and cutoff values,
can be achieved with the following command:
alpha_rarefaction.py -i otu_table_mc2_w_
tax.biom -o alpha_rarefaction/ -t rep_set.tre
-m Mapping_file.txt
2.13 Measuring Beta Beta diversity provides insights into qualitative differences between
Diversity communities [11]. For example, comparisons can be made between
subjects in different treatment groups or across time series. Some
metrics include phylogenetic information, with or without abun-

dance weightings, while others do not require phylogeny. By
default QIIME utilizes weighted and unweighted UniFrac dis-
tances (a phylogenetic metric) [12]. The list of available beta diver-
sity metrics can be called using the command beta_diversity.py -s.
The workflow script beta_diversity_through_plots.py generates a
randomly and evenly subsampled OTU table, measures beta diver-
sity distances between samples, performs principle coordinates
analyses (PCA), and generates plots (see Note 12). For phylogeny-
based metrics, a tree file must also be supplied. To generate prin-
ciple coordinates plots using default settings, run the following
command:
beta_diversity_through_plots.py -i otu_ta-
ble_mc2_w_tax.biom -o beta_diversity/ -t rep_
set.tre -m Mapping_file.txt -e 100
2.14 Concluding While taxonomic profiling of microbial communities is a valuable

Remarks research tool, it has limitations. In order to correctly interpret
community structure and diversity data, it is important to gain an
understanding of the technology and tools, both sequencing and
bioinformatics, that lead to generation of the final data set. All
steps in the path, from experimental design through sequencing to
data analysis, offer a range of options, and this chapter provides a
start point from which researchers can progress.
3 Notes
1. The 16S rRNA gene V4 hypervariable region is a good default

choice for two main reasons. The primers employed by the
HMP and EMP groups allow identification of a wide range of
prokaryotes and archaea. The amplicon generated is relatively
short (approximately 250 bp), thus allowing high-quality reads
with low error rates across the entire region.
2. The most widely used analysis pipelines incorporate a host of
requisite-affiliated software packages (called dependencies)
and currently only run via the command line (i.e., manual text-
based commands). Thus, the operator must have at least some
familiarity with command line inputs and outputs. A useful
place to start, if you have no previous command line experi-
ence, is at linuxcommand.org (http://linuxcommand.org/
lc3_learning_the_shell.php).
3. The installation and operation of bioinformatics software can
be challenging to a computing novice. A local information
technology expert may be a first port of call; however, the
QIIME developers provide an excellent forum service
(https://groups.google.com/forum/#!forum/qiime-forum)
where installation questions and/or script queries will be

promptly answered. The QIIME web site also has extensive
help and tutorial options (https://github.com/biocore/
qiime/wiki/Getting-help).
4. Data may be provided in a variety of formats from different
sequence providers. One of the most useful raw data formats is
the .fastq file, which combines nucleotide sequence and
sequence quality data in a single file and can be used directly as
input to the QIIME pipeline.
5. The FastQC Help section provides detailed descriptions of
each analysis module and common reasons why data may raise
warning (or fail) flags. For amplicon sequencing, one of the
most valuable outputs is the per base sequence quality mod-
ule. This can inform the user regarding a sensible truncation
point should read quality not extend to the end of the ampli-
con or if paired-end reads are not being utilized.
6. As an example, when sequencing a 16S rRNA gene V4 ampli-
con, the full amplicon length is approximately 253 bp. Using
a 250-base paired-end sequencing approach will generate
nearly full-length reads from both directions. Generally speak-
ing, sequence quality drops off toward the end of a read, and
this is especially true for the second (or reverse) read. The
high sequence quality at the start of both forward and reverse
reads can be used to correct any poor-quality bases toward the
end of both reads, thus generating a combined high-quality
sequence. The approach can also be used to obtain full cover-
age of an amplicon that is longer than the read chemistry.
For example, the V3-V4 amplicon is approximately 470 bp.
Using a 300-base paired-end approach, the combined forward
and reverse reads cover the entire amplicon with an overlap
of 130 bp.
7. Databases should be considered as live resources, and the
latest version available should be used. QIIME supports the
Greengenes [13] database (current version is 13.8) and also
has links to the more recently updated SILVA [14] database
(current supported version is release 119).
8. Open-reference sequence clustering is the option currently
favored by the QIIME developers. This approach can be han-
dled by a modern desktop computer in a number of hours (for
a typical Illumina MiSeq sequence run). Be aware that this
approach can generate thousands of OTUs per run, often con-
siderably higher than expected from the environment being
investigated. Several alternative clustering algorithms, claiming
to generate more biologically relevant numbers of OTUs may
be utilized outside of the QIIME environment. A particularly
good example is the UPARSE pipeline [15], details of which
can be obtained from Robert Edgars web site (http://www.

drive5.com/uparse/).
9. OTU clustering is a particularly complex operation, and the
script described is a workflow and incorporates clustering,
choice of a representative sequence for each cluster, generation
of an OTU map (linking individual sequences with OTU num-
bers), generation of an OTU table, and building of phyloge-
netic trees. Customizing each of these steps is possible by
providing a parameters file (http://qiime.org/documenta-
tion/qiime_parameters_files.html). While default settings are
designed to work well for the majority of situations, it is highly
advisable to gain an understanding of how each step functions
and when changing default parameters may be advantageous.
10. QIIME generates OTU tables in BIOM format [16]. This is
a machine-readable format and needs to be converted to a tab-
delimited text format for viewing with programs such as
Microsoft Excel. BIOM is loaded as a default package with
QIIME, and instructions for file conversion can be found at
the BIOM web site. http://biom-format.org/documenta-
tion/biom_conversion.html.
11. Sequencing depth can vary considerably between samples in a
single sequence run. The heterogeneity in sequencing depth
can be controlled by normalizing or rarefying (the argument
for one or the other is ongoing and complex [17, 18]). Choice
of rarefaction depth will impact the plots produced through
both the alpha and beta diversity workflows and should be
carefully considered. One approach is to obtain a summary of
the OTU table and determine the spread of sequencing depth
across samples. Bear in mind that samples with fewer sequences
than the chosen rarefaction depth will be excluded from beta
diversity plots.
12. Principle coordinates analysis (PCA) plots are visualized via a
web-based program called Emperor. The QIIME developers rec-
ommend using the Google Chrome browser to visualize beta
diversity plots via Emperor. Also note that exporting publication
quality images from Emperor can be difficult, and exports may
require re-annotation in another software package.
References
1. Peterson J, Garges S, Giovanni M, McInnes P, Peavy H, Pontzer C, Portnoy M, Sayre MH,
Wang L, Schloss JA, Bonazzi V, McEwen JE, Starke-Reed P, Zakhari S, Read J, Watson B,
Wetterstrand KA, Deal C, Baker CC, Di Guyer M (2009) The NIH human microbiome
Francesco V, Howcroft TK, Karp RW, Lunsford project. Genome Res 19:23172323
RD, Wellington CR, Belachew T, Wright M, 2. Qin J, Li R, Raes J, Arumugam M, Burgdorf
Giblin C, David H, Mills M, Salomon R, KS, Manichanh C, Nielsen T, Pons N, Levenez
Mullins C, Akolkar B, Begg L, Davis C, F, Yamada T, Mende DR, Li J, Xu J, Li S, Li D,
Grandison L, Humble M, Khalsa J, Little AR, Cao J, Wang B, Liang H, Zheng H, Xie Y, Tap
J, Lepage P, Bertalan M, Batto J-M, Hansen T, 9. Rideout JR, He Y, Navas-Molina JA, Walters
Le Paslier D, Linneberg A, Nielsen HB, WA, Ursell LK, Gibbons SM, Chase J, McDonald
Pelletier E, Renault P, Sicheritz-Ponten T, D, Gonzalez A, Robbins-Pianka A, Clemente
Turner K, Zhu H, Yu C, Li S, Jian M, Zhou Y, JC, Gilbert JA, Huse SM, Zhou H-W, Knight R,
Li Y, Zhang X, Li S, Qin N, Yang H, Wang J, Caporaso JG (2014) Subsampled open-refer-
Brunak S, Dor J, Guarner F, Kristiansen K, ence clustering creates consistent, comprehen-
Pedersen O, Parkhill J, Weissenbach J, Bork P, sive OTU definitions and scales to billions of
Ehrlich SD, Wang J (2010) A human gut sequences. PeerJ 2, e545
microbial gene catalogue established by 10. Edgar RC, Haas BJ, Clemente JC, Quince C,
metagenomic sequencing. Nature 464:5965 Knight R (2011) UCHIME improves sensitiv-
3. Gilbert JA, Meyer F, Jansson J, Gordon J, Pace ity and speed of chimera detection.
N, Tiedje J, Ley R, Fierer N, Field D, Kyrpides Bioinformatics 27:21942200
N, Glckner F-O, Klenk H-P, Wommack KE, 11. Lozupone CA, Knight R (2008) Species diver-
Glass E, Docherty K, Gallery R, Stevens R, gence and the measurement of microbial diver-
Knight R (2010) The Earth Microbiome sity. FEMS Microbiol Rev 32:557578
Project: meeting report of the 1 EMP meet- 12. Lozupone C, Hamady M, Knight R (2006)
ing on sample selection and acquisition at UniFracan online tool for comparing micro-
Argonne National Laboratory October 6 2010. bial community diversity in a phylogenetic
Stand Genomic Sci 3:249253 context. BMC Bioinformatics 7:371
4. Caporaso JG, Kuczynski J, Stombaugh J, 13. McDonald D, Price MN, Goodrich J, Nawrocki
Bittinger K, Bushman FD, Costello EK, Fierer EP, DeSantis TZ, Probst A, Andersen GL,
N, Pea AG, Goodrich JK, Gordon JI, Huttley Knight R, Hugenholtz P (2012) An improved
GA, Kelley ST, Knights D, Koenig JE, Ley RE, Greengenes taxonomy with explicit ranks for
Lozupone CA, McDonald D, Muegge BD, ecological and evolutionary analyses of bacteria
Pirrung M, Reeder J, Sevinsky JR, Turnbaugh and archaea. ISME J 6:610618
PJ, Walters WA, Widmann J, Yatsunenko T,
Zaneveld J, Knight R (2010) QIIME allows 14. Quast C, Pruesse E, Yilmaz P, Gerken J,
analysis of high-throughput community Schweer T, Yarza P, Peplies J, Glckner FO
sequencing data. Nat Methods 7:335336 (2013) The SILVA ribosomal RNA gene data-
base project: improved data processing and
5. Schloss PD, Westcott SL, Ryabin T, Hall JR, web-based tools. Nucleic Acids Res
Hartmann M, Hollister EB, Lesniewski RA, 41:D590D596
Oakley BB, Parks DH, Robinson CJ, Sahl JW,
Stres B, Thallinger GG, Van Horn DJ, Weber 15. Edgar RC (2013) UPARSE: highly accurate
CF (2009) Introducing mothur: open-source, OTU sequences from microbial amplicon
platform-independent, community-supported reads. Nat Methods 10:996998
software for describing and comparing micro- 16. McDonald D, Clemente JC, Kuczynski J,
bial communities. Appl Environ Microbiol Rideout JR, Stombaugh J, Wendel D, Wilke A,
75:75377541 Huse S, Hufnagle J, Meyer F, Knight R,
6. Schloss PD (2010) The effects of alignment Caporaso JG (2012) The biological observa-
quality, distance calculation method, sequence tion matrix (BIOM) format or: how I learned
filtering, and region on the analysis of 16S to stop worrying and love the ome-ome.
rRNA gene-based studies. PLoS Comput Biol Gigascience 1:7
6, e1000844 17. McMurdie PJ, Holmes S (2014) Waste not,
7. Kim M, Morrison M, Yu Z (2011) Evaluation want not: why rarefying microbiome data is
of different partial 16S rRNA gene sequence inadmissible. PLoS Comput Biol 10:
regions for phylogenetic analysis of microbi- e1003531
omes. J Microbiol Methods 84:8187 18. Weiss SJ, Xu Z, Amir A, Peddada S, Bittinger K,
8. He Y, Caporaso JG, Jiang X-T, Sheng H-F, Gonzalez A, Lozupone C, Zaneveld JR, Vazquez-
Huse SM, Rideout JR, Edgar RC, Kopylova E, Baeza Y, Birmingham A, Knight R (2015) Effects
Walters WA, Knight R, Zhou H-W (2015) of library size variance, sparsity, and composition-
Stability of operational taxonomic units: an ality on the analysis of microbiome data. PeerJ
important but neglected property for analyzing Prepr 3:e1408. Doi: org/10.7287/peerj.
microbial diversity. Microbiome 3:20 preprints.1157v1
Chapter 10
Adhesion of Yeast and Bacteria to Oral Surfaces

Richard D. Cannon, Karl M. Lyons, Kenneth Chong,
Kathryn Newsham-West, Kyoko Niimi, and Ann R. Holmes
Abstract
Colonization of surfaces in the human body by microorganisms is an early, essential, step in the initiation
of infectious disease. We have developed in vitro assays to investigate interactions between yeast or bacte-
rial cells and human tissues, fluids, or prostheses. Such assays can be used to identify the adhesins, ligands,
and receptors involved in these interactions, for example, by determining which components of the
microbe or human tissue/fluid interfere with adherence in the assay. The assays can also be applied to find-
ing ways of preventing adhesion, and subsequent disease, by investigating the effects of different condi-
tions and added compounds on adherence in the in vitro assays.
Here we describe assays for measuring adhesion of the oral yeast Candida albicans, a common com-
mensal and opportunistic pathogen, or the bacterium Staphylococcus epidermidis, which is not normally
pathogenic but is known to form biofilms on medical prostheses. The assays described belong to two
approaches to investigating adhesion and biofilm formation: (1) retention at a fixed time point following
liquid washes and (2) retention against a continuous flow of medium.
Key words Candida albicans, Staphylococcus epidermidis, Adhesion, Biofilm, Saliva, Colonization,
Epithelial cells, Silicone, Hydroxyapatite, Polymethyl methacrylate
1 Introduction
The oral cavity provides many surfaces and niches which are
colonized by a variety of microorganisms that form complex bio-
films [1]. While these microbial communities and biofilms are
often nonpathogenic, certain microorganisms play important roles
in oral diseases such as dental caries, periodontitis, and oral candi-
dosis. It is therefore important to investigate the multiple micro-
bial/host adhesion interactions within the oral cavity, such as those
that promote development of the oral biofilm dental plaque and
those that facilitate colonization by pathogenic microorganisms
implicated in oral diseases [2]. In some instances it may be possible
to preclude diseases by inhibiting microbial adhesion and preventing
colonization. Many bacteria, fungi, and viruses adhere to host tissues
and other substrates by lectin-like interactions, and carbohydrates
165
166 Richard D. Cannon et al.
are under development as drugs to interfere with the microbial-host

interaction [3, 4]. Candida albicans is a commensal yeast normally
present in the oral cavity in low numbers relative to bacterial
species. It is also an opportunistic pathogen that causes oral candi-
dosis, and interfering with its adhesion has the potential to prevent
such infections [5]. Probiotics, which are used to reduce or pre-
vent colonization by pathogenic microbes, can act by inhibiting
adhesion [6].
In order to develop novel antimicrobial strategies that prevent
or reduce adhesion, microbial adhesion must be assessed and mon-
itored using appropriate assays. In this chapter, we describe the
assays used in our laboratory to measure the adhesion of the oral
yeast C. albicans and the skin commensal bacterium Staphylococcus
epidermidis to different model surfaces that represent those found
in the oral environment. Our assays use radiolabeling of the yeast/
bacteria, or microscopy, to detect and quantify the interactions
with various substrates. Although we focus on two species, the
assays described are applicable to studies of the adhesion of other
oral microorganisms.
There are two main categories of adhesion assays: one mea-
sures microbial retention at a fixed time point following liquid
washes; the other measures adhesion and retention against con-
tinuous liquid flow in real time. We describe examples of both. An
important aspect of both assays is the nature of the force challeng-
ing the retention of the microbe on, or with, the substrate: in the
fixed time point static assay, this is the wash condition before adhe-
sion is measured; for the flow assay, it is the shear force applied as
adhesion is measured. Selection of the type of assay to be used
depends on the in vivo system being modeled. In the oral cavity,
where initial adherence is often a matter of retention on a saliva-
coated surface, both types of assay have relevance. The use of a
parallel plate flow chamber with continuous liquid flow [7] has the
advantage that adhesion can be studied quantitatively with in situ,
real-time, observation, without the use of radioactive compounds.
It is also possible to control the hydrodynamic conditions in terms
of shear rate, flow velocity, and Reynolds number, which deter-
mine the mechanism of mass transport. In addition, the passage of
the substratum through air-liquid interfaces for rinsing, fixation,
and/or staining purposes can be avoided in continuous flow sys-
tems. The non-flow systems have the advantages of not requiring
complex equipment and allowing multiple assays in small volumes,
yet they remain relevant to in vivo conditions.
In general, radiolabeling is our first choice for detecting or
quantifying adhesion. This approach provides high sensitivity and
reproducibility compared to quantifying attachment by either
microscopy or viable cell count. In many cases assays can be
designed for a microtiter plate format to enable simplification of
the protocol and to reduce reagent use. However, assays based on
Adhesion of Yeast and Bacteria to Oral Surfaces 167
radiolabeling require specialized handling and equipment, and the

final assay described uses crystal violet (CV) staining to quantify
C. albicans biofilm formation on dental acrylic. CV staining is less
sensitive than radiolabeling but has the advantages of simplicity
and low cost [8] for quantifying the greater cell mass found in bio-
film assays.
We describe several representative assays of microbial adhesion
developed in our laboratory. The assays measure adhesion of C.
albicans in static assays to the following substrates: (1) saliva-coated
hydroxyapatite (HA) [9], (2) immobilized saliva proteins (C. albi-
cans overlay assay) [10, 11], (3) epithelial cells following saliva
treatment of the yeast cells [12], and (4) saliva-coated medical
grade silicone [13] or saliva-coated denture prosthetic material
(Holmes, Lyons et al. unpublished data). Other assays measure
adhesion of S. epidermidis to denture prosthetic material under (1)
static or (2) flow conditions (Lyons et al. unpublished data) and C.
albicans biofilm formation on saliva-coated denture acrylic
(Newsham-West et al. unpublished data).
2 Materials
2.1 Radiolabeling 1. Equipment (additional to usual laboratory equipment): scintil-

of Yeast and Bacterial lation counter with a microplate and microfuge tube capability
Cells and Cell Culture (see Note 1). Anaerobic jar for the cultivation of bacteria, and
GasPak EZ anaerobic container system (Becton Dickinson &
Co [BD], Franklin Lakes, NJ, USA).
2. Radiochemicals (see Notes 2 and 3): (a) 35S-methionine: Yeast
cells are labeled with 35S-methionine (EasyTag L-[35S]-
Methionine, stabilized aqueous solution; PerkinElmer Life and
Analytical Sciences, Inc. Waltham, MA, USA). (b)
3
H-thymidine: Bacterial cells are labeled with 3H-thymidine
([methyl-3H] Thymidine, aqueous solution, sterilized; GE
Healthcare UK Ltd, Chalfont St Giles, UK). Both are stored
refrigerated. Concentrations of radiochemicals required for
each adherence assay vary and are detailed in Subheading 3.
3. Yeast strains. For most assays, several yeast species and strains,
including laboratory strains as well as clinical isolates, have
been used successfully. The most commonly used strain in our
laboratory is C. albicans ATCC 10261 (American Type Culture
Collection, Manassas, VA, USA). The materials and methods
described are applicable for other yeast species.
4. Media and buffers. All media and buffers are made with water
purified by reverse osmosis with a resistivity of 1015 M cm.
5. Yeast culture media: (a) Yeast extract peptone dextrose (YPD)
liquid medium is used to prepare the strains for storage and
contains (per liter) 10 g yeast extract (BD), 20 g Bacto Peptone
(BD), and 20 g glucose (final concentration 111 mM). For

agar plates, 20 g agar is added. (b) Glucose salts biotin (GSB)
liquid medium is used to prepare inocula and contains per liter:
1 g (NH4)2SO4 (7.57 mM), 2 g KH2PO4 (14.7 mM), 50 mg
MgSO47H2O (0.2 mM), 50 mg CaCl22H2O (0.34 mM),
0.05 mg biotin, and 20 g glucose (111 mM) (see Note 4).
6. Bacterial strains. For the assays described, we used S. epidermi-
dis strains ATCC 12228, ATCC 14990, and ATCC 49134 and
clinical isolates from the oronasal cavity of an individual fol-
lowing a maxillary resection. However, we have successfully
used strains of other bacterial species, such as Staphylococcus
aureus and various oral streptococcal species, in similar adhe-
sion assays, and the method of radiolabeling the bacterial cells
is the same.
7. Bacterial culture medium: (a) Brain heart infusion-yeast extract
(BHY) liquid medium contains (per liter) 37 g brain heart
infusion (BD) and 5 g yeast extract (BD); (b) tryptone yeast
extract (TY) medium: dissolve the following in 987.5 mL
water, 5 g tryptone, 5 g yeast extract, 4 g K2HPO4, autoclave,
then add 12.5 mL filter-sterilized 40 % (w/v) glucose (final
concentrations in medium: K2HPO4: 23 mM; glucose:
27.8 mM); (c) TY-B consists of sterile TY and BHY mixed
aseptically 20:1 (v/v).
2.2 Blot Overlay 1. Equipment (additional to usual laboratory equipment) that

Assay to Demonstrate may be required: Polyacrylamide gel electrophoresis (PAGE)
Adhesion of Yeast apparatus and electro-transfer (Western blotting) apparatus
Cells to Immobilized (see Note 5); end-over-end mixer; access to dark room facili-
Proteins ties; and photographic equipment and chemicals.
2. Yeast strains and media: C. albicans ATCC 10261 is grown in
GSB and labeled with 35S-methionine as described in
Subheading 3.1.
3. PAGE separation of proteins: Materials required include (1)
separating buffer: 18.2 g Tris (final concentration 1.5 M),
0.4 g SDS (final concentration 14 mM) per 100 mL water;
adjust to pH 8.8 with HCl. Filter solution through Whatman
No. 1 paper and store at 4 C; (2) stacking buffer: 6.1 g Tris
(final concentration 0.5 M), 0.4 g SDS (final concentration
14 mM) in 100 mL water; adjust pH to 6.8 with HCl; filter
solution through Whatman No. 1 paper and store at 4 C; (3)
acrylamide-Bis solution (40 %, 37.5:1; Bio-Rad, Hercules, CA,
USA) (hazardsee Note 6); (4) Tetramethylethylenediamine
(TEMED; Bio-Rad); (5) 10 % ammonium persulfate (see Note 7);
(6) loading dye (contains in 10 mL: 50 mg bromophenol blue,
3 mL water, 10 L 1 M NaOH, and 7 mL glycerol); (7) sample
buffer (10 mL stock contains: 2.5 mL stacking gel buffer,
0.2 mL 2-mercaptoethanol, 1 mL 20 % (w/v) SDS, and 6.3 mL
water), store as frozen aliquots; (8) Bradford protein assay kit

(Bio-Rad); (9) pre-stained protein markers (see Note 8) of size
range appropriate for proteins to be separated (Invitrogen,
Carlsbad, CA, USA); (10) electrophoresis running buffer
(contains per liter: 14.4 g glycine (192 mM), 3.03 g Tris
(25 mM); 1 g SDS (3.5 mM)); (11) Coomassie Blue protein
stain, contains per liter: 0.2 g Coomassie Blue R250, 400 mL
methanol, 500 mL water, and 100 mL glacial acetic acid; (12)
destain solution, contains per liter: 200 mL methanol, 700 mL
water, 100 mL glacial acetic acid.
4. Electroblotting of PAGE-separated proteins and blot overlay:
Materials required include (1) nitrocellulose membrane
(Hybond ECL, GE Healthcare, UK); (2) transfer buffer (con-
tains per liter: 3.03 g Tris (25 mM), 14.4 g glycine (192 mM),
200 mL methanol (20 % v/v) (see Note 9); (3) KCl buffer
(1 L) made up of solution A containing in 800 ml water: 0.27 g
KH2PO4 (final concentration in complete buffer 2 mM),
0.46 g K2HPO4 (final concentration in complete buffer
2.6 mM), 0.373 g KCl (final concentration in complete buffer
5 mM), pH 6.5, and solution B containing 0.15 g CaCl22H2O
(final concentration in complete buffer 1 mM) in 200 mL
water; solutions A and B are autoclaved separately before mix-
ing; (4) blocking buffer: 5 % (w/v) bovine serum albumin
(BSA, Sigma-Aldrich, St Louis, MO, USA) in KCl buffer; (5)
blot staining solution: 0.2 % Ponceau S (Sigma-Aldrich) in 1 %
(v/v) acetic acid; (6) phosphate-buffered saline (PBS) contains
per liter: 8.5 g NaCl (145.5 mM), 0.3 g KH2PO4 (2.2 mM),
0.6 g Na2HPO4 pH 7.5; (4.2 mM); (7) Tween 20 (Bio-Rad).
5. Detection of radioactivity by autoradiography: Dried mem-
branes are incubated with X-ray film (e.g., Amersham Hyperfilm
max or similar film sensitive to 35S radiation).
2.3 Adhesion of 1. Equipment (additional to usual laboratory equipment) that may

C. albicans Cells be required: scintillation counter with microfuge tube capability.
to Saliva-Coated 2. Yeast strains and media: C. albicans ATCC 10261 is grown in
Hydroxyapatite GSB and labeled with 35S-methionine as described in
Subheading 3.1.
3. Buffers: KCl buffer (see Subheading 2.2, item 4).
4. Saliva: Whole saliva (approximately 10 mL, unstimulated) is
collected on ice from five donors, and an equal amount from
each donor is pooled. The pooled saliva sample is clarified by
centrifugation at 12,000 g for 15 min, and the supernatant
is mixed with an equal volume of KCl buffer. Proteinase
inhibitors are added (use a commercially available cocktail
such as SigmaFAST [Sigma-Aldrich]). Careinfectious disease
hazardtake standard precautions for handling and disposal
of human biological material.
5. Hydroxyapatite beads: Portions (12 mg) of hydroxyapatite

beads (Bio-Rad) are hydrated prior to use in the adhesion assay
by static incubation in 1.5 mL microfuge tubes in 0.5 mL KCl
buffer at 4 C for 16 h.
6. Scintillation fluid (e.g., Optiphase 3, PerkinElmer).
2.4 Adhesion 1. Equipment (additional to usual laboratory equipment) that

of Saliva-Treated may be required: biohazard hood with sterile air flow; inverted
C. albicans Cells microscope; hemocytometer; CO2 incubator; confocal micro-
to Epithelial Cells scope (see Note 10); 12 channel multichannel pipette; scintil-
lation counter with a microplate capability.
2. Yeast strains and media: C. albicans ATCC 10261 is grown in
GSB and labeled with 35S-methionine as described in
Subheading 3.1.
3. Epithelial cells and culture media: Human epithelial cell lines
HEp-2 and A549 were obtained from the European Collection
of Cell Cultures, Centre for Applied Microbiological Research,
Salisbury, UK. Cells are grown in Eagles MEM medium
(Invitrogen) supplemented with 10 % fetal calf serum (FCS;
Invitrogen) and 1 % glutamine (recently thawed commercially
available solution: 200 mM, Sigma-Aldrich, stored at 20 C
as 100 stock). Confluent cells are prepared for subculture
with trypsin-EDTA (commercially available solution contain-
ing 0.05 % trypsin and 0.53 mM EDTA: [Invitrogen], stored
at 20 C).
4. Saliva samples are prepared as described above (see
Subheading 2.3, item 4).
5. Sterile tissue culture flasks for tissue culture maintenance:
Nunclon cell culture flasks, area 80 cm2 (Nunc, Thermo
Fisher Scientific, Roskilde, Denmark).
6. Sterile 96-well flat bottom microtiter culture plates with lids
for adherence assays with radiolabeled cells (Nunc).
7. Sterile 24-well flat bottom culture plates with lids (Nunc) for
confocal microscopy with fluorescein isothiocyanate (FITC)-
labeled cells.
8. Buffers: (1) Artificial saliva buffer (ASB) constituted to mimic
the ionic composition of saliva contained per liter1.36 g
KH2PO4 (10 mM), 0.29 g CaCl22H2O (2 mM), 1.36 g KCl
(18.2 mM), 0.028 g NaSCN (0.35 mM), 0.63 mg NaF
(15 M), 2.18 g NaHCO3 (26 mM); pH6.5), and glucose
(1 g/L (5.6 mM)); (2) PBS; (3) carbonate/bicarbonate buffer
pH 9.5 (contains per liter: 3.18 g Na2CO3 (30 mM), 5.88 g
NaHCO3 (70 mM); see Note 11).

of C. albicans or may be required: scintillation counter with a microfuge tube
S. epidermidis Cells capability; microtome.
to Saliva-Coated 2. Yeast strains and media: C. albicans ATCC 10261 is grown in
Medical Grade Silicone GSB and labeled with 35S-methionine as described in
or to Denture Subheading 3.1.
Prosthetic Materials 3. Bacterial strains and medium: S. epidermidis is grown in TY-B
(Subheading 2.1, item 7) and radiolabeled with 3H-thymidine
as described in Subheading 3.1.
4. Silicone: Medical grade silicone is cut from standard, single
consistency silicone blocks (Silimed Silastic, InterMed Medical
Ltd, Auckland, New Zealand) into 3 cm 3 cm 1 cm blocks
which are fixed onto plastic tissue embedding cassettes (Tissue-
Tek Uni-Cassette, Sakura Finetek USA, Inc., Torrance, CA,
USA) using Araldite Ultra Clear epoxy adhesive (Selleys Pty.
Ltd., Padstow, NSW, Australia). The fixed silicone block is
sliced (1 cm 3 cm) to a thickness of 300 m using a micro-
tome. The silicone slices are then cut by hand with sharp blade
to make three 1 cm 1 cm 300 m coupons from each slice
(see Note 12).
5. Denture prosthetic materials: Commonly used denture pros-
thetic materials, which are also used to make interim and final
obturator denture prostheses, have been tested in our labora-
tory using this assay. They include heat polymerized poly-
methyl methacrylate (Vertex regular and Vertex Implacryl,
Vertex Dental, Zeist, the Netherlands); chairside cold-cure
polymethyl methacrylate (Kooliner, GC America Inc., IL,
USA; Vertex Castapress, Vertex Dental; and New Rimseal,
Bosworth Company, IL, USA); tissue conditioners (Viscogel,
Dentsply; Softliner, GC America Inc.); silicone-based chairside
lining material (Silagum, DMG, Hamburg, Germany); and
laboratory-polymerized silicone (Molloplast-B, Bolton Dental
Manufacturing, Ontario, Canada). Coupons of these materials
are made using a gypsum mold with 20 patterns measuring
5 2 mm. Where appropriate, the materials are polymerized in
the molds according to the manufacturers recommendations.
After removal of the coupons from the gypsum mold, excess
material is removed using a laboratory handpiece (KaVo EWL
K9; KaVo Dental GmbH, Biberach, Germany) and a tapered
carbide bur (GEBR Brasseler GmbH & Co KG, Lemgo,
Germany) prior to adhesion testing.
6. Saliva and saliva wash: Saliva is prepared as described in
Subheading 2.3, item 4; saliva wash (see Note 13) is prepared
by rinsing the mouth for 30 s with food grade water (20 mL)
before collection into a 50 mL Falcon tube.

of S. epidermidis Cells may be required: parallel plate flow chamber and inverted
to Denture Prosthetic phase microscope (see Notes 14 and 15).
Materials Under Flow 2. Bacterial strain: see Subheading 2.1, item 6.
Conditions 3. Flow adhesion buffer (FAB) contains per liter: 3.73 g KCl
(50 mM), 0.17 g K2HPO4 (1.0 mM), 0.14 g KH2PO4
(1.0 mM), 0.14 g CaCl2 (anhydrous) (1.26 mM); pH 6.8.
4. Gypsum molds.
5. Standard glass microscope slides (75 mm 25 mm 1 mm and
75 mm 25 mm 0.8 mm).
6. Vaseline.
7. Dental yellow stone.
8. Dual-channel variable speed pump (Manostat Vera varistaltic
pump, Cole-Parmer, Vernon Hills, IL, USA).
9. Tubing, 0.89 mm diameter (platinum-cured silicone tubing,
Cole-Parmer).
10. Digital SLR camera (Nikon D70S digital SLR camera, Nikon
Corporation, Tokyo, Japan) with microscope mountings.
11. Image analysis software, Fovea Pro 4 (Reindeer Graphics, Inc.,
Asheville, NC, USA), which operates in the Adobe Photoshop
CS2 environment on an Apple Macintosh computer.
2.7 C. albicans 1. Equipment (additional to usual laboratory equipment) that

Biofilm Formation may be required: Synergy 2 (Biotek) microtiter plate reader;
on Denture Acrylic shaking incubator (Bio-Line, Edwards Instrument Company,
Flag Strips Suspended Narellan, NSW, Australia); spectrophotometer (Ultrospec
in Microtiter Wells 6300pro, Amersham Biosciences, NJ, USA); denture vacuum
flask; and plaster impression materials for molding and heat
curing acrylic substrates.
2. Yeast strain and media: C. albicans ATCC 10261 is used for all
biofilm experiments, and cells for initial inocula are grown on
yeast extract peptone dextrose (YPD) agar plates as described
in Subheading 2.1, item 5. Inocula for biofilm experiments are
prepared in Roswell Park Memorial Institute formulation
(RPMI) 1640 medium (autoclavable RPMI 1640; Sigma-
Aldrich) supplemented with 0.2 % glucose and 0.165 M
3-(N-morpholino) propanesulfonic acid (MOPS) as a buffer-
ing agent. The pH is adjusted to pH 7.0 using NaOH prior to
making the required volume and autoclaving. For biofilm for-
mation, the RPMI medium is supplemented with additional
(1.8 % w/v) glucose.
3. Plaster mold: Negative plaster molds are fabricated from Victor
dental plaster (Henry Schein New Zealand, Auckland, New
Zealand) in metal investment flasks.
4. Acrylic material: The substrate for biofilm formation is fabricated

in heat polymerized polymethyl methacrylate (Vertex
Regular Crystal Clear dental acrylic; Vertex Dental B.V., Zeist,
the Netherlands) (see Note 16).
5. Crystal violet: 0.5 % (w/v) crystal violet (Sigma-Aldrich) is dis-
solved in water, filtered through a 0.44 m filter, and diluted
with distilled water to a final concentration of 0.1 % (w/v).
6. Acetic acid: 10 % (v/v: 5 mL glacial acetic acid (Merck,
Darmstadt, Germany) in 45 mL distilled water).
7. Saliva: Whole saliva is collected, processed, and diluted in an
equal volume of KCl buffer (Subheading 2.2, item 4) as
described in Subheading 2.3, item 4.
3 Methods
3.1 Radiolabeling Yeast and bacteria for use in adhesion experiments are conveniently
of Yeast and Bacterial stored as glycerol stocks at 80 C. These cells can then be used to
Cells, and Cell Culture inoculate medium for growth of cells and for subsequent
radiolabeling.
3.1.1 To Prepare Inocula 1. Yeast cells are streaked on YPD agar plates and incubated at
for Pre-culture of Yeast or 30 C for 24 h. Bacterial cells are streaked on BHY agar plates
Bacteria and incubated in an anaerobic jar at 37 C for 24 h.
2. Cells are removed from the surfaces of the YPD or BHY agar
plates and resuspended in YPD or BHY containing 15 % (v/v)
glycerol (approximately 1 mL which will provide enough inoc-
ula for at least 20 experiments), respectively, and stored at
80 C in 50 L volumes in microfuge tubes.
3. Yeast or bacterial cells for inocula are stored at 80 C in YPD
or BHY containing 1015 % (v/v) glycerol, respectively, in
50 L volumes in microfuge tubes.
4. Yeast inocula are tested by inoculating 50 mL GSB or YPD in
250 mL sterile conical flasks with 510 L inoculum (diluted
in sterile medium if required) and measuring the growth of
cells (optical density at 540 nm [OD540] using a pre-calibrated
spectrophotometer (see Note 17)) during incubation at 30 C
with shaking (200 rpm).
5. The inoculum volume used can be adjusted to get the required
level of growth in a certain volume of medium after a particular
incubation period.
6. Bacterial inocula are tested by inoculating 1.5 mL BHY in a
microfuge tube with 15 L inoculum (diluted in sterile
medium if required) and measuring the growth of cells (optical
density at 600 nm [OD600]) during static incubation at 37 C.
7. The inoculum volume used can be adjusted to get the required

level of growth in a certain volume of medium after a particular
incubation period.
3.1.2 Preparation 1. Cell cultures (50 mL, in 250 mL flask) are inoculated and
of C. albicans Cells grown in GSB at 30 C with shaking for approximately 16 h to
Radioactively Labeled a cell concentration of approximately 2.0 106 cells per mL, as
with 35S-Methionine determined by measurement of OD540 using a spectrophotom-
eter and reference to a standard curve (see Note 18).
2. 35
S-methionine (2 L, 20 Ci; 1175 Ci mmol1; see Note 19)
is added to the flask which is incubated at 30 C with shaking
for a further 2 h to allow incorporation of the radiolabel into
the cells.
3. The cells are harvested by centrifugation (1500 g, 5 min) and
washed twice in 10 mL of adhesion assay buffer (e.g., KCl buf-
fer), and following a determination of cell concentration by
measuring the OD540 of a suitable dilution of cells, the cells are
resuspended to the final cell concentration specified for the
particular assay.
3.1.3 Preparation 1. Cell cultures (3 mL; two full microfuge tubes) are grown at
of S. epidermidis Cells 37 C for 16 h in TY medium containing [methyl-3H]-thymi-
Radioactively Labeled dine (10 L, 10 Ci, added to each tube; 85 Ci mmol1).
with 3H-Thymidine 2. Cells are harvested by centrifugation (4000 g, 5 min) and
washed twice in KCl buffer by centrifugation before resus-
pending at the final cell concentration specified for the particu-
lar assay (determined by OD600 measurement, as described
above for yeast cells).
3.2 Blot Overlay In this assay, the protein to which adherence is to be determined
Assay to Investigate (e.g., a protein present in human saliva, see Note 20) is first sub-
Adhesion of Yeast jected to PAGE separation, before electroblotting onto a nitrocel-
Cells to Immobilized lulose membrane which is then incubated with radiolabeled yeast
Proteins cells. The protein bands to which the radiolabeled yeast have
adhered are detected by autoradiography.
3.2.1 SDS-PAGE Analysis 1. Set up two SDS-PAGE gels using separating gels with an acryl-
amide concentration appropriate for the size of the protein to
be detected (e.g., for binding of C. albicans yeast cells to
salivary proline-rich proteins, 10 % gels are prepared).
2. The saliva or saliva rinse samples are diluted 1:1 with SDS-PAGE
sample buffer and heated (80 C, 10 min) before loading onto
the replicate gels (usually 15 g total protein per lane, see Note
21). On each gel, load protein standards (5 L) so that the
molecular weight of separated proteins can be estimated.
3. The gels are placed in the apparatus (in the Bio-Rad apparatus
the two gels are run back-to-back), submerged in running
buffer, and subjected to direct current at a fixed voltage of

100 mV (caredangerous voltagealways disconnect power
before disassembling apparatus) for approximately 90 min
(until the blue dye front has reached the bottom of the gel).
4. Both gels are removed from the apparatus; one gel is stained to
show the positions of the individual protein bands and the rep-
licate gel is used for electroblotting.
5. Gel staining: One of several techniques can be applied depend-
ing on the expected concentration of the particular proteins in
the sample. We describe here Coomassie Blue staining which is
usually sufficiently sensitive. The gel is submerged in the stain
solution and incubated at room temperature in a closed box
with gentle agitation for at least 1 h (and up to 16 h if evapora-
tion is prevented) and then developed by incubation at room
temperature in destain solution with gentle agitation for
3060 min or longer if required (see Note 22).
6. Gel banding patterns are recorded by photography on a light
box, and protein bands can be quantified using commercially
available computer programs or freely available software such
as NIH Image J (http://rsb.info.nih.gov/ij/index.html).
3.2.2 Electroblotting 1. Blotting is best done with a freshly run gel.

2. Pre-wet the cassette sponge pads, two pieces of 3MM chroma-
tography paper, and the nitrocellulose membrane (cut to the
same size as the gel) with transfer buffer.
3. Carefully use one piece of pre-wetted 3MM paper to remove
the gel from the glass plate of the PAGE apparatus and place
on top of one pre-wetted sponge pad and assemble the sand-
wich within the cassette in the following order: sponge pad,
3MM paper, gel, nitrocellulose, 3MM paper, and second
sponge pad, and place in the electroblot apparatus and cover
with transfer buffer.
4. The apparatus usually has an ice pack in the tank and is run
below 10 C (e.g., in a cold room or refrigerated cabinet).
5. Voltage applied is usually 100 mV for 90 min, but conditions
can be varied for different sized proteins (longer for larger
proteins which do not transfer easily; see Note 23).
6. Dismantle blotting apparatus and peel blot away from gel
(the pre-stained markers allow visual confirmation of transfer)
and store the blot between blotting paper or similar at 4 C
until needed.
3.2.3 Radiolabeled Yeast 1. Block nonspecific protein binding sites on the nitrocellulose
Overlay membrane by incubating with BSA (5 % [w/v] in KCl buffer)
for 2 h at room temperature with reciprocal shaking (50
60 min1). Dimensions of a suitable container (plastic or glass)
should be sufficient in cross section to fit the blot with a space

of approximately 1 cm around it, and with a depth of 57 cm.
2. Remove the blocking solution, wash the blot with 50 mL KCl
buffer, and submerge the blot in KCl buffer (30 mL) contain-
ing radiolabeled yeast cells at 1.1 107 cells/mL.
3. Incubate at 4 C with reciprocal shaking (5060 min1) for
16 h.
4. Remove yeast suspension (careradioactive materialdispose
of according to appropriate regulations) and wash membrane
four times (1 min) with PBS (40 mL) with gentle side-to-side
manual agitation.
5. Remove blot with tweezers and air-dry on blotting paper prior
to exposure to X-ray film. An example of an overlay blot is
shown in Fig. 1.
Fig. 1 Representative blot overlay assay autoradiogram showing adhesion of

35
S-methionine-radiolabeled C. albicans ATCC 10261 cells to PAGE-separated
saliva samples. Samples were separated on a 10 % polyacrylamide gel and elec-
troblotted onto nitrocellulose before incubation with radiolabeled yeast cells and
autoradiography. Lane 1, whole saliva; lane 2, sample extracted from saliva-
coated dental acrylic; lane 3, control sample from untreated dental acrylic.
Arrows (lane 1) indicate four components of whole saliva to which C. albicans
adheres that have been identified as salivary basic proline-rich proteins (PRPs)
[10]. It can be seen that these proteins are not present in the sample of salivary
pellicle from dental acrylic (lane 2)
3.3 Adhesion of In this assay [9], hydroxyapatite beads are used as a model of the
C. albicans Cells tooth surface, which is always coated in saliva (even after tooth
to Saliva-Coated cleaning procedures, the tooth surface is rapidly coated with a sali-
Hydroxyapatite vary pellicle) [14].
1. The KCl buffer surrounding hydrated hydroxyapatite beads
(12 mg) in a 1.5 mL microfuge tube, prepared as described in
Subheading 2.3, item 5, is aspirated and replaced with 1 mL
saliva (see Subheading 2.3, item 4) diluted 1:1 in KCl buffer.
2. The tubes are incubated at 22 C with end-over-end mixing
(612 rpm) for 1 h.
3. The saliva solution is aspirated, the beads washed three times
with KCl buffer (1 mL each time), and the KCl buffer
aspirated.
4. The beads are then incubated with 1 mL KCl buffer containing
0.1 % (w/v) BSA with end-over-end mixing at 22 C for 1 h in
order to block sites which bind proteins non-specifically.
5. The beads are then washed once with KCl buffer (1 mL) and
then 0.9 mL KCl buffer and 0.1 mL radiolabeled cells
(3.0 107 cells/mL, in KCl buffer) are added to each tube.
6. The tubes are incubated at 22 C with end-over-end mixing
(612 rpm) for 90 min.
7. The liquid containing unattached cells is aspirated (care
radioactive materialdispose of according to appropriate reg-
ulations) and the beads are washed three times with KCl buffer
(1 mL).
8. Scintillation fluid (1 mL) is added to the tubes and bead-
associated radioactivity measured (see Notes 24 and 25).
3.4 Adhesion In this assay, monolayers of cultured epithelial cells are used in a
of Saliva-Treated model of C. albicans adherence to human mucosal surfaces. We
C. albicans Cells have used cell lines from culture collections rather than primary
to Epithelial Cells cell cultures (e.g., of oral epithelial cells), but the methods described
could be applied to primary cell monolayers. In order to mimic
intraoral conditions, yeast cells are pretreated with saliva before
measuring adherence to the epithelial monolayers. Standard
96-well microtiter well culture plates are used for adherence assays
using radiolabeled cells, and 24-well culture plates containing ster-
ile glass coverslips are used for confocal microscopy analysis of
adherence.
3.4.1 Epithelial Cell 1. Cultures are passaged using standard aseptic techniques in a
Monolayers biological safety cabinet and maintained in tissue culture flasks
at 37 C in an atmosphere of 5 % CO2. Flasks contain 3050 mL
medium.
2. Cells are subcultured when confluent, as observed using an

inverted microscope (every 23 days). Confluent monolayers
are treated with trypsin/EDTA (4 mL) until the cells start to
lift off the flask surface (25 min in the CO2 incubator). The
cells are resuspended in complete MEM (10 mL), centrifuged,
and then resuspended in 5 mL complete MEM, before the cell
concentration is measured by counting in a hemocytometer.
3. For adherence or confocal assays, a cell suspension is diluted to
approximately 2 105 cells/mL in complete MEM, and 100 or
500 L is seeded in wells of the 96- or 24-well plates, respec-
tively (13 mm diameter sterile coverslips are first added to the
24-well plates) (see Note 25). The plates are incubated at
37 C in a 5 % CO2 atmosphere for 24 h to allow the formation
of confluent monolayers, which are washed once with ASB
(100 L) before addition of ASB (50 L or 250 L to 96- or
24-well plates, respectively) and yeast cells (50 L or 250 L,
respectively).
3.4.2 Adherence Assay 1. Washed radiolabeled C. albicans cells (2.2 106 cells/mL;
Conditions 1.0 mL in microfuge tubes) are pretreated with saliva (diluted
2060 % in ASB) at room temperature for 1 h with end-over-
end mixing.
2. Cells are washed three times in ASB and 50 L added to qua-
druplicate wells in 96-well microtiter plates containing epithe-
lial cell monolayers in 50 L ASB (final volume of 100 L per
well, 1 106 yeast cells).
3. The plates are incubated at 37 C in an atmosphere of 5 % CO2
for 1.5 h. The liquid in the wells is then aspirated and the
monolayers washed three times (see Note 26) with pre-warmed
PBS (100 L) before air-drying and addition of 100 L
Optiphase 3 scintillation fluid.
4. Determine bound radioactivity, and hence number of bound
C. albicans cells, in each well by scintillation detection as above.
3.4.3 Confocal 1. Yeast cells grown as for radiolabeling (but without addition of
35
Microscopy S methionine) are washed in water and resuspended
(2.2 106 mL1; 10 mL) in carbonate/bicarbonate buffer pH
9.5 in a universal bottle wrapped in foil to exclude light.
2. Freshly weighed FITC powder (1 mg) is added and the cell
suspension stirred for 1 h.
3. Cells are washed three times in ASB and 250 L volumes incu-
bated with epithelial cell monolayers in wells of 24-well plates
at 37 C in an atmosphere of 5 % CO2 for 1.5 h.
4. The monolayers are washed three times with pre-warmed PBS
and inverted onto a drop of PBS on a glass microscope slide for
confocal microscopy (plates and slides are protected from light
Fig. 2 Confocal microscopy (split field) showing adhesion of C. albicans ATCC 10261 yeast cells to a monolayer
of cultured epithelial cells (HET1-A cell line kindly provided by Dr. C.C. Harris, Laboratory of Human
Carcinogenesis, NCI, NIH, Bethesda, MD, USA). Yeast cells were labeled with FITC, washed, and incubated with
confluent monolayers on glass coverslips in 24-well plates, before washing and mounting on glass slides for
confocal microscopy. (A) Fluorescence micrograph showing bound yeast cells (bar = 25 m); (B) same field of
view by phase contrast microscopy showing both epithelial (Ep) and C. albicans (Ca) cells
as much as possible). Confocal settings: FITC excitation is at

494 nm and emission is at 520 nm.
5. An example of C. albicans cells adhering to human epithelial
cells is given in Fig. 2.
3.5 Adhesion of These assays that model the initial adherence of microbial cells to
C. albicans or voice or dental prostheses use a similar approach for each of the
S. epidermidis Cells materials used (adhesion under non-flow, static, conditions). Small
to Saliva-Coated rectangular slices (coupons) of medical grade silicone, or small
Medical Grade Silicone molded pieces of denture prosthetic material, are incubated with
or to Denture radiolabeled yeast or bacterial cells in microfuge tubes. We pretreat
Prosthetic Materials the materials with saliva or saliva wash to mimic in vivo conditions.
1. Individual silicone or dental material coupons, prepared as
described in Subheading 2.5, are incubated in triplicate in glass
tubes (of a diameter such that both faces of the coupon are
exposed and the coupon is submerged completely) with
1.0 mL of saliva (or saliva wash or control solutions) with gen-
tle agitation at room temperature for 2 h.
2. Coupons are then washed three times with KCl buffer
(1.0 mL) and transferred to a microfuge tube (again such
that both faces of the coupons are exposed and submerged)
containing 0.9 mL KCl buffer.
25
C. albicans cells bound (x 104)

radiolabelled
20
15
35S-methionine
10
0
0 0.5 1 1.5 2 2.5
35S- methionine radiolabelled C. albicans
cells added in assay (x 106)
Fig. 3 Adhesion of 35S-methionine-radiolabeled C. albicans ATCC 10261 cells to

dental acrylic (Rimseal, polymethyl methacrylate) showing saturation of binding
sites with an increasing number of cells added to the assay. Radiolabeled cells,
1.0 mL at different concentrations in KCl buffer, were incubated end-over-end with
coupons of dental acrylic that had been pretreated with saliva (50 % in KCl buffer
for 2 h) in microfuge tubes at room temperature for 2 h. Coupons were washed and
placed in fresh tubes containing scintillant before determination of coupon-
associated radioactivity by scintillation counting. The numbers of cells bound were
calculated from a standard curve. Results shown are the means SE of quadrupli-
cate determinations from a representative experiment repeated twice
3. To each tube, 35S-methionine-radiolabeled yeast cells (0.1 mL,

1.1 106 cells) or 3H-thymidine-radiolabeled bacterial cells are
added (0.1 mL, 1.1 106 cells) and the tubes are incubated
end-over-end at room temperature for 2 h (see Note 27).
4. Coupons are washed (1 mL KCl buffer 3) then transferred to
a fresh microfuge tube containing 1.0 mL scintillant and radio-
activity measured (see Note 24). An example of C. albicans
adhesion to polymethyl methacrylate is given in Fig. 3.
3.6 Adhesion Parallel plate flow chambers are used in dynamic studies of cell
of S. epidermidis adhesion under well-defined shear forces [7] and the apparatus was
to Denture Prosthetic adapted for the investigation of adherence of bacteria to the den-
Materials Under Flow ture materials listed in Subheading 2.5. These materials were indi-
Conditions vidually constructed as there is no commercial source for pre-made
materials in the dimensions required.
3.6.1 Bacteria: Bacteria are grown in BHY (50 mL in 50 mL bottle) without

(S. epidermidis) added radiolabel at 37 C for 18 h and washed twice in FAB before
resuspending cells to a concentration of 3 108 cells/mL, prior to
circulation through the flow chamber.
3.6.2 Preparation A thin film of each material is manufactured and attached to a glass
of Denture Prosthetic microscope slide using a gypsum mold and the denture flasking
Material Surfaces and pressing method [15].
1. The 75 mm 25 mm 1 mm glass slides are flasked in denture
flasks in gypsum. In the lower denture flask, a microscope slide
is placed flat and embedded in the gypsum.
2. Once the gypsum is set, a second slide is placed on top of the
embedded slide and stuck down with a small amount of wax. A
thin layer of Vaseline is applied to the surface of the stone to
allow easy separation of the flasks.
3. The denture flask is then topped with yellow stone and left to
set. Once the stone is set, the flasks are separated, giving flask
halves with microscope slides embedded in the lower and
opposing upper segments of the flasks which are the gently
scrubbed with dishwashing liquid and boiling water to clean
the microscope slides and remove the Vaseline.
4. The glass slide in the lower flask is removed and replaced with
a glass microscope slide measuring 75 mm 25 mm 0.8 mm.
The slides are then dried, and separating fluid applied to the
stone surface surrounding the microscope slides for easy sepa-
ration after processing.
3.6.3 Parallel Plate Flow 1. The parallel plate flow chamber is mounted on the microscope
Chamber Setup stage.
2. Prior to adhesion testing, the parallel plate flow chamber plates
are cleaned with a commercially available disinfectant (Virkon,
DuPont, Wilmington, DE, USA), rinsed thoroughly with
water, then ethanol, and finally deionized water.
3. The glass plate and the denture prosthetic material plate are
placed in the flow chamber.
4. Laminar fluid flow is achieved in the middle of the flow
chamber by the slope of the inlet and outlet channels of the
flow chamber.
5. With this system it is possible to directly monitor with a micro-
scope, in real time, the initial adhesion of bacteria to the bottom,
denture material, plate in a field of view of 0.5 mm.
3.6.4 Bacterial 1. Initially, FAB is recirculated using a dual-channel variable

Deposition speed pump which recirculates the bacterial and/or yeast
suspension through 0.89 mm diameter tubing at a rate of
0.9 mL/min for 15 min.
2. FAB containing the bacterial suspension is then recirculated
through the flow chamber for 2 h enabling bacteria to adhere
to the denture material. Photographic images of the center
region of the substratum are taken every minute to measure
Fig. 4 Images showing the adhesion over time (three time points shown) of S.
epidermidis cells to dental acrylic (heat processed polymethyl methacrylate) in a
parallel plate flow chamber
adherence to the denture material. An example of S. epidermidis

adhesion to polymethyl methacrylate with time is shown in
Fig. 4.
3. After 2 h, flow is switched to buffer only for 30 min to remove
non-adhering bacteria, with continued photography each min-
ute. The chamber is then drained thus passing an air-liquid
interface over the substratum surface and the adhering
microorganisms.
4. Pre- and post-drainage images are compared to determine the
number of bacteria that are detached by the surface tension
force resulting from the passage of an air-liquid interface.
5. All experiments are performed at room temperature in tripli-
cate with separate bacterial cultures.
6. To determine the influence of saliva on the rate of adhesion,
the denture prosthetic material slide is incubated in clarified
whole saliva (see Subheading 2.3, item 4) diluted 50 % in KCl
buffer for 2 h prior to mounting in the flow chamber.
7. The number of microorganisms adhering to the bottom plate

of the flow chamber in the digital photographs is measured
using image analysis software. The images of the adhering bac-
teria are discriminated from the substratum background by a
single grey-value threshold yielding a binary black and white
image, and the number of adhering bacteria in each image is
counted using the software. The deposition rate is calculated
based on the number of microorganisms adhering per unit
time and area.
8. Control experiments are conducted with bacteria adhering to
bare glass. All adhesion experiments are done in triplicate with
freshly cultured bacteria.
3.7 C. albicans In this assay a biofilm is formed on an acrylic substrate suspended

Biofilm Formation and agitated within a liquid environment which more closely mim-
on Denture Acrylic ics oral conditions than static assays of biofilm growth.
Flag Strips Suspended
in Microtiter Wells
3.7.1 Preparation C. albicans cells from single colonies on YPD agar plates are used
of Inocula for Biofilm to inoculate 1.0 mL RPMI (0.2 % glucose) in sterile 10 mL plastic
Experiments tubes. The tubes are incubated at 35 C, with agitation at 150 rpm
for 18 h using a shaking incubator (Bio-Line, Edwards Instrument
Company, Narellan, NSW, Australia). The concentration of cells is
calculated (see Notes 17 and 18) by measuring the OD600 using a
spectrophotometer (Ultrospec 6300pro, Amersham Biosciences,
NJ, USA). To prepare the inoculum for the biofilm experiments,
these cells are diluted in standard RPMI (0.2 % glucose) to give a
cell concentration of 1 107 cells/mL.
3.7.2 Fabrication Acrylic strips with a design giving six linked flags of acrylic of the
of Acrylic Flag Strips dimensions indicated in Fig. 5a are fabricated in a denture impres-
sion pot (see Note 16) using a template milled from stainless steel
(see Note 28). The bar at the top allows suspension of the flags in
microtiter plate wells, such that they do not touch the well bottom,
and the biofilm assay liquid extends 4 mm up the acrylic flag, giv-
ing each flag a surface area of 53 mm2 on which the biofilm could
form. This design was adopted as it gives a large surface area for
biofilm growth without restriction of fluid flow or volume and six
biofilms can be assayed simultaneously (see Note 29). The bar at
the top is an efficient means to transfer the biofilms through the
experiment stages with minimal disturbance and is narrow enough
for the microtiter plate lid to fit on top and maintain sterility within
the microtiter plate (Nunc). Fabricated strips are polished with 400
grit sandpaper followed by 800 grit sandpaper before sterilization by
immersion in chlorine solution for 15 min (as per removable
denture pre-delivery clinical protocols) using 2 500 mg tablets of
Fig. 5 Acrylic flag strips for biofilm formation. (a) The stainless steel template (T) used to fabricate acrylic flag
strip (F) used as the substrate for C. albicans biofilms grown submerged in the wells of microtiter plates. (b)
Images of flag strips in place within microtiter plates. The space between the flag and well wall is 0.72 mm,
and between the flag and well floor is 3.2 mm
anhydrous sodium dichloro-1,3,5-triazinetrione (Guest Medical

Ltd., Aylesford, UK) in 40 mL filtered deionized water. Acrylic
flag strips are then stored in sterile deionized water at room
temperature until needed (see Note 30).
3.7.3 Coating of Acrylic Sterile acrylic flag strips are submerged in 200 L saliva (prepared
Flags with Saliva as described in Subheading 2.3, item 4) distributed in wells of
sterile microtiter plates (see Fig. 5b) with incubation at 37 C for
30 min.
3.7.4 Initial Deposition 1. Saliva-coated acrylic flag strips are rinsed in growth medium
of Yeast Cells on Acrylic (RPMI 1640), by transferring the flag strips with sterile twee-
Flags zers from wells containing saliva to wells containing growth
medium (200 L), twice.
2. The flag strips are then suspended in microtiter plate wells (see
Note 29) each containing 200 L of RPMI seeded with 1 107
yeast cells/mL as described in Subheading 3.7.1.
3. Lids are placed on the microtiter plates which are incubated at

37 C with agitation (150 rpm, shaking incubator) for 12 h to
create an even dispersion of cells through the medium and to
ensure maximum contact of the inoculum with the flags of
acrylic.
3.7.5 Biofilm Formation 1. The inoculated strips are transferred under sterile conditions to
further sterile microtiter plates containing fresh growth
medium (RPMI 1640 supplemented with 1.8 % glucose) where
they are incubated with a more gentle agitation (50 rpm).
2. Every 12 h, strips are sampled for staining with crystal violet
(see below), and every 24 h the remaining strips are transferred
to further sterile microtiter plates containing fresh growth
medium. In this way, biofilm growth can be assessed at 12, 24,
36, 48, 60, and 72 h timepoints.
3.7.6 Biofilm 1. Biofilms formed on the flag strips are fixed by suspending them
Quantification Using in microtiter plate wells containing 200 L of 99 % (v/v) meth-
Crystal Violet (CV) Staining anol. The strips are removed after 15 min, and the attached
biofilms allowed to air-dry at room temperature. The flag strips
are then immersed in CV 0.1 % (w/v) (200 L) in the wells of
a fresh microtiter plate and incubated at 37 C for 20 min. The
strips are then removed from the wells, gently washed by dip-
ping twice in sterile, deionized water, and air-dried. The dye is
released from the stained biofilm by suspending the flag strips
in microtiter wells containing 200 L of acetic acid 10 % (v/v)
for 30 min. The flag strips are removed, and the solution in the
microtiter plate wells agitated to mix the contents of each well
thoroughly before the absorbances of the resultant solutions of
released dye are measured in a Synergy 2 (Biotek) microtiter
plate reader at OD590. Results are recorded using Gen5
Microtiter Data Collection Analysis software.
2. A representative example of a biofilm growth curve, quantified
with the CV assay, is given in Fig. 6.
4 Notes
1. Scintillation counter: We use a 1450 MicroBeta TriLux

Scintillation counter (LKB, Wallac Oy, Turku, Finland).
2. Safe handling of radioactive materials is very important. The
radioactive isotopes used in the assays described (35S and 3H)
do not produce penetrating radiation and can be handled in a
normal laboratory as long as the appropriate regulatory body
requirements are met. The principal hazard is ingestion, and
therefore avoidance of skin contact is the most important safety
Fig. 6 Representative biofilm growth curve for C. albicans ATCC 10261 grown in
RPMI (supplemented with 1.8 % glucose) quantified by the CV assay. The values
represent the means ( standard deviations) of three separate experiments
stepuse double gloves and appropriate personal protective

equipment including gown and face shield. When first opening
a vial of 35S-methionine, open it in a fume hood in case there is
some release of volatile radioactive material.
3. Incorporation of the radiolabel into microbial cells: We have
conducted experiments that have shown that the incorporation
of radiolabeled methionine or thymidine into yeast or bacterial
cells, respectively, is stable under the experimental conditions
used and that negligible radiolabel is lost from cells during the
experiments. The specific radioactivity of the cells can be cal-
culated by measuring the radioactivity of known numbers of
cells. In our experiments the radiolabeled yeast cells had a
mean activity of 32 18 cells cpm1 and 18 7 cells cpm1 for
bacteria.
4. GSB medium can be conveniently made up from concentrated
stocks of solutions stored at 4 C. For example, a 5 stock of
the salts solution and a 1000 stock of biotin can be diluted as
required and glucose added at the required concentration
before autoclaving in a suitable culture or storage vessel.
Although biotin is a vitamin and can be unstable at high tem-
peratures, sufficient functional biotin is retained after autoclav-
ing to support yeast growth.
5. The equipment required for electrophoresis and blotting is
commercially available; for electrophoresis, we use a mini-
PROTEAN apparatus (Bio-Rad) and a Mini Trans-Blot Cell
(Bio-Rad) for electroblotting. Direct current is supplied by a
PowerPac Basic Power Supply (Bio-Rad).
6. Acrylamides are neurotoxinshandle with care. Purchase

ready-made solutions rather than making stock solutions to
avoid weighing solid acrylamide. Wear gloves for all stages of
gel preparation.
7. Ammonium persulfate (10 % w/v) keeps for several weeks at
4 C. Best practice is to store it frozen in small aliquots (approx.
0.5 mL) which can be thawed as required.
8. Pre-stained markers are useful for checking success of protein
transfer during electroblotting and also for orientation of blots
following the development of the overlay autoradiograph.
9. Transfer buffer: Always dissolve the Tris and glycine in water
(approximately 600 mL) before adding the methanol and
making up to volume (adding methanol first gives an insoluble
precipitate).
10. Confocal microscope: We used a Bio-Rad MRC-600 (Bio-Rad
Microscience Ltd., Hemel Hempstead, UK) with a krypton/
argon laser (which produced excitation lines at 488, 568, and
647 nm) and a Nikon Diaphot Inverted Microscope (Nikon
Instruments Inc., Melville, NY).
11. Carbonate/bicarbonate buffer: Prepare as two solutions (each
of 0.1 M)can be mixed in different ratios to give a range of
pH values.
12. Place silicone slice on clear smooth surface (e.g., plastic Petri
dish) over a ruled template so that the correct size can be cut.
13. Saliva wash samples are used rather than saliva for individuals
(e.g., voice prosthesis patients who have had head and neck
radiotherapy) with poor saliva flow and/or saliva with high
mucin content. Such wash samples are much simpler to han-
dle in the laboratory, and we have still been able to detect pro-
teins involved in adherence in these saliva wash samples.
14. Parallel plate flow chamber: The parallel plate flow chamber is
commercially available (Glycotech, Rockville, MA, USA) and
is constructed of stainless steel (dimensions: l w h = 200 mm
42 mm 10 mm) and contains two glass plates (dimensions
l w h 76 mm 26 mm 1 mm), separated by 0.6 mm using
a Teflon spacer, that constitute the top and bottom plates of
the chamber.
15. We used an Olympus IMT-2 inverted phase contrast micro-
scope, Olympus Corporation, Tokyo, Japan, with an ultra-long
working distance 40 objective (Olympus ULWD-CD Plan
40) for observation of bacteria, and a 10 objective (Olympus
SPlan 10) to observe yeast.
16. Acrylic adhesion substrates are fabricated as per complete
dentures. Vertex acrylic is poured into molds and the flasks closed
under hydraulic pressure (to 50 kg/cm2) before immersion of
the investment flask into hot water (95 C) for 20 min until
complete polymerization of the acrylic.
17. We use a Shimadzu spectrophotometer at OD540 for yeast cells
and OD600 for bacterial cells.
18. We construct a standard curve relating OD540 values to C. albicans
cells per mL by measuring the OD540 with a spectrophotome-
ter and measuring the cell concentration for an appropriate
dilution of the same culture with a hemocytometer.
19. The half-life of 35S is approximately 3 months. With prepara-
tions of methionine that have been stored for periods longer
than a few weeks, the amount of radiolabel added can be
increased up to 4 or 5 L.
20. We have used the same technique to demonstrate adhesion of
C. albicans to various proteins, including recombinant pro-
teins that have been cloned and expressed in Escherichia coli.
The technique was first developed to demonstrate the adher-
ence of C. albicans to salivary proline-rich proteins [10].
21. Total protein estimation is done using a Bradford protein assay
kit (Bio-Rad).
22. Destaining Coomassie-stained PAGE gels: A small pad of sponge
material (e.g., from packaging material) added to the destain
with the gel helps to remove background stain from the gel.
23. Care must be taken when electroblotting small proteins as they
may transfer rapidly and can be lost through the nitrocellulose
if electroblotting is continued for too long. Conversely large
proteins may not transfer well and require prolonged
electroblotting.
24. Static electricity may be generated while handling the microfuge
tubes with rubber gloves, and this may give false scintillation
counter reading. Take a second reading 24 h after the initial
reading to allow static electricity to discharge.
25. To avoid cross-talk in the scintillation counter, leave some
wells unfilled: Use only alternate columns of wells and alter-
nate rows of wells.
26. Aspirating and washing tissue culture monolayers: Aspirate
from the edge of the monolayer gently with manual multi-
channel pipette. Also, add wash solution gently down side of
well wallreverse plate 180 for alternate washes. It is impor-
tant to pre-warm the wash solution to 37 C to prevent loss of
tissue culture cells.
27. Ensure that the silicone coupon is completely submerged in
the solution containing the radiolabeled cells for the entire
duration of incubation to allow cell adhesion to the entire
surface of the silicone coupons. Also check that the liquid
moves up to the top of the tube and back during each rotation.
If not (this is possibly a surface tension effect), stop rotation,

remove and invert the tubes a few times, replace, and restart.
Repeat if necessary.
28. The impressions prepared from the metal template can only be
used a limited number of times, therefore several metal tem-
plates should be prepared in order to produce acrylic flag strips
in batches of six.
29. When microtiter plates are incubated at 37 C for extended
periods, evaporation from the wells on the edge of the plate
occurs. Therefore the six-flag acrylic strips are placed in
microtiter plate rows B-G avoiding the wells around the edge
of the plate which are filled with uninoculated growth medium
to buffer evaporation from the inner wells.
30. If acrylic flag strips are stored for any extended period (>2
days) prior to biofilm formation, acrylic substrates are re-
sterilized by immersion in chlorine solution (2 500 mg tab-
lets of anhydrous sodium dichloro-1,3,5-triazinetrione (Guest
Medical, UK) in 40 mL filtered deionized water) for 15 min
followed by quick immersion in absolute ethanol (100 %) and
brief flaming.
Acknowledgments
We are grateful to Andrew McNaughton (Otago Centre for

Confocal Microscopy) for undertaking the confocal microscopy
analysis and to Steve Swindells (University of Otago School of
Dentistry) for the assistance with the fabrication of the acrylic tem-
plates used in the biofilm assay. We thank Professor Henk Busscher
(Department of Biomedical Engineering, University of Groningen,
the Netherlands) for the advice on the parallel plate flow apparatus.
We gratefully acknowledge funding from the New Zealand Lottery
Grants Board, the New Zealand Dental Association Research
Foundation, and the University of Otago. K. N.-W. is grateful to
the Sir John Walsh Research Institute for the award of a Fuller
Scholarship.
References
1. Jenkinson HF, Lamont RJ (2005) Oral micro- 4. Sharon N (2006) Carbohydrates as future
bial communities in sickness and in health. anti-adhesion drugs for infectious diseases.
Trends Microbiol 13:589595 Biochim Biophys Acta 1760:527537
2. Kolenbrander PE (2000) Oral microbial com- 5. Cannon RD, Holmes AR, Mason AB, Monk
munities: biofilms, interactions, and genetic BC (1995) Oral Candida: clearance, coloniza-
systems. Annu Rev Microbiol 54:413437 tion, or candidiasis? J Dent Res 74:11521161
3. Moscona A (2008) Medical management of 6. Meurman JH (2005) Probiotics: do they have
influenza infection. Annu Rev Med 59: a role in oral medicine and dentistry? Eur
397413 J Oral Sci 113:188196
7. Busscher HJ, van der Mei HC (2006) Microbial streptococci is promoted by selective adsorp-
adhesion in flow displacement systems. Clin tion of salivary proteins to the streptococcal
Microbiol Rev 19:127141 cell surface. Microbiology 146:4148
8. Peeters E, Nelis HJ, Coenye T (2008) 12. Holmes AR, Bandara BM, Cannon RD (2002)
Comparison of multiple methods for quantifi- Saliva promotes Candida albicans adherence
cation of microbial biofilms grown in microti- to human epithelial cells. J Dent Res 81:
ter plates. J Microbiol Methods 72:157165 2832
9. Cannon RD, Nand AK, Jenkinson HF (1995) 13. Holmes AR, van der Wielen P, Cannon RD,
Adherence of Candida albicans to human sali- Ruske D, Dawes P (2006) Candida albicans
vary components adsorbed to hydroxylapatite. binds to saliva proteins selectively adsorbed to
Microbiology 141:213219 silicone. Oral Surg Oral Med Oral Pathol Oral
10. OSullivan JM, Cannon RD, Sullivan PA, Radiol Endod 102:488494
Jenkinson HF (1997) Identification of salivary 14. Dawes C (2008) Salivary flow patterns and the
basic proline-rich proteins as receptors for health of hard and soft oral tissues. J Am Dent
Candida albicans adhesion. Microbiology Assoc 139(Suppl):18S24S
143:341348 15. Morrow RM, Rudd KD, Rhoads JE (1986)
11. OSullivan JM, Jenkinson HF, Cannon RD Dental laboratory techniquescomplete den-
(2000) Adhesion of Candida albicans to oral tures. Mosby, St Louis, MO, USA
Chapter 11
Quantitative Analysis of Periodontal Pathogens

Using Real-Time Polymerase Chain Reaction (PCR)
M Jos Marin, Elena Figuero, David Herrera, and Mariano Sanz
Abstract
The quantitative polymerase chain reaction (qPCR) is a variant of PCR aimed to detect and quantify a
targeted DNA molecule through the addition of probes labeled with fluorescent molecules that emit fluo-
rescence within each amplification cycle, what results in fluorescence values proportional to the amount of
accumulated PCR product. This chapter presents the detailed procedures for quantification of different
periodontal pathogens (Porphyromonas gingivalis, Aggregatibacter actinomycetemcomitans, Tannerella for-
sythia, Campylobacter rectus, and Fusobacterium spp.) using qPCR. It also includes the description of the
most frequent problems encountered and how to solve them. In addition, a detailed protocol for multiplex
qPCR to detect and quantify P. gingivalis and A. actinomycetemcomitans is included.
Key words Quantitative PCR (qPCR), TaqMan, Primers, Probe, Multiplex qPCR, Aggregatibacter
actinomycetemcomitans, Porphyromonas gingivalis, Tannerella forsythia, Campylobacter rectus,
Fusobacterium spp
1 Introduction
Periodontitis is a chronic inflammatory disease of multifactorial eti-

ology, being the presence of specific bacteria residing in the sub-
gingival biofilm, the primary etiological factor. This subgingival
biofilm is a complex microbiota where more than 700 bacterial
species have been detected [1], although, only a limited number of
these species were shown to be a risk factor for the initiation or
progression of periodontitis, namely, Aggregatibacter actinomy-
cetemcomitans, Porphyromonas gingivalis, and Tannerella forsythia
[2]. The presence of these periodontal pathogens in the gingival
crevicular fluid (GCF) has been associated with disease sites as
well as with healthy sites, although in this latter case only in low
numbers. Furthermore, the presence of these, so-called, peri-
odontal pathogens has been identified in the blood of patients
with periodontitis, mainly associated with periodontal interven-
tions, but also after normal daily life activities, such as mastication [3],
191
192 Ma Jos Marin et al.
tooth brushing [4], or dental flossing [5]. These bacterial species

are able to reach the blood stream and may colonize distant sites
and facilitate the development and progression of atherosclerotic
lesions [6].
Anaerobe culturing methods, in which samples harvested from
the GCF [79] or blood [5, 1014] are directly cultivated in
enriched media agar plates have been the gold-standard method
for detecting periodontal pathogens. However, these techniques
have relevant limitations, namely, the need of experienced person-
nel, relatively time consuming, and the difficulty of detecting slow-
growing or uncultivable oral bacterial species. In order to overcome
these limitations, different technologies have been proposed,
including those based in molecular biology. Among them, poly-
merase chain reaction (PCR) is considered the most sensitive
method available for DNA sequence detection [15]. PCR allows
for the analysis of large number of samples in a short period of
time, and it has demonstrated high sensitivity in detecting peri-
odontal pathogens such as P. gingivalis, A. actinomycetemcomitans,
or T. forsythia [1518], thus confirming its validity in periodontal
microbiological diagnosis [9] and its potential role in bacteremia
studies [19].
The base of PCR resides in its ability to amplify one copy of a
DNA template by several million-fold in a simplified and auto-
mated fashion. By conventional PCR, the millions of copies of this
particular DNA sequence can be detected through electrophoresis
in agarose gel. However, it only provides qualitative information
on the targeted bacteria, and therefore, their use for diagnostic
purposes may be limited. Quantitative PCR (qPCR) is a variant of
PCR that allows the quantification of a targeted DNA molecule.
Two main types of chemistries have been used in these assays:
SYBR Green I dye and TaqMan assay. SYBR Green I dye is a
highly specific double-stranded DNA-binding dye, which allows
for the detection of product accumulation during the PCR pro-
cess. It will detect all double-stranded DNA, including nonspe-
cific reaction products [15]. By contrast, the TaqMan assay, also
known as fluorogenic 5 nuclease assay, quantifies through the
addition of a probe labeled with fluorescent molecules that emit
fluorescence within each amplification cycle [20, 21]. The base of
this assay is the presence of a TaqMan probe, which has a reporter
fluorescent dye attached to its 5-end and a quencher dye attached
to its 3-end. The probe hybridizes to the target DNA, and during
the extension phase of PCR, it is cleaved by the 5 3 exonucle-
ase activity of Taq DNA polymerase, separating the fluorophore
and the quencher, what results in fluorescent values proportional
to the amount of accumulated PCR product [20, 21]. A qPCR
device (thermocycler) collects fluorescence data for each sample
during each PCR cycle, and the first cycle at which the instrument
Periodontal Pathogens by qPCR 193
Table 1
Commonly used terms employed in qPCR assays
Threshold Level at which a significant change in fluorescence is detected (green line). The
software associated to the thermocycler can calculate the threshold cycle (auto)
Ct/Cp Number of PCR cycles needed to reach a set threshold fluorescence signal level (point
which the fluorescence signal crosses a defined threshold)
Also called Cp (cross point cycle) for LightCycler terminology. Ct inversely correlates
with initial template concentrations (amounts)
Baseline Background noise level before a significant amplification occurs (315 cycles)
NTC No template control: monitors contamination and primerdimer formation that could
produce false-positive results
detects fluorescence intensity (greater than background fluores-

cence) is termed the threshold cycle (Ct) or the cross point cycle
(Cp), which is directly correlated to the starting target concentra-
tion for the DNA sample (Table 1). Specific software will compare
the Ct/Cp values of the unknown samples to those of positive
controls with known concentrations of the targeted DNA (the
standard curve) to determine the starting concentration of each
unknown.
Multiplex qPCR (m-qPCR) allows the simultaneous amplifica-
tion of different target sequences in a single reaction, thereby
detecting and quantifying simultaneously different bacteria within
the same sample [22]. This is extremely important when the sam-
ple volumes available for processing are limited. Moreover, multi-
plexing reduces analytical costs, improves turnaround time,
expands testing capability and capacity, and adds data richness to
analysis [23, 24]. In m-qPCR, multiple probes are selected on the
bases of different fluoresce wavelength, so it is possible to distin-
guish signals from different primer/probe sets. Both qPCR and
m-qPCR have clear advantages such as excellent sensitivity, repro-
ducibility of assays, increased specificity by the specific hybridiza-
tion between the probe and the target, ability to monitor the
increasing amounts of amplicon as they accumulate, and no needed
post-amplification processing and increased automation.
This chapter presents the detailed procedures for quantifica-
tion of the periodontal pathogens P. gingivalis, A. actinomycetem-
comitans, T. forsythia, Campylobacter rectus, and Fusobacterium
spp. using qPCR. It also includes the description of the most fre-
quent problems encountered and the solutions to solve them. The
m-qPCR presented enables the simultaneous quantification of A.
actinomycetemcomitans and P. gingivalis. This assay may be used to
quantify pure cultures, GCF samples, blood samples, or any other
biological samples (solid).
2 Materials
2.1 Samples 1. Gingival crevicular fluid (GCF), blood, or any other biological
and Positive Controls samples (solid).
2. Sterilized microcentrifuge tubes (1.5 mL).
3. Standard reference strains: P. gingivalis (ATCC 33277), A.
actinomycetemcomitans (DSM 8324), T. forsythia (ATCC
43037), C. rectus (ATCC 33238), Fusobacterium nucleatum
(DMSZ 20482).
4. Blood agar plates.
5. Autoclaved microcentrifuge tubes (2 mL).
6. Brain Heart Infusion (BHI) medium.
7. Phosphate buffer saline (PBS) pH 7.4. Stock PBS: 80 g NaCl,
2 g KCl, 9.17 g Na2HPO4, and 2 g KH2PO4. Make up to 1 L
with distilled H2O, adjust to pH 7.3. Store at 4 C.
8. UV-1800 UVVis spectrophotometer.
9. Jars.
2.2 DNA Extraction A different working area and different laboratory tools (pipettes,
racks) are required for DNA extraction, which should be indepen-
dent from the general or PCR procedures (see Note 1).
1. In case of biological samples, a mechanical homogenizer (IKA-
Werke, Stanfen, Germany).
2. MoIYsis Complete5 (Molzym Gmbh & Co. KG., Bremen,
Germany) (see Note 2).
3. PCR-grade water (Roche Diagnostic GmbH, Penzberg,
Germany).
4. Thermo-shaker for microtubes (Lan Technicsmixing Block,
Labolan, Navarra, Spain).
5. Microcentrifuge (Thermo Scientific, Madrid, Spain).
7. Nanodrop ND-1000 spectrophotometer.
2.3 qPCR A different working area and different laboratory tools (pipettes,
Amplification racks) are required for qPCR amplification, which should be inde-
pendent from the general or DNA extraction procedures (see Note 1).
1. PCR-grade water (Roche Diagnostic GmbH, Penzberg,
Germany).
2. Ice.
3. TaqMan master mixture: LC 480 Probes Master (Roche
Diagnostic GmbH, Mannheim, Germany).
4. Fluorogenic probes (100 pmol/L) (Invitrogen, Life

Technologies, Spain):
(a) A. actinomycetemcomitans: 5-6FAM-AGA ACT CAG
AGA TGG GTT TGT GCC TTAGGG-TAMRA-3.
(b) P. gingivalis: 5-6FAM-CACTGAACTCAAGCCCGG
CAGTTTCAA-TAMRA-3; C. rectus: 5-6FAM-TCC
GTG CCA GCA GCC GC-TAMRA-3.
(c) T. forsythia: 5-6FAM-CCC GCA ACA GAG GGA TAA
CCC GG-TAMRA-3.
(d) Fusobacterium spp.: 5-6FAM CTCTACACTTGTA
GTTCCG-TAMRA-3.
(e) In the case of m-qPCR: Fluorogenic Probe (100 pmol/
L) of P. gingivalis: 5-HEX5-CACTGAACTCAAGCCC
GGCAGTTTCAA-BBQ-3.
5. Primers (25 M in H2O) designed to 16S ribosomal RNA
(rRNA) gene (Invitrogen, Life Technologies, Spain):
(a) A. actinomycetemcomitans forward: 5-GAA CCT TAC
CTA CTC TTG ACA TCC GAA-3; reverse 5-TGC AGC
ACC TGT CTC AAA GC-3.
(b) P. gingivalis forward: 5-GCGCTCAACGTTCAGCC-3;
reverse: 5-CACGAATTCCGCCTGC-3; C. rectus for-
ward: 5-TTT CGG AGC GTA AAC TCC TTT TC-3;
reverse: 5-CGC TTG CAC CCT CCG TAT-3.
(c) T. forsythia forward: 5-GGG TGA GTA ACG CGT ATG
TAA CCT-3; reverse: 5-ACC CAT CCG CAA CCA ATA
AA-3.
(d) Fusobacterium spp. forward: 5-GGATTTATTGGGC
GTAAAGC-3; reverse: 5-GGCATTCCTACAAAT
ATCTACGAA-3.
6. Thermocycler: LightCycler 480 II (Roche Diagnostic GmbH,
Mannheim, Germany).
7. LC 480 multiwell plate 96 or LC 480 multiwell plate 384.
8. qPCR adhesive clear seals.
9. Barrier tips for pipettes.
11. Sodium hypochlorite (NaOCl) 10 %.
12. Termi-DNA-Tor: less than 25 % 2-propanol, less than 45 %
1-propanol, less than 0.3 % formaldehyde, less than 0.6 % gly-
oxal, less than 0.2 % glutaraldehyde (Biotools, B & M Labs,
S.A., Madrid, Spain).
13. Extracted DNA coming from 1 mL of viable bacteria in BHI
medium (P. gingivalis, A. actinomycetemcomitans, T. forsythia,
and C. rectus) containing approximately 109 colony-forming

units (CFU)/mL to prepare standard curves.
14. Extracted DNA from samples.
3 Methods
3.1 Sample 1. GCF samples should be taken with sterile medium paper points
Collection (Maillefer, Ballaigues, Switzerland) and transferred into empty
sterilized microcentrifuge tubes (1.5 mL).
2. Follow standard procedures to harvest blood samples or any
other biological samples (solid).
3.2 Positive Controls 1. Grow the bacteria on blood agar plates under anaerobic condi-
(Standard Curve) tions (10 % H2, 10 % CO2, and balanced N2) at 37 C for
2472 h.
2. Inoculate the bacteria in 5 mL of BHI medium and incubate
under anaerobic conditions for 2448 h (depending on the
bacterial species) in jars to reach an exponential growth phase
(as measured by spectrophotometry at optical density [OD]
550 nm).
3. Prepare tenfold dilutions of each bacterial species on PBS,
plate them on blood agar plates, and incubate under anaerobic
conditions (10 % H2, 10 % CO2, and balanced N2) at 37 C for
2472 h to determinate CFU/mL (optimal concentration:
approximately 109 CFU/mL).
3.3 DNA Extraction All extraction procedures should be done in a specific laboratory
equipped with ultraviolet light during night to prevent any poten-
tial contamination (see Note 1).
3.3.1 DNA Extraction 1. Centrifuge 1 mL of bacteria cells in BHI medium at 13,000 g

from Pure Cultures for 1 min, containing approximately 109 CFU/mL.
2. Recover the bacteria after removing carefully the supernatant
by pipetting.
3. Follow the instructions detailed in the protocol of MoIYsis
Complete5.
4. The DNA concentration is measured on a spectrophotometer.
5. The DNA dissolved in sterile water could be stored frozen
(20 C) until analysis.
3.3.2 DNA Extraction 1. Samples may be frozen at 80 C until analyses, although it is

from GCF Samples recommended to analyze them not further than 6 months after
they were collected.
2. Re-suspend paper points in 1000 L of water and vortex them
for 2 min at maximal setting.
3. Remove paper points.

4. Centrifuge the vial at 13,000 g for 3 min.
5. Discard the supernatant.
6. Process the resultant with the commercial kit for DNA extrac-
tion following manufacturers instructions starting in step 6
from Small Size Sample DNA Isolation protocol (1 mL
Liquid).
7. The final elution of DNA should be done in 100 L of water.
3.3.3 DNA Extraction 1. Bacterial DNA can be extracted from 1 mL of blood.

from Blood Samples 2. Blood needs to be immediately processed with the commercial
kit for DNA extraction following manufacturers instructions.
3.3.4 DNA Extraction 1. Biological solid samples should be homogenized before DNA
from Any Other Biological extraction. This could be done with a mechanical homogenizer
Solid Samples, until a uniform suspension is obtained. If needed, the suspen-
For Example, Atheromatous sion buffer from the extraction kit might be used during the
Plaques homogenization process.
2. This suspension is processed with the commercial kit for DNA
extraction following manufacturers instructions.
3.4 Preparation 1. In sterile 1.5-mL microcentrifuge tubes, perform tenfold serial

of Standard Curves dilutions of each standard bacterial DNA previously extracted
for qPCR on sterile water with a range of 100109 CFU/mL of each
bacteria.
3.5 qPCR Assay 1. In a sterile 2-mL microcentrifuge tube, mix the components in
the following order for each analyzed bacteria: master mix,
probe, primers, and sterile water. Table 2 shows the required
volume and concentration of each component (see Note 3).
2. For multiplex qPCR, mix in a same 2 mL microcentrifuge tube
master mix, A. actinomycetemcomitans and P. gingivalis probe,
and primers in the volumes above indicated and complete to a
final volume of 20 L with sterile water.
3. Place the reaction mix on ice and protect it from light until
use.
4. For each sample, transfer 15 L of PCR reaction mix to the
associated wells in a 96-well reaction plate. If a qPCR system
with a 384-well sample block were used, the final reaction vol-
ume would be 10 L (see Note 4).
5. Add 5 L (2.5 L in case of multiplex qPCR) of DNA of stan-
dard curve, unknown samples, and water (NTC: no template
control) as duplicates to each well in the reaction plate. Care
should be taken to pipette accurately into the wells, as small
variations will affect the assay (see Note 5).
Table 2
Volume and concentration of each component of the qPCR assay
Bacteria Final concentration (nM) Volume (L)

Master mix 2 1 10
Probe 100 pmol/L Aa 200 0.04
Pg 300 0.06
Cr 100 0.02
Tf 200 0.04
Fspp 300 0.06
Primers forward/ Aa 300/300 0.24/0.24
reverse 25 M Pg 300/300 0.24/0.24
Cr 900/900 0.72/0.72
Tf 300/300 0.24/0.24
Fspp 600/600 0.48/0.48
Sterile water Aa 4.48
Pg 4.46
Cr 3.54
Tf 4.48
Fspp 3.98
Primer and probe concentration optimization of each bacterial species for qPCR is
mandatory to identify the ideal concentration combination for these assays. Aa
Aggregatibacter actinomycetemcomitans, Pg Porphyromonas gingivalis, Cr Campylobacter
rectus, Tf Tannerella forsythia, Fspp Fusobacterium species
6. Centrifuge the plate briefly to correct any adherent drop and

bottom bubble problems (see Note 6).
7. Load the plate into the thermocycler and run the following
TaqMan protocol: initial amplification cycle of 95 C for
10 min (denaturation), followed by 40 cycles at 95 C for 15 s
and 60 C for 1 min (extension).
3.6 Data Analysis 1. Analyze the data viewing the amplification plots for the entire
plate, setting the baseline and threshold values, and using the
standard curve.
2. Check the no template control (NTC) wells for any amplifica-
tion. There should be no amplification (see Note 7).
3. Ensure that the efficiency of amplification of the standard curve
(control template) is 90100 % (3.6 slope 3.3).
4. Determine the concentration of each sample based on data
from standard curves (Figs. 1 and 2) (see Notes 8 and 9).
4 Notes
1. Contamination is a common problem associated to PCR

procedures, resulting in the appearance of an amplification
product in the negative controls, which lack the template DNA.
Fig. 1 Amplification plot of a calibration curve and unknown samples
36 Y=-3.376.33*log[X] + 41
34
Standard curve samples
32
Cross point (Cp)
30
28
26
24
22
20
18
16
1 2 3 4 5 6 7 8
Log (DNA concentration_CFU/ml) from standard curve
Fig. 2 Plot of the crossing points (Cp) or threshold cycles (Ct) against the log of initial DNA concentration. CFU
colony-forming units
In order to prevent contamination during all steps of the

qPCR procedure, the following recommendations should be
followed:
(a) PCR should be performed in a separate laboratory with its
own set of equipment (vortex, centrifuge). This laboratory
should include an ultraviolet (UV) light within the lamps
that cover the working areas and outer surfaces of equipment
(racks, pipettes, etc.) that is used overnight to sterilize

the area.
(b) All working surfaces and pipettes should be cleaned before
and after each use with 10 % sodium hypochlorite (NaOCl),
which inactivates pathogenic agents and destroys nucleic
acids. Afterwards, the pipettes should be sprayed with
Termi-DNA-Tor for DNA removal.
(c) Work under sterile conditions during the whole process,
from extraction to the amplification assay.
(d) Use gloves, face mask, and head caps to reduce contamina-
tion from facial skin and hair cells, and change gloves
whenever you suspect any possible contamination.
(e) Store the master mix, primers stock, probe, and water in
small aliquots in separate sections of the freezer. Discard
aliquots of primers stock and probe after use.
(f) Keep reactions and components capped as much as
possible.
(g) Prepare tenfold serial dilutions of the DNA used as tem-
plate (standard curve) at a bench different to the qPCR
area.
(h) Always use sterile water (PCR grade, Roche Diagnostic) to
dissolve extracted DNA (with MoIYsis Complete5 kit) to
avoid false amplifications due to contamination.
(i) Always use sterile water to re-suspend primers stock,
probes, serial dilutions from bacterial DNA (standard
curves), and NTC to add in well reaction plates.
(j) Keep the extracted DNA that is being pipetted as far away
as possible from the reaction plate.
(k) Locate the wells containing NTCs as far away as possible
from positive controls, and test samples to prevent acci-
dental cross contamination.
(l) At the end of qPCR, tips and tubes used in the procedure
must be discarded in a laboratory different to that where
DNA extraction and amplification procedures were
performed.
2. Different chemical, enzymatic, or physical methods have been
used to obtain DNA in sufficient quantity and quality for its
subsequent analysis. However, the MoIYsis Complete5 method
has shown the best results in the purification and quantifica-
tion of DNA from target subgingival periodontal pathogens
from GCF samples, pure bacterial cells, or blood samples.
3. Centrifuge all qPCR reaction components just before assem-
bling reactions to briefly mix.
4. Do not write on the surface of the plate over well positions, as
this will interfere with fluorescence reaction and reading.
5. Standard curves and NTCs (in duplicates) should be prepared

on every plate.
6. Seal the plate using the adhesive clear seals without touching
the cover itself. Seal the reaction plate, and centrifuge at low
speed for 1 min to bring all reaction components together and
eliminate air bubbles.
7. There should be no amplification with NTCs, but if a value of
Ct/Cq is obtained, it will define the detection limit of your
assay.
8. A dilution series of known template concentrations is used to
establish a standard curve. The log of each known concentra-
tion in the dilution series (x-axis) is plotted against the Ct/Cq
value for that concentration (y-axis). This results in a straight
line that has, at least, seven orders of magnitude, and linear
regression analysis permits the calculation of DNA concentra-
tion of any unknown target relative to that standard curve
(Fig. 2). This standard curve reveals the amplification efficiency
of the reaction (slope) and gives some indication of its sensitiv-
ity (y-intercept).
9. The slope of the standard curve describes the kinetics of the
PCR amplification. It indicates how quickly the amount of tar-
get nucleic acid (NA) can be expected to increase with the
amplification cycles. The slope of standard curve is referred to
as the efficiency of the amplification reaction. A perfect ampli-
fication reaction would produce a standard curve with an effi-
ciency of 2, because the amount of target NA would double
with each amplification cycle. The PCR efficiency can easily be
calculated using the formula 101/slope. The efficiency of the
PCR should be between 90 and 100 % (3.6 slope 3.3). If
the efficiency is 100 %, the Ct/Cp values of the tenfold dilu-
tion will be 3.3 cycles apart (there is a twofold change for each
change in Ct/Cp).
References
1. Aas JA, Paster BJ, Stokes LN, Olsen I, Dewhirst 5. Crasta K, Daly CG, Mitchell D, Curtis B,
FE (2005) Defining the normal bacterial flora of Stewart D, Heitz-Mayfield LJ (2009)
the oral cavity. J Clin Microbiol 43:57215732 Bacteraemia due to dental flossing. J Clin
2. Consensus Report (1996) Periodontal dis- Periodontol 36:323332
eases: pathogenesis and microbial factors. Ann 6. Figuero E, Sanchez-Beltran M, Cuesta-
Periodontol 1:926932 Frechoso S, Tejerina JM, del Castro JA,
3. Forner L, Larsen T, Kilian M, Holmstrup P Gutierrez JM, Herrera D, Sanz M (2011)
(2006) Incidence of bacteremia after chewing, Detection of periodontal bacteria in atheroma-
tooth brushing and scaling in individuals with tous plaque by nested polymerase chain reac-
periodontal inflammation. J Clin Periodontol tion. J Periodontol 82:14691477
33:401407 7. Oteo A, Herrera D, Figuero E, O'Connor A,
4. Hartzell JD, Torres D, Kim P, Wortmann G Gonzalez I, Sanz M (2010) Azithromycin as an
(2005) Incidence of bacteremia after routine adjunct to scaling and root planing in the treat-
tooth brushing. Am J Med Sci 329:178180 ment of Porphyromonas gingivalis-associated
periodontitis: a pilot study. J Clin Periodontol detection of Porphyromonas gingivalis in

37:10051015 subgingival plaque samples. J Clin Microbiol
8. Herrera D, Alonso B, Leon R, Roldan S, Sanz 41:49504954
M (2008) Antimicrobial therapy in periodonti- 17. Lau L, Sanz M, Herrera D, Morillo JM, Martin
tis: the use of systemic antimicrobials against C, Silva A (2004) Quantitative real-time poly-
the subgingival biofilm. J Clin Periodontol merase chain reaction versus culture: a com-
35(8 Suppl):4566 parison between two methods for the detection
9. Sanz M, Lau L, Herrera D, Morillo JM, Silva A and quantification of Actinobacillus actinomy-
(2004) Methods of detection of Actinobacillus cetemcomitans, Porphyromonas gingivalis and
actinomycetemcomitans, Porphyromonas gingi- Tannerella forsythensis in subgingival plaque
valis and Tannerella forsythensis in periodontal samples. J Clin Periodontol 31:10611069
microbiology, with special emphasis on 18. Doungudomdacha S, Rawlinson A, Walsh TF,
advanced molecular techniques: a review. J Clin Douglas CW (2001) Effect of non-surgical
Periodontol 31:10341047 periodontal treatment on clinical parameters
10. Lockhart PB, Brennan MT, Sasser HC, Fox and the numbers of Porphyromonas gingivalis,
PC, Paster BJ, Bahrani-Mougeot FK (2008) Prevotella intermedia and Actinobacillus actino-
Bacteremia associated with toothbrushing and mycetemcomitans at adult periodontitis sites.
dental extraction. Circulation 117:31183125 J Clin Periodontol 28:437445
11. Maestre JR, Mateo M, Sanchez P (2008) 19. Figuero E, Lindahl C, Marin MJ, Renvert S,
Bacteriemia secundaria a procedimientos Herrera D, Ohlsson O, Wetterling T, Sanz M
odontologicos periodontales. [Bacteremia after (2014) Quantification of periodontal patho-
periodontal procedures]. Rev Esp Quimioter gens in vascular, blood, and subgingival sam-
21:153156 ples from patients with peripheral arterial
12. Lafaurie GI, Mayorga-Fayad I, Torres MF, disease or abdominal aortic aneurysms.
Castillo DM, Aya MR, Baron A, Hurtado PA J Periodontol 85:11821193
(2007) Periodontopathic microorganisms in 20. Mackay J (2007) Introduction to kinetic (real-
peripheric blood after scaling and root planing. time) PCR. Methods Mol Biol 353:167176
J Clin Periodontol 34:873879 21. Mackay J, Landt O (2007) Real-time PCR
13. Kinane DF, Riggio MP, Walker KF, MacKenzie fluorescent chemistries. Methods Mol Biol
D, Shearer B (2005) Bacteraemia following 353:237261
periodontal procedures. J Clin Periodontol 22. Bolivar AMR, Agustina; Garcia-Lugo, Pablo
32:708713 (2014) PCR and PCR-multiplex: critical
14. Daly CG, Mitchell DH, Highfield JE, parameters and standardization protocol.
Grossberg DE, Stewart D (2001) Bacteremia Avances en Biomedicina 3:2325
due to periodontal probing: a clinical and 23. Squeri R, La Fauci V, Cannavo G, Lo Giudice
microbiological investigation. J Periodontol G, Sindoni L (2006) Identification of the
72:210214 microorganisms responsible for periodontopa-
15. Morillo JM, Lau L, Sanz M, Herrera D, Martin thy by Multiplex RT-PCR. J Prev Med Hyg
C, Silva A (2004) Quantitative real-time poly- 47:142145
merase chain reaction based on single copy 24. Morikawa M, Chiba T, Tomii N, Sato S,
gene sequence for detection of periodontal Takahashi Y, Konishi K, Numabe Y, Iwata K,
pathogens. J Clin Periodontol 31:10541060 Imai K (2008) Comparative analysis of puta-
16. Boutaga K, van Winkelhoff AJ, Vandenbroucke- tive periodontopathic bacteria by multiplex
Grauls CMJE, Savelkoul PHM (2003) polymerase chain reaction. J Periodontal Res
Comparison of real-time PCR and culture for 43:268274
Chapter 12
Methods to Study Antagonistic Activities Among

Oral Bacteria
Fengxia Qi and Jens Kreth
Abstract
Most bacteria in nature exist in multispecies communities known as biofilms. In the natural habitat where
resources (nutrient, space, etc.) are usually limited, individual species must compete or collaborate with
other neighboring species in order to perpetuate in the multispecies community. The human oral cavity is
colonized by >700 microbial species known as the indigenous microbiota. This indigenous flora normally
maintains an ecological balance through antagonistic as well as mutualistic interspecies interactions.
However, environmental perturbation may disrupt this balance, leading to overgrowth of pathogenic spe-
cies which could in turn initiate diseases such as dental caries (tooth decay) and periodontitis (gum dis-
ease). Understanding the mechanisms of diversity maintenance may help developing novel approaches to
manage these polymicrobial diseases. In this chapter, we focus on a well-characterized form of biochemi-
cal warfare: bacteriocins produced by Streptococcus mutans, a primary dental caries pathogen, and hydro-
gen peroxide (H2O2) produced by several oral commensal streptococci. We will describe detailed
methodologies on the competition assay, isolation, purification, and characterization of bacteriocins.
Key words Bacteriocins, Hydrogen peroxide (H2O2), Oral streptococci, Streptococcus mutans,
Interspecies competition, Biofilms, Luciferase reporter
1 Introduction
Most of the antibiotics we use today are produced by microbes,

and it is estimated that >99 % of bacterial species in nature produce
some type of antibiotics [13]. Although the ecological role of
these antibiotics is less studied, it is clear that their production is
for the protection of the producing species against other microbes
[2]. Bacteriocins are peptide antibiotics. Unlike the traditional
antibiotics, which are produced as secondary metabolites, bacte-
riocins are synthesized ribosomally. In general, there are two types
of bacteriocins, the lantibiotics and the non-lantibiotics. The lanti-
biotics are extensively modified peptides, containing dehydrated
threonine and serine residues and thioether bridges [4], while the
non-lantibiotics are unmodified peptides, which are comprised of
one or two components for activity. Bacteriocin production appears
203
204 Fengxia Qi and Jens Kreth
to be prevalent; nearly all sequenced bacterial genomes encode

bacteriocin-like genes, although most of them have not been
characterized.
Inter-species interactions among microbial species within the
same communities are well- documented phenomena in scientific
literature. The dental biofilm is a good model system for studying
interspecies interactions owing to its vast biodiversity (>700 bacte-
rial species) [58], high cell density (1011 cells/g wet weight) [9],
and easy accessibility. In addition, the oral cavity is an environment
with constant cycles of feast and famine and fluctuations of pH due
to food intake from the host. The high density and diversity of oral
biofilm community members coupled with a limited food supply
creates an environment that is conducive to fierce competition for
available resources. Oral streptococci seem to play an important
role in the initial colonization of the tooth surface and they consti-
tute about 80 % of early oral biofilm bacteria [10]. The same colo-
nization niche and a similar metabolic profile of oral streptococci
certainly played an important role in the evolution of anti competi-
tor activities. An interesting antagonistic relationship has been
investigated in detail between Streptococcus mutans and several
other commensal streptococci including Streptococcus sanguinis,
Streptococcus gordonii, and Streptococcus oligofermentans [11].
S. mutans is considered a major pathogen causing human den-
tal caries (also known as tooth decay) [12]. S. mutans is a copious
producer of both types of bacteriocins (named mutacins) [13]. S.
sanguinis, S. gordonii, and S. oligofermentans are abundant oral
commensals. Except for the reported association of some oral
streptococcal commensals with bacterial endocarditis they are con-
sidered benign, or even beneficial with regard to dental caries
[14, 15]. Our group has investigated the mechanisms of interspe-
cies interaction among oral bacteria including S. mutans, S. sangui-
nis, and S. gordonii for over a decade. We have shown that mutacin
production by S. mutans and hydrogen peroxide (H2O2) produc-
tion by oral commensals plays an important ecological role in the
competition between these species in the oral biofilm [16, 17].
Furthermore, in recent years the production of H2O2 by several
oral streptococci has been shown to influence other species [18].
For example, Pseudomonas aeruginosa, a pathogen associated with
cystic fibrosis-related pulmonary infections, can be successfully
antagonized by Streptococcus parasanguinis through H2O2 produc-
tion, although the mechanism of inhibition is different than that
described for the direct H2O2-dependent inhibition of S. mutans.
The antagonizing effect on P. aeruginosa is caused by an H2O2 and
nitrite dependent production of the reactive nitrogenous interme-
diate peroxynitrite [19]. Since the occurrence of oral streptococci
in the lungs of CF patients has been linked to improved lung func-
tion, antagonizing activities of streptococci, in general, could be
associated with the severity of polymicrobial diseases. The
Methods to Study Antagonistic Activities Among Oral Bacteria 205
techniques described in this chapter were developed from these

studies; however, they can be easily adapted to studying interspe-
cies interactions among other species.
2 Materials
2.1 Bacteriocin 1. BHI or TH agar plates: Dissolve 37 g/L Brain-Heart Infusion

Assay (BHI; Becton Dickinson) or 30 g/L Todd-Hewitt broth (TH;
Becton Dickinson) in deionized water (DI H2O), and add 15 g
bacteriological agar. Autoclave at 121 C for 30 min. Let cool to
~55 C and pour plates. Half-strength BHI or TH contains
18.5 g/L (BHI) or 15 g/L (TH), respectively (see Note 1).
2. BHI or TH soft agar: Same as above, but use 7.5 g agar. After
autoclaving, dispense 4 mL aliquots into glass tubes, and store
at 4 C. Before use, melt the agar in boiling water, or in a
microwave oven inside a water-filled beaker.
2.2 Biofilm Assay 1. Lab-Tek II Chamber Slide System (Nalge Nunc International;
and Confocal Laser Naperville, IL, USA).
Scanning Microscopy 2. CellTrackerTM Orange CMTMR (5-(and-6)-(((4-chloro-
methyl)benzoyl)amino)tetramethylrhodamine) (Molecular
Probes, Eugene, OR, USA), store at 70 C.
3. Sucrose (20 % stock) in DI H2O, filter-sterilized (0.22 m fil-
ter). Do not autoclave.
4. Confocal Laser Scanning Microscope.
2.3 H2O2 Detection 1. 10 mM phosphate buffer (pH 7.4): make 100 mM stock solu-
with Indicator Plates tion by mixing 19 mL of 100 mM monobasic sodium phos-
phate and 81 mL of 100 mM dibasic sodium phosphate. Filter
2.3.1 Enzymatic H2O2
sterilize (0.22 m membrane filter) and store at room
Detection
temperature.
2. o-Dianisidine dihydrochloride (ICN, Aurora, OH, USA).
3. Horseradish peroxidase (Pierce, Rockford, IL, USA).
4. Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA).
5. Leuco crystal violet (Sigma-Aldrich), dissolve powder directly
into BHI agar medium (after autoclaving) and pour plates.
6. 30 % H2O2 (Sigma-Aldrich).
7. CO2 incubator for aerobic incubation.
2.3.2 Nonenzymatic H2O2 Alternatively, a nonenzymatic plate assay can be used. The detec-
Detection tion of bacterial H2O2 production in this assay is dependent on the
reaction of hexacyanoferrate(III) and iron(III) in aqueous solution
producing a blue precipitate of Prussian blue in the presence of
H2O2 [20].
1. Dissolve 1 g of iron (III) chloride hexahydrate in 50 mL

H2O. Separately, dissolve potassium hexacyanoferrate (III) in
50 mL H2O.
2. Mix both solutions slowly after they are completely dissolved.
3. Prepare 900 mL BHI (37 g/900 mL) in H2O.
4. Mix 900 mL BHI with the 100 mL iron (III) chloride hexahy-
drate + potassium hexacyanoferrate (III) solution.
5. Add 15 g agar and autoclave. Allow agar to cool and pour
plates.
2.4 Isolation 1. Pharmacia AKTA Purifier (GMI) instrument.

and Purification 2. Trifluoroacetic Acid (TFA), make a 0.1 % solution with HPLC-
of Bacteriocin grade DI H2O, store at room temperature.
3. Methanol (HPLC grade), make 85 % solution with HPLC-
grade DI H2O. Store at room temperature.
4. AcetonitrileHPLC grade.
5. Chloroform.
6. Urea, make 5 M with DI H2O, store at room temperature.
2.5 Derivatization 1. 100 % Ethanol.

of Lantibiotics 2. 5 M NaOH in DI H2O, store at room temperature.
3. Ethanethiol.
4. Glacial acetic acid.
2.6 Cloning 1. Restriction enzymes (New England Biolabs, Ipswich, MA,

and Other Genetic USA), store at 20 C.
Techniques 2. T4 DNA ligase (New England Biolabs), store at 20 C.
3. Elongase enzyme mix (a mixture of Pyrococcus sp. thermostable
DNA polymerase and Taq DNA polymerase) (Life Technologies).
Store at 20 C.
4. TOPO TA Cloning kit, (Life Technologies). Store at 20 C.
5. E. coli DH5 competent cells, (Life Technologies). Store at
70 C.
6. Ampicillin, 100 mg/mL stock dissolved in 50 % (v/v) ethanol,
store at 20 C, use at 100 g/mL final concentration.
7. Kanamycin, 100 mg/mL stock dissolved in DI H2O, store at
20 C, use at 100 g/mL final concentration.
8. Spectinomycin, 150 mg/mL stock in DI H2O, store at 20 C,
use 150 g/mL final concentration.
9. LB broth (Becton Dickinson).
10. Agar (Becton Dickinson).
3 Methods
3.1 Competition 1. Since most, if not all, bacteriocins are produced under high cell
Assay on Plate Culture density, plate cultures are usually used to analyze interspecies
competition. Here, we use an example of competition between
S. mutans and S. sanguinis. The assay can be done by inoculat-
ing either species first as the early colonizer, then inoculating
the other species after overnight growth as the late colonizer.
Additionally, one could inoculate both species at the same
time, i.e., a simultaneous antagonism experiment.
2. Usually, an overnight culture is adjusted to an optical density
at 600 nm (OD600) of 0.5 in 50 % BHI and 10 L is spotted
onto half-strength (50 %) BHI plates as the early colonizer.
3. After an overnight incubation, 10 L of the competing species,
also adjusted to the same OD600, is spotted beside the early
colonizer as the late colonizer, or both species are inoculated at
the same time beside each other (simultaneous antagonism).
The plates are further incubated at 37 C anaerobically over-
night before cell growth is inspected. A typical outcome
between a pair of true competitors is illustrated in Fig. 1. In this
example, when the bacteriocin gene from S. mutans is inacti-
vated (Mut), S. sanguinis is no longer inhibited.
3.2 Competition 1. For competition assays in biofilms, overnight cultures of S.

Assay in Biofilms mutans or S. sanguinis are diluted 1:100 in 50 % BHI plus 0.1 %
sucrose and inoculated into a Lab-Tek II Chamber Slide.
2. The cultures are incubated at room temperature for 3 h to
allow cell attachment before the competing species is inocu-
lated, or both species are inoculated at the same time.
3. The biofilm is grown for 16 h at 37 C as a static culture.
CellTracker Orange is used to label all cells for 2 h before con-
focal microscopy.
4. For microscopy the Lab-Tek II Chamber Slide System is
modified: the objective slide is replaced by a thin cover-slide
Fig. 1 Interspecies competition assay between S. mutans (Sm) and S. sanguinis

(Ss). Mut+ = wild-type mutacin producer; Mut = mutacin mutant
for proper CLSM microscopy since most microscope lenses

have a shorter working distance and image acquisition would
be obscured by the thick objective slide.
5. CLSM is performed with a microscope equipped with detectors
and filter sets for monitoring red fluorescence (excitation wave-
length 540580 nm (560 CWL), dichroic mirror wavelength:
595 nm (LP), barrier wavelength 600660 nm (630 CWL)).
Images might be obtained with a 10 0.3 Plan-Neofluar and a
40 1.4 Plan-Neofluar oil objective.
3.3 H2O2 1. H2O2 production by S. sanguinis plays an important role in

Production Assay interspecies competition with S. mutans [16]. H2O2 produc-
tion is oxygen dependent and produced as by-product during
the enzymatic action of the pyruvate oxidase, SpxB [18].
The production of H2O2 by S. sanguinis in aerated liquid culture
is measured as follows: Samples (1 mL) are taken at the desired
time-points, centrifuged (16,000 g) for 5 min, and trans-
ferred (0.2 mL) to a new incubation tube.
2. A reaction solution is prepared fresh for each experiment
(0.8 mL of 10 mM phosphate buffer [pH 7.4] with 0.16 mM
o-dianisidine dihydrochloride, 1.2 g/mL of horseradish per-
oxidase, 0.02 % Triton X-100) and added to the reaction mix-
ture followed by incubation at 37 C for 20 min.
3. The absorbance at 570 nm is determined, and the concentra-
tion is calculated from a standard curve prepared for each
experiment from a 30 % H2O2 stock solution ranging from 0 to
500 nmol/mL.
4. To measure the effect of S. mutans on the H2O2 production of
S. sanguinis, an overnight culture of S. sanguinis is diluted to
107 cells/mL (OD600 = 0.025) and incubated aerobically at
37 C. After two doubling times, the cells are washed twice
with BHI and the OD600 is adjusted to 0.2.
5. One milliliter of the cell suspension is transferred to a tube,
and 1 mL of either BHI or S. mutans cell suspension
(OD600 = 0.2) is added. The cells are further incubated either as
a planktonic culture or as a cell pellet (16,000 g for 1.5 min)
with medium for 2 h before the H2O2 concentration is mea-
sured with the culture supernatant.
6. For the determination of H2O2 production on the plate, 10 L
of peroxidase (64 g) is added to a half-strength BHI plate
containing 1 mg/mL leuco crystal violet. After the liquid is
absorbed into the agar, 5 L of S. sanguinis is inoculated at the
same spot. After overnight incubation in a CO2 incubator, the
plate is inspected for the development of a purple color on and
around the colony.
3.4 Bacteriocin 1. To isolate bacteriocin producing bacteria from saliva, unstimu-

Activity Assay lated whole saliva is collected by asking the volunteers to
by Deferred expectorate into a sterile 1.5-mL microcentrifuge tube.
Antagonism (Plate 2. The saliva is first diluted 1:10 in phosphate buffered saline
Overlay) (PBS), and cells are dispersed by vortexing for 1 min. A tenfold
serial dilution is performed with the cell suspension, and a por-
tion (100 L) of each dilution is plated on BHI or TH plates.
3. The plates are incubated for 2 days in an anaerobic chamber
with 90 % N2, 5 % CO2, and 5 % H2 at 37 C, or in a candle jar
(microaerophilic conditions) in a regular 37 C incubator.
Plates with well-separated single colonies can be overlaid
directly with an indicator strain, or colonies can be transferred
to a new plate with toothpicks and grown for 12 days until
colonies are formed, and then the plate overlaid with an indica-
tor strain.
4. The indicator strain is grown in BHI or TH broth overnight,
and 0.5 mL of the overnight culture is mixed with 4 mL of
melted BHI or TH soft agar cooled to ~50 C. The mixture is
then poured onto the plate and incubated overnight under the
same conditions. A zone of inhibition of the indicator strain
suggests production of a bacteriocin (Fig. 2).
3.5 Isolation 1. Since most bacteriocins are produced when cell density is high,
of Bacteriocin plate culture is used initially to isolate bacteriocin (see Note 2).
In the case of mutacin I [21], TH plates are made, which con-
tain 0.3 % agarose in place of agar. A sterile PHWP membrane
(0.5 m pore size, Millipore) is placed on the plate surface, and
an overnight culture of the producing strain is spread onto the
membrane. The plate is incubated for 2 days for the bacterial
lawn to form on the membrane, and the membrane is then trans-
ferred onto a new plate. This process is repeated up to 8 transfers
or until the bacterial lawn stops to produce bacteriocin. This
should be tested with the overlay assay on a separate plate.
2. The spent plate is frozen at 70 C and thawed quickly at
60 C in a water bath. Upon freezing-and-thawing, the agarose
would disintegrate to release the liquid content containing the
bacteriocin. The liquid phase is separated from the agarose
debris by centrifugation (20,000 g for 30 min).
Fig. 2 Bacteriocin production assay

Fig. 3 Bacteriocin titer determination
3. Mutacin is extracted from the liquid phase by equal volumes of

chloroform. The emulsion at the chloroformaqueous inter-
face, which contains mutacin I, is collected by centrifugation,
and the pellet dried under a stream of air.
4. The pellet is suspended in 5 M urea. To assay for activity of the
crude mutacin extract, a twofold dilution of the crude extract
is made and 10 L of each dilution is spotted onto a pre-dried
TH plate. After the liquid spot is dry, the plate is overlaid with
the indicator strain (see Subheading 3.1), and the plate is incu-
bated overnight at 37 C anaerobically. One arbitrary unit of
activity is defined as the highest dilution that exhibits a clear
zone of inhibition of the indicator strain (Fig. 3).
3.6 Purification 1. Since all bacteriocins are small peptides, reverse phase HPLC is
of Bacteriocin generally used for purification. In the case of mutacin I and III
[22], the crude extract is applied to a Source 15RPC column
and eluted with a fragmented gradient of buffer A (0.1 % TFA)
and buffer B (0.085 % TFA in 80 % methanol) with the AKTA
purifier and the UNICORN control system (Amersham
Pharmacia Biotech, Piscataway, NJ) (see Note 3).
2. A 1-mL eluent is collected and tested for activities using the
methods described in Subheading 3.2.
3. The active fractions are pooled and dried in a lyophilizer. The
pellet is redissolved in 0.25 % TFA and subjected to a second
round of purification with the same column and protocol.
4. The single active peak fraction is collected, dried in a lyophilizer,
and used for sequence analysis and electrospray-ionization
mass spectrometry (EIMS). A typical HPLC profile is presented
in Fig. 4.
3.7 Sequencing 1. For nonmodified bacteriocins, a simple N-terminal peptide

of the Purified sequencing can be performed using automated Edman degra-
Bacteriocin dation by any protein sequencing service.
2. For lantibiotics, chemical modifications of the peptide should
be made to reduce the thioether bridges and dehydrated amino
mAU %B
1000 Activity analysis
80
7 5
6
Fractions 60
Absorbance
Fragmented elution gradient
% Buffer B
500
40
20
0
F3 F1 2 3 4 5 6 7 8 9 Waste 0
0.0 10.0 20.0 30.0 ml
Fractions of eluent collected
Fig. 4 HPLC profile of mutacin I
acids prior to sequencing via automated Edman degradation

procedures (see Note 4).
3. For chemical modification, 50 g of purified mutacin I is
dried under vacuum and resuspended in 90 L of a derivatiza-
tion mixture consisting of 280 L of ethanol, 200 L of
water, 65 L of 5 M sodium hydroxide, and 60 L of ethane-
thiol. The reaction proceeds at 50 C for 1 h under nitrogen
and is then stopped by the addition of 2 L of acetic acid.
The reaction mixture is dried under vacuum and washed
three times with 50 % ethanol. The pellet is resuspended in
10 L of 50 % acetonitrile with 1 % formic acid for EIMS
analysis and N-terminal peptide sequencing by Edman
degradation.
3.8 Isolation 1. After sequencing the bacteriocin peptide, the structural gene
of Bacteriocin can be isolated via a circular PCR strategy (see Fig. 5). Generally,
Structural Genes a pair of degenerate primers is designed based on the peptide
by Reverse Genetics sequence and the codon usage of the producing strain. One
(See Note 10) primer (reverse) is pointing upstream from the 5 portion of
the derived DNA fragment and the other (forward) faces
downstream from the 3 portion of the DNA fragment.
Peptide sequence
Degenerate primer
reverse forward
Universal primer
Digested chromosomal DNA

Digested plasmid
Self ligation Digested chromosomal DNA
ligation
PCR PCR
sequencing sequencing
Circular PCR SSP-PCR
Fig. 5 Strategies for cloning a bacteriocin gene via reverse genetics
2. The chromosomal DNA of the producing strain is digested to

completion with a panel of restriction enzymes and self-ligated.
The ligation mixtures are used as templates in PCR reactions
with the reverse and forward primers (see Notes 57).
3. The PCR products are then cloned and sequenced. The
upstream and downstream sequences could be distinguished at
the unique restriction site where the chromosomal DNA is
initially cut. With the upstream and downstream sequences
available, the structural gene can be re-confirmed by regular
PCR using primers designed based on the upstream and down-
stream sequences.
4. Alternatively, the structural gene can be obtained by a single
specific primer PCR (SSP-PCR) (see Fig. 5). In this strategy, a
degenerate primer is designed based on the peptide sequence.
5. The chromosomal DNA is digested by a set of restriction
enzymes to completion, and the same set of enzymes are used
to digest a commonly-available cloning vector such as pUC or
pBluescript vectors.
6. The same enzyme digested chromosomal DNA and the plasmid
is ligated, and the ligation mixture is used as template for PCR
with the specific primer and one of the universal primers.
7. The PCR product is then sequenced using the universal

primer.
8. For both strategies, 1 g of chromosomal DNA is normally
used for digestion in 20 L reaction mixture. The PCR condi-
tions are 94 C for 4 min, 50 C for 1 min, and 72 C for
5 min for 1 cycle; 94 C for 1 min, 50 C for 1 min, and 72 C
for 3 min for 25 cycles; a final cycle at 94 C for 1 min,
50 C/1 min, and 72 C/10 min. The PCR reaction contains
a mixture of Taq and Pfu DNA polymerases and to ensure high
processivity and fidelity.
3.9 Mutagenesis 1. To study the function of the bacteriocin in interspecies compe-

via Single- tition, an isogenic strain defective in bacteriocin production is
and Double-Crossover needed. Two strategies can be used to inactivate the bacterio-
cin biosynthesis gene by homologous recombination: (a)
single-crossover insertional inactivation, and (b) allelic replace-
ment via a double-crossover mechanism (Fig. 6).
2. Single-crossover is utilized for mutagenesis of a gene, which is
either in a single gene system or is the last gene in a multigene
operon. In general, a ~300 bp internal fragment of the target
gene is amplified by PCR using primers with restriction sites
incorporated at the 5-ends. The PCR product is then digested
with the appropriate restriction enzyme and ligated into a
suicide vector (i.e., pFW5) digested with the same enzyme.
The ligation mixture is transformed into E. coli, and the recom-
binant plasmid is isolated from positive clones.
3. The plasmid is then transformed into the bacteriocin-produc-
ing strain via natural transformation or electroporation depend-
ing on the specific strain. Transformants are propagated on
selective agar plates and tested for bacteriocin production by
using the plate overlay method (see Note 8).
4. A double-crossover strategy is used to specifically inactivate,
usually by insertion of an antibiotic cassette, individual or mul-
tiple genes in an operon in order to avoid polar effects on the
downstream gene. The antibiotic cassette typically contains its
own promoter but lacks a transcription terminator. For double-
crossover, the simplest method is 3-piece PCR ligation. Briefly,
a 1-kb fragment of the upstream and downstream regions of
the target gene, as well as the antibiotic cassette, is amplified by
PCR. In the primers at the junction of each fragment, an
18-nucleotide overlapping sequence is incorporated in the
primer, which is homologous to the antibiotic cassette sequence
(see Note 9).
5. After the first PCR, the three fragments are purified with a
commercial PCR purification kit (e.g., Qiagen Qiaquick kit)
and eluted with 40 L elution buffer. One microliter of each
Fig. 6 (a) Insertional inactivation by single-crossover integration. (b) Construction of an allelic replacement
construct via a 3-piece PCR ligation strategy
fragment is then mixed and used as template for a second PCR

employing the two primers on both ends of the ligated fragment.
The PCR product is purified using a spin column and trans-
formed directly into the wild-type bacteriocin producer strain.
Transformants are selected on selective plates and subsequently
tested for bacteriocin production.
3.10 Gene 1. The firefly luciferase is a good reporter for quantification of

Expression Analysis promoter activities. For constructing a reporter gene fusion,
by Reporter Fusions the promoter region of the target gene is amplified from chro-
mosomal DNA with primers incorporating restriction enzyme
sites and cloned into plasmid pFW5-luc [23]. pFW5 is a sui-
cide plasmid, which replicates in E. coli but not in S. mutans
unless it is integrated into the chromosome via homologous
recombination at the promoter locus.
2. After integration into the chromosome, the promoter activity
of the target gene under different conditions can be monitored
by measuring the luciferase activity. The luciferase activity
should be normalized to cell density or protein content of the
sample.
3.11 Luciferase 1. To test for luciferase activity, 25 L of 1 mM d-luciferin (Sigma;

Assay Using Live Cells St Louis, MO) suspended in 100 mM citrate buffer, pH 6, is
added to 100 L of the cell culture.
2. To ensure sufficient levels of intracellular ATP pools for lucif-
erase activity, cells are recharged with 1 % glucose for 10 min
prior to luciferin addition.
3. Luciferase activity is measured by using a TD 20/20 luminom-
eter (Turner Biosystems; Sunnyvale, CA).
4 Notes
1. Bacteriocin production is sensitive to the growth conditions

and the detection requires a sensitive indicator strain. The
absence of a zone of inhibition when tested under a particular
condition does not necessarily mean the strain is a nonpro-
ducer. One needs to test the producer bacterium under differ-
ent conditions and employing different indicator strains. Our
experience suggests that bacteriocin production is a stress
response. Therefore, growth of the strain in a very rich medium
such as BHI tends to inhibit bacteriocin production. Hence,
diluted BHI or TH could be used. Interestingly, a very
nutrient-poor medium is also not conducive to bacteriocin
production. Another important observation is that bacteriocin
production is more prevalent among newly-isolated clinical
strains, and the ability tends to diminish or disappear upon
repeated laboratory passage.
2. Although some bacteriocins can be produced by planktonic
(broth) culture grown to late logarithmic or early stationary
phase, the production level is usually lower than that observed
on solid media. For initial isolation, it is important that the
yield is high. Another advantage of using a culture grown on
an agar plate is that an overlay can be made on a parallel plate

culture to verify that the bacteriocin is indeed produced. Using
the filter membrane as a supporting substratum for the bacte-
rial lawn favors bacteriocin production. In our experience, the
highest yield is obtained at passages 25, and the yield
declines after passage 6, possibly due to the aging of the bac-
terial population. Another advantage of using a membrane is
that no subsequently filtering of the supernatant is necessary,
because there is no bacterial cell contamination of the agarose
plate. The membrane allows bacteriocins to diffuse into the
medium underneath while preventing bacterial cells from
going through.
3. As mutacins I and III are fairly hydrophobic molecules, chlo-
roform is used for their extraction. For bacteriocins that are
less hydrophobic, ammonium sulfate ([NH3]2SO4) precipita-
tion can be used. For HPLC-based purification, a linear gradi-
ent is typically used initially to determine at what fraction the
active component is eluted, then a fragmented elution is used
to further separate other components from the active one if the
active peak does not appear to be pure (a pure peak is usually
smooth and symmetrical in shape).
4. N-terminal peptide sequencing using automated Edman deg-
radation chemistry is blocked by dehydrated amino acids or
thioether bridges. If this happens during sequencing of the
purified bacteriocin, it would suggest that the bacteriocin is a
lantibiotic. The number of dehydrated amino acids and thio-
ether bridges can be deduced by comparing the molecular
mass of the peptide and its modified form by EIMS.
5. The circular PCR or SSP-PCR strategy is used because it is
normally difficult to obtain a full-length sequence of the bac-
teriocin peptide, and this is especially true for lantibiotic pep-
tides. However, a 6-7 amino acid sequence from the
N-terminus is relatively easy to obtain. Therefore, by using
these strategies, forward and reverse degenerate primers can
be designed based on this short sequence in order to fish out
the structural gene.
6. An important factor to consider using circular PCR is the con-
centration of the digested chromosomal DNA for self-ligation.
To facilitate self-ligation (intramolecular ligations), less DNA
is better. Our experience is to set up a series of ligations with
different concentrations of DNA and use 1 L of each concen-
tration in PCR experiments.
7. Available sequenced genomes can be used to screen for bacterio-
cin genes, if the producer bacterium is known (http://www.ncbi.
nlm.nih.gov/genomes/lproks.cgi).
8. One caveat for the single-crossover strategy is that if the gene
is too small, like in the case of most bacteriocin structural
genes, it will be difficult to inactivate by the single-crossover.

In this case, an allelic replacement strategy should be used.
9. One caveat for the double-crossover strategy is that the expres-
sion pattern of the downstream genes may be changed because
all of them will be transcribed from the antibiotic cassette pro-
moter. This could create a problem for genes whose expression
pattern is important for its biological function. One strategy to
overcome this problem is to include a transcription terminator
at the 3-end of the antibiotic cassette followed by the native
promoter of the target gene. An alternative (and cleaner) strat-
egy is to use a markerless in-frame deletion system. So far,
there has not been an ideal in-frame deletion system available.
We have created a galactose-based in-frame deletion system for
use in S. mutans [24]. This system can be used for selected
genes as the required deletion of the galKT genes (to facilitate
selection) may obscure other phenotypes.
References
1. Klaenhammer TR (1988) Bacteriocins of lactic 10. Rosan B, Lamont RJ (2000) Dental plaque
acid bacteria. Biochimie 70:337349 formation. Microbes Infect 2:15991607
2. Riley MA, Wertz JE (2002) Bacteriocin diver- 11. Kreth J, Merritt J, Qi F (2009) Bacterial and
sity: ecological and evolutionary perspectives. host interactions of oral streptococci. DNA
Biochimie 84:357364 Cell Biol 28:397403
3. Riley MA, Wertz JE (2002) Bacteriocins: evo- 12. Loesche WJ (1986) The identification of bac-
lution, ecology, and application. Ann Rev teria associated with periodontal disease and
Microbiol 56:117137 dental caries by enzymatic methods. Oral
4. Sahl HG, Bierbaum G (1998) Lantibiotics: Microbiol Immunol 1:6572
biosynthesis and biological activities of 13. Merritt J, Qi F (2012) The mutacins of
uniquely modified peptides from gram-positive Streptococcus mutans: regulation and ecology.
bacteria. Ann Rev Microbiol 52:4179 Mol Oral Microbiol 27:5769
5. Aas JA, Paster BJ, Stokes LN, Olsen I, Dewhirst 14. Becker MR, Paster BJ, Leys EJ, Moeschberger
FE (2005) Defining the normal bacterial flora ML, Kenyon SG, Galvin JL, Boches SK,
of the oral cavity. J Clin Microbiol 43: Dewhirst FE, Griffen AL (2002) Molecular
57215732 analysis of bacterial species associated with
6. Paster BJ, Boches SK, Galvin JL, Ericson RE, childhood caries. J Clin Microbiol 40:
Lau CN, Levanos VA, Sahasrabudhe A, 10011009
Dewhirst FE (2001) Bacterial diversity in 15. Caufield PW, Dasanayake AP, Li Y, Pan Y, Hsu
human subgingival plaque. J Bacteriol J, Hardin JM (2000) Natural history of
183:37703783 Streptococcus sanguinis in the oral cavity of
7. Paster BJ, Olsen I, Aas JA, Dewhirst FE (2006) infants: evidence for a discrete window of infec-
The breadth of bacterial diversity in the human tivity. Infect Immun 68:40184023
periodontal pocket and other oral sites. 16. Kreth J, Merritt J, Shi W, Qi F (2005)
Periodontol 2000 42:8087 Competition and coexistence between
8. Socransky SS, Haffajee AD, Cugini MA, Smith Streptococcus mutans and Streptococcus sanguinis
C, Kent RL Jr (1998) Microbial complexes in in the dental biofilm. J Bacteriol 187:
subgingival plaque. J Clin Periodontol 71937203
25:134144 17. Qi F, Chen P, Caufield PW (2001) The group
9. Hamilton IA (2000) Ecological basis for dental I strain of Streptococcus mutans, UA140, pro-
caries. In: Kuramitsu HK, Ellen RP (eds) Oral duces both the lantibiotic mutacin I and a non-
bacterial ecology. Horizon Scientific Press, lantibiotic bacteriocin, mutacin IV. Appl
Wymondham, Norfolk, UK, pp 215275 Environ Microbiol 67:1521
18. Zhu L, Kreth J (2012) The role of hydrogen synthesis genes. Appl Environ Microbiol
peroxide in environmental adaptation of oral 66:32213229
microbial communities. Oxid Med Cell Longev 22. Qi F, Chen P, Caufield PW (1999) Purification
2012:717843 of mutacin III from group III Streptococcus
19. Scoffield JA, Wu H (2015) Oral streptococci and mutans UA787 and genetic analyses of muta-
nitrite-mediated interference of Pseudomonas cin III biosynthesis genes. Appl Environ
aeruginosa. Infect Immun 83:101107 Microbiol 65:38803887
20. Giacaman RA, Torres S, Gomez Y, Munoz- 23. Podbielski A, Spellerberg B, Woischnik M,
Sandoval C, Kreth J (2015) Correlation of Pohl B, Lutticken R (1996) Novel series of
Streptococcus mutans and Streptococcus sangui- plasmid vectors for gene inactivation and
nis colonization and ex vivo hydrogen peroxide expression analysis in group A streptococci
production in carious lesion-free and high car- (GAS). Gene 177:137147
ies adults. Arch Oral Biol 60:154159 24. Merritt J, Tsang P, Zheng L, Shi W, Qi F
21. Qi F, Chen P, Caufield PW (2000) Purification (2007) Construction of a counterselection-
and biochemical characterization of mutacin I based in-frame deletion system for genetic
from the group I strain of Streptococcus mutans, studies of Streptococcus mutans. Oral Microbiol
CH43, and genetic analysis of mutacin I bio- Immunol 22:95102
Chapter 13
Natural Transformation of Oral Streptococci by Use

of Synthetic Pheromones
Gabriela Salvadori, Roger Junges, Rabia Khan, Heidi A. mdal,
Donald A. Morrison, and Fernanda C. Petersen
Abstract
The discovery that Streptococcus pneumoniae uses a competence-stimulating peptide (CSP) to induce com-
petence for natural transformation, and that other species of the mitis and the anginosus streptococcal
groups use a similar system, has expanded the tools to explore gene function and regulatory pathways in
streptococci. Two other classes of pheromones have been discovered since then, comprising the bacterio-
cin-inducing peptide class found in Streptococcus mutans (also named CSP, although different from the
former) and the SigX-inducing peptides (XIP), in the mutans, salivarius, bovis, and pyogenes groups of
streptococci. The three classes of peptide pheromones can be ordered from peptide synthesis services at
affordable prices, and used in transformation assays to obtain competent cultures consistently at levels usu-
ally higher than those achieved during spontaneous competence. In this chapter, we present protocols for
natural transformation of oral streptococci that are based on the use of synthetic pheromones, with exam-
ples of conditions optimized for transformation of S. mutans and Streptococcus mitis.
Key words Oral streptococcus, Streptococcus mutans, Streptococcus mitis, Competence, Natural trans-
formation, Competence-stimulating peptide (CSP), SigX-inducing peptide (XIP)
1 Introduction
Natural genetic transformation is widespread in oral streptococci,

including the mutans, the salivarius, the mitis, and the anginosus
groups [13]. The mechanisms depend on streptococci entering
competence, a physiological state triggered by pheromones, and
characterized by the induction in expression of several gene clus-
ters, including those essential for DNA uptake and recombination.
Streptococcal competence pheromones can be divided into
three classes: (1) competence stimulating peptides (CSPs), (2) SigX-
inducing peptides (XIPs), and (3) bacteriocin-inducing peptides
with competence-inducing activity. The first class, used by the mitis
and anginosus groups, is sensed by a two-component system com-
prising a histidine kinase (ComD) that binds the extracellular
219
220 Gabriela Salvadori et al.
cognate CSPs, leading to the phosphorylation of the response regu-

lator ComE [4, 5]. The XIP class is used by most other streptococci,
and is transported inside the cells by Opp oligopeptide permeases,
leading to the activation of intracellular Rgg-type regulators [69].
The last class, comprising bacteriocin-inducing peptides, is only
known in S. mutans. Although primarily involved in bacteriocin
induction, these peptides, also named CSPs, do induce competence,
but in contrast to the CSP-class in the mitis and anginosus groups,
they act indirectly by stimulating the XIP-class [10, 11]. All classes
of pheromones lead, ultimately, to the induction of the alternative
sigma factor SigX, which then induces the expression of competence
genes required for DNA uptake and recombination.
Spontaneous competence development under laboratory con-
ditions is in practice not always reliable, and subtle changes in fac-
tors such as medium batch, medium composition, pH, and growth
temperatures, may dramatically affect competence levels [12, 13].
The introduction of synthetic pheromones in protocols for natural
transformation has overcome many of these difficulties, by bypass-
ing stringent conditions for production and secretion of natural
pheromones, and by prolonging the transient nature of the compe-
tent state in streptococci [5]. The higher transformation efficiencies
and reproducible results usually achieved with synthetic phero-
mones can be further optimized by the choice of growth medium
and DNA donor, to reach levels that allow the practical use of mark-
erless strategies for direct genome editing, as recently demonstrated
for S. mutans and S. pneumoniae (see Chapter 14) [14].
This chapter describes transformation protocols for oral strep-
tococci based on competence stimulated by synthetic pheromones.
We present examples of optimized assays for S. mutans UA159
(XIP and bacteriocin-inducing classes of pheromones) and the S.
mitis type strain CCUG 31611T (CSP class of pheromone), and
general protocols with CSP that have been used for other strains of
the mitis group, as well as for the anginosus group. These can be
used as starting points for transformation of other streptococcal
strains and species for which optimal protocols have not yet been
established.
2 Materials
2.1 Competence 1. Agar plates: Todd-Hewitt Broth (THB) 30 g/L (Difco

Induction Using Laboratories, Detroit, MI, USA). Add 15 g/L of agar to the
Synthetic Peptides medium (VWR Chemicals, Radnor, PA, USA), and autoclave
at 121 C for 15 min. For selective agar plates, the appropriate
antibiotic(s) should be added to the medium once it has cooled
to below 60 C, and the media should be stored under condi-
tions appropriate for the antibiotics (see Note 1).
2. Liquid media: For transformation of S. mutans: Tryptone Soya
Broth (Thermo Scientific, Waltham, MA, USA)) 30 g/L is
Natural Transformation 221
used for stimulation with CSP. For stimulation with XIP,

chemically defined medium (CDM) [15], prepared from con-
centrated stock solutions with 1 % glucose is used for growth,
stock preparation of streptococcal pre-cultures and transfor-
mation assays; for transformation of the anginosus group:
THB-HS broth comprising Todd Hewitt Broth (THB; Difco
Laboratories) supplemented with 210 % (v/v) heat-inacti-
vated horse serum (HS); for transformation of Streptococcus
mitis: C + YYB [16], a chemically defined medium derived from
the previously described C + Y medium [17], but with an
increased concentration of yeast extract and bovine serum
albumin; for other oral streptococci: THY (see Notes 24).
3. Synthetic peptides: Synthetic pheromones specific for the strain
or species used may be ordered from peptide synthesis services.
Examples of pheromones that have been used in transforma-
tion of streptococci are shown in Table 1. A stock solution of
CSP is prepared by resuspending the lyophilized peptide in
distilled water to a concentration of 10 mM, and storing it at
Table 1
Sequence of CSP and XIP from selected strains of oral streptococci for which the synthetic
pheromones have been shown to induce competence
Group Strain CSP sequencea XIP sequence

Mutans S. mutans UA159G SGSLSTFFRLFNRSFTQA GLDWWSL
(18-CSP)
SGSLSTFFRLFNRSFTQALGK
(21-CSP)
Salivarius S. salivarius JIM8777/ PYFTGCL
CCHSS3G PFFMIYY
S. vestibularis ACTC 49124/
FO396
Mitis S. mitis CCUG 31611TG EIRQTHNIFFNFFKRR
S. gordonii Challis CH1G DVRSNKIRLWWENIFFNKK
S. gordonii NCTC 7865T, DIRHRINNSIWRDIFLKRK
S. sanguinis SK36G DLRGVPNPWGWIFGR
S. oligofermentans LMG DSRNIFLKIKFKKK
21535T DLRNIFLKIKFKKK
S. cristatus ACTC 51100G DKRLTYFITNLFPKRKK
S. infantis ATCC 700779G EMRLPKILRDFIFPRKK
S. oralis COL19G
Anginosus S. anginosus NCTC 10713T DSRIRMGFDFSKLFGK
S. constellatus NCTC 11325T DSRIRMGFDFSKLFGK
S. intermedius NCTC 11324T DSRIRMGFDFSKLFGK
T type strain
G genome sequence available
a
In S. mutans the CSP belongs to the class of bacteriocin-inducing peptides, and differs from the CSP class identified in
the mitis and anginosus groups
20 C. Working solutions of 10 or 100 M are aliquoted and

stored at 20 C. Lyophilized XIP (GenScript, Piscataway, NJ,
USA) is reconstituted with 20 L dimethyl sulfoxide (DMSO)
(Sigma-Aldrich, St. Louis, MI, USA) to which 1 mL distilled
water is added to give a final concentration of 10 mM (stored
at 20 C). Working stocks are prepared at 100 M by dilu-
tion in distilled water (see Note 5). For the CSP and XIP
sequences, see Subheading 3.3, step 4.
4. DNA donor: Amplicons and replicative plasmid used in the
experiments are listed in Table 2. For maximal recovery of
transformants, use purified DNA at saturation levels (see
Notes 6 and 9).
5. DNaseI (Roche, DNaseI recombinant, 10 U/L): DNaseI is
used to degrade DNA not taken up by the cells. This step is
usually omitted if the aim is to obtain maximal number of
transformants.
3 Methods
Compared with protocols based on spontaneous competence, the

use of synthetic pheromones leads often to higher transformation
efficiencies, a better control over the time of competence
Table 2
List of strains, primers, plasmids, and amplicons used in this study
Strain Description
S. mitis CCUG31611T Wild-type S. mitis biovar 1 type strain, corresponds to NCTC 12261T
S. mutans UA159 Wild-type S. mutans UA159
MI074 CCUG31611, but SM12261_0092::Kan; KanR
SM045 UA159, but dexA::kan; KanR
Primers Sequence (5 to 3)
FP906 ATTCACCCCAAAAAGTGCTG
FP907 ATAATATGCGGACGCTGAGG
FP1163 CATCTTGATAGCGTGGCTCA
FP1166 TTGAATTGAGACGGATTGGA
Plasmid Marker
pVA838 ErmR
Amplicon Description
aRJ02 FP906/FP9076.3 kbKanR, from SM045
aRJ21 FP1163/FP11666.9 kb - KanR, from MI074
development, and may extend transformation to a wider range of

strains. The peptides are stable, and offer a straightforward
approach for transformation of strains in which the pheromone
sequence is known.
We present two protocols for transformation of S. mutans, the
first optimized for transformation by XIP using CDM as the
growth medium [14]. The high efficiency of this method is par-
ticularly useful for direct markerless construction of mutants (see
Chapter 14). The second protocol involves the use of synthetic
CSP (bacteriocin-inducing class) in rich medium. Although only a
fraction of the S. mutans population becomes competent in rich
medium, compared with stimulation by XIP in CDM, transforma-
tion can reach levels sufficiently high for marker-based construc-
tions [11, 18]. The advantage of using rich medium such as TSB is
that there is no need for the availability of an extensive list of ingre-
dients such as that used for CDM preparation and it is much sim-
pler to prepare.
Genetic manipulation of S. mitis by natural transformation is
known to be challenging. Here, we present a procedure for S. mitis
transformation in THB-HS broth and an optimized protocol using a
semi-defined medium known to support endogenous competence.
In the mitis group, CSPs often vary among the different spe-
cies and in some cases among different strains as well. Thus, it may
be necessary to characterize the CSP-encoding gene in strains for
which the CSP sequence is unknown. Strategies that have been
used for CSP gene identification based on PCR amplification and
sequencing are therefore presented.
3.1 Transformation Given the influence of (1) as yet undefined environmental condi-
Efficiency/Kinetics tions, (2) the transient nature of competence, and (3) strain to
Protocol Using strain variations in transformation efficiencies, one may choose to
Synthetic XIP in CDM run kinetic experiments before establishing routine transformation
in S. mutans protocols. The following protocol has been used to determine the
kinetics of competence development in S. mutans UA159 using
synthetic XIP and PCR amplicon with antibiotic marker as donor
DNA (see Fig. 1).
1. Stock cultures are stored at 80 C or 20 C in 30 %
glycerol.
2. Plate the desired bacteria on THB agar plates. Incubate at
37 C overnight in 5 % CO2.
3. Preparation of pre-cultures: Inoculate 310 colonies using a
sterile transfer loop in 5 mL of CDM. Incubate at 37 C in 5 %
CO2 for 34 h or until absorbance at 600 nm (A600) = ~0.5.
Store the cells in 15 % glycerol at 70 C or use it directly in
the transformation experiments (see Note 7).
4. Dilute the pre-culture culture 1:10 in CDM (A600 = ~ 0.05)
and incubate at 37 C in air. Thaw the specific CSP on ice
while waiting for the next step.
Fig. 1 Kinetics of Streptococcus mutans UA159 competence development in the presence of (a) synthetic XIP
during growth in CDM and (b) synthetic CSP in TSB. Transformation used the 7-kb PCR amplicon (aRJ02) as
donor DNA. The dots represent transformation efficiency values and the triangles are the corresponding absor-
bance values at 600 nm (A600) of the growing culture corresponding to the time at which DNA was added, and
are averages of three replicates. XIP or CSP was added at time 0. For each time point, DNase I was added
after 20 min DNA exposure, and incubation proceeded for 40 min before plating the culture on nonselective
and selective antibiotic plates. Bars represent standard error of the mean (SEM)
5. After 1.52 h incubation (A600 = ~0.080.1 or CFU = ~7 106/

mL), add XIP (see Table 1) to the culture to a final concentration
of 1 M, and continue incubating at 37 C in air (see Note 8).
6. Add 50100 ng donor DNA to a 100 L aliquot of cells at dif-
ferent time points (see Fig. 1). Mix it by gently tapping the base
of the microfuge tube or gently stir with a micropipette tip.
Incubate the culture at 37 C for 20 min (see Notes 911).
7. Add 200 L of warm CDM containing 20 U/mL DNaseI
(Roche, DNaseI recombinant, 10 U/L) to each aliquot of
the competent cells, and proceed with incubation at 37 C in
air for another 40 min.
8. Spread 20200 L from serial dilutions of the transformation
mix on selective or nonselective THB agar plates. Wait for the
liquid to dry on the agar. Invert the plates and incubate them
at 37 C in 5 % CO2 for 2448 h.
9. Count the colonies in each plate and calculate the transforma-
tion efficiency using the formula below:
CFU transformants 100

Transformation efficiency ( % ) =
Total CFU
10. Select at least three colonies (putative transformants) for fur-

ther characterization. The colonies should be individually
transferred to separate tubes with fresh TSB containing the
appropriate antibiotic and incubated at 37 C in 5 % CO2 for
24 h.
11. Screen the colonies to verify whether the DNA construct was
correctly incorporated in the mutant. Methods based on PCR
using primers designed to provide amplicons that distinguish
the mutants from the wild-type may be used for this purpose.
Carry out the next two steps if the mutants are used in down-
stream applications.
12. Plate the selected bacteria in the presence of the antibiotic and
incubate them at 37 C in 5 % CO2 for 2448 h. Ensure at least
two passages in antibiotic-containing media (see Note 12).
13. Grow the selected bacteria in THB until exponential phase is
reached, and store the culture in 15 % glycerol at 20 C or
70 C.
3.2 Transformation These are simplified protocols that we have used in experiments in
Protocols for which determination of the kinetics of competence are not the
Downstream focus, or in which acceptable efficiency levels are obtained without
Applications Using the need for further optimization steps.
Synthetic Pheromones
3.2.1 Streptococcus This protocol has been used for transformation of different S.
mutans mutans strains using DNA with antibiotic marker (see Note 13).
Transformation Using 1. Follow steps 1 and 2 described in Subheading 3.1.
Synthetic CSPs 2. Inoculate 310 colonies using a sterile transfer loop in 5 mL of
in Rich Medium TSB. Incubate at 37 C in 5 % CO2 for 34 h (A600 = ~0.5).
Store the cells in 15 % glycerol at 80 C or use it directly in
the transformation experiments.
3. Dilute the pre-culture 1:10 in TSB and prepare 5001000 L
aliquots in 1.5 mL Eppendorf tubes. Add 18-CSP to a final
concentration of 50 nM and donor DNA, and incubate at
37 C in air for 2.53 h (see Notes 4 and 9).
4. Follow the steps 813 described in Subheading 3.1.
3.2.2 Streptococcus Here we present two protocols: one optimized protocol in semi-
mitis defined medium, and other in THB-HS medium. By following the
protocol in semi-defined medium using a 7-kb PCR amplicon as
donor DNA, we achieved 3040 % transformation efficiency while
in THB-HS the efficiency was only ~0.01 % (see Fig. 2) (see Note 4).
Transformation Using This protocol has been optimized for transformation of the S. mitis
Synthetic CSP type strain CCUG 31611 (corresponds to NCTC 12261). Pre-
in C + YYB Medium culture preparation:
1. Plate the desired bacteria in a blood plate without antibiotics.
Incubate at 37 C overnight in 5 % CO2.
2. Pre-culture preparation: Inoculate 310 colonies using a sterile
transfer loop in 5 mL of TSB. Incubate at 37 C in 5 % CO2 for
34 h (A600 = ~0.5). Store the cells in 15 % glycerol at 80 C or
use it directly in the transformation experiments (see Note 7).
Fig. 2 S. mitis transformation in THB-HS or C + YYB using pVA838 (1 g/mL) or a

7 kb amplicon (aRJ21) (150 ng/mL) as donor DNA, and CSP at a final concentra-
tion of 300 nM. Average of 3 replicates. Error bars represent standard error
Transformation:
1. Inoculate 10 L of the pre-culture in 900 L C + YYB. The vol-
umes can be scaled up but the 1:100 dilution should be
maintained.
2. Incubate at 37 C in 5 % CO2 for 23 h (A600 = ~0.04).
3. Add CSP at a final concentration of 300 nM and donor DNA (see
Note 14). Saturated concentrations for DNA are in the range of
100150 ng per mL for PCR amplicons and above 10 g for
genomic DNA and above 1 g for plasmid DNA (pVA838).
4. Incubate at 37 C in air for 3 h (see Note 8).
5. Follow the steps 813 as described in Subheading 3.1 for S.
mutans.
Transformation Using Pre-culture preparation:

Synthetic CSP
1. Inoculate 5 mL THY with ~10 L of the stock culture using a
in Rich Medium
sterile transfer loop, and incubate at 37 C overnight in 5 %
CO2 microaerophilic atmosphere.
2. Dilute the ON culture 1:100 in THY and incubate at 37 C in
a 5 % CO2 atmosphere for 45 h (A600 = ~0.3). Store the cells in
10 % glycerol at 70 C, or use it directly in the transformation
experiments.
Transformation:
1. Inoculate 100 L of the pre-culture in 900 L THB-HS. The
volumes can be scaled up but the 1:10 dilution should be
maintained.
2. Add CSP at a final concentration of 200 nM and transforming

DNA at the concentration described above. Incubate at 37 C
in air for 34 h (see Note 8).
3.2.3 The Anginosus This protocol, slightly modified from Gaustad and Morrison [19],
Group: Streptococcus has allowed transformation of S. intermedius, S. anginosus, and S.
intermedius, Streptococcus constellatus (see Notes 4 and 15). This protocol uses THB-HS, but
anginosus, and TSB can also be used [20].
Streptococcus constellatus
1. Stock cultures are stored at 80 C or 20 C in 30 %
glycerol.
2. Inoculate 5 mL THB-HS with ~10 L of the stock culture
using a sterile transfer loop and incubate at 37 C overnight in
a 5 % CO2 (in air) atmosphere.
3. Dilute the overnight culture 1:10 in THB-HS and prepare
5001000 L aliquots in 1.5 mL Eppendorf tubes. Incubate at
37 C in air for 11.5 h. Thaw the specific CSP on ice while
waiting for the next step (see Note 16).
4. Add the CSP and DNA donor to the culture to a final concen-
tration of 2050 nM and allow growing at 37 C in air for 1 h,
followed by 37 C in 5 % CO2 for 13 h.
3.2.4 Other Oral The protocol used for the anginosus group (Subheading 3.2.3)
Streptococci may also be applicable to S. gordonii, S. sanguinis, and other oral
streptococci [19]. The following modified assay that allows S. gor-
donii stocks to be stored in frozen aliquots, and directly applied in
competence experiments has been suggested [21]:
Pre-culture preparation:
1. Add 10 g/mL synthetic CSP to overnight cultures grown at
37 C in THY supplemented with chloramphenicol (5 g/
mL).
2. Freeze the cells in 100 L aliquots following the addition of
10 % glycerol.
Transformation:
1. Add 900 L THY to 100 L of the frozen cell aliquot.
2. Add transforming DNA, and incubate the cells at 37 C for 3 h.
3.3 Synthetic CSPs: Verify whether the pheromone for the strain you will be working
Sequence with has been previously identified. Some of the CSPs and XIPs
Identification that have been used for transformation of oral streptococci are pre-
sented in Table 1. For updated information on other strains, search
gene/protein databases such as Entrez. If the pheromone in your

chosen strain has not yet been identified, you may use PCR to
amplify and sequence the gene for the competence pheromone.
This information is then used to predict the pheromone amino
acid sequence for your strain.
1. In S. mutans, the comC and comS genes encoding the CSP and
XIP pheromones, respectively, are highly conserved [22]. Thus,
the 18-CSP and the XIP for S. mutans UA159 shown in Table 2
should function for most, if not all, other S. mutans strains. In
the mitis group, there is a large variation of CSP pheromones,
particularly among S. mitis strains, whereas in the anginosus
group of streptococci different species may produce the same
pheromone. In these two groups, the CSP-encoding (comC)
gene has been identified by using primers annealing to con-
served Arg-tRNA and Glu-tRNA genes flanking the comCDE
operon [4]. In this case, the primer pair t-Arg
5-GGCGGTGTCTTAACCCCTTGACCAACGGACC and
t-Glu 5-CATAGCTCAGCTGGATAGAGCATTCGCCTTC
is expected to amplify a fragment of approximately 2.52.6 kb
[4]. For sequencing, the final reaction volume of 25 L should
contain 200300 ng chromosomal DNA, Taq Polymerase, or a
high-fidelity DNA polymerase such as Pfu at the recommended
concentration, 0.2 mM dNTP, 1 PCR buffer containing stan-
dard MgCl2 concentration, and 0.1 M of each forward and
reverse primer. The following cycle parameters for amplification
of comC in S. mutans may be used: 94 C for 3 min; 36 cycles
of 94 C for 30 s, 60 C for 30 s and 72 C for 30 s; and a final
polymerization step of 72 C for 7 min. Adjust the PCR condi-
tions for amplification of the comCDE operon in the other oral
streptococci by replacing the 30 s extension time at 72 C with
2 min during the 36 cycles (see Note 17).
2. Verify the presence of the amplified product by electrophoresis
in a 0.7 % (w/v) agarose gel, and determine the comC sequence
in the amplified product, which by using the primers indicated
above is within ~350 bp from the 5-end in other
streptococci.
3. To predict the CSP sequence you may align the sequence of
your amplified product with previously identified comC genes.
Most often, the deduced CSP peptide sequence, when trans-
lated from your PCR product, is preceded by a double glycine
cleavage site. Alignment of ComC with peptides of known
cleavage sites may help define the mature peptide sequence
[4]. Many peptide synthesis services are currently available,
making the acquisition of the peptides convenient and
affordable.
4 Notes
1. Brain Heart Infusion (BHI), TSB, or blood agar plates may

also be used. The following antibiotic concentrations were
used: kanamycin (Kan) 500 g/mL, and erythromycin (Erm)
10 g/mL.
2. Competence for natural transformation persists for a longer
time in CDM using XIP when compared to TSB with CSP (see
Fig. 1). Other media that have been used in S. mutans CSP
transformation assays include BHI-HS [23] and THY-HS
[24]. For markerless constructions we recommend to trans-
form S. mutans in CDM with XIP as detailed in Chapter 14.
3. We have had a previous experience in which a particular TSB
batch resulted in very poor streptococcal transformation effi-
ciencies, a problem that was solved by simply changing the
batch of TSB used.
4. In choosing the medium for transformation, one important
factor to consider is the mechanism used by streptococci to
sense the different pheromones. For instance, when using XIP,
a peptide-free medium is advantageous because other peptides
present in the medium may compete with XIP for transport
into the cells.
5. Crude desalted XIP of >80 % purity is usually highly active.
6. Specific commercial DNA purification kits can be used for iso-
lation of plasmids, PCR amplified fragments, or genomic
DNA, following the recommended DNA isolation protocols.
Lysis procedures for oral streptococci should take into consid-
eration the rigidity of the streptococcal cell wall. For S. mutans,
we usually incubate the cells (up to 109 cells) for 20 min in the
presence of 100 U/mL mutanolysin and 20 mg/mL lyso-
zyme. In trying alternative methods, the purity of the isolated
DNA, which may impact on transformation efficiency, should
be considered.
7. It is our experience that pre-cultures prepared from fresh colo-
nies give higher transformation efficiencies, although we have
not assessed it systematically (see Subheading 3.1).
8. Although streptococcal transformation protocols usually rec-
ommend growth of the cells in 5 % CO2, we usually obtain
higher transformation efficiency levels when the cells are sim-
ply grown in air.
9. The transient nature of competence should be considered in
kinetic studies. In S. mutans, the competent state may often
last for more than 4 h in CDM, but in TSB transformation is
finished around this time (see Fig. 1). In S. mitis, for instance,
competence lasts for about 1 h in C + YYB (data not shown).
The onset of competence may also vary, with S. gordonii exhib-

iting an almost immediate response to CSP [25], whereas in S.
mutans a delay in competence induction response is observed
[23, 26]. This is because the CSP in S. mutans acts first by
stimulating bacteriocin response before competence is
triggered.
10. Saturating levels of amplicon DNA donor for S. mutans were
experimentally determined and a final concentration of 75 ng/
mL was shown to saturate the reaction [14].
11. The use of positive and negative controls is highly recom-
mended, particularly during the construction of new mutants.
A negative control will provide information on the selective
activity of the antibiotics. Positive controls are usually DNA
donor with an antibiotic resistance marker that is known to
transform the target strain. The routine use of the same posi-
tive control allows comparison of the transformation efficiency
between different experiments.
12. The repeated passage in media with the relevant antibiotic is
conducted to avoid carryover of nontransformed bacteria. This
step is particularly pertinent when using selection markers con-
ferring resistance to antibiotics that are bacteriostatic (i.e.,
those that inhibit the growth of the cells without killing them).
For selection of markerless mutants, we refer to Chapter 14.
13. The synthetic 18-CSP (SGSLSTFFRLFNRSFTQA), analo-
gous to the peptide found in the supernatant of S. mutans,
induces maximal competence at 20 nM [27], but it may be
used at concentrations as high as 1000 nM without compro-
mising transformation efficiency. The synthetic 21-CSP
(SGSLSTFFRLFNRSFTQALGK) predicted from the comC
sequence has often been used, but a delay in competence may
be observed because this form of the peptide needs further
processing by S. mutans into the 18-CSP active form.
14. S. mitis CCUG 31611T synthetic CSP (EIRQTHNIFFNFFKRR)
has shown to induce maximal number of transformants by
using a final concentration of 200300 nM (data not shown).
15. We have used this protocol to obtain consistent transformation
levels of the S. intermedius, S. anginosus, and S. constellatus
type strains, and S. intermedius CCUG 28205.
16. The type strains of S. intermedius, S. anginosus, and S. contella-
tus respond to the same pheromone (see Table 1). However,
analyses of recent genome sequence for these species reveal that
this is not always the case. This information should be checked
before deciding which pheromone is going to be used.
17. The PCR strategy to identify the CSPs will depend on the
presence of the flanking regions annealing to the specified
primers. Different sequences or gene arrangements may, there-

fore, escape detection. Note that among the transformable oral
streptococci, it is only in S. mutans that the comC gene is not
flanked by the Arg-tRNA and Glu-tRNA genes.
Acknowledgments
This work was partially supported by the National Science

Foundation, grant no. MCB-1020863, by the Faculty of Dentistry,
University of Oslo, and by the Norwegian surveillance system for
antibiotic resistance in microbes (Norsk overvkingssystem for anti-
biotikaresistens hos mikroberNORM).
References
1. Federle MJ, Morrison DA (2012) One if by 10. Lemme A, Grobe L, Reck M, Tomasch J,
land, two if by sea: signalling to the ranks with Wagner-Dobler I (2011) Subpopulation-
CSP and XIP. Mol Microbiol 86:241245 Specific transcriptome analysis of competence
2. Fontaine L, Wahl A, Flechard M, Mignolet stimulating peptide induced Streptococcus
J, Hols P (2015) Regulation of competence mutans. J Bacteriol 193:18631877
for natural transformation in streptococci. 11. Son M, Shields RC, Ahn SJ, Burne RA, Hagen
Infection, genetics and evolution. J Mol SJ (2015) Bidirectional signaling in the com-
Epidemiol Evol Genet Infect Dis 33:343360 petence regulatory pathway of Streptococcus
3. Johnsborg O, Eldholm V, Hvarstein LS mutans. FEMS Microbiol Lett 362. Doi:
(2007) Natural genetic transformation: preva- 10.1093/femsle/fnv159
lence, mechanisms and function. Res Microbiol 12. Chen JD, Morrison DA (1987) Modulation
158:767778 of competence for genetic transformation in
4. Hvarstein LS, Hakenbeck R, Gaustad P Streptococcus pneumoniae. J Gen Microbiol
(1997) Natural competence in the genus 133:19591967
Streptococcus: evidence that streptococci can 13. Son M, Ghoreishi D, Ahn SJ, Burne RA,
change pherotype by interspecies recombina- Hagen SJ (2015) Sharply tuned pH response of
tional exchanges. J Bacteriol 179:65896594 genetic competence regulation in Streptococcus
5. Morrison DA (1997) Streptococcal competence mutans: a microfluidic study of the environ-
for genetic transformation: regulation by pep- mental sensitivity of comX. Appl Environ
tide pheromones. Microb Drug Resist 3:2737 Microbiol 81:56225631
6. Khan R, Rukke HV, Ricomini Filho AP, 14. Morrison DA, Khan R, Junges R, Amdal HA,
Fimland G, Arntzen MO, Thiede B, Petersen Petersen FC (2015) Genome editing by natural
FC (2012) Extracellular identification of genetic transformation in Streptococcus mutans.
a processed type II ComR/ComS phero- J Microbiol Methods 119:134141
mone of Streptococcus mutans. J Bacteriol 15. Chang JC, LaSarre B, Jimenez JC, Aggarwal
194:37813788 C, Federle MJ (2011) Two group A strep-
7. Mashburn-Warren L, Morrison DA, Federle tococcal peptide pheromones act through
MJ (2010) A novel double-tryptophan pep- opposing Rgg regulators to control biofilm
tide pheromone controls competence in development. PLoS Pathog 7:e1002190
Streptococcus spp. via an Rgg regulator. Mol 16. Stevens KE, Chang D, Zwack EE, Sebert ME
Microbiol 7858978606 (2011) Competence in Streptococcus pneumoniae
8. Fontaine L, Boutry C, de Frahan MH, Delplace B, is regulated by the rate of ribosomal decoding
Fremaux C, Horvath P, Boyaval P, Hols P (2010) errors. mBio 2:doi: 10.1128/mBio.00071-11
A novel pheromone quorum-sensing system con- 17. Lacks S, Hotchkiss RD (1960) A study of
trols the development of natural competence in the genetic material determining an enzyme
Streptococcus thermophilus and Streptococcus sali- in Pneumococcus. Biochim Biophys Acta
varius. J Bacteriol 192:14441454 39:508518
9. Monnet V, Juillard V, Gardan R (2016) Peptide 18. Reck M, Tomasch J, Wagner-Dobler I (2015)
conversations in Gram-positive bacteria. Crit The alternative sigma factor SigX controls bac-
Rev Microbiol 42:339351 teriocin synthesis and competence, the two
quorum sensing regulated traits in Streptococcus tolerance in Streptococcus mutans UA159.

mutans. PLoS Genet 11, e1005353 Infect Immun 74:16311642
19. Gaustad P, Morrison DA (1998) Induction 24. Li YH, Lau PC, Lee JH, Ellen RP, Cvitkovitch
of transformation in streptococci by synthetic DG (2001) Natural genetic transformation
competence stimulating peptides. Methods of Streptococcus mutans growing in biofilms.
Cell Sci 20:6570 J Bacteriol 183:897908
20. Petersen FC, Pecharki D, Scheie AA (2004) 25. Vickerman MM, Iobst S, Jesionowski AM,
Biofilm mode of growth of Streptococcus inter- Gill SR (2007) Genome-wide transcriptional
medius favored by a competence-stimulating changes in Streptococcus gordonii in response
signaling peptide. J Bacteriol 186:63276331 to competence signaling peptide. J Bacteriol
21. Warren TK, Lund SA, Jones KF, Hruby DE 189:77997807
(2007) Comparison of transformation proto- 26. Kreth J, Merritt J, Shi W, Qi F (2005)
cols in Streptococcus gordonii and evaluation of Co-ordinated bacteriocin production and com-
native promoter strength using a multiple-copy petence development: a possible mechanism
plasmid. Can J Microbiol 53:417426 for taking up DNA from neighbouring species.
22. Palmer SR, Miller JH, Abranches J, Zeng L, Mol Microbiol 57:392404
Lefebure T, Richards VP, Lemos JA, Stanhope 27. Petersen FC, Fimland G, Scheie AA (2006)
MJ, Burne RA (2013) Phenotypic hetero- Purification and functional studies of a
geneity of genomically-diverse isolates of potent modified quorum-sensing peptide
Streptococcus mutans. PLoS One 8, e61358 and a two-peptide bacteriocin in Streptococcus
23. Ahn SJ, Wen ZT, Burne RA (2006) Multilevel mutans. Mol Microbiol 61:1322
control of competence development and stress 1334
Chapter 14
Markerless Genome Editing in Competent Streptococci

Roger Junges, Rabia Khan, Yanina Tovpeko, Heidi A. mdal,
Fernanda C. Petersen, and Donald A. Morrison
Abstract
Selective markers employed in classical mutagenesis methods using natural genetic transformation can
affect gene expression, risk phenotypic effects, and accumulate as unwanted genes during successive muta-
genesis cycles. In this chapter, we present a protocol for markerless genome editing in Streptococcus mutans
and Streptococcus pneumoniae achieved with an efficient method for natural transformation. High yields of
transformants are obtained by combining the unimodal state of competence developed after treatment of
S. mutans with sigX-inducing peptide pheromone (XIP) in a chemically defined medium (CDM) or of S.
pneumoniae with the competence-stimulating peptide (CSP) together with use of a donor amplicon carry-
ing extensive flanking homology. This combination ensures efficient and precise integration of a new allele
by the recombination machinery present in competent cells.
Key words Pheromone, Competence, Natural transformation, Markerless mutagenesis, Streptococcus

mutans, Streptococcus pneumoniae, Streptococcus, XIP, CSP
1 Introduction
Genome editing is a powerful tool for the analysis of gene function

and regulatory pathways in many organisms. In the bacteria, natural
genetic transformation can provide a direct route between synthetic
DNA constructs and the cell genome, via DNA uptake and homolo-
gous recombination. As the efficiency of this process is often low,
many routine strategies for mutagenesis by this route employ a selec-
tive marker linked to the desired mutation, allowing recovery of rare
recombinants simply by use of selective agar medium. While invalu-
able for mutagenesis in organisms with low transformation efficiency,
such markers unfortunately carry unwanted information, which
can potentially alter the organisms gene expression and phenotype.
In addition, during successive cycles of mutagenesis different markers
accumulate, compounding their side effects.
To create a simple method of markerless genome editing in
S. mutans UA159 or in laboratory strains of S. pneumoniae, it was
233
234 Roger Junges et al.
necessary to raise the efficiency of transformation close to unity.

In S. mutans, the recently described competence pheromone
known as the sigX-inducing peptide (XIP) [1, 2] has the potential
of stimulating development of competence in all cells in a popula-
tion, in contrast to the bimodal response to another S. mutans
competence pheromone, CSP [36]. In addition, the XIP-induced
competent state is unusually persistent in S. mutans, lasting for
hours and accompanied by a reduced apparent growth rate [7]. In
S. pneumoniae, high rates of transformation are obtained by treat-
ment with pneumococcal CSP [8] and the use of large DNA frag-
ments. As previously shown, transformation efficiency increases
dramatically for genomic donor fragments larger than 1 kb,
approaching a maximum only above 10 kb [9, 10].
By combining the aforementioned key factors with use of PCR
amplicon donors targeted to a single genomic site, an significant
increase in transformation efficiency from 0.1 to 1 % to rates higher
than 30 % was obtained in both S. mutans and in S. pneumoniae
[11, 12]. Recovery of the desired mutant can be accomplished by
use of a simple PCR step using specific primers that distinguish the
mutant from the parental alleles. Given that efficiencies above 30 %
are routinely obtained, the screening of one or two dozen colonies
is normally sufficient.
This chapter describes a protocol that incorporates these key
factors to achieve markerless genome editing in a S. mutans refer-
ence strain or in S. pneumoniae.
2 Materials
2.1 Competence 1. S. mutans UA159 (ATCC 700610) is a type strain. The S.

Induction pneumoniae strains are derivatives of the lab strain Rx-1 [9].
and Transformation 2. Agar plates: Tryptic Soy Broth (TSB) 30 g/L (Becton
Dickinson, Franklin Lakes, NJ, USA). Add 15 g/L of agar to
the medium (VWR Chemicals, Radnor, PA, USA) and auto-
clave at 121 C for 15 min. For selective plates, the appropriate
antibiotic should be added to the medium once it has cooled
below 60 C and the plates should be stored under appropriate
conditions for the antibiotics (see Note 1).
3. Liquid media: Chemically defined medium (CDM) [6], pre-
pared from stable concentrated stock solutions with 1 % glu-
cose is used for growth, stock preparation, and transformation
assays (see Note 2). Tryptic Soy Broth (TSB, Becton Dickinson,
Franklin Lakes, NJ, USA) is used for growth, transformation
assays, preparation of stocks, and serial dilutions.
4. Synthetic peptides: Synthetic pheromones can be ordered from
custom peptide synthesis services. Lyophilized synthetic XIP
(GLDWWSL) is reconstituted with 20 L dimethyl sulfoxide
Genome Editing In Streptococci 235
(DMSO) (Sigma-Aldrich, St. Louis, MI, USA) to which 1 mL

of distilled water is added to give a final concentration of
10 mM (stored at 20 C). Working stocks are prepared at
100 M by dilution in distilled water (see Note 3). Synthetic
CSP1 (EMRLSKFFRDFILQRKK) or CSP2 (EMRIS
RIILDFLFLRKK) for S. pneumoniae are dissolved in water at
250 g/mL.
5. Donor DNA is prepared as PCR amplicons with homologous
flanking regions 23 kb each. Positive control donors can be
prepared by PCR amplification of a selective marker flanked by
23 kb on both sides.
2.2 PCR 1. DNA polymerase: Q5 High-Fidelity DNA polymerase kit

(New England Biolabs, Ipswich, MA, USA) is used for con-
struction of amplicons (see Note 4). For colony screening,
TrueStart Hot Start Taq DNA polymerase (Thermo Fisher
Scientific, Waltham, MA, USA) is used.
2. PCR: Any conventional PCR cycler can be used. In the exam-
ples shown, the Eppendorf Mastercycler Gradient PCR
Thermal Cycler (Eppendorf, Hamburg, Germany) was used.
2.3 Agarose Gel 1. Agarose: SeaKem LE Agarose (Lonza, Basel, Switzerland) is

Electrophoresis used to prepare gel for electrophoresis. The concentration of
agarose is adjusted according to expected fragment size follow-
ing the recommendations of the manufacturer.
2. TAE buffer: Tris-acetate-EDTA (TAE) buffer containing
40 mM Tris, 20 mM acetic acid, and 1 mM EDTA is used to
prepare the gel and for running the electrophoresis.
3. DNA stain: SYBR safe DNA gel stain in 10,000 DMSO
(Thermo Fisher Scientific, Waltham, MA, USA) is incorpo-
rated into the gel at a 1:10,000 proportion with TAE buffer.
2.4 Primers Oligonucleotide primers are obtained (deprotected and desalted)

from commercial synthesis services. Primers are dissolved in molec-
ular biology-grade (sterile deionized) water to obtain a stock con-
centration of 100 M, and stored at 20 C.
3 Methods
The following protocol was established with the objective of

achieving sufficiently high yields of transformants to allow conve-
nient direct editing of the genome without resort to selective mark-
ers. The nature of each sequence alteration hinges on the research
objective and can vary from single base substitutions to gene dele-
tions, rearrangements, or replacements (see Subheading 3.4). In S.
mutans, the basis for the method is the extended maintenance of a
Fig. 1 Comparison of Kinetics of sigX expression in CDM and TSB. Strain SM068
was grown in 200-L volumes of (a) CDM or (b) TSB in a 96-well plate in ambi-
ent air at 37 C with or without 1 M XIP (black) or 250 nM CSP-18 (grey),
respectively. The sigX expression (sigX reporter luminescence measured as
relative light units, RLU) relative to the optical density at 600 nm (OD600) of the
culture ((c) for CDM, (d) for TSB) was monitored periodically. The results shown
are the averages of three replicates (SEM)
high level of competence in CDM, which allows a steady accumu-

lation of recombinants during prolonged exposure to a high-MW
donor DNA (see Fig. 1).
3.1 Construction 1. Design primers to amplify two sections in the locus where the
of Markerless mutation will be inserted (Fig. 2a). For illustrative applications
Amplicons see Subheading 3.4. In the case illustrated in Fig. 2, the region
marked in red will be deleted. Primers P1 and P4 are 1822 bp
Fig. 2 Workflow for markerless genome editing. Amplicon design and construction (ac) are prepared accord-
ing to the desired mutation, exemplified in this diagram as a gene deletion (marked in red). This step is fol-
lowed by a highly efficient (>35 %) method for natural transformation (d). Two dozen colonies recovered from
the transformation step are screened with specific primers that amplify fragments with different sizes in the
mutant when compared to the parent (e). The mutant is then isolated (f, g) and the mutation is confirmed with
DNA sequencing (h)
in length and have a melting temperature close to 60 C. Base

composition should be 4060 % G + C and, if possible, the
3-terminal base should be a G or a C. Primers should be
designed to avoid primer dimers or hairpin structures. Primers
P2 and P3 overlap and their design is restricted to the specific
sequence in the region where the mutation will be inserted.
Firstly, design the primer to match a region flanking the desired
mutation. Secondly, add to this sequence a similar-sized
sequence present in the antisense strand, to overlap with the
other amplicon (Fig. 2a). Independently, both of these sequences
that will form one primer should have 1618 bp and a melting
temperature close to 60 C. The final sequences of P2 and P3
will have 3436 bp each. It is crucial that the flanking regions of
homology in each of the amplicons extend for 23 kb.
2. Amplify each fragment (a + b) with a proofreading DNA poly-
merase enzyme that is optimized for accurate amplification of
long fragments (see Note 4). Confirm the presence of the cor-
rect fragment by gel electrophoresis.
3. Design two primers that will be used to connect and amplify
the fragments (P5 and P6) (Fig. 2b). Using nested primers
(P5 and P6), in contrast to using the original outer primers
(P1 and P4), greatly increases the specificity of the amplifica-
tion [13]. As before, utilization of a proofreading DNA poly-
merase enzyme that is optimized for long fragments is essential
(see Note 5).
4. Confirm the amplification of the correct fragment by gel elec-
trophoresis. Fragments a and b (Fig. 2a) are occasionally
observed but weaker than the band for the final amplicon.
5. Utilize the final amplicon as donor DNA to transform the
recipient strain (Fig. 2c).
3.2 Markerless 1. Stock cultures prepared in CDM (OD600 0.5) from fresh colo-
Transformation nies are stored with 15 % glycerol at 80 C (see Note 6).
Protocol Using XIP 2. Dilute the frozen stock 1:10 in fresh CDM; the initial
in S. mutans OD600 = ~0.05 (see Note 7).
3. Incubate at 37 C until OD600 = 0.1 (approximately 2 h).
4. Add XIP to final concentration of 1 M.
5. Add 50100 ng of the donor amplicon per mL of culture and
incubate at 37 C for 3 h in a closed 1.5-mL Eppendorf tube
(see Note 8).
6. Prepare serial dilutions of the competent culture in TSB; spread
100 L of the following dilutions on TSB agar plates104,
105, and 106.
7. Incubate the plates for 24 h at 37 C with 5 % CO2 (see Note 9).
8. Pick 24 isolated colonies using sterile inoculating loop or
pipettor tips; resuspend each in 10 L of sterile distilled water.
Prepare a streak plate from 2 L of each suspension on a new

TSB or THB agar plate (see Note 10).
9. Prepare a PCR reaction with the following components to test
each colony for the intended genome modification (see Note 11).
During screening, P7 and P8 (Fig. 2e) should amplify bands
with different sizes in the mutant when compared to the
parent, allowing for differentiation (see Note 12).
Template (from suspended colony) 0.2 L
Forward primer 0.06 L
Reverse primer 0.06 L
10 Reaction buffer 1 L
10 mM dNTP 0.2 L
Nuclease-free water 7.28 L
25 mM MgCl2 1 L
Hot start Taq DNA polymerase 0.2 L
Total 10 L
10. Thermal cycling parameters:
Description Temperature Time

Initial denaturation 95 C 3 min
25 cycles 95 C 30 s
55 C 30 s
72 C 50 s
Final extension 72 C 5 min
11. Analyze the products of each PCR reaction by gel electropho-
resis. The results will be either negative (showing only the
band for the parent allele), pure positive (showing only the
band for the mutant allele), or mixed (showing both bands)
(see Fig. 3). If a desired transformant is identified, a subclone
can be cultured from the corresponding streak plate after 24 h
of incubation at 37 C, 5 % CO2 (Fig. 2f). If there are no pure
positive but only mixed and negative reactions, pick several
colonies from the streak plate made from a mixed colony and
repeat the same screening procedure.
12. Repeat the screening procedure twice to allow for complete
segregation of the mutant.
13. Prepare stocks by growing the selected bacteria in TSB over-
night and then store the culture with 15 % glycerol at 20 C
or 80 C (Fig. 2g).
14. A confirmatory PCR should be conducted once the stock is
made to confirm that only the mutant allele is present. Final
confirmation of the mutation in the target gene is done by
sequencing (Fig. 2h).
Controls Exemplary Colonies
M WT NYT1 A B C D E Strain
+ - + - + - + - + - + - + - Primer pair
WT mut WT WT mut mut WT Genotype
Fig. 3 Gel analysis of PCR products from exemplary colonies examined for transfer of suppressor single-base
substitution in S. pneumoniae. +, primer pair specific for WT sequence; , primer pair specific for mutant
sequence. M, molecular weight standard; WT, parent strain; NYT1, suppressor strain; AE, colonies tested;
genotype, wild type (WT), or mutant (mut)
3.3 Markerless 1. Stock cultures prepared in TSB (OD550 0.2) and stored with
Transformation 15 % glycerol at 80 C are diluted 1:100 in fresh TSB and
Protocol Using CSP incubated at 37 C until OD550 0.05.
in S. pneumoniae 2. To 1 mL of culture in a screw-cap tube, add CSP to a final
concentration of 250 ng/mL, bovine serum albumin (BSA) to
0.04 %, CaCl2 to 0.5 mM, 100 ng of donor amplicon DNA,
and incubate for 80 min at 37 C.
3. Prepare serial dilutions of the competent culture in TSB; spread
100 L of the following dilutions on TSB agar plates: 105,
106, and 107. Incubate the plates for 16 h at 37 C.
4. Design screening primers to distinguish donor from recipient
alleles.
5. Pick 25 isolated colonies using sterile loops or needles; resus-
pend each in 10 L in water. Prepare a streak plate with 2 L
of each suspension on a new TSB plate, and follow steps 914
in Subheading 3.2.
3.4 Examples In the applications of this method sketched in Fig. 2, three genes
of Applications in S. mutans and one gene in S. pneumoniae were targeted for vari-
ous alterations by use of sequences from GenBank accession num-
bers NC_004350 and NC_003098. Strains and specific primer sets
for each case are listed in Tables 1, 2, 3, 4, and 5.
3.4.1 Example 1. The method can be used to invert small sequences in the genome.
Eight-Basepair Inversion In this example, the objective was to investigate a promoter region
of comE putatively recognized by SigX [3] in S. mutans UA159, by
making an 8-bp inversion. The steps of Subheading 3.1 were fol-
lowed for the creation of the amplicon that was used to transform
SM068 and SM091 (psigX::luc reporter derivatives of UA159)
into SM177 and SM179 (see Table 1; Fig. 4).
Table 1
List of strains
Name Description Source

Streptococcus mutans
UA159 Wild-type reference strain ATCC 700610
SM059 UA159, but pcipB-luc::spc; SpcR [1]
SM068 UA159, psigX-luc; SpcR [1 ]
SM089 UA159, but comS:: ery; ErmR [1 ]
SM091 SM068, but comS:: ery; SpcR ErmR [1 ]
SM134 SM089, but PcipB-luc::spc; ErmR SpcR SM089 SM059
SM177 SM068, but comE SigX-box inversion; SpcR SM068 aRJ04
SM179 SM091, but comE SigX-box inversion; SpcR ErmR SM091 aRJ04
SM188 SM068, but c_105 1-bp substitution; SpcR SM068 aRJ17
SM189 SM091, but DR comE; SpcR ErmR SM091 aRJ18
SM190 SM134, but DR comE; ErmR SpcR SM134 aRJ18
Streptococcus pneumoniae
NYT1 CP2137, but comW, rpoD-L363F; SmR CmR KanR [12]
NvR ErmR TcR
CP2137 cps, comA; SmR CmR [12]
CP2451 CP2137, but rpoD-L363F; SmR CmR CP2137 NYT1
Table 2
Primers for construction of S. mutans strains SM177 and SM179
Primer Sequence (5 to 3) Template Amplicon product

FP916 GAGATGGGCTTTTTGGATGA UA159 Segment 1
FP936 TCTACTAACTTAATAACCCTACTTATC
TGCGAATAATATAATCAGATGATTAAGCAT
FP917 TGCGGTCTATTGACCTCCTC UA159 Segment 2
FP935 ATGCTTAATCATCTGATTATATTATTCGCAGA
TAAGTAGGGTTATTAAGTTAGTAGA
FP009 TATGGACCAAGAAATGCTGT Segment 1+ Overlap PCRa
FP947 GCCCCCTTTATGGAACAAAT Segment 2 5757 bpaRJ04
FP918 GCATAGGTGAGTCAAAGTGGTT SM177 and 194-bp allele-
FP919 CTAACTTAATAACCCTACTTATCTGCGA SM179 specific product
FP920 AAGCAGTAATGCTAATGCTGTTAATC UA159 347-bp allele-
FP921 CTAACTTAATAACCCTACTTATCTATTCGCA specific product
Underlined letters represent inversion mutation
a
Complementary primers FP935 and FP936 carry an inverted SigX-box of comE. FP009/FP947 were used to link seg-
ments 1 and 2, creating the final 5,757-bp amplicon aRJ04, which was used to transform strains SM068 and SM091,
creating strains SM177 and SM179, respectively. Primer pairs FP918/FP919 and FP920/FP921 were used to detect
the inversion in SM177 and SM179
Complementary primers (P2 and P3) carrying the inversion

were designed and respectively matched by primers located in the
flanking regions (P1 and P4) to create two segments of 4.3 kb
(P1/P2) and 4 kb (P3/P4).
Table 3
Primers for construction of S. mutans strain SM188
Primer Sequence Template Amplicon product

FP894 TCCGGATGCAGAAGGTATTC UA159 Segment 1
FP895 CAATAAAAGTTCTCACCCAATCTGGA
FP896 TCCAGATTGGGTGAGAACTTTTATTG UA159 Segment 2
FP897 CATCCTGCCGTTCCTATCAT
FP937 TGTCCCGCTGGATACAGATT Segment 1 + Segment 2 Overlap
FP938 TGTCCCGCTGGATACAGATT PCRaaRJ17
FP898 GGTTGATTGGGTTTTTGTGG UA159 444-bp allele-
FP899 TTTTTATGCTTTTCAATAAAAGTTCTA specific product
FP900 CGGATTGGATTGGGAGACTA SM188 681-bp allele-
FP901 TTTTTATGCTTTTCAATAAAAGTTCTC specific product
FP902 CTCTAAGACTAATCCAGATTGGGTT U159 364-bp allele-
FP903 GCGAGTTTCAAAAAGGAAGC specific product
FP904 CTCTAAGACTAATCCAGATTGGGTG SM188 552-bp allele-
FP905 GGCAGACAGCTTCTTTGGTC specific product
Underlined letters represent the 1-base substitution
a
Complementary primers FP895 and FP896 carry the single base substitution in c105. FP937/FP938 were used to
overlap and link segments 1 and 2, creating the final amplicon aRJ17, which was used to transform strain SM068, creat-
ing mutant SM188. Primers FP898-FP905 were used to detect the base substitution in SM188
Table 4
Primers for construction of S. mutans strains SM189 and SM190
Primer Sequence Template Amplicon product

FP1050 AATATAAAAGGGAGCGATGAAACTT UA159 Segment 1
FP1051 GATAAGCAATAGATATAGCCTTCTTT
GATCATGTTC
FP1052 GAACATGATCAAAGAAGGCTATATCT UA159 Segment 2
ATTGCTTATC
FP1053 GTAGCTATTTTGTCCTAAACGGTCA
FP1054 TGATTGTTTTTGTGGTATCTGCTAA Segment Overlap PCRaaRJ18
FP1055 TTTACACAAGCTTTGGGAAAATAAG 1 + segment 2
FP1056 CAACGGCTGATTAACAGAAAA SM189 and 309-bp product in UA159;
FP968 TCATTCTAGTGATAATAAACATTTTGC SM190 but 272-bp product in
mutants SM189 and
SM190
a
Complementary primers FP1051 and FP1052 carry the deleted direct repeat sequence of comE. FP1054/FP1055 were
used to overlap and link segments 1 and 2, creating the final amplicon product aRJ18, which was used to transform
strains SM091 and SM134, creating strains SM189 and SM190, respectively. FP1056 and FP968 were used to detect
the deletion in mutants SM189 and SM190
Table 5
Primers used for the construction of S. pneumoniae mutant CP2451
Primer Sequence (5 to 3) Template Amplicon product

YT30 GACAGGCTTTGAGTCTCTTGATGG NYT1 5.5 kb rpoD
YT31 CGGACGCTCAAACTTGGCTAATTC
YT49 CAAGTCGTAGCAAACCGC CP2137 (control) 500-bp WT allele specific product
YT51 CACGGTAAGCACCTGAAAC NYT1
YT50 CAAGTCGTAGCAAACCGT CP2137 500-bp mutant allele specific
YT51 CACGGTAAGCACCTGAAAC NYT1 (control) product
Underlined letters represent the 1-base substitution
Fig. 4 Genomic changes made in four application examples of direct editing. In S. mutans, (a) inversion of a
comE SigX-box, (b) deletion of two direct repeats, and (c) substitution of a single base to introduce a stop
codon within gene smut_orf_c105. In S. pneumoniae, (d) substitution of a single base causing a Leu363 Phe
replacement in RpoD
Nested primers (P5 and P6) 5.7 kb apart (see Fig. 4) were used
to connect the two segments with overlapping PCR, creating a
final amplicon product having the mutation in its center.
This final product was used to transform S. mutans (see
Subheading 3.2).
Twelve colonies were screened. Seven were mixed; four
were pure mutant clones (33 %). Pure mutant colonies were iso-
lated, re-screened, and stocked as strains SM177 and SM179
(Table 1).
3.4.2 Example 2. To investigate the function of two direct repeats located close to
Thirty-Nine-Basepair the putative promoter region of comE (Fig. 2) in S. mutans UA159,
Deletion primers for PCR (P2 and P3) were designed to flank the region
selected for deletion, and respectively matched by primers located
in distal flanking regions (P1 and P4) to create two segments of
2.9 kb (P1/P2) and 2.8 kb (P3/P4).
Nested primers (P5 and P6) 5.1 kb apart were used to connect
the two segments with overlapping PCR, creating the final ampli-
con product containing a central deletion.
The final amplicon product was used to transform S. mutans
strains SM091 and SM134. Among 16 colonies screened, 4 were
mixed (25 %), and 12 were negative. To allow segregation, these 4
colonies were grown in TSB the next day, plated, re-selected, and
re-screened twice in order to isolate the pure mutant clone, creating
strains SM189 and SM190.
3.4.3 Example 3. To investigate the function of an open-reading frame (smut_

Single-Base Substitution orf_1_105) that overlaps SMU_60 in S. mutans UA159, a base
in S. mutans substitution was designed to create a stop codon in the former
ORF but retain unaltered translation of the latter ORF.
Complementary primers (P2 and P3) carrying the single-base
substitution were designed and respectively matched by primers
located in the flanking regions (P1 and P4) to create two segments
of 3.9 kb (P1/P2) and 3.8 kb (P3/P4).
Nested primers (P5 and P6) 7.5 kb apart were used to connect
the two segments by overlapping PCR, creating a final amplicon
product having the mutation in its center. The final amplicon
product was used to transform S. mutans UA159 (see Note 13).
The screening primers in this case required use of touch-
down PCR to achieve discriminatory specificity (see Note 14 and
15). The annealing temperature was reduced gradually from 68
to 63 C during the first 15 cycles and maintained at 63 C dur-
ing remaining cycles. Otherwise the program was the same as
described above (Subheading 3.2, step 10). A mixed colony was
re-streaked, and a verified pure mutant subclone was retained as
strain SM188 (Table 1). The mutation was confirmed by DNA
sequencing.
3.4.4 Example 4: Single-amino acid substitutions in the S. pneumoniae primary

Single-Base Substitution sigma factor (RpoD) can bypass the need for the critical ComW
in S. pneumoniae component during transformation [12]. To investigate the effect
of the corresponding single-base substitution, a 5.5-kb region
around rpoD of strain NYT1 (Table 1) was amplified using prim-
ers YT30 and YT31 (Table 5), centered on the mutant base
(Fig. 4d).
Mutant sequences were amplified, purified, and transformed
into S. pneumoniae strain CP2137. Using a 69 C annealing tem-
perature during the colony-screening PCR, the screening yielded
50 % pure transformants based on amplification from the primer set
complementary to the mutant sequence.
One colony was streaked out and 10 subclones from this streak
were again tested by PCR. Although 9 of 10 colonies again showed
the mutant sequence, one showed the WT sequence, indicating
the need to re-streak and pick isolated colonies. A single subclone
with the mutant sequence was reconfirmed by DNA sequencing
and named CP2451.
4 Notes
1. Other suitable agar plates with media supporting growth of

streptococci such as Brain Heart Infusion (BHI) or Todd Hewitt
Broth (THB) may also be used. Antibiotics were used at the fol-
lowing concentrations: kanamycin (Kan), 500 g/mL; erythro-
mycin (Erm), 10 g/mL; spectinomycin (Spc), 500 g/mL.
2. CDM is stored in closed bottles at 4 C for up to 4 weeks.
3. XIP or CSP peptides received as crude desalted product of
>80 % purity are routinely highly active.
4. It is important to adjust the annealing temperature for each
pair of primers when using this kit.
5. To increase the efficiency of the PCR reaction with nested
primers, it is important to use less than 20 ng of fragments a
and b in a 50-L reaction.
6. The most important factor here is to grow pre-cultures from
fresh colonies that were plated just a day before. Streak the
strain on a TSB plate and incubate overnight at 37 C in CO2.
Next day, resuspend a group of colonies in 10 mL liquid CDM
medium (initial OD600 = 0.05 to 0.1), grow approximately 3 h
in a capped tube or in CO2 until mid-log phase (OD600 = 0.5).
Add 1/5 volume of glycerol, prepare aliquots of 1 mL, and
store at 80 C. We observed a remarkable increase in transfor-
mation efficiencyfrom ~7 % to >30 % - when we started
preparing pre-cultures this way.
7. Dilution of pre-cultures may be optimized down to 1:100,000

to increase the proportion of transformants.
8. We found 75 ng to be a saturating amount of donor amplicon.
Transformation assays were usually conducted in closed 1.5-mL
Eppendorf tubes. It is not recommended to increase the time of
incubation with the donor amplicon beyond 3 or 4 h.
9. Efficiency varies depending on the locus of the amplicon, for
unknown reasons.
10. Given that transformation efficiencies are high, it is also possi-
ble to screen fewer colonies. We recommend using 24 as a
good resource to estimate the efficiency of the experiment and
also ensure recovery of a mutant.
11. If the transformation has lower efficiency, a group of colonies
can be collected for PCR screening. Once a group containing
a positive colony is identified, it can be streaked again to isolate
individual subclones for screening.
12. In addition, as a second step of screening, a second pair of prim-
ers, P9 and P10 can be designed with at least one of them in,
e.g. the deleted region, therefore binding to the DNA of the
parent but not to the mutant. This allows for an independent
confirmation of the deletion (Fig. 2e).
13. If the nature of the mutation provides fragments with the same
MW in the mutant and the parent allele, two pairs of primers
can be designed here, the first pair with one of the primers
binding only to the parent, and the second with one of the
primers binding exclusively to the mutant. It is convenient,
however, to design these for fragments with different MW for
differential identification.
14. When designing codon changes it is important to check for
rare codon usage [14].
15. Touchdown PCR involves decreasing the annealing tempera-
ture in small increments and provides specificity by favoring
the specific base pairing between primer and template [15].
Acknowledgments
This work was partially supported by the National Science

Foundation, grant no. MCB-1020863, by the Faculty of Dentistry,
University of Oslo, and by the Norwegian surveillance system for
antibiotic resistance in microbes (Norsk overvkingssystem for anti-
biotikaresistens hos mikroberNORM). We thank Kunal Desai for
assistance with exploratory experiments.
References
1. Khan R, Rukke HV, Ricomini AP, Fimland G, mutans in a chemically defined medium.
Arntzen MO, Thiede B, Petersen FC (2012) J Bacteriol 194:37743780
Extracellular identification of a processed type 8. Johnston C, Campo N, Berge MJ, Polard P,
II ComR/ComS pheromone of Streptococcus Claverys JP (2014) Streptococcus pneumoniae, le
mutans. J Bacteriol 194:37813788 transformiste. Trends Microbiol 22:113119
2. Mashburn-Warren L, Morrison DA, Federle 9. Cato A Jr, Guild WR (1968) Transformation
MJ (2010) A novel double-tryptophan peptide and DNA size: I. Activity of fragments of
pheromone controls competence in defined size and a fit to a random double cross-
Streptococcus spp. via an Rgg regulator. Mol over model. J Mol Biol 37:157178
Microbiol 78:589606 10. Morrison DA, Guild WR (1972)
3. Reck M, Tomasch J, Wagner-Dobler I (2015) Transformation and deoxyribonucleic acid size:
The alternative sigma factor SigX controls bac- extent of degradation on entry varies with size
teriocin synthesis and competence, the two of donor. J Bacteriol 112:11571168
quorum sensing regulated traits in Streptococcus 11. Morrison DA, Khan R, Junges R, Amdal HA,
mutans. PLoS Genet 11, e1005353 Petersen FC (2015) Genome editing by natu-
4. Son M, Ghoreishi D, Ahn SJ, Burne RA, ral genetic transformation in Streptococcus
Hagen SJ (2015) Sharply tuned pH response mutans. J Microbiol Methods 119:134141
of genetic competence regulation in 12. Tovpeko Y, Morrison DA (2014) Competence
Streptococcus mutans: a microfluidic study of for genetic transformation in Streptococcus
the environmental sensitivity of comX. Appl pneumoniae: mutations in sigmaA bypass the
Environ Microbiol 81:56225631 comW requirement. J Bacteriol 196:
5. Son MJ, Ahn SJ, Guo Q, Burne RA, Hagen SJ 37243734
(2012) Microfluidic study of competence regu- 13. Szewczyk E, Nayak T, Oakley CE, Edgerton
lation in Streptococcus mutans: environmental H, Xiong Y, Taheri-Talesh N, Osmani SA,
inputs modulate bimodal and unimodal expres- Oakley BR (2006) Fusion PCR and gene tar-
sion of comX. Mol Microbiol 86:258272 geting in Aspergillus nidulans. Nat Protoc
6. Chang JC, LaSarre B, Jimenez JC, Aggarwal 1:31113120
C, Federle MJ (2011) Two group A strepto- 14. Sharp PM, Bailes E, Grocock RJ, Peden JF,
coccal peptide pheromones act through oppos- Sockett RE (2005) Variation in the strength of
ing Rgg regulators to control biofilm selected codon usage bias among bacteria.
development. PLoS Pathog 7, e1002190 Nucleic Acids Res 33:11411153
7. Desai K, Mashburn-Warren L, Federle MJ, 15. Korbie DJ, Mattick JS (2008) Touchdown
Morrison DA (2012) Development of compe- PCR for increased specificity and sensitivity in
tence for genetic transformation of Streptococcus PCR amplification. Nat Protoc 3:14521456
Chapter 15
Tools andStrategies forAnalysis ofGenome-Wide

andGene-Specific DNA Methylation Patterns
AniruddhaChatterjee, EuanJ.Rodger, IanM.Morison,
MichaelR.Eccles, andPeterA.Stockwell
Abstract
DNA methylation is a stable epigenetic mechanism that has important roles in the normal function of a cell
and therefore also in disease etiology. Accurate measurements of normal and altered DNA methylation
patterns are important to understand its role in regulating gene expression and cell phenotype. Remarkable
progress has been made over the last decade in developing methodologies to investigate DNA methyla-
tion. The availability of next-generation sequencing has enabled the profiling of methylation marks at an
unprecedented scale. Several methods that were previously used to profile locus-specific methylation have
now been upgraded to a genome-wide scale using high-throughput sequencing or array platforms.
However, because there are so many techniques available, researchers are faced with the challenge of
assessing the potential merits or limitations of each technique and selecting the appropriate method for
their analysis. In this review we discuss the strengths and weaknesses of genome-wide and gene-specific
analysis tools for interrogating DNA methylation. We particularly focus on the design and analysis strate-
gies involved. This review will provide a guideline for selecting the appropriate methods and tools for
large-scale and locus-specific DNA methylation analysis.
Key words Epigenetics, DNA methylation, Bisulfite sequencing, RRBS, WGBS, 450K, Next-
generation DNA sequencing, CpG island, Differential methylation, Alignment
1 Introduction
Methylation of the cytosineguanine dinucleotide in DNA is a stable

epigenetic mark that plays an important role in regulating gene
expression and determining the phenotype of a cell. The role of DNA
methylation is fundamental to cellular functions such as genomic
imprinting, X chromosome inactivation, tissue differentiation,
phenotypic plasticity, and disease susceptibility [17].
The advent of next-generation sequencing has provided an
unparalleled opportunity to profile DNA methylation patterns at
single-base resolution and whole genome scale [8]. Multiple new
techniques have been developed in the last 5 years to study
249
250 Aniruddha Chatterjee et al.
genome- wide DNA methylation that employ microarray and

next-generation sequencing (NGS) platforms. Although the tech-
niques for profiling other genomic features (such as gene muta-
tions and expression) are relatively standardized, there are diverse
and evolving methods available for genome-wide DNA methyla-
tion analysis. There are at least 20 sequencing-based methods for
the analysis of DNA methylation [9]. Array-based, capture, or
enrichment-based methods also exist for genome-wide DNA
methylation profiling. Therefore, selection of the appropriate
methodological platforms can be challenging. Further, the data
analysis process substantially differs for each technique [9]. Analyses
of gene-specific methylation patterns are relatively simple: bisulfite
sequencing (Sanger sequencing) and several PCR-based methods
are the primary methods for regional investigation. However, sev-
eral variants of these methods are available and several new devel-
opments allow high resolution, multiple region- and multiple
sample-based investigation of gene/region-specific DNA methyla-
tion. Decisions regarding the analytical method will depend on
cost, number of samples to be investigated and the type of analysis
needed to answer a biological question. Therefore, it is important
to understand the benefits, potential biases and resource require-
ments of particular techniques that suit the research objectives.
In this review, we discuss the strengths and weaknesses of
genome-wide and gene-specific analysis tools for interrogating DNA
methylation. We particularly focus on design and analysis strategies
involved for profiling the mammalian methylome (mainly human).
2 Whole Genome andGenome-Wide Analysis ofDNA Methylomes
Progressive increases in sequencing throughput have enabled DNA

methylation analysis at whole genome or genome-wide (i.e., almost
every CpG site in the genome) scale or reduced representation scale
(selected CpG sites providing snapshots of thousands of regions in
the genome). Here we discuss the basic principles, data processing,
and analysis tools for large-scale DNA methylation profiling.
2.1 Principles Whole genome or genome-wide techniques employ a common

ofGenome-Wide principle of analysis; i.e., a local treatment of the genome to distin-
Methylation Profiling guish between methylated and unmethylated sites followed by a
strategy of global investigation of these modified sites. The global
investigation approaches are generally next-generation sequencing
or array platforms. For local treatment there are three main
approaches. First, restriction endonucleases are used to cleave
single- or double-stranded DNA at specific recognition nucleotide
sequences. The presence of methyl groups can block the cleavage
by specific restriction enzymes. Broadly, two types of enzymes
are used in methylation studies. Methylation-sensitive enzymes
(e.g., McrBC, HpaII, Hin6I, and AciI) that recognize specific
Tools and Strategies for Mining DNA Methylomes 251
DNA sequences that are unmethylated, and methylation-insensitive

enzymes (e.g., ApeKI, MspI, and TaqI) that cleave particular DNA
sequences regardless of methylation status of the recognition
sequence. Second, bisulfite conversion: treating DNA fragments
with sodium bisulfite before PCR analysis is a gold standard
method for detecting methylation in a genome. Frommer and col-
leagues recognized the utility of this technique for DNA methyla-
tion analysis [10]. Sodium bisulfite treatment of DNA converts
cytosine (C) residues to uracil (U), but leaves 5-methylcytosine
residues unchanged. One note is that sodium bisulfite treatment
does not allow to distinguish between 5-methylcytosine and
5-hydroxymethylcytosine [11]. Third, affinity enrichment: the
affinity enrichment method involves application of an antibody
(specific for methylated cytosines) to enrich for methylated regions
in the genome by immunoprecipitating genomic DNA [12]. After
antibody incubation, the selected DNA can be hybridized to custom
arrays or sequenced to perform a global analysis.
2.2 Major There are at least 20 different techniques (or their variants) for
Techniques forLarge- genome-wide DNA methylation profiling (details of the widely
Scale (Genome-Wide) used genome-wide techniques are given in Table1). Here, we discuss
DNA Methylation more frequently used techniques that are sensitive, reproducible,
Analysis and provide single nucleotide methylation information.
Table 1
Details of major techniques for large-scale methylation profiling
Technique Genome treatment Input DNA Resolution CpG coverage

WGBS Bisulfite 10ng Single base >90% (>26 million)
PBAT Bisulfite 125pg10ng Single base 90% (26 million)
RRBS Restriction enzyme+bisulfite 10ng2g Single base 13% (4 million)
MRE-Seq Restriction enzyme 13g Peak height/relative 6% (1.7 million)
MeDIP-Seq Affinity enrichment 50ng5g Peak height/relative 85% (24 million)
MBD-Seq Affinity enrichment 1g Peak height/relative 60% (17 million)
MethylCap-Seq Affinity enrichment 1g Peak height/relative 80% (22 million)
Infinium Site-specific probes+bisulfite 500ng Single base 2% (0.6 million)
BeadChip
450K
CHARM 2.0 Restriction enzyme 2g Peak height/relative 19% (6 million)
NOMe-Seq GpC 25g Single base NA
methyltransferase+bisulfite
ChIP-BS-seq Chromatin capture+bisulfite 60ng Single base NA
Agilent Target enrichment+bisulfite 1g Single base 13% (3.7 million)
SureSelect
2.2.1 Whole Genome Whole genome bisulfite sequencing (WGBS) is perhaps the most
Bisulfite Sequencing powerful method to interrogate the methylome as it potentially
(WGBS) allows investigation of every CpG site in the genome (typically
2022 million CpGs are covered in the mappable human genome).
WGBS libraries can be prepared similarly as for normal whole
genome sequencing, except with the additional step of bisulfite
conversion. Variants of WGBS approaches are T-WGBS
(transposase-based library construction) and PBAT (post-bisulfite
adaptor tagging, which can be performed with as little as 125pg of
input DNA) [13].
2.2.2 Reduced Reduced representation bisulfite sequencing (RRBS) is an alterna-

Representation Bisulfite tive to WGBS and shares many conceptual similarities. In terms of
Sequencing (RRBS) library preparation, RRBS differs in two aspects from WGBS.Firstly,
instead of sonication, genomic DNA is digested with a methylation
insensitive restriction enzyme to enrich for CpG sites in the
digested fragments [14]. Secondly, a size-selection (generally
40220bp) step is performed on the digested fragments (each
fragment contains at least one CpG site) that reduces the require-
ment of extensive sequencing. If a size-selection of 40220bp is
performed then for human, mouse, and zebrafish, ~2.3, 1.4, and
2.2% of the genomes are represented, respectively. The method
enriches for CpG sites (5.7-fold for the human genome) [15],
CpG islands (~30-fold for human) [16] and gene promoters and
gene bodies. However, the representation of repeat elements and
enhancers is low in RRBS. This reproducible technique [1722]
has so far been applied to generate genome-wide methylation maps
for mammals, including humans [15, 20], mice [16] and rats [23],
and other vertebrates such as zebrafish [24, 25]. Variants of
standard RRBS methods are double restriction RRBS (for exam-
ple, with ApeKI and MspI together that allows increased CpG cov-
erage), multiplexing RRBS (mRRBS) [26, 27], laser-capture
microdissection RRBS (with ~1ng of input DNA) and single cell
or scRRBS.The steps of data processing in RRBS are similar to
those of WGBS.
2.2.3 Human- The HumanMethylation450 BeadChip (referred to as 450K from

Methylation450 BeadChip this point forward) is the most popular method for genome-wide
(450K) DNA methylation profiling at this stage and it provides informa-
tion on 482,421 CpG sites [28]. Probe extension assays are used
to distinguish between methylated and unmethylated template
DNA at each of these sites [28]. Data processing can be performed
using several publicly available R analysis packages (see
Subheadings2.3 and 3 for details).
2.2.4 Cross-Talk The reason for the popularity of the 450K platform is that it allows
BetweenWGBS, RRBS, investigation of multiple samples at low cost and data analysis
and450K is easier. If hundreds of samples are used then good quality
information will certainly be obtained for >450,000 CpG sites

for each of the samples every time the experiment is repeated. For
RRBS, obtaining data for each sample for the same site is not certain
as it will depend on the sequence coverage and library preparation
variation (e.g., PCR bias). However, 450K arrays allow investiga-
tion of only 1.7% CpG sites in the genome and they suffer from
the problem of not including enough CpG sites for many genes (in
comparison, RRBS investigates ~10 times more CpG sites than
450K). In addition, allelic information cannot be obtained from
the 450K platform unlike RRBS and WGBS where sequences are
obtained and allele-specific methylation (ASM) or the methylation
contribution of subpopulations of cells (in the case of cancer studies)
can be determined. At time of writing, 450K is only available for
analysis of human samples; in contrast, if a reference genome is
available, RRBS and WGBS can provide methylomes at base-
resolution DNA for any organism.
In WGBS, two lanes of sequencing using the Illumina HiSeq
system are required to obtain tenfold average coverage of CpG
sites (i.e., 445 million paired-end reads) for one sample. Currently,
this would cost ~US$6000, which is the equivalent to eight RRBS
samples. Also in comparison to RRBS, WGBS generates a large
volume of raw data that can be challenging to analyze. Interestingly,
a recent analysis of 42 datasets of different human cell types con-
cluded that WGBS is generally inefficient as >70% of sequenced
reads did not yield useful information about CpG methylation
[29]. Taken together, the choice between these two techniques
will depend on the type of information required from the investi-
gation, the number of samples to be processed, cost, and availabil-
ity of resources for obtaining and analyzing data.
2.2.5 Enrichment-Based Enrichment-based methods include Methylation-dependent

Techniques Immunoprecipitation followed by sequencing (MeDIP-Seq),
methyl-CpG binding domain (MBD) protein-enriched genome
sequencing (MBD-seq), methyltransferase-directed Transfer of
Activated Groups followed by sequencing (mTAG), and
Methylation-sensitive Restriction Enzyme digestion followed by
sequencing (MRE-seq). However, it is now established that these
techniques have limitations in that: (1) they do not allow investiga-
tion of single CpG sites or provide information about the relative
abundance of DNA methylation in large genomes, (2) the effi-
ciency of immunoprecipitation depends on sequence specificity
and CpG density; therefore the results for the methylation status of
a region are affected by the efficiency of pull down and major com-
putational adjustments are needed to minimize the bias due to
varying CpG density in the genome [8], and (3) if samples have
copy number variation, then additional sequencing of control and
further normalization is required. Large-scale studies have demon-
strated enrichment-based methods have lower accuracy compared
to RRBS and WGBS [30, 31].
2.2.6 Combinatorial Recently, combinatorial techniques have been developed to analyze

Techniques toProfile additional epigenomic marks along with DNA methylation. For
Multiple Epigenomic Marks example, the Nucleosome Occupancy and Methylome sequencing
(NOMeseq) method determines the position of nucleosomes and
DNA methylation at the same time. In NOMeseq, nuclei are
treated with GpC methyltransferase (M.CviPI) before bisulfite
conversion, followed by sequencing. This approach allows the
analysis of DNA methylation status and chromatin status (nucleo-
some occupancy) together. Another technique in this category is
Chromatinimmunoprecipitation bisulfite sequencing (ChIPBS
seq), which allows investigation of histone modification marks and
DNA methylation on the same DNA.For this method, immuno-
precipitation of chromatin with an antibody or methyl-binding
protein is performed first (to capture a specific chromatin protein
or histone mark) followed by bisulfite sequencing of the same
enriched DNA.
2.3 Processing Data from sequencing services are in the form of fastq files, which
andAnalysis contain a header line, the sequence read, and the quality of each
ofGenome-Wide DNA sequence. Illumina NGS platforms are often used for methylation
MethylationData work and return millions of reads for each sample. It is common
practice to use Illumina chemistry to yield at least 100125bp
reads at acceptable quality and this may be done with single-ended
or paired-end runs.
After some preprocessing, the methylation status is established
by mapping bisulfite-converted sequence reads back to the refer-
ence genome using mapping programs that allow for C to T con-
versions. Measuring CpG site methylation can be done on an
individual site-by-site basis, but it is more commonly integrated
over a window to smooth site-to-site variation. Analysis of WGBS
usually employs window lengths of 1000bp. RRBS methylation
analysis is appropriately performed on windows that are defined by
MspI fragments [32].
2.3.1 Assessment
andProcessing Pipelines for methylation sequence data usually begin by checking
of Sequencing-Based the run quality with applications like FastQC (Fastqc tools: http://
MethylationData www.bioinformatics.bbsrc.ac.uk/projects/fastqc). Besides indicat-
ing the overall quality, FastQC gives a per base sequence content
plot, which can be informative for the effectiveness of bisulfite con-
version, especially for RRBS where a very characteristic signature
results from the MspI C/CGG cleavage sites (see Note 1).
Sequence reads can be quality trimmed (e.g. fastq_quality_
trimmer URL: http://hannonlab.cshl.edu/fastx_toolkit/) to
truncate each read where the quality falls below a certain thresh-
old. Q20 is commonly used for the limita level where there is a
1% chance of a base read error. However, this strategy can reduce
the length of reads excessively, when there is a single lower quality
sequencing cycle, so a preferred method is to hard trim all of the
reads at a length where the quality is deteriorating below Q20 as

indicated by the per base sequence quality plot of FastQC.
The application fastx_trimmer is one such method.
Adaptor sequence trimming should be performed to ensure
that the adaptor sequences used in Illumina chemistry have not
caused mapping problems. This is desirable for WGBS in which
some shorter genomic fragments may be generated, but it is essen-
tial for RRBS where 40220bp size selection followed by 100 or
125 sequence cycles ensures that a significant proportion of reads
will run into the adaptor. For RRBS, a further complication is that
sequence library preparation inserts a C at the 3-end of each frag-
ment immediately before the trailing adaptor. For reads that con-
tain adaptors, both the adaptor and the preceding C must be
trimmed. The DMAP [32] application cleanadaptors is one
which performs this operation; the usual strategy is to trim back
3bp from the start of the adaptor [15] (see Note 2).
Since next-generation sequencing data files tend to be large,
they are usually stored in a compressed format. For example, fastq
files are gzip compressed to be fastq.gz, making them about one
third the size. Most applications for the steps above can work with
compressed data interchangeably.
2.3.2 Alignment ofReads Mapping bisulfite reads to a reference genome requires aligners
forSequencing-Based DNA that map C residues in reads to Cs in the reference while mapping
Methylation Data Ts to either reference Cs or Ts. For mapping complementary reads,
the A and G matches must be similarly treated. A selection of
commonly-used tools for aligning bisulfite reads are listed in Table2.
Two strategies exist for this: use of aligners that incorporate map-
ping asymmetry (e.g., BSMAP [33], RMAPBS [14], BRAT [34])
or pre-conversion of the reference genome and the sequence reads
to C-to-T converted forms and mapping with normal sequence
aligners. Major aligners that use pre-conversion of the genome are
Bismark [35] (see Notes 3 and 4), BS-Seeker [36], and MethylCoder
[37]. These alignment programs accomplish genome mapping
with bowtie1 [38], bowtie2 [39], Burrows Wheeler alignment
(BWA) [40], or their own implementation of established align-
ment methods. For example, BRAT-BW [41] which implements a
modified BWA method.
Sequence reads generated with ABI SOLiD technologies are
usually presented in a colorspace format, which encodes the transi-
tions between nucleotide pairs as a string of numerical values. This
complicates the mapping of bisulfite reads where C/T transitions
must be tolerated. None-the-less, some aligners have been devel-
oped to work with this representation (B-SOLANA [42], PASS-bis
[43], BATMETH [44]). However, the use of ABI SOLiD technol-
ogy in genome-wide DNA methylation analysis is limited and
therefore interest in these aligners might decrease in future.
Table 2
Tools for alignment of genome-scale bisulfite-sequenced reads
Aligner Language based on Requirement Notes URL

BSMAP C++, modified 64-bit Linux SAM output https://code.google.
SOAP systems com/p/bsmap/
RMAPBS C++, modified 64-bit Linux Equal length reads http://rulai.cshl.edu/
RMAP systems rmap/
Bismark Perl & bowtie [2] Pre-convert SAM, BAM & http://www.
reference native outputs bioinformatics.
genome babraham.ac.uk/
Samtools package projects/bismark/
BS-Seeker Python & bowtie Pre-convert SAM, BAM & http://pellegrini.mcdb.
[2] reference native outputs ucla.
genome edu/BS_Seeker2/
BatMeth C, perl Uses colorspace Basespace & https://code.google.
format data colorspace com/p/batmeth/
PASS-bis C++ Uses colorspace Prebuilt indices http://pass.cribi.unipd.
format data it/cgi-bin/pass.pl
MethylCoder Python Bowtie, python Superceded by https://github.com/
libraries bwa-meth brentp/methylcode
bwa-meth Python BSA, samtools BAM file output https://github.com/
brentp/methylcode
BRAT C++ 64-bit, Linux/ Proprietary output, http://compbio.cs.ucr.
Unix <=4.2Gb genome edu/brat/
BRAT_BW C++ 64bit, Linux/ Own BW aligner, http://compbio.cs.ucr.
Unix same limits as edu/brat/
BRAT
B-SOLANA Python Bowtie, samtools SOLiD colorspace https://code.google.
aligner, com/p/bsolana/
preindexing
Some aligners will map paired end reads but with longer read
lengths and, especially for RRBS, short library insert lengths cause
overrepresentation of the read portions where read 1 and read 2
overlap. For this reason, it is common practice to use only single-
ended reads for methylation analysis.
2.3.3 Processing Several tools have been developed to assist with Infinium
andAnalysis of450K Data HumanMethylation450 beadarray data analysis (Table3). For
many of these tools an automated option is available to process
large amounts of data. The majority of these tools are implemented
in the R platform and detailed descriptions of individual tools are
available elsewhere [45]. For 450K data, it is important to perform
Table 3
Tools for analyzing differential methylation patterns from genome-wide data
Tool Based on Notes/comments/suitable for References

Tools for sequencing-based genome-wide data
DMAP C Differential methylation analysis for WGBS and RRBS and [32]
novel fragment-based approach for RRBS.AVOVA, Chi
Square test, and Fishers exact tests are implemented.
BSmooth R DMR identification uses localized smoothing algorithm [70]
BiSeq R Uses a hierarchical procedure to trim significantly DMRs [71]
following FDR correction
methylSig R Uses a beta binomial model to test for DMCs or DMRs [93]
RADMeth DMCs or DMRs determined using beta binomial regression [94]
methylKit R Incorporates logistic regression-based DMC analysis, sample [65]
clustering, and visualization
Tools for 450K analysis
RnBeads R Comprehensive analysis of DNA methylation data from a wide [95]
range of assays, providing highly annotated reports
Minfi R Includes preprocessing capabilities and both bump hunting and [54]
block finding capabilities for identification of DMRs
MethyLumi R Provides R classes for quality control, normalization, and [53]
analysis of 450K array data
wateRmelon R Preprocessing of 450K data incorporates background [56]
adjustment and between-array quantile normalization
ChAMP R Integrates several methods for analysis of batch effects, DMRs, [96]
and copy number aberrations
DMRcate R Ranks DMRs based on tunable kernel smoothing of the [97]
differential methylation signal
missMethyl R Normalization, removal of unwanted variation in differential [98]
methylation analysis, gene set analysis for 450K
careful processing of the raw data before differential methylation

analysis is performed, as there are several sources of variation in the
450K platform that could result in significant bias in downstream
analysis. Here, we summarize the sources of bias and steps required
for processing of raw 450K data.
2.3.3.1 Filtering There are several main sources of problematic probes in the 450K
Low-Signal, Cross- platform, which need to be removed before the next step is per-
Reactive, andSNP- formed. Raw IDAT files provide a P-value to denote the detection
Associated Probes confidence of a probe. For quality control, probes that do not pass
a detection limit (for example P>0.05 or P>0.01) due to a low
quality signal, spatial artifacts, or low performance of a particular
experiment, are filtered out [46].
The bisulfite conversion process reduces sequence complexity
and this leads to increased cross-reactivity in the probes. Infinium
probes are designed to hybridize to one location on the genome;
however, due to cross-reactivity some probes can co-hybridize to
multiple sequences (generally target repetitive sequences). The loss

of specificity can lead to wrong methylation calls for downstream
analysis [47]. Two studies have provided different estimates of the
number of cross-reactive probes in the 450K platform. In silico
analysis has suggested that 8.6% of the probes are cross-reactive
[48], whereas, Bowtie mapping of the probe sequences (allowing
up to two mismatches) indicated that 140,000 probes mapped
(almost 29% of all the probes) to multiple locations in the human
genome (hg19) [49]. Therefore, it is important to remove the
cross-reactive probes before further analysis. Analysis tools such as
minfi provide options for filtering out these probes (Table3).
It is estimated that ~4.3% of 450K probes are located in docu-
mented SNPs [48]. These SNPs can provide three different meth-
ylation calls for one site based on the allele of an individual; 100%
methylation (homozygous first allele, C/C), 50% (heterozygous
allele, mix of C and T) or 0% (homozygous second allele, T/T).
SNPs can also affect hybridization, thereby biasing the quantifica-
tion of methylation. Therefore, it is advisable to remove these
SNPs during preprocessing steps. The 450K analysis tools (Table3)
provide options to exclude them.
2.3.3.2 Two-Assay The HumanMethylation450 BeadChip is an extension of The

Design of450K BeadChip HumanMethylation27 BeadChip assay. However, there is a funda-
mental difference in the design of these assays. The 27K assay used
Infinium I probe chemistry, where two bead types for one CpG site
were used (one for the methylated and one for the unmethylated
alleles, one in the red channel and one in the green channel). The
signal intensities for both alleles at a site were measured on the
same color channel (Cy3 green for some loci and Cy5 red for oth-
ers). 28% of the probes (135,476 CpG sites) in the 450K platform
come from the Infinium I assay. The other 72% (350,036 CpG
sites) in the 450K platform employ Infinium II chemistry where
one bead per probe is used and two different color channels
represent the two different probe-extension products (Cy3-labeled
guanine for methylated alleles and Cy5-labeled adenine for
unmethylated alleles) [28]. This two-assay design and the different
range of methylation values displayed by these two types of probes
leads to a source of variation and therefore normalization within
the array is necessary. Two main steps can be considered for 450K
array normalization: (1) background and color bias correction
(Subheading2.3.3.3) and (2) Infinium I/II-type bias adjustment
(Subheading2.3.3.4).
2.3.3.3 Background For Infinium I, it has been observed that the methylation values
andDye-Bias Correction provided in the red or the green channel differ substantially. In
Infinium II probes, the methylated and unmethylated alleles are
measured in two different channels resulting in a reduction in the
dynamic range of the beta values (methylation values). For the
Infinium I assay, each CpG is evaluated in the same color channel

and the intensity information from the unused color channel can
be used for background correction. In HumanMethylation450K
Beadchips, 850 internal control probes are included (613 negative
control probes for measuring background intensity and 186 con-
trol nonpolymorphic probes to measure color differences between
channels) [50]. Due to these issues, background correction and
dye-bias adjustments are required for the 450K platform. Several R
packages provide analysis suites (methylumi, RnBeads, and minfi)
for background and color bias adjustments.
2.3.3.4 Type I/II Probe As a result of differences in probe design, as described above
Correction (Subheading 2.3.3.2), the methylation values for type I and II
probes can vary, and to account for these differences several nor-
malization methods have been implemented by different tools.
Although it is difficult to recommend a particular normalization
method, multiple methods can be explored and overlapped to get
a sense of the effectiveness of normalization [51]. Several normal-
ization methods for 450K data analysis are used: (1) noob (normal-
exponential convolution using out-of-band probes) [52], as
implemented by the methylumi package [53], scales the internal
controls (red and green channel). (2) Subset-quantile for Within
Array Normalization (SWAN), which is implemented by the popu-
lar minfi package [54]. (3) The Beta MIxture Quantile normaliza-
tion (BMIQ) [55] method. The Lumi package provides an effective
strategy of normalization by combining quantile normalization
with the BMIQ approach [51]. (4) The data-driven normalization
approach introduced in the wateRmelon package [56]. (5)
Normalization methods based on peak height correction as imple-
mented in the IMA package [57]. RnBeads provides an option to
perform any one of the approaches (1) to (4), if desired (Table3).
2.3.3.5 Representation DNA methylation data from 450K arrays are represented as beta
ofMethylation inthe450K values (see Note 5), which are defined as the ratio of the methyl-
Platform: Beta Value vs. ated signal over the total signal (methylated+unmethylated). Beta
MValue values range from 0 (completely unmethylated) to 1 (completely
methylated). However, the potential for variation is unequal for
highly methylated, unmethylated, and intermediate methylated
CpG sites (referred as heteroscedasticity). Values close to beta=0
or beta=1 have a low standard deviation as methylation tends to be
tightly clustered around these extremes, whereas the standard
deviation for intermediate methylation (beta=0.5) is higher. An
alternative approach is to express methylation as M values (see
Note 6). The M values can be derived by performing a logistic
transformation of the beta values. This transformation is linear for
intermediate methylation but affords a better resolution of hypo-
and hypermethylation. In an initial study it was demonstrated that
the M-value method improved the detection rate and the true
positive rate of differential methylation for both highly methylated

and highly unmethylated CpG sites [58]. However, a later study
(with a higher sample number and replicates) showed that the dif-
ference between the beta and M value methods are negligible in
detecting differential methylation; however, with a small sample-
size the M value method performed slightly better [59]. Perhaps
for statistical modeling it is more appropriate to use M values, as it
is less likely to show characteristic bias towards methylated or
unmethylated sites; however, for methylation analysis beta values
are preferable to use as they are conceptually simple to interpret in
biological contexts.
2.3.3.6 Emerging Several additional tools and approaches are being developed to
Aspects in450K Data gain a more inclusive and comprehensive view of the DNA meth-
Analysis ylation changes in a cell using 450K analysis. Three aspects are of
particular interest. Firstly, the DNA methylation pattern of every
cell type is unique and the use of mixed cell types (such as whole
blood) can confound findings due to cellular heterogeneity. Several
statistical methods have been developed based on previous data
and the minfi and RnBeads packages provide analysis suites to eval-
uate the contribution of different cell types in whole blood. Two
additional tools, EWasher (a package written in Perl) [60] and
RefFreeEWAS (an R-package) [61], also facilitate analysis of cell
type composition. However, it is important to be cautious while
using these approaches as these algorithms can over-correct the
data and therefore can obscure methylation changes of interest.
A second aspect is prediction of epigenetic age. Horvath etal.
[62] analyzed 8000 samples (using array-based methylation data)
and proposed that 353 CpG sites that are a predictor of age. They
also proposed that an epigenetic clock or age of tissue could dif-
fer from chronological age. This difference could be due to the
environmental exposure of an individual. An epigenetic clock is a
tool that allows prediction of the epigenetic age of a sample using
450K data. Hannum etal. have used another methylation dataset
and algorithm for prediction of age [63]. Integration of these tools
could be used to further explore aging and DNA methylation.
Thirdly, 450 K (or any other bisulfite conversion-based
method) does not distinguish between 5-methylcytosine (5mC)
and 5-hydroxymethylcytosine (5hmC). A recent method has been
developed that adapts oxidative bisulfite (oxBS) chemistry to spe-
cifically detect both 5mC and 5hmC in a single workflow using
450K BeadChips, called oxBS-450K [64].
2.3.4 Identification In bisulfite sequencing methods, methylation levels of a genomic

ofDifferential Methylation region or site are calculated from the counts of methylated and
unmethylated reads. The next step is to identify differences in meth-
ylation between different groups and conditions (for example, dis-
ease group vs. control group). For analyzing differential methylation,
several approaches can be used for the unit of analysis (Fig.1).
2.3.4.1 Analysis Unit A simple and commonly used approach is to analyze each CpG site
forDifferential Methylation in the investigated samples and then identify differentially methyl-
Single CpG Site ated sites, referred to as differentially methylated positions (DMPs),
Approach or differentially methylated CpGs (DMCs). For analyzing array-
based data (such as 450K) it is necessary to interrogate single CpG
sites because for many genes or regions there is only information
about one CpG available in array platforms. Further, if multiple
CpGs are covered for a gene, in many cases these CpGs are far
apart from each other and therefore using a single CpG site as the
unit of analysis is the common approach for these types of data. A
DMC approach is perhaps more useful when a small number of
CpG sites are analyzed (Fig.1). For single CpG analysis, a widely
used tool is methylKit [65], an R package that applies a Fishers
exact test or logistic regression to calculate p-values, which are
adjusted to q-values for multiple test correction using a SLIM
approach [66].
DMR (Differentially For techniques such as WGBS or RRBS where millions of CpG
Methylated Region) sites are investigated (e.g., in humans, WGBS covers ~30 million
Approach CpG sites, while RRBS covers ~4 million CpG sites), investigation
of every single CpG site as an independent unit of analysis can
greatly increase the false discovery rate. This is due to the fact that
Fig. 1 Major analysis approaches for genome-wide DNA methylation analysis. There are several approaches
for analyzing differential methylation between different groups and conditions. These approaches differ based
on the unit of analysis: (1) the single CpG site approach independently analyzes each CpG site in investigated
samples; (2) for RRBS, MspI-digested fragments can be used as the unit of analysis (implemented in DMAP
[32] package); (3) a common approach for large bisulfite sequencing data is to investigate regions with fixed
size genomic windows. It is possible to use sliding windows based on user-specified criteria
variation at single sites is greater than that of a contig of sites since

the relatively lower coverage per site increases the sampling-based
variation [67]. Further, statistically, the number of pieces of infor-
mation available for a single CpG site is lower than for a region and
therefore comparison of a single site is less robust than that of a
region.
Therefore, a common approach for large bisulfite sequencing
data is to investigate larger regions and identify differentially meth-
ylated regions or DMRs (Fig.1). Several studies have used a fixed
or sliding window approach where the genome is divided into
intervals (typically 1000bp length) and differential methylation is
detected in these regions [68, 69]. The DMAP package provides a
flexible option of defining any length of tiled window for analyzing
WGBS or RRBS data (Table3). In addition to the straightforward
tiled approach, several variations and additional criteria can be
imposed for detection of DMRs. For example, for WGBS data,
BSmooth performs local averaging and sample-wise smoothing of
methylation values after alignment and methylation estimates by
read position [70]. BSmooth applies numerous CpG-wise t-tests
and based on a t-test threshold, differentially methylated regions
(DMRs) are defined. Another tool, called BiSeq, which can be
used for RRBS or WGBS data, performs smoothing of methylation
data within CpG clusters considering spatial dependence [71].
Differential methylation is then detected in CpG clusters; the false
discovery rate (FDR) is controlled and finally DMR boundaries are
defined. However, BiSeq considers the spatial arrangement of CpG
sites in the genome and defines CpG clusters by specifying a num-
ber of frequently covered CpG sites (option: min.sites) that are
close to each other (option: max.dist) and uses these clusters for
subsequent analysis. Imposing flexible criteria for defining CpG
clusters (i.e., lowering min.sites value) in an analysis would result
in a higher number of analyzable CpG clusters and a higher num-
ber of DMRs, but might increase the chances of false positives.
DMF (Differentially The tiled DMR approach is well suited for WGBS, but for RRBS,
Methylated Fragment) where only 2.5% of the genome is sequenced, the majority of the
Approach windows will be empty or have partial inclusion of fragments.
Further, if a small region is variably/differentially methylated
between individuals, use of a 1000bp or longer window might
dilute this variation [67] and therefore might be not be detected if
a large window size is used. For RRBS, a new MspI fragment-based
approach for investigating DNA methylation was introduced by
the DMAP package. This approach is conceptually similar to the
DMR approach, but instead of fixed-length windows, MspI-
digested fragments of 40220bp lengths are used as the unit of
analysis (Fig.1). After identifying DMFs, it is possible to use a tiled
approach to identify dense clusters with DMFs. Effective use of
DMF has been recently demonstrated for profiling human neutro-
phil methylomes [72].
3 Tools forVisualization, Integration, andMiscellaneous Analysis ofDNA

Methylation Data
Due to the complexity of profiling DNA methylation patterns in

the genome, the development of new tools has been an active area
of research. Initially, after the advent of NGS technologies several
alignment tools were developed and in recent years a focus has
been to develop user-friendly, web-based tools for easy analysis of
DNA methylation data for bench scientists. Details of these
resources can be found in Table4.
3.1 Visualization For visualization of DNA methylation data, two excellent tools are
ofDNA Methylation SeqMonk and Integrative Genomics Viewer (IGV). SeqMonk is a
graphical Java application and freely distributed for a wide range of
computer platforms (Microsoft Windows, Linux, MacOS X, and as
Java source code). SeqMonk is pre-configured with a significant set
of genomic sequences and their annotations. When a genome is
loaded, either de novo or as a project, the displays are capable of
showing genes, mRNAs and exons but the displayed information is
variably configurable depending on requirements. Furthermore,
information from SeqMonk feature tables can be used to identify
proximal genes and CpG features of candidate variable regions.
Apart from visualization, SeqMonk also allows differential meth-
ylation analysis using its quantitation pipelines. IGV is a desktop
application designed to enable interactive exploration of large-scale
genomic and epigenomic data sets [73]. IGV is written in the Java
programming language and supports all major operating systems
(Windows, Mac, and Linux). Using a drop-down menu, it is pos-
sible to load a genome of choice. IGV also enables visualization of
several different datasets together. For visualization of individual
read or allelic methylation pattern, IGV accepts aligned and sorted
BAM or SAM files as input (see Notes 710).
3.2 Epigenomic Several browsers have been developed for the interactive explora-
Browsers forExploring tion of epigenomic data. Although UCSC and Ensembl browsers
Public Datasets are suitable for any type of genomics analysis, some browsers have
been designed especially for epigenetics. For example, the WashU,
EpiViz, and NCBI epigenomics browsers. These tools provide
opportunities for integrating user-generated DNA methylation
data with other epigenomic marks (such as histone marks, DNAase
hypersensitive sites, etc.) to understand the relationship of identi-
fied altered DNA methylation marks with other epigenomic fea-
tures. In the last 3 years, massive amounts of epigenomic data have
been released by ENCODE and the Roadmap Epigenomics proj-
ect. Downloading and analysis of each of these datasets can pose
significant challenges; however, these browsers allow for easy visu-
alization of these datasets and can help in hypothesis generation.
Table 4
264
User-friendly tools for integrative and comprehensive analysis of the epigenome
Tool Input file Suitable for URL

Visualization tools genome-wide methylation data
SeqMonk Various file format supported. e.g., SAM/ Analysis and visualization of bisulfite sequence http://www.bioinformatics.babraham.
BAM or text files data ac.uk/projects/seqmonk/
IGV Sorted BAM or SAM files after alignment Visualization of bisulfite sequence data at the https://www.broadinstitute.org/igv/
sequence read level
IGB Sorted BAM or SAM files after alignment Visualization of bisulfite sequence data at the http://bioviz.org/igb/index.html
sequence read level
Spark BAM, SAM, .txt and other formats Clustering (k-means) of epigenomic data, such http://www.sparkinsight.org
Aniruddha Chatterjee et al.
as chromatin marks from Chip-Seq

experiments
MethVisual Multiple or separate FASTA files or gff Alignment, quality control, visualization and https://www.bioconductor.org/
files statistical analysis packages/release/bioc/html/
methVisual.html
Browsers for epigenomic analysis
UCSC genome Gene or genome-coordinate or custom Data extraction from table browser and http://genome.ucsc.edu/index.html
browser file by users (bed, bedGraph, Bigwig, exploration using custom tracks
and other formats)
WashU epigenome Gene or genome-coordinate or custom Comparing user data to public datasets http://epigenomegateway.wustl.edu
browser file by users (multiple cell type and tissue data is available)
VizHub Visualize sequencing data from Roadmap http://vizhub.wustl.edu
Epigenomics project
NCBI epigenomics Gene or genome-coordinate Visualize and download public epigenomic http://www.ncbi.nlm.nih.gov/
datasets epigenomics
Ensembl Similar functionality like UCSC browser.
Datasets and annotations could be
downloaded using BioMart function
1000 methylomes Gene or genome-coordinate 1000 DNA methylation projects and functional http://cbbiweb.uthscsa.edu/
enrichment analysis kMethylationWS/ws
EpiViz Gene or genome-coordinate Visualization of functional genomics data. http://epiviz.github.io
Provides option to make user-friendly plots
from the data.
Genomic Genome-coordinate Statistical analysis of genomic interval data https://hyperbrowser.uio.no/hb/
hyperbrowser
Tools for integrative epigenetic analysis
Epiexplorer Genome-coordinates in bed format Integration and analysis of users data with http://epiexplorer.mpi-inf.mpg.de
ENCODE data
ChromHMM Genome-coordinates or users data Integrates multiple chromatin datasets for http://compbio.mit.edu/
several histone modification marks to identify ChromHMM/
combinatorial and spatial chromatin marks
MAPPER Genome-coordinates Transcription factor binding factor analysis http://genome.ufl.edu/mapper/
EpiGRAPH Genome-coordinates or users data Integrated analysis of DNA methylation and http://epigraph.mpi-inf.mpg.de/
histone data using public datasets WebGRAPH/
Databases for DNA methylation related analysis
MethDB Online form to query genes or regions Obtaining processed methylation data of many http://www.methdb.de
published papers
MethBank Genome-coordinates or gene names Exploration of genome-wide methylomes of http://www.dnamethylome.org
gametes and early embryos in model
organisms (zebrafish and mouse)
Miscellaneous user-friendly tools for methylome analysis
Methylation Tab-separated file containing methylation Generates plots and statistical summaries of http://gattaca.imppc.org:3838/
plotter value up to 100 CpGs in 100 different DNA methylation status in small datasets methylation_plotter/
samples
LiftOver tool from List of genome coordinates Converts genome coordinates from one build https://genome.ucsc.edu/cgi-bin/
UCSC to another hgLiftOver
Methclone BAM files Allelic methylation analysis using next- https://code.google.com/p/
generation sequencing data methclone
Bis-SNP Sequenced read level data Identifies nucleotide polymorphisms (SNPs) in http://epigenome.usc.edu/
bisulfite sequence data publicationdata/bissnp2011
Tools and Strategies for Mining DNA Methylomes
265
3.3 Miscellaneous EpiExplorer is a web server-based tool that enables an interactive

Tools forDNA platform for exploring large-scale genomic or epigenomic datasets
Methylation Analysis for the human and mouse genomes [74]. The input file for
EpiExplorer is a list of genomic regions (for example, regions iden-
tified as differentially methylated in a users study, in BED format).
EpiExplorer allows analysis of the association of several genomic
features with user-specified genomic regions. Some of the key fea-
tures that can be analyzed on the EpiExplorer webserver are histone
modifications (ENCODE), DNase I-hypersensitive sites (ENCODE),
DNA methylation (ENCODE), chromatin state segmentation,
transcription factor binding sites (ENCODE), lamina- associated
domains, repeat elements, and conservation (46-way by PhastCons).
Another useful tool is Epigram that predicts histone modifications
and DNA methylation patterns from DNA motifs. Recently, several
tools have been developed for region-specific plotting of genome-
wide DNA methylation data (Table4). One such tool is methclone,
which analyzes DNA methylation changes using a composition
entropy difference calculation and a measure of shift in clonality in
alleles [75]. This allows the comparison of clonal or allelic methyla-
tion changes. Methclone accepts aligned bam files and is relatively
easy to use; however, processing speed could be a limiting factor if
a large number of samples is analyzed.
4 Validation or Gene-Specific Analysis ofDNA Methylation
Similar to genome-wide techniques, several techniques are available

to study the DNA methylation status of specific genes or candidate
regions. These techniques are based on two major components:
bisulfite conversion and PCR.The choice of these techniques
mainly depends on three factors. First, how many genes are required
to be investigated. Second, for validation studies, what kind of
methylation values/differences need to be validated. Third, the
type of information required from the validation experiments.
These techniques are relatively easy to implement in a typical molec-
ular genetics laboratory without specialized equipment. Further,
the process for these types of data analyses is relatively easy and
several interactive tools are available to assist with data analysis with-
out the need for extensive bioinformatics knowledge (Table5).
4.1 PCR Followed Bisulfite sequencing PCR (referred to as BSP) was first described
by Sequencing-Based almost two decades ago and is one of the first (and still widely
Techniques used) techniques described for analyzing region-specific DNA
methylation status [76]. This approach involves PCR amplifying a
gene/region of interest using bisulfite primers followed by
sequencing the PCR product. There are two ways by which the
sequencing can be performed. The PCR product can be directly
Table 5
Tools and resources for analyzing gene-specific DNA methylation patterns
Tool Suitable for Source

Primer design and related tools
Methprimer The most widely used program for primer http://www.urogene.org/
design methprimer/
Bisulfite primer Designed by Zymo. Functionality similar to http://www.zymoresearch.com/
seeker methprimer tools/bisulfite-primer-seeker
BiSearch Checks for mispriming sites and allows in http://bisearch.enzim.hu
silico tests to assess amplification efficiency
Beacon Useful for designing primers for methyLight www.premierbiosoft.com/molecular_
designer experiments beacons/index.html
BioWord This is a Microsoft Word add-in that allows http://sourceforge.net/projects/
reverse complementation, alignment, and bioword/
different editing functions helpful to
evaluate designed primers
MethBlast In silico evaluation of oligonucleotide http://medgen.ugent.be/
sequence similarities to bisulfite modified methBLAST/methBLAST_cs.php
genome sequences
Allows one to blast the designed primers to
computationally evaluate the potential for
multiple mapping/annealing of bisulfite
converted sequences
Gene-specific data analysis tools
4 Peaks Bisulfite sequence visualization tool http://nucleobytes.com/4peaks/
BISMA Analysis of locus-specific bisulfite sequence http://biq-analyzer.bioinf.mpi-inf.
data and graphical output of the results mpg.de/download.php
BDPC A tool based on BISMA.Provides an http://services.ibc.uni-stuttgart.de/
integrated framework for analyzing data BDPC/
from bisulfite sequencing experiments
BiQ analyzer Comprehensive analysis tool for bisulfite http://biq-analyzer.bioinf.mpi-inf.
sequencing data analysis mpg.de/download.php
ESME The special feature of ESME is its ability to www.epigenome.org/index.
normalize for background noise and php?page=download
unconverted sequence. Particularly useful
for direct-BSP data normalization
QUMA Tool to align, visualize, and quantify bisulfite http://quma.cdb.riken.jp/
sequence data
CpGassoc R tool for analysis and visualization for http://genetics.emory.edu/conneely
regional DNA methylation
sequenced (generally using the Sanger sequencing method,

referred as direct-BSP) or the PCR products can be cloned
followed by sequencing (cloning-BSP). The PCR products are
generally cloned into a suitable plasmid vector which is then trans-
formed into competent Escherichia coli cells. Several transformants
are picked and purified independently for sequencing. The direct-
BSP is less laborious and can provide quick validation results; how-
ever, due to the mixture of alleles and heterogeneous methylation
patterns, direct-BSP leads to noisy sequencing results and it is rela-
tively hard to resolve intermediate levels of methylation using
direct-BSP.On the other hand, cloning-BSP requires more effort
and time, but it provides information for each clone (allele) and
the sequences obtained in this method are relatively clean for
interpretation.
Another variant of BSP is pyrosequencing, which has been
extensively used for validation of genome-wide methylation data.
In this method, after bisulfite conversion, target regions are ampli-
fied similar to BSP; however, one of the amplification primers is
biotinylated at its 5-end. In the next step, purification is performed
(streptavidin-coated beads) and only the DNA strand containing
biotin is retained. Finally, a sequencing primer is used to perform
sequencing-by-synthesis and nucleotide incorporation is moni-
tored in real-time and the light intensity of enzymatic conversion
of the released pyrophosphate in each step is recorded (pyrogram).
The methylation status of each CpG is determined from the inten-
sity ratio of T and C [77].
4.2 Targeted Several new methods have been developed that allow targeted
High-Throughput investigation of hundreds of regions simultaneously by combining
Sequencing the principle of BSP with high-throughput sequencing or digital
technologies. One such approach is BiSulfite Amplicon Sequencing
(BSAS) [78]. In this method, multiple target regions are amplified
using bisulfite primers, as similarly described in BSP.Next, these
amplicons are indexed (i.e., adding linker sequences to recognize
these sequences post-sequencing). Finally, the barcoded and ampli-
fied regions are sequenced on a small benchtop sequencer (such as
Illumina MiSeq instruments that are capable of producing ~10
million reads of at least 250bp each). Another method for targeted
methylation sequencing is Bisulfite padlock probes (BSPP) [79].
In this method, a padlock probe (with a common linker flanking
both sides of the sequence) anneals to bisulfite-converted genomic
DNA and then extends to form circularized DNA.In the next
step, circularized DNA targets are PCR-amplified with barcoded
primers followed by high-throughput sequencing [79]. Further,
microdroplet PCR technology has been applied for sensitive target-
specific DNA methylation analysis. Bisulfite treatment followed by
microdroplet PCR with next-generation sequencing was demon-
strated to provide high quality methylation information on 2100
genes [80]. This method is automated and currently can be used

to profile 22,000 targets [9]. Application of digital PCR tech-
niques for methylation analysis will be an interesting development
in the future, as these applications have important clinical implica-
tions for translation of methylation research.
4.3 Exclusive Two techniques that are described in this section are methylation-
PCR-Based specific PCR (MSP) and MethyLight. MSP follows a two-tube
Techniques approach, where a region of interest is separately amplified with
two primer sets; one that binds to methylated DNA and another
that binds to unmethylated DNA [81]. The products are then
visualized on a gel to see the unmethylated or methylated status of
the investigated sequence. The MSP technique is qualitative; how-
ever methods have been developed to quantify methylation in real
time using MSP.Two methods that enable quantitative analysis of
MSP are Sensitive Melting Analysis after Real Time-Methylation-
Specific PCR or SMART-MSP (qPCR measurement of product
intensity with a melting step for detecting incompletely bisulfite
converted PCR amplicons) [82] and MethylQuant, which measures
the increased fluorescence from SYBR Green I for quantification
[83]. MSP is a convenient method for methylation analysis and can
very quickly be performed in a standard laboratory setting;
however, it is only a good method if the methylation status of two
samples is binary (i.e., one is completely methylated and the other
is completely unmethylated). This method is not very sensitive for
resolving intermediate methylation or small methylation differ-
ences between two samples.
For gene expression analysis and genotyping hydrolysis (e.g.,
TaqMan), probes have been used for 1-2 decades. MethyLight
uses hydrolysis probe technology to determine the methylation
status of a target region [84]. MethyLight reactions have several
variations [85]; however, in the most frequently used method, two
primers are used for amplification and the hydrolysis probe is
designed to bind to methylated DNA.In this method, a reference
gene is also used to normalize the results, as is used for standard
qPCR.MethyLight can provide quantitative information in real
time. However, this technique is relatively expensive compared to
other BSP methods (as hydrolysis probes are expensive, particu-
larly if many genes are investigated) and suffers from similar prob-
lems to MSP in that this method is less reliable for detecting
intermediate or heterogeneous methylation and more suitable for
analyzing completely methylated alleles.
4.4 Mass Array- The Sequenom EpiTyper assay is one of the most widely used
Based Sequenom methods for analyzing regional DNA methylation. The initial pro-
cess in the Sequenom assay is similar to BSP but it uses a
T7-promoter tagged reverse primer. This product is transcribed
invitro using the T7 primer and polymerase enzyme. Next these
products are cleaved using RNase A, providing numerous small

fragments based on the sequence. When these fragments contain
the complement of a CpG, the mass difference between the G or A
can be detected using Matrix-Assisted Laser Desorption/Ionization
Time Of Flight (MALDI-TOF) spectrometry to measure individual
CpG-site methylation [86]. Sequenom provides a user-friendly data
analysis option (in Microsoft Excel sheets) and is commonly used as
a validation tool for DNA methylation analyses due to its reputation
of high precision and reproducibility [8791].
4.5 Recommen- Bisulfite primers are generally designed to amplify bisulfite-

dations and converted DNA irrespective of its CpG methylation status. The
Considerations for sequences to which the primers bind do not contain CpG dinu-
Bisulfite-Based cleotides. More importantly, the primer design programs (for
Targeted DNA example, MethPrimer) assume that non-CpGs are unmethylated
Methylation Analysis and therefore will be seen as T (after bisulfite conversion) in the
PCR step. However, depending on the cell type, non-CpGs can
be methylated and therefore it could affect the results of bisulfite
PCRs. Several studies have shown that bisulfite primers can pref-
erentially amplify methylated or unmethylated alleles and there-
fore can bias the methylation calls. This is likely due to the
sequence differences of methylated and unmethylated DNA [85].
Primers (designed using MethPrimer) have shown bias toward
the unmethylated DNA and therefore underestimate true meth-
ylation values [92]. To eliminate these biases, primers can be
designed in regions devoid of cytosines or with degenerate bases
(G/A) to account for both C (methylated) or U (unmethylated),
referred to as methylation- insensitive primers [92]. The data
analysis process for bisulfite sequencing is straightforward and
can be done without extensive bioinformatics knowledge.
However, for intermediate methylation where the signal for
Sanger sequencing is a mix of C/T, methods for resolving the
peak height to derive the methylation status of the locus are not
yet optimal. Also, better normalization methods for noisy
sequences need further development.
5 Conclusions
Interest in epigenetics research has sharply increased with recent

advances in genome sequencing. The development of new tech-
niques for accurate measurement of DNA methylation has always
been an active area of research and we believe new methods or vari-
ants of existing methods will continue to emerge. It will be impor-
tant to develop simple and convenient DNA methylation measuring
techniques for marker testing and clinical use. There will be particu-
lar focus on building user-friendly tools that are able to integrate
different layers of epigenomic data. These developments are likely

to be directed towards sequencing-based techniques as the cost of
sequencing and library preparation is steadily decreasing. Further,
low cell number or single-cell DNA methylation analyses will become
an increasingly important focus in coming years as these techniques
have considerable implications in understanding epigenetic heteroge-
neity, especially in cancer cells.
6 Notes
Commands given here are indicative of how to perform a task,

alternatives exist for most of these operations.
1. Run FastQC on a raw data file (Read1.fastq.gz):
fastqc --outdir QC Read1.fastq.gz
which will write the output to a directory QC.The default is
to write to the current directory along with the source file.
2. Remove adaptor sequences (in a file contam.fa) and hard trim
reads to 85bp:
cleanadaptors -I contam.fa -z -l 85 -x 20
-t 3 \
-F Read1.fastq.gz -Z -o Read1_adtrim_l85.
fastq
where -z indicates the following file will be gzip compressed
and -Z uncompressed. Trimming is directed by -l 85 and -x 20
will omit any reads shorter than 20bp. Removal of inserted C
residues during ligation steps in library preparation are directed
by -t 3 which will trim back 3 residues before any adaptor
sequences. The uncompressed fastq file is suitable input for
mapping.
3. Build the Bismark index for bowtie, in a directory (GRCh37)
containing genome fasta files for each chromosome:
bismark_genome_preparation GRCh37
which will find all .fa or .fasta files in GRCh37 and create a
directory Bisulfite_genome containing the pre-converted
sequence index files. This process will take some time. The
command assumes that bowtie and/or bowtie2 is installed on
the users path.
4. Run Bismark to map a file of sequence reads (Read1_adtrim_
l85.fastq) to the converted genome:
bismark -n 1 GRCh37 Read1_adtrim_l85.fastq
which will create the file Read1adtrim_l85.fastq_bismark.bam
and a file Read1_adtrim_l85.fastq_bismark_mapping_report.
txt, giving statistics for the mapping. The command assumes
that the Samtools package has been installed on the system
with the executables on the users path in order for Bismark to
produce the output BAM file. Other outputs (native or SAM),

selected by command line options, do not require Samtools.
5. The formula below is used to calculate beta methylation values
in array-based platforms:
max ( M , 0 )
b = log 2
max ( M ,0 ) + max (U , 0 ) + 100
6. The formula below is used to calculate M values from array-

based methylation data and the relationship of beta values and
M values is also shown:
max ( M , 0 ) + 1
M = log 2
max (U , 0 ) + 1
b
M = log 2
1- b
7. Command for sorting a SAM file:

sort -k 3,3 -k 4,4n alignedRRBS.sam>aligne-
dRRBS.sam.sorted
8. Command for sorting a BAM file (sampleX_bismark.bam):
samtools sort sampleX_bismark.bam sampleX_
bismark_sort
which will produce the file sampleX_bismark_sort.bam.
9. Command for converting a SAM to BAM file (sampleX_bis-
mark.sam):
samtools view -bS sampleX_bismark.sam -o
sampleX_bismark.bam
which is identical to:
samtools view -bS sampleX_bismark.sam>sam-
pleX_bismark.bam
10. Command for converting a BAM to SAM file:
samtools view -h sampleX_bismark.bam -o sam-
pleX_bismark.sam
which is identical to:
samtools view -h >
sampleX_bismark.bamsam-
pleX_bismark.sam
Acknowledgments
AC and MRE would like to thank New Zealand Institute for

Cancer Research Trust, and IM would like to thank Gravida
(formerly NRCGD) for their support. We would like to apologize
to other research groups whose work we could not cite due to
context and space limitations.
References
1. Carrel L, Willard HF (2005) X-inactivation user-friendly alignment tool for reduced repre-
profile reveals extensive variability in X-linked sentation bisulfite sequencing. Bioinformatics
gene expression in females. Nature 28:430432
434:400404 15. Chatterjee A, Stockwell PA, Rodger EJ,
2. Rollins RA, Haghighi F, Edwards JR, Das R, Morison IM (2012) Comparison of alignment
Zhang MQ, Ju J, Bestor TH (2006) Large- software for genome-wide bisulphite sequence
scale structure of genomic methylation pat- data. Nucleic Acids Res 40, e79
terns. Genome Res 16:157163 16. Meissner A, Mikkelsen TS, Gu H, Wernig M,
3. Suzuki MM, Bird A (2008) DNA methylation Hanna J, Sivachenko A, Zhang X, Bernstein
landscapes: provocative insights from epig- BE, Nusbaum C, Jaffe DB, Gnirke A, Jaenisch
enomics. Nat Rev Genet 9:465476 R, Lander ES (2008) Genome-scale DNA
4. Igarashi J, Muroi S, Kawashima H, Wang X, methylation maps of pluripotent and differenti-
Shinojima Y, Kitamura E, Oinuma T, Nemoto ated cells. Nature 454:766770
N, Song F, Ghosh S, Held WA, Nagase H 17. Baranzini SE, Mudge J, van Velkinburgh JC,
(2008) Quantitative analysis of human tissue- Khankhanian P, Khrebtukova I, Miller NA,
specific differences in methylation. Biochem Zhang L, Farmer AD, Bell CJ, Kim RW, May
Biophys Res Commun 376:658664 GD, Woodward JE, Caillier SJ, McElroy JP,
5. Chatterjee A, Morison IM (2011) Monozygotic Gomez R, Pando MJ, Clendenen LE, Ganusova
twins: genes are not the destiny? Bioinformation EE, Schilkey FD, Ramaraj T, Khan OA,
7:369370 Huntley JJ, Luo S, Kwok PY, Wu TD, Schroth
6. Chatterjee A, Eccles MR (2015) DNA meth- GP, Oksenberg JR, Hauser SL, Kingsmore SF
ylation and epigenomics: new technologies and (2010) Genome, epigenome and RNA
emerging concepts. Genome Biol 16:103 sequences of monozygotic twins discordant for
multiple sclerosis. Nature 464:13511356
7. Chatterjee A (2012) Conference scene: epi-
genetic regulation: from mechanism to inter- 18. Bock C, Kiskinis E, Verstappen G, Gu H,
vention. Epigenomics 4:487490 Boulting G, Smith ZD, Ziller M, Croft GF,
Amoroso MW, Oakley DH, Gnirke A, Eggan
8. Laird PW (2010) Principles and challenges of K, Meissner A (2011) Reference maps of
genome-wide DNA methylation analysis. Nat human ES and iPS cell variation enable high-
Rev Genet 11:191203 throughput characterization of pluripotent cell
9. Plongthongkum N, Diep DH, Zhang K (2014) lines. Cell 144:439452
Advances in the profiling of DNA modifica- 19. Gertz J, Varley KE, Reddy TE, Bowling KM,
tions: cytosine methylation and beyond. Nat Pauli F, Parker SL, Kucera KS, Willard HF,
Rev Genet 15:647661 Myers RM (2011) Analysis of DNA methyla-
10. Frommer M, McDonald LE, Millar DS, Collis tion in a three-generation family reveals wide-
CM, Watt F, Grigg GW, Molloy PL, Paul CL spread genetic influence on epigenetic
(1992) A genomic sequencing protocol that regulation. PLoS Genet 7, e1002228
yields a positive display of 5-methylcytosine 20. Gu H, Bock C, Mikkelsen TS, Jager N, Smith
residues in individual DNA strands. Proc Natl ZD, Tomazou E, Gnirke A, Lander ES, Meissner
Acad Sci U S A 89:18271831 A (2010) Genome-scale DNA methylation map-
11. Rodger EJ, Chatterjee A, Morison IM (2014) ping of clinical samples at single-nucleotide reso-
5-hydroxymethylcytosine: a potential therapeu- lution. Nat Methods 7:133136
tic target in cancer. Epigenomics 6:503514 21. Smallwood SA, Tomizawa S, Krueger F, Ruf
12. Mukhopadhyay R, Yu W, Whitehead J, Xu J, N, Carli N, Segonds-Pichon A, Sato S, Hata K,
Lezcano M, Pack S, Kanduri C, Kanduri M, Andrews SR, Kelsey G (2011) Dynamic CpG
Ginjala V, Vostrov A, Quitschke W, Chernukhin island methylation landscape in oocytes and
I, Klenova E, Lobanenkov V, Ohlsson R preimplantation embryos. Nat Genet 43:
(2004) The binding sites for the chromatin 811814
insulator protein CTCF map to DNA 22. Steine EJ, Ehrich M, Bell GW, Raj A, Reddy S,
methylation- free domains genome-wide. van Oudenaarden A, Jaenisch R, Linhart HG
Genome Res 14:15941602 (2011) Genes methylated by DNA methyl-
13. Miura F, Ito T (2015) Highly sensitive tar- transferase 3b are similar in mouse intestine
geted methylome sequencing by post-bisulfite and human colon cancer. JClin Invest
adaptor tagging. DNA Res 22:1318 121:17481752
14. Xi Y, Bock C, Muller F, Sun D, Meissner A, Li 23. Hartung T, Zhang L, Kanwar R, Khrebtukova I,
W (2012) RRBSMAP: a fast, accurate and Reinhardt M, Wang C, Therneau TM, Banck
MS, Schroth GP, Beutler AS (2012) 35. Krueger F, Andrews SR (2011) Bismark: a
Diametrically opposite methylome- flexible aligner and methylation caller for
transcriptome relationships in high- and low- seq
bisulfite- applications. Bioinformatics
CpG promoter genes in postmitotic neural rat 27:15711572
tissue. Epigenetics 7:421428 36. Chen PY, Cokus SJ, Pellegrini M (2010) BS
24. Chatterjee A, Ozaki Y, Stockwell PA, Horsfield seeker: precise mapping for bisulfite sequenc-
JA, Morison IM, Nakagawa S (2013) Mapping ing. BMC Bioinformatics 11:203
the zebrafish brain methylome using reduced 37. Pedersen B, Hsieh TF, Ibarra C, Fischer RL
representation bisulfite sequencing. Epigenetics (2011) MethylCoder: software pipeline for
8:979989 bisulfite-treated sequences. Bioinformatics
25. Chatterjee A, Stockwell PA, Horsfield JA, 27:24352436
Morison IM, Nakagawa S (2014) Base- 38. Langmead B, Trapnell C, Pop M, Salzberg SL
resolution DNA methylation landscape of (2009) Ultrafast and memory-efficient align-
zebrafish brain and liver. Genomics Data ment of short DNA sequences to the human
2:342344 genome. Genome Biol 10:R25
26. Boyle P, Clement K, Gu H, Smith ZD, Ziller 39. Langmead B, Salzberg SL (2012) Fast gapped-
M, Fostel JL, Holmes L, Meldrim J, Kelley F, read alignment with Bowtie 2. Nat Methods
Gnirke A, Meissner A (2012) Gel-free multi- 9:357359
plexed reduced representation bisulfite 40. Li H, Durbin R (2009) Fast and accurate short
sequencing for large-scale DNA methylation read alignment with Burrows-Wheeler trans-
profiling. Genome Biol 13:R92 form. Bioinformatics 25:17541760
27. Chatterjee A, Rodger EJ, Stockwell PA, Weeks 41. Harris EY, Ponts N, Le Roch KG, Lonardi S
RJ, Morison IM (2012) Technical consider- (2012) BRAT-BW: efficient and accurate map-
ations for reduced representation bisulfite ping of bisulfite-treated reads. Bioinformatics
sequencing with multiplexed libraries. 28:17951796
JBiomed Biotech 2012:741542
42. Kreck B, Marnellos G, Richter J, Krueger F,
28. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle Siebert R, Franke A (2012) B-SOLANA: an
B, Le JM, Delano D, Zhang L, Schroth GP, approach for the analysis of two-base encoding
Gunderson KL, Fan JB, Shen R (2011) High bisulfite sequencing data. Bioinformatics
density DNA methylation array with single 28:428429
CpG site resolution. Genomics 98:288295
43. Campagna D, Telatin A, Forcato C, Vitulo N,
29. Ziller MJ, Gu H, Mller F, Donaghey J, Tsai Valle G (2013) PASS-bis: a bisulfite aligner
LT, Kohlbacher O, De Jager PL, Rosen ED, suitable for whole methylome analysis of illu-
Bennett DA, Bernstein BE, Gnirke A, Meissner mina and SOLiD reads. Bioinformatics
A (2013) Charting a dynamic DNA methyla- 29:268270
tion landscape of the human genome. Nature
500:477481 4 4. Lim JQ, Tennakoon C, Li G, Wong E, Ruan Y,
Wei CL, Sung WK (2012) BatMeth: improved
30. Bock C, Tomazou EM, Brinkman AB, Muller mapper for bisulfite sequencing reads on DNA
F, Simmer F, Gu H, Jager N, Gnirke A, methylation. Genome Biol 13:R82
Stunnenberg HG, Meissner A (2010)
Quantitative comparison of genome-wide 4 5. Morris TJ, Beck S (2015) Analysis pipelines
DNA methylation mapping technologies. Nat and packages for Infinium
Biotech 28:11061114 HumanMethylation450 BeadChip (450k)
data. Methods 72:38
31. Matarese F, Carrillo-de Santa Pau E,
Stunnenberg HG (2011) 4 6. Dedeurwaerder S, Defrance M, Bizet M,
5-Hydroxymethylcytosine: a new kid on the Calonne E, Bontempi G, Fuks F (2014) A
epigenetic block? Mol Syst Biol 7:562 comprehensive overview of Infinium
HumanMethylation450 data processing. Brief
32. Stockwell PA, Chatterjee A, Rodger EJ, Bioinform 15:929941
Morison IM (2014) DMAP: differential meth-
ylation analysis package for RRBS and WGBS 4 7. Chen YA, Lemire M, Choufani S, Butcher DT,
data. Bioinformatics 30:18141822 Grafodatskaya D, Zanke BW, Gallinger S,
Hudson TJ, Weksberg R (2013) Discovery of
33. Xi Y, Li W (2009) BSMAP: whole genome cross-reactive probes and polymorphic CpGs in
bisulfite sequence MAPping program. BMC the Illumina Infinium HumanMethylation450
Bioinformatics 10:232 microarray. Epigenetics 2013(8):203209
34. Harris EY, Ponts N, Levchuk A, Roch KL, 48. Price EM, Cotton AM, Lam LL, Farre P,
Lonardi S (2010) BRAT: bisulfite-treated reads Emberly E, Brown CJ, Robinson WP, Kobor
analysis tool. Bioinformatics 26:572573 MS (2013) Additional annotation enhances
potential for biologically-relevant analysis of studies using the illumina infinium platform.
the Illumina Infinium HumanMethylation450 BMC Bioinformatics 13:59
BeadChip array. Epigenetics Chromatin 6:4 60. Zou J, Lippert C, Heckerman D, Aryee M,
49. Zhang X, Mu W, Zhang W (2012) On the Listgarten J(2014) Epigenome-wide associa-
analysis of the illumina 450k array data: probes tion studies without the need for cell-type
ambiguously mapped to the human genome. composition. Nat Methods 11:309311
Front Genet 3:73 61. Houseman EA, Molitor J, Marsit CJ (2014)
50. Xu Z, Niu L, Li L, Taylor JA (2015) ENmix: a Reference-free cell mixture adjustments in
novel background correction method for analysis of DNA methylation data.
Illumina HumanMethylation450 BeadChip. Bioinformatics 30:14311439
Nucleic Acids Res 44, e20 62. Horvath S (2013) DNA methylation age of
51. Marabita F, Almgren M, Lindholm ME,
human tissues and cell types. Genome Biol
Ruhrmann S, Fagerstrom-Billai F, Jagodic M, 14:R115
Sundberg CJ, Ekstrom TJ, Teschendorff AE, 63. Hannum G, Guinney J, Zhao L, Zhang L,
Tegner J, Gomez-Cabrero D (2013) An eval- Hughes G, Sadda S, Klotzle B, Bibikova M,
uation of analysis pipelines for DNA methyla- Fan JB, Gao Y, Deconde R, Chen M, Rajapakse
tion profiling using the Illumina I, Friend S, Ideker T, Zhang K (2013)
HumanMethylation450 BeadChip platform. Genome-wide methylation profiles reveal
Epigenetics 8:333346 quantitative views of human aging rates. Mol
52. Triche TJ Jr, Weisenberger DJ, Van Den Berg Cell 49:359367
D, Laird PW, Siegmund KD (2013) Low-level 64. Stewart SK, Morris TJ, Guilhamon P, Bulstrode
processing of illumina infinium DNA methyla- H, Bachman M, Balasubramanian S, Beck S
tion BeadArrays. Nucleic Acids Res 41, e90 (2015) oxBS-450K: a method for analysing
53. Davis S, Du P, Bilke S, Triche T Jr, Bootwalla hydroxymethylation using 450K BeadChips.
M (2015) Methylumi: handle illumina meth- Methods 72:915
ylation data. R Package Version 2160 2015 65. Akalin A, Kormaksson M, Li S, Garrett-
54. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd- Bakelman FE, Figueroa ME, Melnick A, Mason
Acosta C, Feinberg AP, Hansen KD, Irizarry CE (2012) methylKit: a comprehensive R
RA (2014) Minfi: a flexible and comprehensive package for the analysis of genome-wide DNA
bioconductor package for the analysis of infin- methylation profiles. Genome Biol 13:R87
ium DNA methylation microarrays. 66. Wang HQ, Tuominen LK, Tsai CJ (2011)
Bioinformatics 30:13631369 SLIM: a sliding linear model for estimating the
55. Teschendorff AE, Marabita F, Lechner M, proportion of true null hypotheses in datasets
Bartlett T, Tegner J, Gomez-Cabrero D, Beck with dependence structures. Bioinformatics
S (2013) A beta-mixture quantile normaliza- 27:225231
tion method for correcting probe design bias in 67. Ehrlich M, Lacey M (2013) DNA methylation
illumina infinium 450k DNA methylation and differentiation: silencing, upregulation and
data. Bioinformatics 29:189196 modulation of gene expression. Epigenomics
56. Pidsley R, Y Wong CC, Volta M, Lunnon K, 5:553568
Mill J, Schalkwyk LC (2013) A data-driven 68. Bock C, Beerman I, Lien WH, Smith ZD, Gu
approach to preprocessing illumina 450K H, Boyle P, Gnirke A, Fuchs E, Rossi DJ,
methylation array data. BMC Genomics Meissner A (2012) DNA methylation dynam-
14:293 ics during invivo differentiation of blood and
57. Wang D, Yan L, Hu Q, Sucheston LE, Higgins skin stem cells. Mol Cell 47:633647
MJ, Ambrosone CB, Johnson CS, Smiraglia 69. Li Y, Zhu J, Tian G, Li N, Li Q, Ye M, Zheng
DJ, Liu S (2012) IMA: an R package for high- H, Yu J, Wu H, Sun J, Zhang H, Chen Q, Luo
throughput analysis of Illuminas 450K infin- R, Chen M, He Y, Jin X, Zhang Q, Yu C, Zhou
ium methylation data. Bioinformatics 28: G, Sun J, Huang Y, Zheng H, Cao H, Zhou X,
729730 Guo S, Hu X, Li X, Kristiansen K, Bolund L,
58. Du P, Zhang X, Huang CC, Jafari N, Kibbe Xu J, Wang W, Yang H, Wang J, Li R, Beck S,
WA, Hou L, Lin SM (2010) Comparison of Wang J, Zhang X (2010) The DNA methy-
beta-value and M-value methods for quantify- lome of human peripheral blood mononuclear
ing methylation levels by microarray analysis. cells. PLoS Biol 8, e1000533
BMC Bioinformatics 11:587 70. Hansen KD, Langmead B, Irizarry RA (2012)
59. Zhuang J, Widschwendter M, Teschendorff BSmooth: from whole genome bisulfite
AE (2012) A comparison of feature selection sequencing reads to differentially methylated
and classification methods in DNA methylation regions. Genome Biol 13:R83
71. Hebestreit K, Dugas M, Klein HU (2013) ing methylation of specific cytosines within the
Detection of significantly differentially methyl- genome. Nucleic Acids Res 32, e168
ated regions in targeted bisulfite sequencing 84. Eads CA, Danenberg KD, Kawakami K, Saltz
data. Bioinformatics 29:16471653 LB, Blake C, Shibata D, Danenberg PV, Laird
72. Chatterjee A, Stockwell PA, Rodger EJ, Duncan PW (2000) MethyLight: a high-throughput
EJ, Parry MF, Weeks RJ, Morison IM (2015) assay to measure DNA methylation. Nucleic
Genome-wide DNA methylation map of human Acids Res 28, E32
neutrophils reveals widespread inter- individual 85. Hernandez HG, Tse MY, Pang SC, Arboleda
epigenetic variation. Sci Rep 5:17328 H, Forero DA (2013) Optimizing methodolo-
73. Robinson JT, Thorvaldsdottir H, Winckler W, gies for PCR-based DNA methylation analysis.
Guttman M, Lander ES, Getz G, Mesirov JP Biotechniques 55:181197
(2011) Integrative genomics viewer. Nat 86. Coolen MW, Statham AL, Gardiner-Garden
Biotechnol 29:2426 M, Clark SJ (2007) Genomic profiling of CpG
74. Halachev K, Bast H, Albrecht F, Lengauer T, methylation and allelic specificity using quanti-
Bock C (2012) EpiExplorer: live exploration tative high-throughput mass spectrometry:
and global analysis of large epigenomic datas- critical evaluation and improvements. Nucleic
ets. Genome Biol 13:R96 Acids Res 35, e119
75. Li S, Garrett-Bakelman F, Perl AE, Luger SM, 87. Christensen BC, Kelsey KT, Zheng S,
Zhang C, To BL, Lewis ID, Brown AL, Houseman EA, Marsit CJ, Wrensch MR,
DAndrea RJ, Ross ME etal (2014) Dynamic Wiemels JL, Nelson HH, Karagas MR, Kushi
evolution of clonal epialleles revealed by meth- LH, Kwan ML, Wiencke JK (2010) Breast can-
clone. Genome Biol 15:472 cer DNA methylation profiles are associated
76. Clark SJ, Harrison J, Paul CL, Frommer M with tumor size and alcohol and folate intake.
(1994) High sensitivity mapping of methylated PLoS Genet 6, e1001043
cytosines. Nucleic Acids Res 22:29902997 88. Breitling LP, Yang R, Korn B, Burwinkel B,
77. Tost J, Gut IG (2007) Analysis of gene-specific Brenner H (2011) Tobacco-smoking-related
DNA methylation patterns by pyrosequencing differential DNA methylation: 27K discovery
technology. Methods Mol Biol 373:89102 and replication. Am JHum Genet
78. Masser DR, Berg AS, Freeman WM (2013) 88:450457
Focused, high accuracy 5-methylcytosine 89. Figueroa ME, Lugthart S, Li Y, Erpelinck-
quantitation with base resolution by benchtop Verschueren C, Deng X, Christos PJ, Schifano
next-generation sequencing. Epigenetics E, Booth J, van Putten W, Skrabanek L,
Chromatin 6:33 Campagne F, Mazumdar M, Greally JM, Valk
79. Diep D, Plongthongkum N, Gore A, Fung PJ, Lwenberg B, Delwel R, Melnick A (2010)
HL, Shoemaker R, Zhang K (2012) Library- DNA methylation signatures identify biologi-
free methylation sequencing with bisulfite cally distinct subtypes in acute myeloid leuke-
padlock probes. Nat Methods 9:270272 mia. Cancer Cell 17:1327
80. Komori HK, LaMere SA, Torkamani A, Hart 9 0. Farthing CR, Ficz G, Ng RK, Chan CF,
GT, Kotsopoulos S, Warner J, Samuels ML, Andrews S, Dean W, Hemberger M, Reik W
Olson J, Head SR, Ordoukhanian P, Lee PL, (2008) Global mapping of DNA methylation
Link DR, Salomon DR (2011) Application of in mouse promoters reveals epigenetic repro-
microdroplet PCR for large-scale targeted bisul- gramming of pluripotency genes. PLoS Genet
fite sequencing. Genome Res 21:17381745 4, e1000116
81. Herman JG, Graff JR, Myohanen S, Nelkin 9 1. Liang P, Song F, Ghosh S, Morien E, Qin M,
BD, Baylin SB (1996) Methylation-specific Mahmood S, Fujiwara K, Igarashi J, Nagase H,
PCR: a novel PCR assay for methylation status Held WA (2011) Genome-wide survey reveals
of CpG islands. Proc Natl Acad Sci U S A dynamic widespread tissue-specific changes in
93:98219826 DNA methylation during development. BMC
Genomics 12:231
82. Kristensen LS, Mikeska T, Krypuy M, Dobrovic
A (2008) Sensitive melting analysis after real 9 2. Fuso A, Ferraguti G, Scarpa S, Ferrer I,
time-methylation specific PCR (SMART- Lucarelli M (2015) Disclosing bias in bisulfite
MSP): high-throughput and probe-free quan- assay: MethPrimers underestimate high DNA
titative DNA methylation detection. Nucleic methylation. PLoS One 10, e0118318
Acids Res 36, e42 9 3. Park Y, Figueroa ME, Rozek LS, Sartor MA
83. Thomassin H, Kress C, Grange T (2004) (2014) MethylSig: a whole genome DNA
MethylQuant: a sensitive method for quantify- methylation analysis pipeline. Bioinformatics
30:24142422
94. Dolzhenko E, Smith AD (2014) Using beta- (2014) ChAMP: 450k chip analysis methyla-
binomial regression for high-precision differ- tion pipeline. Bioinformatics 30:428430
ential methylation analysis in multifactor 97. Peters TJ, Buckley MJ, Statham AL, Pidsley R,
whole-genome bisulfite sequencing experi- Samaras K, V Lord R, Clark SJ, Molloy PL
ments. BMC Bioinformatics 15:215 (2015) De novo identification of differentially
95. Assenov Y, Muller F, Lutsik P, Walter J, methylated regions in the human genome.
Lengauer T, Bock C (2014) Comprehensive Epigenetics Chromatin 8:6
analysis of DNA methylation data with 98. Phipson B, Maksimovic J, Oshlack A (2015)
RnBeads. Nat Methods 11:11381140 missMethyl: an R package for analyzing data
96. Morris TJ, Butcher LM, Feber A, Teschendorff from Illuminas HumanMethylation450 plat-
AE, Chakravarthy AR, Wojdacz TK, Beck S form. Bioinformatics 32:286288
Chapter 16
Generating Multiple Base-Resolution DNA Methylomes

Using Reduced Representation Bisulfite Sequencing
Aniruddha Chatterjee, Euan J. Rodger, Peter A. Stockwell,
Gwenn Le Me, and Ian M. Morison
Abstract
Reduced representation bisulfite sequencing (RRBS) is an effective technique for profiling genome-wide
DNA methylation patterns in eukaryotes. RRBS couples size selection, bisulfite conversion, and second-
generation sequencing to enrich for CpG-dense regions of the genome. The progressive improvement of
second-generation sequencing technologies and reduction in cost provided an opportunity to examine the
DNA methylation patterns of multiple genomes. Here, we describe a protocol for sequencing multiple
RRBS libraries in a single sequencing reaction to generate base-resolution methylomes. Furthermore, we
provide a brief guideline for base-calling and data analysis of multiplexed RRBS libraries. These strategies
will be useful to perform large-scale, genome-wide DNA methylation analysis.
Key words Epigenetics, DNA methylation, Reduced representation bisulfite sequencing, Multiplexed,
Second-generation sequencing, CpG island, DMAP
1 Introduction
DNA methylation, which is prevalent in all vertebrates, is a funda-

mental mechanism for regulating gene function [1, 2]. DNA
methylation has a crucial role in gene silencing, tissue differentia-
tion, genomic imprinting, X chromosome inactivation, phenotypic
plasticity, and disease susceptibility [39]. DNA methylation is
considered to be an excellent potential marker for diseases includ-
ing oral cancers [10]. In mammals, DNA methylation occurs
almost exclusively at CpG dinucleotides and approximately 70 %
of the mammalian genes are associated with CpG rich regions (i.e.,
CpG islands) [11]. It is now well documented that the methylation
status of promoter-associated CpG islands can govern gene expres-
sion by altering the binding of transcription activators or by chang-
ing the chromatin conformation [5]. Therefore, profiling the
methylation status of CpG islands is an important aspect of under-
standing the role of epigenetic events in mammalian genomes.
279
Bisulfite treatment-based methods remain a de facto gold

standard for profiling CpG methylation status [12]. Whole
genome bisulfite sequencing (WGBS) is considered to be the
most effective technique as it potentially allows investigation of
every CpG site in the genome. However, WGBS comes with a
high cost (approximately $5500 USD for tenfold average cover-
age of the genome) and a large volume of raw data that can be
challenging to analyze. Interestingly, a recent analysis of 42 data-
sets of different human cell types concluded that WGBS is gen-
erally inefficient as >70 % sequenced reads did not provide useful
information about CpG methylation [13]. Compared to WGBS,
reduced representation bisulfite sequencing (RRBS) provides a
suitable alternative for generating base-resolution DNA methy-
lomes at a reduced cost.
In RRBS, genomic DNA is first digested with a restriction
enzyme that contains a CpG dinucleotide in its recognition motif
and is insensitive to the methylation status of the CpG sites in the
genome [14]. By size-selecting a small fraction of the digested
fragments (each fragment contains at least one CpG site), RRBS
reduces uninformative sequences and the requirement of extensive
sequencing; for example, human, mouse, and zebrafish MspI RRBS
libraries represent ~2.3, 1.4, and 2.2 % of the genomes respectively
with a size selection of 40220 bp [15]. The method enriches for
CpG sites (5.7-fold) [16], CpG islands (~30-fold) [17], gene pro-
moters and enhancer elements [18]. RRBS is reproducible and has
been widely used by many groups world-wide [1924]. The tech-
nique has so far been applied to generate genome-wide methyla-
tion map for mammals, including human [16, 22], mice [17] and
rats [25], and other vertebrates such as zebrafish [15, 26].
The output from second-generation sequencers is progressively
increasing. For example, the Illumina GAII sequencers (in 2009)
were able to produce 2030 million single-ended, bisulfite con-
verted sequence reads per lane, whereas, the Illumina HiSeq 2500
sequencers (in 2013) can produce 180 million RRBS reads per lane.
This advance allows multiple reduced representation genomes to be
sequenced in a single flow cell, while maintaining sufficient reads
and CpG coverage. We have previously demonstrated the strategies
and modifications for sequencing and base-calling of multiplexed
libraries on an RRBS background [27]. Here we describe an updated
and extended protocol for multiplexed RRBS method. In addition,
we provide a brief guideline for base-calling, quality assessment,
mapping, and downstream analysis of RRBS data.
2 Materials
1. QIAamp DNA Mini Kit (QIAGEN GmbH, Germany).

2. Phosphate buffered saline, PBS: 137 mM NaCl, 2.7 mM KCl,
10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4.
Generating Multiple Methylomes Using RRBS 281
3. Absolute ethanol, analytical grade.

4. TE buffer: 10 mM TrisHCl, 1 mM EDTA, pH 8.0.
5. Qubit fluorometer (Life Technologies).
6. Nanodrop ND-1000 spectrophotometer (Nanodrop Products,
Wilmington, DE, USA).
7. MspI: 20 U/L, with compatible restriction buffer.
8. Distilled, deionized purified water (e.g., Milli-Q).
9. PCR thermocycler.
10. QIAquick and MinElute PCR purification kits (QIAGEN).
11. TruSeq Nano DNA sample preparation kit (Illumina).
12. XC loading dye 6: 0.25 % (w/v) xylene cyanol, 40 % (w/v)
sucrose.
13. NuSieve GTG agarose (Lonza).
14. TAE buffer 0.5: 20 mM Tris-acetate, 0.5 mM EDTA with
0.0024 % ethidium bromide.
15. Mupid-exU electrophoresis tank (Takara, Japan).
16. 25-bp DNA ladder: 0.1 g/L.
17. UV gel imaging system.
18. Scalpel blades.
19. QIAquick gel extraction kit (QIAGEN).
20. 100 % Isopropanol, analytical grade.
21. EZ DNA methylation kit (Zymo Research).
22. PfuTurbo Cx Hotstart DNA polymerase (Agilent
Technologies).
23. PCR-grade dNTPs: each at 2.5 mM.
24. Precast TBE 420 % gradient gels and electrophoresis tank
(Criterion, Bio-Rad).
25. TBE buffer: 45 mM Tris-borate, 1 mM EDTA.
26. SYBR green nucleic acid gel stain (Life Technologies).
27. 2100 Bioanalyzer and high sensitivity DNA kit (Agilent
Technologies).
28. HiSeq2000/2500 sequencing system (Illumina).
3 Methods
This protocol describes the preparation of multiplexed DNA libraries

and provides a brief guideline of key steps involved in base-calling,
downstream processing, and analysis of RRBS data. Precautionary
measures are required to avoid contamination of samples: bench
surfaces, pipettes, and equipment should be thoroughly cleaned
regularly with decontamination solution; use sterile low
DNA-binding plasticware, disposable gloves, and sterile aerosol

barrier pipette tips; change pipette tips between samples; use
Milli-Q purified water and sterilized glassware to make buffers.
Carry out all procedures at room temperature unless otherwise
specified. The protocol described here takes approximately 55 h
(i.e., 3.5 days in laboratory setting) from genomic DNA extraction
to the purified RRBS libraries.
3.1 DNA Purification 1. Use the QIAamp DNA mini kit according to the manufacturers
instructions for specific cell or tissue type, incubating over-
night (~16 h) in a proteinase-containing solution at 56 C on
a heating block. Elute the DNA using 100 L of TE buffer
(see Note 1).
2. Quantify DNA using a Qubit fluorometer and estimate DNA
quality using a Nanodrop spectrophotometer, according to the
manufacturers instructions (see Note 2).
3.2 MspI Digestion 1. To a PCR tube, add the following for each sample and mix:
(a) 2.0 g of genomic DNA (see Note 3).
(b) 160 U of MspI restriction endonuclease.
(c) 4 L of 10 NEB Buffer 4.
(d) Milli-Q H2O to 40 L final volume.
2. Incubate overnight (~16 h) at 37 C in the thermocycler.
3. Use the QIAquick PCR purification kit according to the man-
ufacturers instructions, eluting in 60 L TE buffer.
3.3 End Repair 1. In a PCR tube, add 40 L TruSeq End Repair mix 2 (ERP2)
from the TruSeq DNA sample preparation kit to the digested
genomic DNA (from Subheading 3.2) and mix.
2. Incubate for 30 min at 30 C in the PCR thermocycler.
3. Use the MinElute PCR purification kit according to the manu-
facturers instructions, eluting in 17.5 L TE buffer.
3.4 A-Tailing 1. In a PCR tube, add 12.5 L TruSeq A-Tailing mix (ATL)
TruSeq DNA sample preparation kit to the end-repaired
genomic DNA (from Subheading 3.3) and mix.
2. Incubate in thermocycler as follows:
(a) 30 min at 37 C.
(b) 5 min at 70 C.
(c) 5 min at 4 C.
3. Immediately proceed to the next step.
3.5 Adaptor Ligation 1. In the same PCR tube (from Subheading 3.4), add the following
and mix:
(a) 2.5 L TE buffer.
(b) 2.5 L TruSeq Adaptor Index (choosing a single barcoded

adaptor from the 24 available adaptors, see Note 4).
(c) 2.5 L TruSeq DNA Ligase mix 2 (LIG2).
2. Incubate for 10 min at 30 C in a thermocycler.
3. Immediately after incubation, add 5 L of Stop Ligase mix
(STL) and mix.
facturers instructions, eluting in 18 L TE buffer.
5. Add 2 L of XC loading dye.
3.6 Agarose Gel Size 1. Prepare 3 % (w/v) NuSieve GTG agarose gel in 0.5 TAE buf-
Selection fer with 0.0024 % ethidium bromide according to manufac-
turers instructions (see Note 5). Pour molten gel into a gel
tray with a comb and leave to set for 3060 min.
2. Carefully remove the comb and transfer the tray with set gel
into an electrophoresis tank. Pour fresh 0.5 TAE running
buffer into the tank so that the gel is fully submerged (~1 cm).
Load 6 L of 25 bp DNA ladder into the first lane and 20 L
of the DNA sample (from Subheading 3.5) into the third lane
(with a lane between multiple samples). Run the gel at 50 V for
~90 min (see Note 6).
3. Immediately after the run, carefully transfer the gel onto a clean
surface to excise gel bands. With a clean scalpel blade, make a
straight vertical cut to slice off the DNA ladder lane and transfer
onto a UV gel imager alongside a plastic ruler. Visualize the
DNA ladder lane under UV using the imager and determine the
position (in mm) of the 160 and 340 bp fragments (adaptor
modified sizes that correspond to 40220 bp fragments).
4. With a new blade, make two straight vertical cuts to isolate the
sample lane (should be performed in a UV light devoid environ-
ment). Align the ruler next to the excised sample lane and make
a straight horizontal cut just above the calculated 340 bp mark
and another straight horizontal cut just below the calculated
160 bp mark. Using the edge of the same blade, carefully trans-
fer the excised fragment into a 2 mL plastic centrifuge tube.
5. Using an empty 2 mL tube to zero the electronic scales, deter-
mine the weight of the gel fragment. Use the QIAquick gel
extraction kit according to the manufacturers instructions with-
out heating the gel, eluting in 40 L TE buffer (see Note 7).
3.7 Bisulfite 1. Prepare EZ DNA methylation kit CT conversion reagent

Conversion according to the manufacturers instructions, mixing thor-
oughly until fully dissolved.
2. Add 10 L of M-Dilution Buffer to the 40 L DNA sample
(from Subheading 3.6) and mix. Incubate for 15 min at 37 C
in thermocycler.
3. Add 100 L of the prepared CT conversion reagent to the

sample and mix. Incubate for 18 h at 50 C on thermocycler
(see Note 8).
4. After incubation, place tube on ice for 10 min and proceed
with the EZ DNA methylation kit according to the manufac-
turers instructions to purify the bisulfite-converted DNA,
eluting in 20 L of TE buffer.
3.8 Semi- 1. In a PCR tube, add the following and mix to make a total vol-
Quantitative PCR ume of 25 L and mix well (see Note 9):
(a) 12.3 L of deionized (Milli-Q) water.
(b) 2.5 L of 10 PfuTurbo Cx reaction buffer.
(c) 3 L of 2.5 mM dNTP stock.
(d) 3 L of TruSeq PCR Primer Cocktail.
(e) 3 L of DNA sample (from Subheading 3.7).
(f) 1.2 L of PfuTurbo Cx Hotstart DNA Polymerase.
2. Divide the above mix into two PCR tubes (tube A and tube B,
12 L each) and run on separate PCR thermocyclers using the
following reaction conditions (tube A, n = 15; tube B, n = 20,
see Note 10):
(a) 95 C 2 min.
(b) 95 C 30 s.
(c) 65 C 30 s.
(d) 72 C 45 s.
(e) Return to step 2 for n1 cycles.
(f) 72 C 7 min.
(g) 4 C hold.
3. Following the PCR, add 2 L of XC loading dye to each tube.
Set up a 420 % Criterion precast TBE gradient gel in electro-
phoresis tank according to the manufacturers instructions.
Load 6 L of 25-bp DNA ladder into the first lane and 12 L
of each PCR product into subsequent lanes. Run the gel at
100 V for 100 min.
4. Following the run, add 10 L of SYBR Green stain to 100 mL
of TBE buffer into a light-proof container. Carefully remove
TBE gel from the plastic cassette and immerse in the SYBR
Green mixture and incubate at room temperature with gentle
rocking for 30 min. Visualize the gel using the UV gel imager.
5. To generate enough DNA for sequencing, the library is ampli-
fied with the lowest possible number of cycles. Excessive ampli-
fication of the libraries increases the prevalence of PCR errors
and increases the amplification of short fragments leading to
skewed CpG coverage after sequencing [18]. In addition we
Fig. 1 Semi-quantitative PCR of size-selected RRBS libraries with different cycle numbers. A single 160340 bp
size-selected RRBS library was amplified with 12, 14, 16, 18, 20, and 22 cycles of PCR to determine the optimal
number of cycles for large-scale amplification. PCR products were visualized on a 420 % Criterion gradient
polyacrylamide TBE gel stained with SYBR green nucleic acid gel stain alongside a 25 bp DNA ladder. The shift
of the size-selected library towards high molecular products with increasing cycle numbers can be observed.
The band at ~125 bp in all libraries was possibly due to adaptoradaptor dimerization
have observed that excessive amplification often leads to a shift

of the size selected library to a relatively higher molecular
weight bands, possibly due to concatenation of the fragments
(Fig. 1).
6. Assess the gel and determine the optimal cycle number carefully
for large-scale PCR amplification of libraries (see Note 10).
3.9 Large-Scale PCR 1. In a PCR tube, add the following and mix (125 L total
Amplification volume):
and Second (a) 61.5 L of Milli-Q water.
Size-Selection
(b) 12.5 L of 10 PfuTurbo Cx reaction buffer.
(c) 15 L of 2.5 mM dNTP stock.
(d) 15 L of TruSeq PCR Primer Cocktail.
(e) 15 L of DNA sample (from Subheading 3.7).
(f) 6 L of PfuTurbo Cx Hotstart DNA Polymerase.
2. Run the samples on a thermocycler using the reaction condi-
tions used previously and the estimated optimal cycle number
(in Subheading 3.8).
facturers instructions to purify large-scale amplified libraries,
eluting in 18 L TE buffer.
4. A second round of gel size selection step is performed to

remove primer contamination from the amplified libraries.
Add 2 L of XC loading dye and run on a 3 % NuSieve agarose
gel as before (from Subheading 3.6). Similarly, excise the band
between 160 and 340 bp. Use the MinElute gel extraction kit
according to the manufacturers instructions, eluting in 20 L
TE buffer.
3.10 Quantification, 1. Quantify 2 L of the final DNA library (from Subheading 3.9)
Quality Assessment, using the Qubit fluorometer according to the manufacturers
and Preparation instructions.
for Sequencing 2. For quality assessment, run 1 L of the final DNA library (from
Subheading 3.9) on the 2100 Bioanalyzer using the high sen-
sitivity DNA kit according to the manufacturers instructions
(see Note 11).
3. Dilute the library to a 10 nM stock concentration and freeze at
80 C (see Subheading 3.11).
3.11 Cluster For multiplexed sequencing runs, multiple RRBS libraries (from
Generation Subheading 3.10) are pooled in equimolar ratios for the cluster
for Multiplexed RRBS generation step. Each sample is made into a 2 nM working concen-
tration and then pooled in equal volumes. For five RRBS libraries,
5 L of each library are pooled, making a total volume of 25 L.
10 L from the pooled libraries are then denatured with NaOH
and diluted to a concentration of 8 pM. 120 L of the final diluted
sample is used in the Illumina cBot machine for the cluster genera-
tion step.
3.12 Base-Calling
of Multiplexed RRBS In Illumina HiSeq instruments, by default, base-calling of the
Libraries libraries is performed using the Illumina Real Time Analyzer (RTA)
software which runs on the HiSeq instrument. However, the non-
random base composition at the start of RRBS fragments can con-
found the scheme of RTA base-calling in some multiplexed RRBS
sequencing runs. We have previously shown a comparative example
analysis between RTA and an alternative base-calling operation,
i.e., Illumina off-line base-calling application (OLB) derived data-
sets in a RRBS background [27]. Therefore, we recommend per-
forming post-run standardization using a designated standard lane
with the OLB (see Note 12).
The steps for this operation are described in this section. After
the OLB application is run, the de-multiplexing must be repeated.
It may be necessary for some of these steps to be performed by
operators or administrators of the server facilities, which perform
the first phases of processing of the HiSeq run data. The OLB pro-
cess writes a directory and new base call data to the run data direc-
tory written to file storage by the HiSeq machine. The actual
commands required will depend on whether the output is written
directly to the run data directory (option a) or whether it is neces-

sary to copy the entire directory to the working directory (option
b). Many names for files and directories are examples and are
shown in brackets (<>) and $ > is the computer command line
prompt.
The steps are as follows:
1. Create a working directory in an area where there is write
access and set the default to it:
$>mkdir<myOLBdir>
$>cd<myOLBdir>
2. (a) Create a link to the run data directory:
$>ln s<RunDataLocationOnServer>/<runDataDi
r><localRunData>
or
(b) Copy the run data to the working directory:
$>cd<RunDataLocationOnServer>
$>tar cf -<runDataDir>| (cd<myOLBdir>; tar
xf -)
or an equivalent. The quantity of data will be large: a typical
HiSeq 100 bp single-ended run generates 1.5 Tb of com-
pressed images and base-call data.
3. Either of the above should create a run data directory or a link
to that run data directory in the working directory. Check it
with:
$>ls
4. Configure the OLB Run with:
$> <PathToOLBinstall>/bin/bustard.py <run-
DataDir>/Data/Intensities/ \
--CIF --tiles=s_1,s_2,s_4,s_8
--compression=gzip --control-lane=8 \
--make
for a run in which lanes 1, 2, and 4 are to be processed using
lane 8 as the control. Note that the control lane must be speci-
fied in the set defined by --tiles. The configure process creates
a directory: <runDataDir>/Data/Intensities/Bustard_<date>
_<username>containing scripts to perform the base-calling.
Start the process with:
$>cd<runDataDir>/Data/Intensities/
Bustard_<date>_<username>
$>nohup make -j 8 &
This will set the process running in the background. The
nohup command allows the process to continue if the user logs
off the server and -j 8 directs the make utility to run in parallel
on 8 processor cores. This value should be set as high as pos-
sible for the CPUs available. It is not necessary to remain
logged in during the OLB operation. The progress can be fol-
lowed with the command:
$>tail -f nohup.out
Completion is indicated by the line
Base-calling has completed successfully.
5. Demultiplexing:
Illuminas bcl2FastQ (CASAVA) utility is used to generate files
in fastq format for each of the samples run on a lane. Input to
this is base call files either from the RTA processing or from
OLB processing as above. The index barcode information is
given in a csv (Comma Separated Value)-formatted spread-
sheet file with the following fields:
FCID,Lane,SampleID,SampleRef,Index,Descript
ion,Control,Operator,ProjectID
AC2H00ACXX,1,Smpl_1,Person1,GTCCGC,Sample1,
N,RRBS,AnOperator,test01
AC2H00ACXX,1,Smpl_2,Person2,GAGTGG,Sample2,
N,RRBS,AnOperator,test01
where FCID is the serial number of the flowcell used. Two
example lines follow the header. The format of this file differs
from that produced by the Illumina Experiment Manager appli-
cation: the format is as described in Illumina documentation
(http://supportres.illumina.com/documents/documentation/
software_documentation/bcl2fastq/bcl2fastq_
letterbooklet_15038058brpmi.pdf). In the examples above, the
demultiplexing step will generate fastq files with names prefixed
by Smpl_1 & Smpl_2 contained in a directory named
Project_test01.
We find that different versions of the demultiplexing applica-
tion are required for RTA base-calling since later releases used
compressed base call files whereas those from OLB do not.
RTA demultiplexing now requires v1.8.3 of bcl2FastQ while
OLB base calls require v1.8.2 or earlier.
6. (a) Configuring for OLB output:
The base call files are generated in the directory made in the
OLB step above:
<path2v1.8.2>configureBclToFastq.pl --fastq-
cluster-count 0 \
--input-dir<runDataDir>/Data/Intensities/
Bustard_<date>_<username>/ \
--output-dir=<DeMux_filelocation>--sample-
sheet run_samplesheet.csv
to generate a directory structure<DeMux_
filelocation>containing various scripts.
(b) Configuring for RTA output:
<path2v1.8.3>configureBclToFastq.pl --fastq-
cluster-count 0 \
--input-dir<runDataDir>/Data/Intensities/
BaseCalls/ \
--output-dir=<DeMux_filelocation>--sample-
sheet run_samplesheet.csv
7. Demultiplexing:
cd<DeMux_filelocation>
nohup make -j 8 &
As above. This will start the demultiplexing process which is
likely to take several hours to complete on a full flowcell.
Completion is indicated by the message:
INFO: operation completed successfully
The process will produce a series of directories in < DeMux_
filelocation > set above with each project in a separate direc-
tory with names Project_test01 based on the last field of the
sample sheet. Within each project directory files for each sam-
ple will be in directories Sample_Smpl_1 in files named
Smpl_1_GTCCGC_R1_001.fastq.gz and similarly, containing
compressed fastq read data. The second read files for paired
end data are named _R2_. Command line options for these
commands are fully described in Illumina documentation
(http://supportres.illumina.com/documents/documenta-
tion/software_documentation/bcl2fastq/bcl2fastq_
letterbooklet_15038058brpmi.pdf).
3.13 Assessing Data 1. We use the FastQC (http://www.bioinformatics.babraham.

Quality and Alignment ac.uk/projects/fastqc/) tool to evaluate the quality of the
sequenced data because it is easy to use and fast. Other tools
which can be used for quality evaluation include SolexaQA
[28], FASTX toolkit (http://hannonlab.cshl.edu/fastx_tool-
kit/), PRINSEQ (http://prinseq.sourceforge.net) [29] and
MethyQA [30].
2. Trimming the sequenced read based on Phred score quality,
removal of filled-in CpG bases during library preparation from
the 3-end of the reads and removal of the adaptor sequences
from the reads is performed using our in-house developed
cleanadaptors tool [16]. The following commands will trim
adaptors from reads, leaving no reads shorter than 4 bp, the
length at which Bismark complains:
gzip -dc Smpl_1_GTCCGC_R1_001.fastq.gz | \
cleanadaptors -i adaptors.txt -t 3 -F - | \
cleanadaptors -i adaptors.txt -t 3 -x 4 -F
->Smpl_1_R1_adtrimmed.fastq
where -t 3 trims 3 bases back from the 3 ends of matching
reads in order to remove the C residue inserted during library
preparation. The above commands also decompress the origi-
nal fastq data: a step that is necessary prior to mapping. Similar
operation can be performed using FASTX toolkit, Trim Galore
(www.bioinformatics.babraham.ac.uk/projects/trim_galore/)
or Cutadapt (http://code.google.com/p/cutadapt/).
3. We used Bismark [31] to align RRBS reads for its relatively

unbiased nature [16] and speed of alignment. Bismark pro-
vides information for both CpG and non-CpG (CHG, CHH)
methylation. There are several other powerful bisulfite align-
ment programs which can also be used for similar purpose such
as BSMAP [32], BS Seeker [33], RRBSMAP [14], BatMeth
[34], PASS-bis [35] and SAAP-RRBS [13]. SAAP-RRBS can
also perform methylation calls, annotation of CpG sites, and
visualization (see Note 13).
4. We allow one mismatch (switch in the command n = 1) in the
seed (i.e., in the first 28 bp of the sequenced reads) while per-
forming Bismark alignment instead of the default, which allows
two mismatches. We routinely achieve 6070 % alignment effi-
ciency for human RRBS libraries and 4050 % for zebrafish
while aligning against the whole reference genome (see Note 14)
using 100-bp RRBS reads.
3.14 Output and CpG In sequencing-based methylation analysis, a common step is to fil-
Coverage ter the regions of the genome based on a coverage threshold
After Alignment (information per unit of analysis) to reduce the sampling error and
enhance the precision of the methylation values. Table 1 shows the
number of analysable MspI fragments (which passed a coverage
criteria) and the total number of CpG sites in them for 5 RRBS
libraries to provide examples of the end output from a multiplexed
RRBS pipeline.
3.15 Visualization Integrated Genome Viewer (IGV) enables visualization of meth-

of DNA Methylation ylation status from sequenced reads in the regions of interest [36].
from RRBS Reads Following read alignment by Bismark, BAM (previously used to
produce SAM files) are obtained and numerically sorted by chro-
mosome number and position. BAM files can be sorted using sam-
tools. (http://samtools.sourceforge.net/). Command for sorting
BAM file is: samtools sort bamfile > sortedbamfile. SAM files could
be sorted using the command: sort -k 3,3 -k 4,4n alignedRRBS.
sam > alignedRRBS.sam.sorted. The sorted BAM/SAM files are
imported in to IGV and the genome of choice is loaded using the
drop-down menu of IGV. The reads are visualized in bisulfite
mode with only the CG track on. SeqMonk (http://www.bioin-
formatics.bbsrc.ac.uk/projects/seqmonk) is another user-friendly
tool that allows loading various file formats (e.g., BAM, SAM,
text) to visualize methylation at base-resolution and the displayed
information and feature of SeqMonk is widely configurable
depending on requirements. MethVisual is another tool that allows
visualization of sequenced reads in its bisulfite mode and regional
analysis [30]. Note that RRBS does not allow to distinguish
between 5-methylcytosine and 5-hydroxymethylcytosine [37].
Table 1
MspI fragments with at least two CpGs with coverage of 10 in the multiplexed RRBS samples
Reads after QC check and Unique mapping Total number of Number of

Sample adaptor cleaning (millions) (%) fragmentsa CpG sitesb
X9012 31.3 67.3 307,712 1,996,825
X9014 18.7 64.1 142,108 864,623
X9018 43.8 67.8 347,536 2,138,975
X9019 28.3 66.5 295,926 1,866,592
X9020 17.9 63.4 176,249 1,134,616
aThe fragments that had at least two CpG sites covered by 10 or more sequence reads were included in the analysis.
Similar coverage criteria have been applied on 1000-bp tiling windows for filtering RRBS data [2]
bThe total number of CpG sites contained in those MspI fragments passing the coverage criterion described
3.16 Tools There is no single standard tool for detecting differential methylation
for Detecting as the strategies for investigating differential methylation need to
Differential be customized and modified in regards to the analysis platform
Methylation used, research question, unit of analysis, and the appropriate statis-
from RRBS Data tical tests. Nevertheless, there are some tools that can be used to
perform different types of methylation analysis. SeqMonk is a user
friendly tool that can perform methylation analysis on several fea-
tures, such as CpG Islands, proximal genes, etc. [38]. If detection
of differential methylation at single CpG sites is sought, methylKit,
an R package, can be used [39]. methylKit works directly with
sorted SAM files (sorted based on chromosome and read start).
As Bismark now produces BAM files, these files need to be con-
verted to SAM file to use in methylKit (BAM files can be converted
to SAM files using samtools). methylKit applies Fishers exact or
logistic regression to provide a list of differentially methylated CpG
sites in a pair-wise comparison and is suited for disease vs. control
analysis. However, it is less accurate for sample groups with high
inter-individual variation. BiSeq is another R package which con-
siders spatial dependence of CpG sites and can detect differentially
methylated regions [40].
We have recently developed a differential methylation analysis
package (DMAP) to generate reference DNA methylomes and
identify differentially methylated regions across multiple samples
from RRBS and WGBS data [41]. DMAP directly works with
BAM or SAM files or mix of BAM or SAM files and contains a suite
of statistical tools (Fishers exact, Chi-Square test, and ANOVA/F
test) to detect differentially methylated regions/fragments and
provide information and distances from nearest genes, CpG fea-
tures. DMAP is written in C language for fast operation and allows
flexibility to the end users. The source code and documentation is
freely available from http://biochem.otago.ac.nz/research/

databases-software/. DMAP provides option for tiled window
analysis for WGBS and RRBS data. Further, DMAP implements a
novel analysis approach for RRBS; it allows a fragment (MspI)-
based analysis for RRBS to identify differentially methylated frag-
ments (DMFs) [42]. DMAP is faster in operation compared to the
contemporary tools [41].
4 Notes
1. We have successfully created RRBS libraries from human

peripheral blood neutrophils, human placenta, melanoma cell
lines and formalin-fixed paraffin-embedded (FFPE) samples,
and zebrafish (Danio rerio) liver, brain, and embryos. For com-
plete proteinase digestion, incubate overnight as recommended
by Smith et al. [18] with either QIAamp protease solution or
20 g/mL proteinase K. To maximize elution efficiency, care-
fully pipette TE buffer onto the spin column membrane and
incubate at room temperature for 5 min before centrifugation.
Reapply the eluate to the membrane and repeat centrifugation.
This step helps to elute the remnant DNA from columns. If a
small amount of eluate remains on the inside of the spin col-
umn, reapply to the membrane and centrifuge for 30 s.
2. High quality input DNA is necessary for successful RRBS
library preparation. Nanodrop absorbance at 260 and 280 nm
(A260/280) ratios of between 1.8 and 2.0 are indicative of rela-
tively pure DNA, but quality assessment of samples by agarose
gel electrophoresis is also recommended. Nanodrop absor-
bance measurements are useful as a quick estimate of DNA
concentration, but fluorometric methods such as the Qubit are
more accurate, especially for detecting samples with RNA
contamination.
3. The TruSeq protocol recommends 1 g input DNA. We use
2 g to account for loss of DNA at each elution step and the
degradation expected as part of the bisulfite conversion.
However, others have successfully generated RRBS libraries
using smaller amounts of input DNA [38]. For all incubations
from this point on, the tubes were gently flicked to mix and
briefly spun down in a mini centrifuge. Incubations were per-
formed in a thermocycler to ensure accurate temperature con-
trol. For complete MspI digestion, samples were incubated
overnight.
4. The TruSeq DNA sample preparation kits contain either 24
single index adaptors (low throughput kit) or 96 dual indices
pre-loaded in a 96-well plate (high throughput kit). The adaptors
in the low throughput kit each contain a unique 6-bp index
barcode within its 63-bp sequence. The Illumina HiSeq

2000/2500 sequencers use a red laser to sequence A/C and a
green laser to sequence G/T. It is essential to maintain a bal-
ance between the color channels for each base of the index
reads being sequenced, otherwise they will fail to register.
Therefore, we recommend using the pooling guidelines in the
TruSeq DNA sample preparation guide and the Illumina
Experiment Manager software to determine adaptor combina-
tions suitable for multiplexing.
5. NuSieve GTG agarose is a low melting temperature agarose
for resolving 101000 bp DNA fragments. It requires more
delicate handing than regular agarose and it is recommended
to prepare it according to the manufacturers instructions. A
3 % (w/v) NuSieve gel is relatively brittle.
6. Running a 3 % NuSieve gel higher than 50 V will melt the gel.
7. To prevent degradation, we do not recommend incubating
excised gel slices at 50 C for 10 min as instructed by the
QIAquick and MinElute gel extraction kits. Using gentle agi-
tation, the gel slices will fully dissolve in ~20 min at room
temperature.
8. Previously published methods recommend two rounds of
bisulfite conversion with the QIAGEN Epitect kit [43], but
we found that two rounds of conversion and purification
resulted in significant loss of DNA. We have achieved more
consistent bisulfite conversion of size-selected libraries using
the EZ DNA methylation kit with a single conversion reaction
but a longer incubation of 18 h.
9. The TruSeq protocol recommends the use of DNA Taq poly-
merase but, as reported by Gu et al. [43] we have found
PfuTurbo Cx Hotstart polymerase (1.2 L in 25 L PCR reac-
tion volume) to provide considerably better amplification of
RRBS libraries. Note that our preferred concentration of
PfuTurbo Cx Hotstart is greater than that recommended by
Gu et al. [43]. Furthermore, Gu et al. recommended 1 L
DNA template be included in a 25 L PCR reaction volume
[43]. However, we have demonstrated that increasing the
DNA template (i.e., 3 L DNA template in a 25 L PCR reac-
tion volume) yields better amplification (http://www.chem.
agilent.com/Library/casestudies/Public/PfuTurbo_Cx_
Facilitates_High_Throughput_Methylation_
Studies_5991-1927.pdf).
10. A PCR product with optimal cycle number will have a reason-
able amount of 160340 bp products without nonspecific
amplification (a 125-bp adaptoradaptor dimer band is often
present, but this is removed during the second gel size
selection).
11. Following the Agilent Bioanalyzer run, use the region table
option on the 2100 Expert software to select the 150325 bp
region. This provides the average fragment size within that
region, which is used to calculate the library molarity. Use the
following formula to determine library molarity:
nM = concentration (ng/L)/(average fragment length
(bp) 0.00065)
If the adaptor band (~125 bp) is present in high concentration
in the final library, then it might affect the number of valid
cluster generated during sequencing. Therefore, either another
gel selection or exclusion of these libraries from sequencing is
recommended.
12. In HiSeq machines, by default, base-calling of the libraries is
performed using the RTA software which runs on the HiSeq
processor. However, the nonrandom base composition at the
start of RRBS fragments can confound the fluorophore emis-
sion standardizing set in the first four cycles [27] potentially
causing RTA to produce sub-optimal base-calling in multi-
plexed RRBS sequencing runs. In order to work around this
limitation, Illumina provide strategies to standardize from a
different lane of the flowcell. While this process can be done at
sequencing time by RTA, it is better performed subsequently
using the OLB, which generates new base calls from com-
pressed image files from the HiSeq. In order to make this most
effective, it is best to plan that a lane of normal genomic DNA
is run on at least one lane of the flowcell: the species is not
important since the requirement is for random base distribu-
tion in the first four cycles. Even if this requirement cannot
strictly be met, the process may still improve read qualities
(Fig. 2). The decline in quality by cycle of both A and B in
Fig. 2 is typical of that seen in sequencing RRBS libraries but,
importantly, the usable read length after OLB processing has
been extended by some 20 bp, which can be expected to pro-
duce better unique mapping. In this instance, the entire flow-
cell was of RRBS samples, none-the-less, running OLB using
one of the best lanes as control generated significantly better
reads. In cases where a lane of truly random genomic sequence
can be used, the improvement would be expected to be even
better. In some cases, we have observed that while OLB pro-
cessing may return more reads, the uniquely mapped propor-
tion diminished so that there was no net advantage [27]. Our
recommendation is to try both RTA and OLB base-calling to
see which gives more uniquely mapped reads. The overhead of
multiple mapping runs is not excessive with the tools recom-
mended here.
13. Most of the alignment programs are multithreaded, i.e., uses sev-
eral CPU cores. For example, Bismark uses 4 cores and BSMAP
uses 6. A computer with multiple CPU cores (6 or more) and
a
Quality scores across all bases (Sanger / illumina 1.9 encoding)
40
38
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 910-14 20-24 30-34 40-44 50-54 60-64 70-74 80-84 90-94 100
Position in read (bp)
b Quality scores across all bases (Sanger / illumina 1.9 encoding)

40
38
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 910-14 20-24 30-34 40-44 50-54 60-64 70-74 80-84 90-94 100
Position in read (bp)
Fig. 2 Per base sequence quality of RRBS reads from a zebrafish sample. The image is generated by FastQC
program (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). (a) demultiplexed RRBS library from
RTA base-calling and returning 4,584,091 reads. (b) the same RRBS library demultiplexed after OLB base-
calling using another lane on the flowcell as the control returning 4,583,180 reads. The yellow box plots (red
bar: median, box: interquartile ranges 2575 % and whisker: 1090 % percentile) show the base-calling qual-
ity scores across all sequencing reads of the sample. The blue line indicates the mean quality score. The other
samples had similar per base sequence quality
reasonable RAM (preferably 8 GB or more) is a minimum

requirement for RRBS data analysis. IGV also demands higher
RAM (8 GB or higher) while loading large number of reads for
visualization.
14. During the early days of second-generation sequencing per-
forming alignment against the in silico-generated reduced rep-
resentation genome was an alternative option to save the CPU
time taken to complete mapping. However, this strategy no
longer gives substantial computational advantage due to the
development of fast alignment programs. Further, mapping
against reduced genome could increase false unique mapping,
i.e., a sequenced read which will map multiply to the whole
genome and therefore be discarded in the subsequent analysis
might map uniquely against reduced genome [44].
Acknowledgements
We gratefully acknowledge the help and support of Dr. Robert

Day, Dr. Rebecca Laurie, and Les McNoe of the Otago Genomics
and Bioinformatics Facility (OGBF), Dunedin, New Zealand, dur-
ing the development of this method. This work was supported by
Gravida, National Centre for Growth and Development (formerly
NRCGD) and the Health Research Council (HRC), New Zealand.
A.C. would like to gratefully acknowledge the New Zealand
Institute for Cancer Research Trust (NZICRT) for their support.
References
1. Elango N, Yi SV (2008) DNA methylation and (2008) Quantitative analysis of human tissue-
structural and functional bimodality of verte- specific differences in methylation. Biochem
brate promoters. Mol Biol Evol 25:16021608 Biophys Res Commun 376:658664
2. Bock C, Beerman I, Lien WH, Smith ZD, Gu 7. Chatterjee A, Morison IM (2011) Monozygotic
H, Boyle P, Gnirke A, Fuchs E, Rossi DJ, twins: genes are not the destiny? Bioinformation
Meissner A (2012) DNA methylation dynam- 7:369370
ics during in vivo differentiation of blood and 8. Chatterjee A, Eccles MR (2015) DNA meth-
skin stem cells. Mol Cell 47:633647 ylation and epigenomics: new technologies and
3. Carrel L, Willard HF (2005) X-inactivation pro- emerging concepts. Genome Biol 16:103
file reveals extensive variability in X-linked gene 9. Chatterjee A (2012) Conference scene: epi-
expression in females. Nature 434:400404 genetic regulation: from mechanism to inter-
4. Rollins RA, Haghighi F, Edwards JR, Das R, vention. Epigenomics 4:487490
Zhang MQ, Ju J, Bestor TH (2006) Large- 10. Foy JP, Pickering CR, Papadimitrakopoulou
scale structure of genomic methylation pat- VA, Jelinek J, Lin SH, William WN Jr, Frederick
terns. Genome Res 16:157163 MJ, Wang J, Lang W, Feng L, Zhang L, Kim
5. Suzuki MM, Bird A (2008) DNA methylation ES, Fan YH, Hong WK, El-Naggar AK, Lee JJ,
landscapes: provocative insights from epig- Myers JN, Issa JP, Lippman SM, Mao L,
enomics. Nat Rev Genet 9:465476 Saintigny P (2015) New DNA methylation
6. Igarashi J, Muroi S, Kawashima H, Wang X, markers and global DNA hypomethylation are
Shinojima Y, Kitamura E, Oinuma T, Nemoto associated with oral cancer development.
N, Song F, Ghosh S, Held WA, Nagase H Cancer Prev Res (Phila) 8:10271035
11. Blackledge NP, Klose R (2011) CpG island genetic influence on epigenetic regulation.
chromatin: a platform for gene regulation. PLoS Genet 7:e1002228
Epigenetics 6:147152 22. Gu H, Bock C, Mikkelsen TS, Jger N, Smith
12. Zhou H, Hu H, Lai M (2010) Non-coding ZD, Tomazou E, Gnirke A, Lander ES, Meissner
RNAs and their epigenetic regulatory mecha- A (2010) Genome-scale DNA methylation map-
nisms. Biol Cell 102:645655 ping of clinical samples at single-nucleotide reso-
13. Ziller MJ, Gu H, Mller F, Donaghey J, Tsai lution. Nat Methods 7:133136
LT, Kohlbacher O, De Jager PL, Rosen ED, 23. Smallwood SA, Tomizawa S, Krueger F, Ruf
Bennett DA, Bernstein BE, Gnirke A, Meissner N, Carli N, Segonds-Pichon A, Sato S, Hata K,
A (2013) Charting a dynamic DNA methyla- Andrews SR, Kelsey G (2011) Dynamic CpG
tion landscape of the human genome. Nature island methylation landscape in oocytes and
500:477481 preimplantation embryos. Nat Genet
14. Xi Y, Bock C, Mller F, Sun D, Meissner A, Li 43:811814
W (2012) RRBSMAP: a fast, accurate and 24. Steine EJ, Ehrich M, Bell GW, Raj A, Reddy S,
user-friendly alignment tool for reduced repre- van Oudenaarden A, Jaenisch R, Linhart HG
sentation bisulfite sequencing. Bioinformatics (2011) Genes methylated by DNA methyl-
28:430432 transferase 3b are similar in mouse intestine
15. Chatterjee A, Ozaki Y, Stockwell PA, Horsfield and human colon cancer. J Clin Invest
JA, Morison IM, Nakagawa S (2013) Mapping 121:17481752
the zebrafish brain methylome using reduced 25. Hartung T, Zhang L, Kanwar R, Khrebtukova
representation bisulfite sequencing. Epigenetics I, Reinhardt M, Wang C, Therneau TM, Banck
8:979989 MS, Schroth GP, Beutler AS (2012)
16. Chatterjee A, Stockwell PA, Rodger EJ, Diametrically opposite methylome-
Morison IM (2012) Comparison of alignment transcriptome relationships in high- and low-
software for genome-wide bisulphite sequence CpG promoter genes in postmitotic neural rat
data. Nucleic Acids Res 40:e79 tissue. Epigenetics 7:421428
17. Meissner A, Mikkelsen TS, Gu H, Wernig M, 26. Chatterjee A, Stockwell PA, Horsfield JA,
Hanna J, Sivachenko A, Zhang X, Bernstein Morison IM, Nakagawa S (2014) Base-
BE, Nusbaum C, Jaffe DB, Gnirke A, Jaenisch resolution DNA methylation landscape of
R, Lander ES (2008) Genome-scale DNA zebrafish brain and liver. Genomics Data
methylation maps of pluripotent and differenti- 2:342344
ated cells. Nature 454:766770 27. Chatterjee A, Rodger EJ, Stockwell PA, Weeks
18. Smith ZD, Gu H, Bock C, Gnirke A, Meissner A RJ, Morison IM (2012) Technical consider-
(2009) High-throughput bisulfite sequencing in ations for reduced representation bisulfite
mammalian genomes. Methods 48:226232 sequencing with multiplexed libraries.
19. Baranzini SE, Mudge J, van Velkinburgh JC, J Biomed Biotechnol 2012:741542
Khankhanian P, Khrebtukova I, Miller NA, 28. Cox MP, Peterson DA, Biggs PJ (2010)
Zhang L, Farmer AD, Bell CJ, Kim RW, May SolexaQA: at-a-glance quality assessment of
GD, Woodward JE, Caillier SJ, McElroy JP, Illumina second-generation sequencing data.
Gomez R, Pando MJ, Clendenen LE, Ganusova BMC Bioinformatics 11:485
EE, Schilkey FD, Ramaraj T, Khan OA, 29. Schmieder R, Edwards R (2011) Quality con-
Huntley JJ, Luo S, Kwok PY, Wu TD, Schroth trol and preprocessing of metagenomic datas-
GP, Oksenberg JR, Hauser SL, Kingsmore SF ets. Bioinformatics 27:863864
(2010) Genome, epigenome and RNA 30. Sun S, Noviski A, Yu X (2013) MethyQA: a
sequences of monozygotic twins discordant for pipeline for bisulfite-treated methylation
multiple sclerosis. Nature 464:13511356 sequencing quality assessment. BMC
20. Bock C, Kiskinis E, Verstappen G, Gu H, Bioinformatics 14:259
Boulting G, Smith ZD, Ziller M, Croft GF, 31. Krueger F, Andrews SR (2011) Bismark: a flexi-
Amoroso MW, Oakley DH, Gnirke A, Eggan ble aligner and methylation caller for Bisulfite-
K, Meissner A (2011) Reference maps of Seq applications. Bioinformatics 27:15711572
human ES and iPS cell variation enable high- 32. Xi Y, Li W (2009) BSMAP: whole genome
throughput characterization of pluripotent cell bisulfite sequence MAPping program. BMC
lines. Cell 144:439452 Bioinformatics 10:232
21. Gertz J, Varley KE, Reddy TE, Bowling KM, 33. Chen PY, Cokus SJ, Pellegrini M (2010) BS
Pauli F, Parker SL, Kucera KS, Willard HF, Seeker: precise mapping for bisulfite sequencing.
Myers RM (2011) Analysis of DNA methylation BMC Bioinformatics 11:203
in a three-generation family reveals widespread
34. Lim JQ, Tennakoon C, Li G, Wong E, Ruan Y, regions in targeted bisulfite sequencing data.
Wei CL, Sung WK (2012) BatMeth: improved Bioinformatics 29:16471653
mapper for bisulfite sequencing reads on DNA 41. Stockwell PA, Chatterjee A, Rodger EJ,
methylation. Genome Biol 13:R82 Morison IM (2014) DMAP: differential meth-
35. Campagna D, Telatin A, Forcato C, Vitulo N, ylation analysis package for RRBS and WGBS
Valle G (2013) PASS-bis: a bisulfite aligner data. Bioinformatics 30:18141822
suitable for whole methylome analysis of illu- 42. Chatterjee A, Stockwell PA, Rodger EJ,
mina and SOLiD reads. Bioinformatics Duncan EJ, Parry MF, Weeks RJ, Morison IM
29:268270 (2015) Genome-wide DNA methylation map
36. Robinson JT, Thorvaldsdttir H, Winckler W, of human neutrophils reveals widespread inter-
Guttman M, Lander ES, Getz G, Mesirov JP individual epigenetic variation. Sci Rep
(2011) Integrative genomics viewer. Nat 5:17328
Biotechnol 29:2426 43. Gu H, Smith ZD, Bock C, Boyle P, Gnirke A,
37. Rodger EJ, Chatterjee A, Morison IM (2014) Meissner A (2011) Preparation of reduced rep-
5-hydroxymethylcytosine: a potential therapeu- resentation bisulfite sequencing libraries for
tic target in cancer. Epigenomics 6:503514 genome-scale DNA methylation profiling. Nat
38. Smallwood SA, Kelsey G (2012) Genome-wide Protoc 6:468481
analysis of DNA methylation in low cell num- 44. Akalin A, Garrett-Bakelman FE, Kormaksson
bers by reduced representation bisulfite M, Busuttil J, Zhang L, Khrebtukova I, Milne
sequencing. Methods Mol Biol 925:187197 TA, Huang Y, Biswas D, Hess JL, Allis CD,
39. Akalin A, Kormaksson M, Li S, Garrett- Roeder RG, Valk PJ, Lwenberg B, Delwel R,
Bakelman FE, Figueroa ME, Melnick A, Mason Fernandez HF, Paietta E, Tallman MS, Schroth
CE (2012) methylKit: a comprehensive R GP, Mason CE, Melnick A, Figueroa ME
package for the analysis of genome-wide DNA (2012) Base-pair resolution DNA methylation
methylation profiles. Genome Biol 13:R87 sequencing reveals profoundly divergent epi-
40. Hebestreit K, Dugas M, Klein HU (2013) genetic landscapes in acute myeloid leukemia.
Detection of significantly differentially methylated PLoS Genet 8:e1002781
Chapter 17
A Protocol for the Determination of the Methylation Status

of Gingival Tissue DNA at Specific CpG Islands
Trudy J. Milne
Abstract
Tissue biopsies are very precious. A method that allows the isolation of a high quality and quantity of
genomic DNA, total RNA, and total protein from a single biopsy that is suitable for downstream applications
(e.g., DNA methylation analysis, quantitative PCR, and gel electrophoresis techniques) is very desirable.
The method described here combines a tissue stabilization reagent combined with a spin-column method
for the simultaneous purification of gingival tissue DNA, RNA, and protein. The genomic DNA is then
used for quantitative analysis of DNA methylation using real-time PCR (the qAMP method). Subsequent
data analysis is very straightforward using online software. Future analyses may include the RNA transcript
analysis as well as protein expression levels of genes identified as differentially methylated.
Key words qAMP, Regional DNA methylation, Gingival tissue, Genomic DNA, Total RNA,
Allprotect tissue reagent, AllPrep DNA/RNA/protein kit
1 Introduction
Our genetic code contains all the information required for our
bodies to function. However, it is the epigenetic code (epimean-
ing on or above) that determines when and where our genes
are expressed. Waddingtons original definition of epigenetics in
1942 was that phenotype arises from genotype through pro-
grammed change [1]. The modern definition of epigenetics refers
to the information inherited during cell division other than the
DNA sequence itself. It is these heritable genotypes or epig-
enomic defects, which may arise from environmental stimuli, that
result in a change in the local and global density of DNA methyla-
tion or incorrect histone modification. Alterations to the pro-
gramme that orchestrates gene expression may have implications
in normal human development and disease [24].
DNA methylation occurs via the addition of a methyl group
to a cytosine residue on the linear DNA strand [5, 6]. This addi-
tion only occurs when the cytosine is adjacent to a guanine
299
300 Trudy J. Milne
residue separated by only one phosphate (CpG) [7]. CpGs occur

throughout the genome but less frequently than chance predicts
and tend to occur in clusters at the 5' regulatory regions of genes
(known as CpG islands). The hypermethylation of CpG islands
at the gene promoter region is mostly associated with gene
silencing [8].
The commercially available EpiTect Methyl II PCR system
employs the quantitative analysis of DNA methylation using a real-
time PCR (qAMP) method [9] for the correlation of CpG island
methylation status with biological phenotypes or disease outcomes.
The qAMP method relies on the differential cleavage of target
sequences by two different restriction endonucleases, one that is
methylation-sensitive and the other being methylation-dependent.
The products of the digestion reactions are then used as qPCR
templates to determine the ratio of methylated and un-methylated
DNA at the site of interest.
The following protocol describes the simultaneous isolation of
DNA, RNA, and protein from a single gingival tissue sample and
the determination of the methylation status of a specific region of
a gene of interest. The RNA and protein can be stored for subse-
quent determination of the mRNA and protein levels of the sam-
ple. The protocol combines the proprietary Allprotect reagent
for immediate stabilization of DNA/RNA and protein, the
Allprep DNA/RNA/Protein isolation kit and the EpiTect
Methyl II PCR system.
2 Materials
2.1 Tissue Collection 1. Allprotect tissue reagent (QIAGEN GmbH, Germany).

2. RNase Away (Life Technologies, CA, USA).
2.2 DNA, RNA 1. Allprep DNA/RNA and protein mini kit (QIAGEN).
and Protein 2. Kimble-Chase Kontes Pellet Pestle.
Purification
3. Barrier (filter) tips.
4. DNase-, RNase-, and protease-free water.
5. RNase-Free DNase set (QIAGEN).
6. Ethidium bromide (10 mg/mL).
7. 1 kb Plus DNA markers (Life Technologies).
8. Agarose Loading Dye.
9. Tris-EDTA (TE) buffer (50 stock): Prepare by dissolving Tris
base (242 g/L) in 500 mL water and then add 100 mL 0.5 M
EDTA and 57.1 mL glacial acetic acid. Adjust volume to 1 L
with distilled water.
A Protocol for the Determination of the Methylation Status of Gingival Tissue DNA 301
10. Horizontal electrophoresis system (for example, Sub-Cell GT

Cell, Bio-Rad).
11. Microvolume spectrophotometer (for example, NanoVue,
GE Healthcare).
2.3 Genomic DNA 1. PureLink genomic DNA mini kit (contains RNase A, Life
RNase Treatment Technologies).
2. Microvolume spectrophotometer (NanoVue, GE Healthcare).
2.4 Methylated 1. CpG Methyltransferase (M-SssI, New England Biolabs

Control Genomic DNA NEBM0226S).
2.5 EpiTect Methyl 1. EpiTect Methyl II DNA Restriction kit (QIAGEN 335452).
II PCR System DNA 2. EpiTect PCR Array (QIAGEN 335212A-G and
Methylation Profiling 335222A-G).
3. EpiTect Methyl II PCR Assay (QIAGEN 335002).
4. PCR tubes.
5. FAST 96-well qPCR plates.
6. ABI 7500 FAST qPCR system (Applied Biosystems, CA, USA).
7. SABioscience analysis Microsoft Excel spreadsheet template.
3 Methods
3.1 Tissue Collection 1. The gingival tissue biopsies are collected and immediately sub-
merged in Allprotect solution (0.5 mL or at least ten vol-
umes). The tissue is then stored overnight at 4 C before
long-term storage at 20 C (see Note 1).
3.2 DNA, RNA, The AllPrep DNA/RNA/Protein Mini Purification protocol is

and Protein outlined in the manufacturers manual. Briefly:
Purification
1. -Mercaptoethanol is first added to the buffer RLT
(10 L/1 mL).
2. The tissue is removed from the Allprotect Tissue Reagent
using forceps and placed in a 1.5-mL microfuge tube with the
RLT buffer (350 L). Using fine scissors, cut the tissue into
small pieces before being disrupted using a Kimble-Chase
Kontes Pellet Pestle. The tissue is finally homogenized by
passing it through a blunt 20-gauge (0.9 mm diameter) needle
fitted to an RNase-free syringe (see Note 2).
3. The tissue homogenate is then centrifuged for 3 min at
16,000 g and the supernatant transferred to an AllPrep DNA
spin column. The spin column is then centrifuged for 30 s at
8000 g.
302 Trudy J. Milne
4. The Allprep column can be placed in a new collection tube and

stored at 4 C for later purification of the bound genomic
DNA (gDNA).
5. The flow-through that now contains the RNA and protein is
mixed with 96100 % ethanol (250 L) and mixed well by
pipetting. The sample is then transferred to an RNeasy spin
column and centrifuged for 15 s at 8000 g. The flow-
through is then transferred to a 2-mL tube for protein purifica-
tion at a later stage.
6. The bound RNA is then treated with DNase I treatment (see
Note 3). RW1 wash buffer (350 L) is added to the RNeasy
spin column and centrifuged for 15 s at 8000 g and the
flow-through discarded.
7. The DNase I stock solution (10 L) is diluted in 70 L of buf-
fer RDD by gentle inversion of the tube. The DNase I mix
(80 L) is then carefully applied directly to the RNeasy spin
column and incubated on the benchtop (2030 C) for 15 min.
Following the DNase I incubation, the RNeasy spin column is
washed with Buffer RW1 (350 L) and centrifugation for 15 s
at 8000 g.
8. Two washes with Buffer RPE (500 L) are then performed with
the flow-through discarded after each spin at 8000 g. The
column is then placed in a new 2 mL tube and spun at 16,000 g
to dry the membrane removing any residual Buffer RPE.
9. The RNeasy spin column is then placed into a new 1.5 mL
storage microfuge. RNase-free water (50 L) is added directly
to the spin column membrane and centrifuged for 1 min at
8000 g to elute the RNA. The total RNA can be stored at
20 C or 70 C for up to 1 year.
10. The gDNA bound to the Allprep DNA spin column can now
be purified. Buffer AW1 (500 L) is added to the Allprep
DNA spin column and centrifuged for 15 s at 8000 g.
Discard flow-through and repeat.
11. The Allprep spin column is then placed into a new 1.5 mL
storage microfuge. The genomic DNA is eluted with Buffer
EB (100 L, preheated to 70 C) that is added directly to the
membrane and incubated at room temperature (1525 C) for
2 min followed by centrifugation at 8000 g for 1 min. The
eluted gDNA can be stored in Buffer EB long-term at 20 C.
3.3 Genomic DNA A modified Purelink Genomic DNA kit protocol is used (see Note 4).
RNase Treatment Briefly:
1. To the gDNA (13 g), DNase- and RNase-free water is added
to a final volume of ~60 L.
2. RNase A (20 L, 20 mg/mL) in 50 mM TrisHCl pH 8.0,
10 mM EDTA is added and incubated at 37 C for 30 min.
3. Genomic Lysis/Binding buffer (200 L) is added, mixed well

by pipetting up and down and then incubated at 56 C for
30 min.
4. Ethanol (96100 %, 200 L) is then added to the lysate and
mixed well with vortexing for 5 s.
5. The lysate/ethanol mix is transferred to the spin column in a
collection tube and centrifuged for 1 min at 10,000 g.
6. The collection tube and flow-through is discarded and the spin
column placed in a new collection tube. The bound gDNA is
washed with Wash buffer 1 (500 L) that is removed with cen-
trifugation for 1 min at 10,000 g.
7. The flow-through is discarded and a second aliquot of Wash
buffer 2 (500 L) is added and again centrifuged, for 3 min at
12,000 g.
8. To elute the gDNA, the spin column is transferred to a clean
storage microcentrifuge tube and Purelink Genomic Elution
buffer (50 L) is added to the membrane of the spin column.
Following a 1 min incubation at room temperature, the gDNA
is eluted by centrifuging the spin column for 1 min at
12,000 g.
3.3.1 Yield and Purity This can be done using a spectrophotometer or agarose electro-
Assessment of DNA phoresis (see Note 5).
and RNA
3.3.1.1 RNA and DNA 1. Absorbance values should lie between 0.1 and 1.0 to ensure an
Spectrophotometry optimal measurement. The concentration of RNA = 1 A260
Unit of ssRNA = 40 g/mL H2O. Pure RNA has an A260/A280
ratio of 1.92.1 in 10 mM Tris buffer. The concentration of
gDNA = 1 A260 Unit of dsRNA = 50 g/mL H2O. Pure DNA
has an A260/A280 ratio 1.8. Do not interchange between
spectrophotometers, as values are machine-dependent.
3.3.1.2 gDNA Agarose 1. To make one agarose gel (0.8 % (w/v), 30 mL), agarose
Electrophoresis (0.24 g) is dissolved in TAE buffer (30 mL) by heating slowly
in a microwave oven.
2. When cool (50 C), ethidium bromide (10 mg/mL, 5 L) is
added to the mixture, the agarose is swirled to mix and poured
into an agarose gel tray. A comb is inserted into the gel to cre-
ate wells to allow each sample to be loaded onto the gel. The
gel can be placed at 4 C to speed up the gel-setting process.
3. The sample gDNA (100200 ng) or DNA standard markers
(5 L, 1 g/L) are mixed with 1 L of 6 gel loading dye to
a maximum volume of 10 L.
4. Once set, the agarose gel is placed in the electrophoresis tank
and TAE running buffer is added. The gDNA samples and
304 Trudy J. Milne
Fig. 1 Agarose gel electrophoresis of genomic DNA. Lane (a), Before RNase A
treatment the gDNA can be seen as a strong band of greater than 23 kb (possibly
with protein bound to it) along with lower molecular weight RNA bands. Lane (b),
After RNase A treatment and a second spin-column purification the low molecu-
lar weight RNA bands are no longer present
DNA standard markers are loaded into the gel and run at 80 V
for approximately 1 h or until the bromophenol blue has
migrated almost two-thirds of the gel.
5. The gel can also be post-stained with ethidium bromide
(0.5 g/mL) in TAE buffer for 3060 min.
6. The gDNA bound to the ethidium bromide is then visualized
with a UV transilluminator (Fig. 1).
3.4 EpiTect Methyl The EpiTect Methyl II PCR system uses two different restriction
II PCR System DNA endonucleases whose activities require either the presence or
Methylation Profiling absence of methylated cytosine in their respective recognition
sequences. The products of digestion are then used as qPCR tem-
plates for relative quantitation of whose products for the genes of
interest (see Note 6).
3.4.1 Restriction 1. A mix containing only the gDNA, restriction digestion buffer,
Digestion and RNase- DNase-free water for each sample is prepared
(see Note 7).
2. Four digestion tubes are set up for each sample, a mock
(no enzyme) (Mo), a methylation-sensitive enzyme (Ms), a
methylation-dependent enzyme (Md), and a double digest
(both enzymes) (Msd) according to the manufacturer's

instructions.
3. The gDNA mix (14 L) is then added to each of the restriction
digestion tubes and incubated in a thermal cycler overnight at
37 C.
4. Following heat-inactivation of the enzymes at 65 C for 20 min
the resulting products are ready for qPCR analysis or can be
stored at 20 C.
3.4.2 Real-Time PCR 1. Following the restriction digestion the resultant products are
mixed with an appropriate real-time SYBR Green qPCR mas-
ter mix and directly dispensed either onto an array plate con-
taining pre-aliquoted PCR primers or combined with the
EpiTect Methyl II PCR Assay before being dispensed onto a
96-well plate (see Note 8).
2. The sealed PCR plate is placed into the real-time PCR instru-
ment and the amplification reactions carried out using thermal
cycling parameters specific to your instrument (see Note 9).
3.4.3 Data Analysis 1. On completion of the cycling program, the Cq values are
obtained according to the instructions provided by the manu-
facturer of the real-time PCR instrument (see Note 10).
2. Microsoft Excel spreadsheet data files are uploaded to the
SABioscience website (www.sabiosciences.com) for analysis.
4 Notes
1. Wear gloves at all times. Use only RNase-DNase-free plastic-

ware. Instruments such as the forceps and scissors should be
rinsed in RNase Away. All solutions should be made up in
DNase-RNase-free water.
2. The addition of the Allprotect reagent to the tissue may
make it very firm. Fine scissors may be required to break the
tissue down to a size suitable for the further processing with
the pestle.
3. The on-column DNase I treatment is recommended and
results in a higher yield of RNA compared to treatment with
DNase after purification which may require an ethanol precipi-
tation step.
4. Where possible a spin-column should be employed, as the
return of DNA and RNA is higher than protocols that use
ethanol precipitation.
5. Unlike spectrophotometry, agarose electrophoresis gives an
indication of the integrity of the DNA or RNA.
306 Trudy J. Milne
6. A number of EpiTect Methyl II PCR assays are available for a

96-well plate format. For a single DNA sample (a) a single
gene assay, (b) signature PCR arrays (22 genes) on one
96-plate, and (c) complete PCR arrays (94 genes) on four
96-well plates.
7. The EpiTect methyl II DNA Restriction Kit can be used for
the digestion of up to 12 g of genomic DNA. The manufac-
turer's handbook should be consulted for the amount of
genomic DNA required for each Signature PCR array or single
gene PCR assay. Ensure accurate pipetting of the small 0.5 L
volume. Visually inspect the volume of enzyme each time.
8. Controls for monitoring enzyme digestion efficiency are
included on each array plate; however when single gene assays
are employed, SEC and DEC control assays need to be
purchased.
9. In this laboratory, an Applied Biosystems 7500 Fast Real-time
PCR instrument was used with the following thermal cycling
parameters: 95 C for 10 min (initial denaturation) followed
by three initial cycles of 99 C 30 s and 72 C 1 min, and then
40 dissociation cycles of 97 C 15 s and 72 C 1 min.
10. When comparing multiple plates ensure the settings for all
plates are set the same.
Acknowledgments
This work was supported by a grant from the New Zealand Dental
Association.
References
1. Van Speybroeck L (2002) From epigenesis to 6. Doskocil J, Sorm F (1962) Distribution of
epigenetics: the case of C. H. Waddington. Ann 5-methylcytosine in pyrimidine sequences of
NY Acad Sci 981:6181 deoxyribonucleic acids. Biochim Biophys Acta
2. Holliday R, Pugh JE (1975) DNA modification 55:953959
mechanisms and gene activity during develop- 7. Illingworth RS, Bird AP (2009) CpG islands
ment. Science 187:226232 a rough guide. FEBS Lett 583:
3. Feinberg AP (2007) Phenotypic plasticity and 17131720
the epigenetics of human disease. Nature 8. Deaton AM, Bird A (2011) CpG islands and
447:433440 the regulation of transcription. Genes Dev
4. Williams SD, Hughes TE, Adler CJ, Brook AH, 25:10101022
Townsend GC (2014) Epigenetics: a new fron- 9. Oakes C, La Salle S, Robaire B, Trasler JM
tier in dentistry. Aust Dent J 59(Suppl 1):2333 (2006) Evaluation of a quantitative DNA meth-
5. Wyatt GR (1951) The purine and pyrimidine ylation analysis technique using methylation-
composition of deoxypentose nucleic acids. sensitive/dependent restriction and real-time
Biochem J 48:584590 PCR. Epigenetics 1:146152
Chapter 18
Genome-Wide Analysis of Periodontal and Peri-Implant

Cells and Tissues
Moritz Kebschull, Claudia Hlsmann, Per Hoffmann,
Abstract
Omics analyses, including the systematic cataloging of messenger RNA and microRNA sequences or DNA
methylation patterns in a cell population, organ or tissue sample, are powerful means of generating com-
prehensive genome-level data sets on complex diseases. We have systematically assessed the transcriptome,
miRNome and methylome of gingival tissues from subjects with different diagnostic entities of periodontal
disease, and studied the transcriptome of primary cells ex vivo, or in vitro after infection with periodontal
pathogens. Our data further our understanding of the pathobiology of periodontal diseases and indicate
that the gingival -omes translate into discernible phenotypic characteristics and possibly support an alterna-
tive, molecular classification of periodontitis.
Here, we outline the laboratory steps required for the processing of periodontal cells and tissues for
-omics analyses using current microarrays or next-generation sequencing technology.
Key words Periodontal disease, Gene expression, Transcriptome, MicroRNA, DNA methylation,
Microarray, Next-generation sequencing, Gingiva
1 Introduction
After decades of research confined to the study of candidate sin-

gle molecules or pathways, technologies available today allow for
an unbiased, systematic evaluation of biological information in
cells or tissues of interest on a large or genome-wide scale, and the
associated underlying biology. These approaches are collectively
referenced using the suffix -omics, e.g., gen-omics for the
analysis of the Genome, transcript-omics for the analysis of
(transcribed) messenger RNA, miRN-omics for the study of
micro RNAs, and methyl-omics for the genome-wide assess-
ment of DNA methylation.
-omics studies are a powerful means of generating comprehensive
genome-level data sets on complex diseases and have provided
enormous insights mostly in cancer research [13], but also in
307
308 Moritz Kebschull et al.
Tissue Sample Collecting cells

Harvesting
Bacterial
Homogenization Infection of DCs
RNA-Purification
Commercial Kit or
Phenol-Chloroform-Extraction
RNA-Quantitation
Library
Preparation
cRNA-
Hybridization to
Bead Chip
Validation of
cDNA library
Microarray
Scanning RNA-
Sequencing
Data Analysis
Fig. 1 Workflow in a typical gene expression experiment of gingival tissue or infected dendritic cells
other conditions such as muscular dystrophy [4], Alzheimers dis-

ease and dementia [5, 6], rheumatologic disorders [7, 8], and
asthma [9, 10].
We have adopted an omics-based approach in the study of the
pathobiology of periodontal and peri-implant diseases, starting
with the transcriptome and expanding by integrating other -omes
(Fig. 1). Specifically, we examined gingival tissue transcriptomes in
clinically healthy and periodontitis-affected gingival tissues [11], in
experimental gingivitis [12], in chronic and aggressive periodonti-
tis [13], and in peri-implantitis [unpublished data]. The data were
used to assess the relationship of gene expression with the levels of
subgingival periodontal bacteria [14, 15], and to test whether the
-Omics Analyses of Cells and Tissues 309
clinical entities of chronic and aggressive periodontitis were also

reflected by characteristic differences on the transcriptome level
[13]. These studies identified molecules or pathways with a possi-
ble role in aggressive periodontitis that were then subject to a more
focused characterization, i.e., the activation of natural killer cells by
CRACC [16], and the differential activation of invariant natural
killer T-cells in chronic and aggressive periodontitis [17].
Subsequently, utilizing unsupervised clustering of the transcrip-
tomic datasets, we uncovered novel, molecular classes of peri-
odontitis that also differed in their clinical and microbiological
phenotype [18]. The gingival transcriptomes were supplemented
by the miRNome [19] and methylome (unpublished data).
In another set of studies, we examined whether comprehensive
periodontal therapy may induce changes in gene expression of
peripheral blood mononuclear cells, focusing on the potential of
therapy to promote an antiatherogenic phenotype [20].
We propose that the -omics-based study of gingival or mucosal
tissues, primary cells isolated from subjects with periodontal dis-
ease, or cell culture material from model systems designed to mimic
periodontitis will allow an enhanced understanding of the pathobi-
ology of the periodontal diseases, inform the design of subsequent
studies, and eventually lead to an improved diagnosis and therapy.
Herein, we provide a detailed description of the necessary lab-
oratory steps in order to process gingival tissue samples, peripheral
blood samples and material from a cell culture model or periodon-
tal infections in view of hybridization with full-genome microar-
rays or analysis by next-generation sequencing. We have focused
on the description of the procedures that will likely be performed
by the oral biology researchers themselvesnote that RNA label-
ing and hybridization to microarrays or the construction of
sequencing libraries are routinely performed in core facilities or by
solution providers.
In the following chapters, we provide information on the basic
analysis steps for data from microarray and sequencing experiments
(see Chapter 19 by Kebschull et al., this volume), and on the use of
machine learning tools for the supervised and unsupervised analy-
sis of -omics data (see Chapter 20 by Kebschull et al., this
volume).
2 Materials
2.1 Source Materials (a) RNAlater (Ambion, Houston, TX, USA, #AM7021).
2.1.1 Tissue Samples (b) Eppendorf Biopure Safe-lock 1.5 mL tubes (Eppendorf,
Germany, #22600028) (see Note 1).
Gingival/Mucosal Tissue
Harvesting and Processing
Tissue Disruption (a) AllPrep DNA/RNA/Protein Kit, Qiagen (Germany).

and Homogenization (b) Eppendorf Tubes 5.0 mL, PCR clean (Eppendorf, Germany;
#0030119460).
(c) 2-Mercaptoethanol 99.0%RLT-Lysis-Buffer, AllPrep DNA/
RNA/Protein Kit, Qiagen (Germany).
(d) Homogenizer, e.g., Miccra D-1 (ART Prozess- & Labortechnik
GmbH & Co. KG, Germany).
(e) DNaseRNase-free distilled water.
(f) Eppendorf Safe-Lock Tubes, 2.0 mL (Eppendorf, Germany,
#0030120094).
(g) For alternative protocol (see Subheading 2.2, item b): TRIzol
Reagent (Invitrogen, Carlsbad, CA, USA; #15596-018).
2.1.2 Primary (a) Standard phlebotomy set, e.g., Vacutainer Safety-Lok Blood
Mononuclear Cells Isolated Collection Set (Becton Dickinson, #367283).
from Patient Blood (b) Vacutainer CPT Cell Preparation Tubes 8 mL (BD, #362753).
Blood Collection
Blood Cell Separation (a) Cooled centrifuge with releasable brake, e.g., Centrifuge
5702R (Eppendorf).
(b) 50 mL Falcon tubes.
(c) Phosphate-buffered saline without Ca2+/Mg2+ (Mediatech,
Manassas, VA, USA; #21-031-CV).
(d) Hemocytometer, e.g., improved Neubauer bright-line
(Hausser Scientific, Hersham, PA, USA; #1492).
(e) MACS separation columns (Miltenyi Biotech, Auburn, CA,
USA; #130-042-401).
(f) MACS multistand (Miltenyi; #007331).
(g) AutoMACS rinsing solution, pH 7.2 (Miltenyi,
#130-091-222).
(h) MACS microbeads (Miltenyi).
CD4 (#120-000-440).
CD14 (#120-000-305).
2.1.3 Cultured Cells (a) Dulbeccos Modified Eagle Medium, DMEM, (Invitrogen,
Germany, #41965-039).
Primary Culture
(b) Fetal Bovine Serum (Invitrogen, Germany, #26140-079).
(c) Penicillin-streptomycin (10,000 U/mL) (Invitrogen,
Germany, #15140-122).
(d) rGM-CSF, 50 g, (ImmunoTools, Friesoythe, Germany #
12343125).
(e) Corning CellBIND 24-Well Plates, (Corning, NY, USA,
#3337).
Cell Harvesting (a) TrypsinEDTA (0.05 %), phenol red, (Invitrogen, Germany
and Processing #25300-054).
(b) Eppendorf Biopure Safe-lock 1.5 mL tubes (Eppendorf,
Germany, #22600028).
(c) Corning Small Cell Scraper (Corning, NY, USA, #3010).
(d) Cooled microcentrifuge, e.g., Centrifuge 5415R (Eppendorf,
Hamburg, Germany).
2.2 Extraction (a) Cooled microcentrifuge, e.g., Centrifuge 5415R (Eppendorf,

and Purification Hamburg, Germany).
of Nucleic Acids (and (b) Absolute ethanol.
Protein)
(c) 2-Mercaptoethanol 99.0 %.
2.2.1 Purification (d) AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Germany,
by a Commercial Kit #80004).
(e) RNeasy MinElute Cleanup Kit (Qiagen, Germany, #74204).
(f) Eppendorf Tubes 5.0 mL, PCR clean.
2.2.2 Alternative Protocol (a) Trizol Reagent (Invitrogen, Carlsbad, CA, USA;
#15596-018).
(b) Chloroform.
(c) Cooled microcentrifuge, e.g., Centrifuge 5415R (Eppendorf,
Hamburg, Germany).
(d) Ethanol 99.5 % mol. biol. grade.
(e) Ethanol 75 %.
(f) Glycogen (Invitrogen; #10814-010) adjusted to 5 g/mL
with nuclease-free water (Invitrogen, #10977-015).
(g) 0.1 M sodium citrate in 10 % ethanol.
(h) 8 mM NaOH.
(i) Isopropyl alcohol.
(j) 0.3 M Guanidine hydrochloride in 95 % ethanol.
(k) 1 % SDS.
(l) RNeasy Mini kit (Qiagen, Valencia, CA, USA; #74104).
2.3 Quantitation 1. Spectrophotometer, e.g., NanoDrop ND1000 (Thermo

and Purity Assessment Scientific, Wilmington, DE, USA).
2. Agilent 2100 Bioanalyzer (Agilent Technologies Inc., CA, USA).
3. Agilent RNA 6000 Nano Kit (Agilent Technologies Inc., CA,
USA, #5067-1511).
2.4 High-Throughput 1. Microarray platforms.

Analysis (a) Access to a microarray core facility for hybridization of the
samples to Illumina microarrays (ask for site-specific
instructions for sample preparation, etc.).
2. RNA expression profiling using Illumina HumanHT-12

v4 Expression BeadChips.
(a) TargetAmp-Nano Labeling Kit for Illumina Expression
BeadChip (TAN07924, Epicentre).
(b) Illumina HumanHT-12 v4 BeadChips (BD-103-0204).
3. DNA Methylation profiling using Illumina Infinium
HumanMethylationEPIC BeadChips.
(a) Zymo EZ DNA Methylation Kit (D5001, Zymo Research).
(b) Illumina Infinium HumanMethylationEPIC BeadChip
(WG-317-1002).
4. Next-generation sequencing.
(a) Access to a NGS core facility for the analysis of the samples
on Illumina sequencing machines (ask for site-specific
instructions for sample preparation, etc.).
RNA Sequencing (Illumina platform).
(b) Illumina TruSeq RNA Sample Prep Kit v2 (Illumina, CA,
USA, #RS-122-2001, RS-122-2002).
(c) Illumina NextSeq500 or HiSeq2500/3000/4000 System
(Illumina, CA, USA).
Small RNA Sequencing (Illumina platform).
(d) Illumina TruSeq Small RNA Library Prep Kit
(#RS-200-0012).
(e) Illumina NextSeq500 or HiSeq2500/3000/4000 System
(Illumina, CA, USA).
3 Methods
3.1 Source Materials 1. Harvest a tissue sample according to standard clinical protocols;
place in a 1.5 mL polypropylene tube with 1 mL of RNAlater
3.1.1 Tissue Samples
(see Note 1).
Gingival/Mucosal Tissue 2. Hold at 4 C overnight, subsequently drain and freeze at
Harvesting and Processing 80 C until further processing (see Note 2).
Tissue Disruption 1. For purification of total RNA of a high number of samples in

and Homogenization parallel we prefer using a commercial kit, e.g., AllPrep DNA/
RNA/Protein Kit, Qiagen (see Note 3). The following steps
refer to the manufacturers instructions. If you prefer the alter-
native method by phenolchloroform extraction (see Note 4),
add 1 mL of TRIzol reagent per 100 mg of sample to your
sample and proceed with step c.
2. Prepare lysis buffer before starting the homogenization: add
10 L of 2-mercaptoethanol in 1 mL of RLT buffer.
3. Weigh the frozen tissue sample with a precision balance and

transfer to a 5 mL, precooled polypropylene tube. Work quickly
to avoid thawing during weighing. Add the manufacturers
recommended amount of lysis buffer to each sample and store
on ice.
4. Afterwards, thoroughly homogenize the tissue (three episodes
of 20 s each at full speed see Notes 57). Clean tip with RNase-
free distilled water after each sample to avoid
cross-contamination.
5. Aliquot the homogenized sample in 2 mL nuclease-free tubes
(for phenolchloroform extraction: divide into two tubes),
take one aliquot for further processing and freeze the others at
80 C to keep them as backup. For Column-Extraction:
take care to not overload the binding capacity of the columns
by using more than the recommended amount of starting
material/lysate for further purification.
3.1.2 Primary 1. Phlebotomize according to standard protocols and sample

Mononuclear Cells Isolated approximately 8 mL of blood each into each of four Vacutainer
from Patient Blood CPT tubes (see Note 8).
Blood Collection
Blood Cell Separation 1. Centrifuge tubes for 15 min at 1000 g with centrifuge brake
turned off (see Note 9).
2. Carefully collect the white layer of peripheral blood monocytic
cells using a 5 mL pipette and place in 50 mL Falcon tube.
3. Wash with 50 mL of ice-cold PBS (10 min, 300 g, 4 C),
remove supernatant by aspiration.
4. Wash with 15 mL of ice-cold PBS (10 min, 300 g, 4 C).
5. Resuspend pellet in 10 mL of ice-cold PBS, count the cells
with the hemocytometer.
6. Centrifuge (5 min, 300 g, 4 C), resuspend in PBS to a den-
sity of 107 cells/80 L, keep on ice.
7. Add 20 L of CD4 (or CD14 beads, respectively) per 80 L of
cell suspension, incubate on ice for 15 min.
8. Wash with 10 mL of ice-cold PBS, centrifuge (5 min, 300 g,
4 C), resuspend in 500 L of ice-cold PBS.
9. Place MACS column in multi-stand, wash column twice with
5 mL of autoMACS solution.
10. Apply cell suspension to MACS column, wash twice with 5 mL
of autoMACS, collect flow-through and label tube as CD14.
11. Remove MACS column from separator stand, place in 15 mL
Falcon tube and elute twice with 5 mL of ice-cold autoMACS
using the plunger provided and label tube as CD14+.
12. Using the CD14 cell suspension, proceed accordingly for

CD4 (beginning from step 7).
13. Pellet each cell population by centrifugation (10 min, 300 g,
4 C).
14. Add 0.5 mL of lysis buffer (Qiagen) or (for alternative proto-
col) TRIzol reagent to pellet (approximately 3 volumes of pel-
let), mix well by pipetting. Samples can be stored at 80 C, if
needed (see Note 10).
3.1.3 Cultured Cells 1. Generate dendritic cells from mouse bone marrow progenitor
cells as described elsewhere [17].
Primary Culture
2. Resuspend prepared and washed cells in DMEM (+10 % FCS,
+1 % P/S) supplemented with 20 ng/mL rGM-CSF and
seed 1 106 cells/mL per well into a 24-well plate. Incubate
for 6 days (37 C, 5 % CO2). Wash with 1 mL DMEM (+5 %
FCS, without antibiotics) and replace culture medium every
2 days.
3. For further processing, cells should be 7080 % confluent, oth-
erwise repeat washing step and incubate for 2 more days.
4. For challenge with a periodontal pathogen.
Remove supernatant and wash cells with 1 mL of DMEM
(+5 % FCS, without antibiotics). Add 200 L trypsinEDTA
and incubate plate at 37 C for approximately 5 min. Stop
trypsinization with 600 L of DMEM (+5 % FCS, without
antibiotics) and transfer cell suspension in a 1.5 mL poly-
propylene tube. Centrifuge for 10 min at 280 g.
For infection, resuspend pelleted cells in DMEM (+5 %
FCS, without antibiotics). Adjust cell concentration
adapted to your MOI (multiplicity of infection) and seed
500 L cell suspension in a 24-well plate.
Scrape 2-3 bacterial colonies (see Note 11) off the agar plate
and resupend in DMEM (+5 % FCS, without antibiotics).
Adjust cell concentration with a spectrophotometer (OD
0.7 109 bacteria cells) according to your MOI. Add to the 24
well plate and incubate for 24 h at 37 C, 5 % CO2.
Cell Harvesting 1. Subsequently, remove supernatant and wash cells three times
and Processing with DMEM (+10 % FCS, +1 % P/S). Collect cells with a cell
scraper and transfer into a 1.5 mL polypropylene tube.
Centrifuge for 10 min at 280 g.
2. Add 500 L of Qiagen lysis buffer or TRIzol reagent and
carry on with Subheading 3.4.2. Please note that you use
half the declared amount of substances for RNA extraction
(see Note 12).
3.2 Extraction (a) Centrifuge the lysate at 13,000 g for 5 min, at 28 C to

and Purification remove insoluble material. For extraction of different fractions
of Nucleic Acids (and strictly follow the instructions of the manual. Work as quickly
Protein) as possible in a clean workspace (see Notes 13 and 14).
3.2.1 Purification
(b) Prepare all buffers before starting purification: 8 mg DTT per
by a Commercial Kit
1 mL of ALO, the recommended amount of ethanol (96
100 %) in Buffer RPE, AW1, and AW2.
(c) Place columns in a new 2 mL collection tube.
(d) Transfer supernatant to an AllPrep DNA spin column and spin
for 30 s at 10,000 g. Store columns placed in a new 2 mL
collection tube at 4 C for later DNA purification.
(e) Start with RNA-purification: Add the recommended amount of
ethanol (96100 %) to the flow-through and mix gently by
pipetting up and down. Transfer 700 L to a RNeasy spin column,
centrifuge for 15 s (10,000 g) and repeat this step until the
entire sample has passed through the membrane. Collect the
flow-through for later protein/miRNA purification.
(f) Wash column first with 700 L of Buffer RW1 and afterwards
with 500 L of RPE (spin 15 s at 10,000 g) . Discard the
flow-through after each centrifugation step. Lastly, add 500 L
of RPE and centrifuge for 2 min to dry the membrane. For
eliminating all disturbing remains of buffer, spin additionally in
a new collection tube for 1 min at 10,000 g.
(g) Place column in a clean 1.5 mL collection tube and elute RNA
by adding 3050 L RNase-free water, spin for 1 min at
10,000 g.
(h) miRNA-Purification: add 1 volume of Buffer APP to the
flow-through from step e. Mix well and incubate at room tem-
perature. Stop here and follow the instructions of the supple-
mentary protocol for the miRNA purification by using the
RNeasy MinElute Cleanup Kit, Qiagen (see Note 15).
(i) Centrifuge for 10 min at full speed. Store the pellet for later
protein precipitation. Transfer the supernatant in a 5 mL tube
and add 1 volume of ethanol (100 %). Mix well by pipetting.
Transfer sample step by step to the RNeasy MinElute spin
column, centrifuge for 15 s at 10,000 g and discard the
flow-through after each step.
(j) Place column in a new 2 mL collection tube and wash the
membrane with 500 L of Buffer RPE and subsequently with
500 L of 80 % ethanol, discard the flow-through.
(k) Place again in a new 2 mL collection tube, open the lid and
centrifuge at full speed for 5 min. Discard the flow-through.
(l) Place the column in a new 1.5 mL tube and elute miRNA
with 14 L RNase-free water and spin 1 min. at full speed.
Pay attention that you add the water directly to the center of
the membrane.
(m) For protein precipitation: add 500 L of 70 % ethanol to the
pellet of step i, centrifuge for 1 min at full speed and decant
supernatant. Dry the pellet for 510 min at room
temperature.
(n) Dissolve pellet in 100 L ALO (see Note 16).
3.2.2 Alternative Protocol (a) Add 100 L of chloroform to the sample (volume 500 L in a
(PhenolChloroform fume hood (see Note 17), shake vigorously for approximately
Extraction) 15 s, vortex for 1 min and incubate at room temperature for
2 min. Spin (15 min, 12,000 g, 4 C) to separate the aqueous
and organic layers.
(b) Carefully transfer the upper, colorless aqueous phase contain-
ing the RNA into a new 1.5-mL tube using a pipette with a
1 mL tip, add 4 L of glycogen (5 g/mL) and 250 L of
ethanol (see Note 18). Mix by shaking for 15 s, incubate for
10 min on ice and spin (10 min, 12,000 g, 4 C). From this
point on, all work can be carried out on a laboratory bench,
ideally devoted to RNA work only. Use of a hood further
reduces the risk of contamination with nucleases. Clean work-
space and instruments with RNase Zap according to the manu-
facturers instructions.
(c) Keep the whitish interphase and the red phenolchloroform
phase for the isolation of DNA and protein (overnight storage
at 4 C possible).
(d) Remove supernatant by pouring, wash the white RNA pellet
(should be clearly visible) with 500 L of 80 % ethanol (freshly
prepared from 100 % ethanol and RNase free water), spin
(10 min, 7500 g, 4 C).
(e) Remove supernatant by pipetting, invert the tube to allow the
pellet to dry (for approximately 10 min) (see Notes 1820).
(f) Resuspend the pellet carefully in 100 L of RNase free water
(see Note 21). The extracted total RNA can be stored at
80 C at this point, if needed.
(g) To ensure high quality of the obtained RNA, further purify
using the Qiagen RNeasy Mini Kit. To ensure sufficient RNA
concentration for subsequent reactions, the volume of RNase
free water used to elute the RNA after the column purification
should be based on the type of tissue sampled and the pellet size
after the precipitation, (e.g., 20 L for smaller tissue samples and
monocytes/lymphocytes, 40 L for larger tissue samples).
(h) Prior to the column cleanup, the sample contains total RNA
including microRNAs.
(i) After the extraction of RNA, the lower phase containing DNA
and proteins is processed. First, remove all remaining aequeous
phase.
(j) Add 150 L of 100 % ethanol, invert to mix, incubate for 2 min
at RT, spin to pellet the precipitated DNA (2000 g, 5 min,
4 C).
(k) Remove supernatant for protein isolation, if desired.
(l) Wash DNA pellet with 500 L of 0.1 M sodium citrate/etha-
nol solution. Incubate for at least 30 min at RT, invert from
time to time to mix. Spin (2000 g, 5 min, 4 C), remove
supernatant, repeat wash with sodium citrate.
(m) After removing the sodium citrate supernatant, add 1 mL of
75 % ethanol, incubate for 20 min at RT with occasional mix-
ing by inversion. Spin (2000 g, 5 min, 4 C), remove
supernatant.
(n) Dry pellet (RT, 510 min see Notes 22 and 23). Resuspend in
8 mM NaOH (about 500 L for 50 mg of tissue or per 1 107
cells). Spin down insoluble material (12,000 g, 10 min, 4 C),
transfer DNA to new tube. For long-term storage, adjust pH
to 78 with HEPES.
(o) For subsequent protein isolation, add 750 L of isopropanol
to the supernatant collected after DNA precipitation (step
3.2.2 (k)), incubate (10 min, RT), spin (12,000 g, 10 min,
4 C) to pellet the protein.
(p) Remove supernatant, wash pellet with 0.3 M guanidine hydro-
chloride in 95 % ethanol for 20 min at RT. Spin (7500 g,
5 min, 4 C), remove wash solution, repeat wash twice.
(q) Add 2 mL of 100 % ethanol to pellet, incubate for 20 min at
RT, spin (7500 g, 5 min, 4 C), remove supernatant, air-dry
protein pellet.
(r) Resuspend protein pellet in 1 % SDS. Sample can be warmed
up to 50 C to support dissolving of the pellet (see Notes 24
and 25). Spin down insoluble material (10,000 g, 10 min,
4 C), transfer supernatant to new tube.
3.3 Quantitation 1. After purification measure the quality and quantity of the
and Purity Assessment obtained total RNA and DNA by spectrophotometric analysis.
Typical yields of a 30 mg gingival tissue sample (input for col-
umn extraction) are approximately 4050 g of total RNA, of
smaller tissue samples 2040 g. Yields of phenolchloroform
extraction are expected between 80 and 160 g from larger
tissue samples and 2060 g from smaller tissue samples. The
260280 ratio is typically between 1.9 and 2.1 (see Notes 26
and 27). The samples can be stored at 80 C.
2. Additionally, a further quality check based on a chip-

electrophoresis, e.g., with the Agilent 2100 Bioanalyzer, is
useful to exclude degraded RNA samples, phenol, or salt traces,
to assure the comparison of the samples as well as the repro-
ducibility of the following experiments, such as next-genera-
tion sequencing. A successful library preparation and the
following sequencing strongly depend on the purity of the
RNA. Furthermore, unnecessary expenses and effort can be
reduced that way. Corresponding to the 18S and 28S ribo-
somal subunits, clean RNA samples should have two well-
defined, sharp peaks and a 28S18S rRNA ratio of 2:1. The
RNA integrity number (RIN) of 8-10 defines respectable qual-
ity for further processing (cf. Fig. 2). (see Note 28).
3. Protein quantity is assessed by Bradford assay (see Note 29).
3.4 High-Throughput 1. Use a minimum of 50 ng (or up to 500 ng) of good quality

Analysis (RIN > 7) total RNA as input for the TargetAmp Nano label-
ing kit to produce biotinylated antisense RNA that can be
3.4.1 Microarray
hybridized to the BeadChips. This step is usually automated in
Platforms
the core facility (see Notes 30 and 31).
RNA Expression Profiling 2. Hybridize aRNA with the bead arrays in the core facility.
Using Illumina
Human-HT12 v4
Expression BeadChips
DNA Methylation Profiling 1. Perform bisulfite conversion of unmethylated cytosines into

Using Illumina Infinium uracil, whilst methylated cytosines remain unchanged, using
HumanMethylationEPIC 500 ng of genomic DNA and the EZ kit (see Note 32). This
BeadChips step is usually automated in the core facility.
2. Denature DNA and perform isothermal amplification and
subsequent fragmentation. This step is usually automated in
the core facility.
3. Hybridize the fragmented, bisulfite-converted DNA to bead
chips using an automated slide processor in the core facility.
3.4.2 Next-Generation 1. Concerning your experimental application, the decision for

Sequencing one of the available sequencing technologies should be well-
considered, e.g., the Illumina TruSeq RNA Sample Preparation
RNA Sequencing (Illumina
Kit v2 for preparing the templates sequenced by the Illumina
Platform)
NExtSeq500 orHiSeq2500/3000/4000 (see Note 33).
2. Dilute purified, high-quality total RNA in distilled nuclease-
free water to obtain an amount of 200 ng to a final volume of
50 L and subsequently follow the sample preparation proto-
col very closely (see Notes 3436).
3. Before pooling the libraries, a supplementary quality check is
highly advisable, e.g., with the Agilent 2200 TapeStation
a T1081
[FU]
40
35
30
25
20
15
10
S
18
28
25 200 500 1000 2000 4000 [nt]
Overall Results for sample 9 : T1081

RNA Area: 164,2 RNA Integrity Number (RIN): 10 (B.02.08)
RNA Concentration: 87 ng/l Result Flagging Color:
rRNA Ratio [28s / 18s]: 2,4 Result Flagging Label: RIN:10
Fragment table for sample 9 : T1081

Name Start Size [nt] End Size [nt] Area % of total Area
18S 3.793 5.381 26,9 16,4
28S 7.791 9.407 64,1 39,0
b 83,2
[FU]
10
0
S
18
25 200 500 1000 2000 4000 [nt]
Overall Results for sample 3 : 83,2

RNA Area: 159,7 RNA Integrity Number (RIN): 2.3 (B.02.08)
RNA Concentration: 113 ng/l Result Flagging Color:
rRNA Ratio [28s / 18s]: 0,0 Result Flagging Label: RIN:2.30
Fig. 2 Example for RNA-quantitation by agilent 2100 bioanalyzer (a) Two sharp peaks for the 18S and 28S
subunit are visible. The RNA integrity number (RIN) of 10 indicates a clean and undegraded sample of high
quality, which is useful for further processing. (b) Degraded RNA-sample. High background and absent peaks
represent an inadequate quality which is reflected in a low RIN, here being 2.3
System (see Note 37). A sole band of approximately 260 bp is

expected in a pure, prepared sample.
4. Perform paired end RNA sequencing reaction in Illumina
HiSeq machine in the core facility.
5. The recommended number of paired-end reads for gene
expression profiling is at least 30 million.
Small RNA Sequencing 1. When using total RNA including the small RNA fractions, up
(Illumina Platform) to 1 g (in 5 L of nuclease-free water) of material can be used
(See Notes 38 and 39) as input for the TruSeq library prep kit. If using already puri-
fied small RNAs, 1050 ng are recommended.
2. Perform library preparation following the manufacturer's rec-
ommendations. This step is usually automated in the core
facility.
3. Perform single-read RNA sequencing reaction in Illumina
HiSeq machine in the core facility.
4. The recommended number of reads depends on the applica-
tion. For miRNA expression profiling, 12 million single-end
reads are considered sufficient. For the discovery of novel small
RNAs, significantly higher read numbers (1020 million) are
required.
4 Notes
1. Barcoded tubes can simplify sample identification and storage

of a high number of samples, e.g., Cryo.s Greiner Bio-One,
Germany, #F071080.
2. Alternatively, the samples can be snap-frozen in liquid nitrogen
chair-side and subsequently directly transferred to 80 C. Keep
in mind that handling liquid nitrogen imposes a safety hazard
in a clinical setting. Thus, as RNAlater treatment has proven
to reliably preserve sample RNA, we prefer to use RNAlater
over liquid nitrogen. In addition, keeping RNA stable for up
to several days at ambient temperature, it also allows shipment
of tissue samples from different study centers to a processing
center.
3. We prefer the simultaneous extraction of total RNA, genomic
DNA and protein from the same piece of tissue to allow for the
analysis of different -omes in the same samples. Of course,
there are kits for solely extracting total RNA fractions available
as well.
4. The organic extraction by phenolchloroform is a commonly
used method to isolate total RNA from tissues and is very
cost-efficient. However, it is more time-consuming and needs
more practical knowledge to obtain a high amount of intact

and purified RNA of comparable quality. As phenol has nearly
the same absorbance spectrum of RNA contaminations are
difficult to detect and require additional quality checks. Phenol
(included in TRIzol reagent) is toxic by inhalation, and chloro-
form is considered a potential carcinogen. Thus, a fume hood
(e.g., Safeaire (Fisher Hamilton, Two Rivers, WI, USA) or good
ventilation and appropriate personal safety measures (gloves,
safety glasses, protective clothing) are imperative when handling
these components. Furthermore, phenol-containing waste must
be collected and disposed of separately in many countries.
5. The complete disintegration of the tissue samples by homog-
enization is crucial to obtain optimal RNA yields. Residual
pieces lead to a clogging of the column pores. Check for
remaining intact tissue particles approximately 2 min after
homogenization. If needed, continue to process the sample
until completely homogenized.
6. We advise against the utilization of a sonicator for the lysis of
tissue samples since the considerable heat generated by this
device can result in enhanced RNA degradation.
7. This protocol is optimized for rather large and fibrous tissue
samples (i.e., interdental gingival papillae). To reliably process
considerably smaller samples, we recommend the use of a mor-
tar and pestle to finely pulverize the sample after shock-freezing
in liquid nitrogen. The pulverized sample can be resuspended
in lysis buffer and further processed according to the manufac-
turers instructions.
8. Instead of using BD Vacutainer CPT tubes already containing
Ficoll for cell separation, heparin whole blood can also be
diluted 1:1 with PBS and combined with 3 mL of Ficoll in a
15-mL Falcon tube.
9. Centrifugation without brake is critical for the preparation of
peripheral blood monocytic cells (PBMC). Check in advance,
as not all standard centrifuges bear this feature.
10. The isolation of monocytes and lymphocytes should be per-
formed from freshly collected blood. If needed, the blood sam-
ple can be stored at 4 C for several hours before processing,
but the sample should not be frozen. In case of emergency, the
cells in the white layer isolated by gradient centrifugation
(peripheral blood monocytic cells, PBMC) can be frozen in
standard cell culture freezing medium [e.g., RPMI (Gibco/
Invitrogen) + 10 % fetal bovine serum (Gibco) + 10 % DMSO
(Sigma-Aldrich, St. Louis, MO, USA)], but a significant loss in
RNA yield must be expected.
11. The infection of murine DCs by a periodontal pathogen is a
typical cell culture model for periodontal infections that allows
to test the influence of specific genes and pathways using cells

isolated from knockout or transgenic animals. Alternatively,
periodontal cells of human origin could be used, possibly after
modification using transfection or viral transduction with
shRNA.
12. Cells can also by lysed directly on the plate using 1 mL TRIzol
or lysis buffer for a 100 mm plate.
13. When processing the samples, it is highly advisable to wear
gloves (and change them frequently) to avoid contaminations
with exogenous nucleases. Further, we recommend the use of
certified nuclease-free plasticware and filtered tips. Surfaces
and instruments should be treated with an RNase removal
fluid, e.g., RNase Away (Sigma Aldrich, #83931). If possible,
all RNA-related work should be carried out in dedicated work-
space, preferably a hood or a PCR enclosure (e.g., Labconco
PCR enclosure, Labconco, Kansas City, MI, USA). The sam-
ples should be placed on ice at all times, except when specifi-
cally instructed otherwise. Instead of using ice, we found the
use of laptop coolers (e.g., Nalgene, Fisher, Rochester, NY,
USA) more convenient.
14. Pay attention that you strictly use the final concentrations of
buffers and alcohol as described in the protocol of the kit.
Alterations could disturb a successful binding of the nucleic
acids to the column.
15. If you are interested in simultaneously purifying a fraction of
miRNA, you could combine the AllPrep DNA/RNA/Protein
Mini Kit with RNeasy MinElute Cleanup Kit (Qiagen,
Germany) with a supplementary protocol described by the
manufacturer. The purified fraction contains miRNA as well as
other small RNAs. Of course you can continue with step 3.2.1
(m) if you do not need a fraction of small RNA.
16. If protein is still insoluble, increase amount of Buffer
ALO. Alternatively solve the pellet in 5 % (w/v) SDS or in 8 M
urea. Sonication of the sample (510 s, 60 %, cooling between
each cycle) may be helpful.
17. Phenol (included in TRIzol reagent) is toxic by inhalation, and
chloroform is considered a potential carcinogen. Thus, a fume
hood (e.g., Safeaire (Fisher Hamilton, Two Rivers, WI, USA)
or good ventilation and appropriate personal safety measures
(gloves, safety glasses, protective clothing) are imperative when
handling these components.
18. The precipitation can also be carried out with isopropanol,
resulting in a lower salt content of the pellet. However, we
recommend the use of ethanol, since isopropanol pellets are
more difficult to see and handle. The salts are subsequently
removed by the column-based purification step. Furthermore,
the use of round-bottomed 2-mL tubes (instead of 1.5-mL)

improves the visibility and handling of the obtained pellet.
19. Pay attention not to lose the pellet when inverting the tube.
20. Do not overdry the RNA pellet, as a completely dried out pel-
let is transparent and far more difficult to see. It may also fail to
dissolve thoroughly in subsequent steps.
21. Caution should be taken to completely dissolve the nucleic
acid pellet after the drying step by vigorous pipetting for
approximately 1 min/sample.
22. DNA stored in sodium citrate/ethanol solution can be stored
at RT for up to two hours, DNA in 75 % ethanol up to several
months at 4 C.
23. Do not overdry DNA pellet, e.g., by SpeedVac as it becomes
very difficult to get the DNA in solution again which normally
goes along with DNA degradation.
24. Sometimes, the protein pellet is difficult to dissolve in 1 %
SDS. Alternatively, Hummon and coworkers proposed several
alternative solvents, e.g.,10 M urea, 2 % diethylamine, or 1 %
SDS and 62.5 mM sarkosyl at pH 8.08.8 [21].
25. Samples can/should be stored at 80 C.
(a) After harvesting (drained tissue sample or homogenized
sample in TRIzol reagent).
(b) After extraction of total RNA, miRNA, DNA (DNA sam-
ples can be stored at 4 and 20 C) and protein (protein
samples in SDS can be stored at 20 C).
(c) To collect a number of samples over time. Simultaneous
processing of 68 samples has proven to be safe and
efficient.
26. If the A260A280 ratio is not in the range of 1.92.1 after total
RNA isolation, consider re-cleaning the sample with the
Qiagen kit. A ratio lower than 2.0 indicates a contamination of
protein or phenol. The A260A230 ratio should also be calcu-
lated to check for a contamination by chaotropic salts and
organic compounds.
27. To increase the nucleic acid concentration in the sample, we
recommend to use a vacuum centrifuge (e.g., Vacufuge,
Eppendorf, Hamburg, Germany) at 4 C to pellet the nucleic
acid and resuspend it in an appropriate volume of RNase-free
water.
28. An alternative measure to judge the quality of the total RNA
preparations is to run a formaldehyde 1 % agarose gel and
check the 28S rRNA band (~4.5 kB) and 18S rRNA band
(~1.9 kB). The 28S band should be twice the intensity of the
18S band.
29. For Bradford assays for protein quantification, the total SDS
concentration must be <0.1 %.
30. If RNA yields are significantly lower than 50 ng, the TargetAmp
Pico or another kit performing a double Eberwine reaction
during the labeling process can be utilized. Whilst the Eberwine
reaction is a robust amplification, it still generates a 3-bias
with a significantly smaller size distribution, especially when
performed twice [22].
31. The size distribution of the generated biotinylated aRNA can
be checked using a Bioanalyzer.
32. The bisulfite conversion step is very sensitive to DNA quality
and phenol contamination of the input DNA. Consider a col-
umn cleanup step.
33. Several library kits and platforms for RNA sequencing are
available on the currently growing market; further technolo-
gies are e.g.,SOLiD/Applied Biosystems, Ion Torrent NGS/
Thermo Fisher Scientific or Roche 454 Life Science. Focused
on a specific question, a sequencing experiment should be
thoroughly designed. For example, the choice of sequencing
depth should comply with the experimental aim; for experi-
ments with gingival tissue, we recommend a sequencing depth
of >50 million paired-end reads to ensure an adequate sensitiv-
ity for lower gene expression levels.
34. Work with a multichannel pipette and high precision to avoid
unnecessary manipulations which would add up from step to
step. If you use Illumina preparation kits for the first time we
suggest to start with an input of 12 samples.
35. For handling beads, there are some aspects to consider in order
to purify samples well from rRNA contained therein and in
order to avoid an additional loss of nucleic acid: The magnetic
beads should be warmed up to room temperature and vor-
texed thoroughly immediately before each handling. Use a
suitable magnetic stand for your 96-well PCR plate. After
incubation on the magnetic stand, make sure that the fluid is
completely clear and that a compact pellet has formed before
you continue. Be careful not to disturb the magnetic beads
with tips while pipetting.
36. Stops are possible after second strand synthesize, end repair,
adapter ligation, and the enrichment of DNA fragments.
Covered samples can be stored at 20 C for up to 7 days.
37. We do not recommend a routine quantitation with qPCR as
suggested in the manufacturer's instructions. Though repre-
senting a reliable help for the implementation of sequencing
methods in a laboratory, processing many samples simultane-
ously is rather inconvenient and costly.
Alternatively, another reliable and cheap method for

quantitation is a measurement by fluorescence, e.g., with the
Qubit dsDNA Broad-Range (BR) Assay Kit (Thermo Fisher
Scientific, USA).
38. Most comments for RNA sequencing (see above) are also valid
for small RNA sequencing applications.
39. Safe stopping points are after the reverse transcription and
amplification step, and after the normalization of the libraries
to 2 nM. Samples at these points can be stored for up to 7 days
at 20 C.
Acknowledgments
This work was supported by grants from the German Society for
Periodontology (DG PARO) and the German Society for Oral and
Maxillofacial Sciences (DGZMK) to M.K. and by grants from
NIH/NIDCR (DE015649, DE021820, and DE024735) and by
an unrestricted gift from Colgate-Palmolive Inc. to P.N.P.
References
1. Chung CH, Bernard PS, Perou CM (2002) 7. Thornton S, Sowders D, Aronow B, Witte DP,
Molecular portraits and the family tree of Brunner HI, Giannini EH, Hirsch R (2002)
cancer. Nat Genet 32(Suppl):533540 DNA microarray analysis reveals novel gene
2. Quackenbush J (2006) Microarray analysis expression profiles in collagen-induced arthri-
and tumor classification. N Engl J Med tis. Clin Immunol 105:155168
354:24632472 8. van der Pouw Kraan TC, van Baarsen LG,
3. Hoshida Y, Villanueva A, Kobayashi M, Peix J, Rustenburg F, Baltus B, Fero M, Verweij CL
Chiang DY, Camargo A, Gupta S, Moore J, (2007) Gene expression profiling in rheuma-
Wrobel MJ, Lerner J, Reich M, Chan JA, tology. Methods Mol Med 136:305327
Glickman JN, Ikeda K, Hashimoto M, Watanabe 9. Burke W (2003) Genomics as a probe for dis-
G, Daidone MG, Roayaie S, Schwartz M, Thung ease biology. N Engl J Med 349:969974
S, Salvesen HB, Gabriel S, Mazzaferro V, Bruix 10. Izuhara K, Saito H (2006) Microarray-based
J, Friedman SL, Kumada H, Llovet JM, Golub identification of novel biomarkers in asthma.
TR (2008) Gene expression in fixed tissues and Allergol Int 55:361367
outcome in hepatocellular carcinoma. N Engl 11. Demmer RT, Behle JH, Wolf DL, Handfield M,
J Med 359:19952004 Kebschull M, Celenti R, Pavlidis P, Papapanou
4. Haslett JN, Kunkel LM (2002) Microarray PN (2008) Transcriptomes in healthy and
analysis of normal and dystrophic skeletal mus- diseased gingival tissues. J Periodontol 79:
cle. Int J Dev Neurosci 20:359365 21122124
5. Colangelo V, Schurr J, Ball MJ, Pelaez RP, 12. Jonsson D, Ramberg P, Demmer RT, Kebschull
Bazan NG, Lukiw WJ (2002) Gene expression M, Dahlen G, Papapanou PN (2011) Gingival
profiling of 12633 genes in Alzheimer hippo- tissue transcriptomes in experimental gingivitis.
campal CA1: transcription and neurotrophic J Clin Periodontol 38:599611
factor down-regulation and up-regulation of 13. Kebschull M, Guarnieri P, Demmer RT,
apoptotic and pro-inflammatory signaling. Boulesteix AL, Pavlidis P, Papapanou PN
J Neurosci Res 70:462473 (2013) Molecular differences between chronic
6. Haroutunian V, Katsel P, Schmeidler J (2009) and aggressive periodontitis. J Dent Res
Transcriptional vulnerability of brain regions in 92:10811088
Alzheimers disease and dementia. Neurobiol 14. Papapanou PN, Behle JH, Kebschull M,
Aging 30:561573 Celenti R, Wolf DL, Handfield M, Pavlidis P,
Demmer RT (2009) Subgingival bacterial col- odontitis phenotypes. J Dent Res 93:
onization profiles correlate with gingival tissue 459468
gene expression. BMC Microbiol 9:221 19. Stoecklin-Wasmer C, Guarnieri P, Celenti R,
15. Kebschull M, Papapanou PN (2011) Demmer RT, Kebschull M, Papapanou PN
Periodontal microbial complexes associated (2012) MicroRNAs and their target genes in
with specific cell and tissue responses. J Clin gingival tissues. J Dent Res 91:934940
Periodontol 38:1727 20. Papapanou PN, Sedaghatfar MH, Demmer
16. Kramer B, Kebschull M, Nowak M, Demmer RT, Wolf DL, Yang J, Roth GA, Celenti R,
RT, Haupt M, Korner C, Perner S, Jepsen S, Belusko PB, Lalla E, Pavlidis P (2007)
Nattermann J, Papapanou PN (2013) Role Periodontal therapy alters gene expression of
of the NK cell-activating receptor CRACC peripheral blood monocytes. J Clin Periodontol
in periodontitis. Infect Immun 81:690696 34:736747
17. Nowak M, Kramer B, Haupt M, Papapanou 21. Hummon AB, Lim SR, Difilippantonio MJ,
PN, Kebschull J, Hoffmann P, Schmidt-Wolf Ried T (2007) Isolation and solubilization of
IG, Jepsen S, Brossart P, Perner S, Kebschull M proteins after TRIzol extraction of RNA and
(2013) Activation of invariant NK T cells in DNA from patient material following prolonged
periodontitis lesions. J Immunol 190: storage. Biotechniques 42(467-470):472
22822291 22. Spiess AN, Mueller N, Ivell R (2003) Amplified
18. Kebschull M, Demmer RT, Grun B, Guarnieri RNA degradation in T7-amplification methods
P, Pavlidis P, Papapanou PN (2014) Gingival results in biased microarray hybridizations.
tissue transcriptomes identify distinct peri- BMC Genomics 4:44
Chapter 19
Differential Expression and Functional Analysis of High-

Throughput -Omics Data Using Open Source Tools
Moritz Kebschull, Melanie Julia Fittler, Ryan T. Demmer,
Abstract
Today, omics analyses, including the systematic cataloging of messenger RNA and microRNA sequences
or DNA methylation patterns in a cell population, organ, or tissue sample, allow for an unbiased, compre-
hensive genome-level analysis of complex diseases, offering a large advantage over earlier candidate gene
or pathway analyses. A primary goal in the analysis of these high-throughput assays is the detection of
those features among several thousand that differ between different groups of samples. In the context of
oral biology, our group has successfully utilized omics technology to identify key molecules and pathways
in different diagnostic entities of periodontal disease.
A major issue when inferring biological information from high-throughput omics studies is the fact
that the sheer volume of high-dimensional data generated by contemporary technology is not appropri-
ately analyzed using common statistical methods employed in the biomedical sciences.
In this chapter, we outline a robust and well-accepted bioinformatics workflow for the initial analysis
of omics data generated using microarrays or next-generation sequencing technology using open-source
tools. Starting with quality control measures and necessary preprocessing steps for data originating from
different omics technologies, we next outline a differential expression analysis pipeline that can be used
for data from both microarray and sequencing experiments, and offers the possibility to account for ran-
dom or fixed effects. Finally, we present an overview of the possibilities for a functional analysis of the
obtained data.
Key words Periodontal disease, Gene expression, Transcriptome, microRNA, DNA methylation,
Microarray, Next-generation sequencing, Gingiva, Differential expression analysis, Functional groups
1 Introduction
omics analyses such as the whole-genome assessments using

microarrays or next-generation sequencing outlined in Chapter 18
generate a large number of observations in relatively few samples.
It is generally of major interest to assess which of these features
differ between subgroups of samples defined a priori on the basis
of relevant characteristics, e.g., clinical diagnosis, experimental
treatment, etc. When performing these differential expression
327
analyses of omics data, the researcher is inevitably confronted

with the fact that high-dimensional data sets are difficult to ana-
lyze using traditional statistical approaches. Specifically, the analysis
needs to account for thousands of statistical tests performed simul-
taneously. Additional corrections may be necessary for specific fea-
tures of clinical samples. The amount of resulting data generated
requires unprecedented computational resources in terms of pro-
cessing power, memory, and disk space.
Our group has considerable experience in the analysis of high-
throughput datasets in the context of periodontal infections, e.g.,
the expression profiles or periodontal health and disease [18] or
experimental gingivitis [9].
This chapter describes how to process the raw data provided
by a core facility after hybridization with microarrays or massively
parallel sequencing. We elaborate on typical quality assessments
and preprocessing steps, and then proceed to a common differen-
tial expression analysis workflow using the R/Bioconductor
framework [10] and the limma library [11, 12]. Importantly,
both microarray-based expression values and gene counts created
based on sequencing resultsafter transformation into continu-
ous valuescan be used as source data for this workflow. Using
limma, it is possible to perform a differential expression analysis
correcting for both random effectssuch as the individual sub-
ject in cases where several biologically and statistically dependent
samples originate from the same individualand fixed effects,
e.g., the study center, the surgeon harvesting a biopsy, race and
ethnicity of the subject, or the level of disease severity of the par-
ticular tissue sample as a continuous variable. In a similar fashion,
the library allows not only the assessment of differential expres-
sion in two or more defined groups, but also the identification of
genes that differ significantly in relation to continuous variables,
such as periodontal probing depth, or levels of subgingival peri-
odontal bacteria.
It is important to realize that there is a wealth of software pack-
ages available for the analysis of both array and sequencing datas-
ets, both from commercial providers and open source software. In
this chapter, we have opted to use open source software that is
both universally accessible and well established in the field, which
we have experience using in studies of periodontal cells and tissues.
However, given the rapid evolvement in this field, future modifica-
tions of the workflow are likely.
In the next chapter of this series (see Chapter 20 by Kebschull
et al. of this volume), we expand the basic analyses described here
by using machine learning algorithms on high-throughput data,
both for purposes of supervised classification of a priori labeled
samples, and for unsupervised discovery of new classes.
Differential Expression Analysis of Omics data 329
2 Materials
2.1 Hardware 1. For microarray analysis: A computer with x86-64 compatible

processor(s) running either Linux or Windows or Mac OS
X. RAM >4 GB, about 1 TB free hard drive space.
2. For next-generation sequencing data analysis: A computer with
x86-64 compatible processor(s) running Linux with as many
processor cores as possible (See Notes 1, 13), RAM >32GB,
and several TB of free hard drive space.
2.2 Software 1. The R statistical environment, including the Bioconductor

framework, and the following libraries.
(a) minfi
(b) illuminaio
(c) IlluminaHumanMethylation450kanno.ilmn12.hg19
(d) affy
(e) Rsubreads
(f) edgeR
(g) limma
(h) sva
(i) statmod
2. (Optional, but highly recommended) An integrated program-
ming environment (IDE) for R, e.g., RStudio, or a program-
ming editor, e.g., GNU Emacs/ESS.
3. (Optional, but highly recommended) A version control sys-
tem, e.g., git
4. FastQC software http://www.bioinformatics.babraham.
ac.uk/projects/fastqc/.
5. STAR aligner software [13, 14] https://github.com/alexdo-
bin/STAR.
6. Trimmomatic software [15] http://www.usadellab.org/
cms/?page=trimmomatic.
7. GSEA software [16] http://software.broadinstitute.org/
gsea/index.jsp.
8. Cytoscape [17] http://www.cytoscape.org/.
9. Enrichment Map (Cytoscape plugin) [18] http://www.bader-
lab.org/Software/EnrichmentMap.
10. ErmineJ software [19] http://erminej.chibi.ubc.ca/.
2.3 Manifests, 1. Manifest file for the HT-12 bead arrays from Illuminas website
Annotations, Genome (http://support.illumina.com/array/array_kits/humanht-12_
Files v4_expression_beadchip_kit/downloads.html).
2. The manifest for the methylation arrays is part of the

IlluminaHumanMethylation450kanno.ilmn12.hg19 R package.
3. Genome files, e.g., from Ensembl http://ftp.ensembl.org/
pub/release-84/fasta/homo_sapiens/dna/Homo_sapiens.
GRCh38.dna.primary_assembly.fa.gz.
4. Matching annotation files, e.g., from Ensembl ftp://ftp.
ensembl.org/pub/release-84/gtf/homo_sapiens/Homo_
sapiens.GRCh38.84.gtf.gz.
2.4 Targets File 1. Tab-delimited text (*.txt) or comma-separated text (*.csv) file.
2. One row per sample.
3. Has all technical information.
(a) Lab identifier.
(b) Array number (for microarray data).
(c) Position on bead array (for microarray data).
(d) Batch information.
(e) Possibly also quality information (yield, RIN, etc.).
4. And all phenotypic information of possible value, e.g. (in case
of gingival tissue biopsies).
(a) Demographics (age, gender, race, and ethnicity of study
subject).
(b) Diagnosis.
(c) Systemic conditions.
(d) Local measures of disease at the biopsy site (periodontal
probing depth, clinical attachment level, subgingival levels
of periodontal bacteria associated with the tissue biopsy).
2.5 Raw Data 1. From microarray experiment.

(a) *.idat files for all arrays run.
2. From Next-Generation Sequencing experiment.
(a) *.fastq files for all sequenced samples, de-multiplexed and
adaptor-trimmed by core facility.
3 Methods
3.1 Preprocessing (a) In R, set working directory, and load the limma and the illumi-
of Array Data naio libraries.
> setwd("~/projects/ht12")
3.1.1 HT-12
> library(limma)
Expression Arrays
> library(illuminaio)
(b) Place *.bgx manifest file for the HT-12 bead arrays from
Illuminas website (http://support.illumina.com/array/
Regular probes Negative control probes
8.0
14
7.5
log2 intensities
log2 intensities
12
7.0
10
6.5
8
6.0
6
1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Arrays Arrays
Fig. 1 Boxplots of HT-12 expression array signal intensity before normalization. Note that the very dim array
#6 should be removed
array_kits/humanht-12_v4_expression_beadchip_kit/down-
loads.html) and all *.idat source files in a directory, and read
them into R using limmas read.idat function.
> idatfiles = dir(pattern="idat")
> bgxfile = dir(pattern="bgx")
> raw <- read.idat(idatfiles, bgxfile)
(c) For quality control, plot the average signal intensities for regu-
lar and control probes on the arrays. Very dim arrays are indic-
ative of suboptimal hybridization and should be removed
(Fig. 1).
> pdf("boxplots_preNorm.pdf")
> par(mfrow=c(1,2))
> boxplot(log2(raw$E[raw$genes$Status=="reg
ular",]),range=0, xlab="Arrays",ylab="log2
intensities", main=\
"Regular probes")
> boxplot(log2(raw$E[raw$genes$Status=="neg
ative",]),range=0, xlab="Arrays",ylab="log2
intensities", main\
="Negative control probes")
> dev.off()
(d) To support these observations, check for the proportion of
probes with an expressed callthey should be fairly similar.
> propexp <- propexpr(raw)
(e) Read the targets file.
# required for the read.AnnotatedDataFrame
# function
> library(affy)
> targets <- read.AnnotatedDataFrame(file="t
argets.csv", header=TRUE)
> targets <- pData(targets)
(f) Normalize data using quantile normalization (see Note 2).
> y <- neqc(raw)
(g) Explore similarities and dissimilarities of the samples using

multidimensional scaling (MDS) plots or hierarchical cluster-
ing (and different labels, using variables from the targets file).
These plots allow (1) to explore the data for broad differences
in expression related to the different variables, (2) to identify
possible batch effects, and (3) to detect, in combination with
the measures introduced above, arrays that did not hybridize as
planned.
> library(limma)
> pdf("MDSplots.pdf")
> par(mfrow=c(1,2))
> plotMDS(y,gene.selection="common",
labels=variable1)
> plotMDS(y,gene.selection="common",
labels=variable2)
> dev.off()
> pdf("hierClust.pdf")
> par(mfrow=c(1,2))
> d = dist(t(y$E))
> plot(hclust(d), labels = variable1)
> plot(hclust(d), labels = variable2)
> dev.off()
3.1.2 450k 1. Gather the *.idat source files for all samples, and place a targets
Methylation Arrays file (as *.csv) in the source directory.
2. To load the data into R using the minfi library [20].
# load libraries
> library(minfi)
> library(IlluminaHumanMethylation450kanno.
ilmn12.hg19)
> library(IlluminaHumanMethylation450kmanif
est)
# set working directory and load targets file
> setwd("~/projects/methylation")
> workDir <- "~/projects/methylation"
> targets <- read.450k.sheet(workDir)
# read raw data
> RGset <- read.450k.exp(targets = targets)
3. Check for bad arrays using the detection p-value (see Note 3):
# get detection p vals
> detP <- detectionP(RGset)
> failed <- detP > 0.01
> failed <- colMeans(failed)
> pData(RGset) -> pDataRGSet
> names(failed) <- pDataRGSet$Sample_Name
> write.table(failed, file="failed.txt",
sep="\t")
4. Generate an extensive quality control report for all samples:

> qcReport(RGset, pdf = "qcReport.pdf")
5. Generate beanplots for all samples (see Note 4):
> pdf(file="Beanplot.pdf", 5, 20)
> densityBeanPlot(RGset, sampNames =
pDataRGSet$Sample_Name)
> dev.off()
6. Preprocess and additional quality control using the minfiQC
function (see Note 5):
# normalize raw data
> MSet.Ill <- preprocessIllumina(RGset,
bg.correct=TRUE)
# get information on arrays minfi thinks that
are good or bad
> QCout <- minfiQC(MSet.Ill)
> pdf("QC.pdf")
> plotQC(QCout$qc)
> dev.off()
7. Convert to beta:
> ratioSet <- ratioConvert(MSet.Ill, what =
"both", keepCN = TRUE)
> MSet.Ill.genome <- mapToGenome(ratioSet)
#get beta values for each CpG island (rows)
and label the columns with the sample names
from the targets file
> beta.Ill <- getBeta(MSet.Ill.genome)
> colnames(beta.Ill) <- targets$Sample_Name
3.2 Preprocessing (a) The raw reads are usually provided as FASTQ files.
of Sequencing Data (b) First, a general quality assessment of the millions of raw reads
3.2.1 RNA Seq Data per sample should be performed. We recommend the FastQC
software, a standalone Java program available at http://www.
Quality Control bioinformatics.babraham.ac.uk/projects/fastqc/. FastQC
of the Reads lists the number and length of reads and their quality encod-
ing, and visualizes and judges several quality parameters
(Fig. 2a, b). The results can be viewed using a standard web
browser. Please note that not all failures and warnings dis-
played by FastQC are actually problematic for RNA Seq data
(see Note 69).
> fastqc sample1_read1.fq.gz
Preprocessing (a) Filtering removes entire reads below a certain quality thresh-
old (see Note 10). We recommend the trimmomatic program
http://www.usadellab.org/cms/?page=trimmomatic, a
standalone Java application, because it can filter paired end
reads and is multithreaded, i.e., fast [15]. The following
Fig. 2 FastQC examples: (a) Per base sequence quality. Note how the quality of the base calls decreases
toward the end of the reads. (b) Per base sequence content. For each position in the read, the percentage of
the four bases is plotted. Note the bias in the beginning of the read, a typical phenomenon for Illumina RNA Seq
data caused by random hexamer priming
command runs trimmomatic on a paired end sample, and

produces four output files, two paired ones where the initial
pairs are still intact after filtering, and two unpaired files contain-
ing the data from broken pairs.
> java jar trimmomatic-0.36.jar PE

threads 24 phred33 sample1_read1.fq.gz
sample1_read2.fq.gz output_paired_read1.
fq.gz output_unpaired_read1.fq.gz output_
paired_read2.fq.gz output_unpaired_read2.
fq.gz AVGQUAL:25
(b) Trimming removes bases from the end of the reads, based on
a given length and/or based on a quality threshold. The fol-
lowing command trims bases from the 3-end of the reads that
are below 25, and eventually filters the whole read when it gets
too short by the trimming.
> java jar trimmomatic-0.36.jar PE threads
24 phred33 sample1_read1.fq.gz sample1_
read2.fq.gz paired1.fq.gz unpaired1.fq.gz
paired2.fq.gz unpaired2.fq.gz TRAILING:25
MINLEN:75
In addition, in case FastQC reports adapter contamina-
tions, trimmomatic can remove those using the following
option (see Note 11).
ILLUMINACLIP:TruSeq3-PE.fa:2:30:10
(c) Repeat FastQC evaluation to assess whether the preprocessing
steps were successful.
Alignment (a) The reads are aligned to the genome using the splice-aware
to Reference Genome and very fast STAR aligner [13].
(b) STAR needs at least 32GB of memory for human genome
alignments (see Notes 1, 12, and 15).
(c) STAR is able to take advantage of multiple processing cores of
the computers processor(s). The number of cores to use is up
to 100 % of all present physical cores, oron more recent
machines that allow hyperthreadingup to 200 %. Select the
number of parallel processes using the --runThreadN
<NThreads > option (see Note 13).
(d) The alignment workflow consists of two steps, (1) the genera-
tion of genome index files, and (2) the mapping of the users
reads to the genome.
(e) Generation of index files
Create directory ./genome in STAR directory, and place
the latest ENSEMBL genome sequence in this directory.
> mkdir genome
> cd genome
> wget http://ftp.ensembl.org/pub/re-
lease-84/fasta/homo_sapiens/dna/Homo_sapi-
ens.GRCh38.dna.primary_assembly.fa.gz
> gunzip Homo_sapiens.GRCh38.dna.primary_as-
sembly.fa.gz
Create index using STAR (space requirements ~30 GB).

> STAR --runThreadN 24 --runMode genomeGenerate
--genomeDir ./ --genomeFastaFiles ./Homo_sapiens.
GRCh38.dna.primary_assembly.fa
(f) Mapping of reads.
Download annotation GTF file from the Ensembl ftp
server and place it in the ./genome folder.
> wget ftp://ftp.ensembl.org/pub/re-
lease-84/gtf/homo_sapiens/Homo_sapiens.
GRCh38.84.gtf.gz
Change to the source data directory containing the
FASTQ files and map using STAR and the previously
generated index. Specify where your genome index is
located, how many cores to use, and (for compressed
source files) to use zcat instead of cat to decompress on
the fly (see Notes 1418).
> STAR --runThreadN 24 --genomeDir ~/bin/
STAR/genome --sjdbGTFfile ~/bin/STAR/genome/
Homo_sapiens.GRCh38.84.gtf.gz --readFilesIn
./sample1-read1.fq.gz ./sample1-read2.fq.gz
--readFilesCommand zcat --outFilterType
BySJout --outFilterMultimapNmax 20 --align-
SJoverhangMin 8 -- alignSJDBoverhangMin 1
--outFilterMismatchNmax 999 --alignIntronMin
20 --alignIntronMax 10000 --alignMatesGap-
Max 1000000 --genomeLoad LoadAndKeep --out-
ReadsUnmapped Fastx
The alignment produces the following files.
Log.final.outsummary mapping statistics.
Log.outdetailed log of the run, can be used for
troubleshooting.
Aligned.out.sammain results file, all aligned reads in
SAM format.
SJ.out.tabsplice junctions.
Quantification (a) The SAM file that was produced by STAR is analyzed using the
and Normalization featureCounts function (included in the Rsubreads R library)
[21] to assign reads to genes.
> library(Rsubreads)
> counts <- featureCounts(files=sample1.
sam, annot.ext= ~/bin/STAR/genome/ Homo_
sapiens.GRCh38.84.gtf, isPairedEnd=TRUE,
isGTFAnnotationFile=TRUE, strandSpecific=2,
nthreads=24, useMetaFeatures=TRUE)
(b) The output of the featureCounts function, the object counts, is a

list containing a data frame with annotation information for all
genes and a matrix with raw gene counts for each input library.
(c) For a normalization of gene counts using a scale normalization
approach, we use the TMM normalization method [22] imple-
mented in the DGEList function of the edgeR package in R.
> library(edgeR)
> dge <- DGEList(counts=counts)
> dge <- calcNormFactors(dge)
(d) Subsequently, the read counts should be transformed using the
voom function from the limma library in R/Bioconductor
[23] (see Notes 19 and 20).
> norm <- voom(dge, design, plot=F)
(e) The normalized data can be explored using clustering or MDS
plots as described above for array datasets (Subheading 3.1,
step 7) to assess whether replicates cluster together, and
whether there are obvious batch effects that need to be cor-
rected (see below).
Small RNA Seq Data 1. Quality control, preprocessing.

(a) The assessment of raw data quality and the preprocessing
can be done using the workflow above, using the single
ended input functions. Note that extensive adapter clip-
ping is often necessary in small RNA sequencing experi-
ments (see Note 2528).
2. Alignment.
(a) The data are aligned to the genome using STAR in a simi-
lar fashion as RNA Seq reads. There are, however, some
recommended measures to address the specifics of small
RNAs (see Notes 2224).
Prohibit splicing with alignIntronMax 1.
Generate genome without GTF.
3. Quantification.
(a) Normalized counts can be acquired using featureCounts,
normalized and transformed to continuous data using
voom, as detailed above.
3.3 Differential (a) Limma. This protocol uses limma, a well-established and
Expression Analysis powerful R package originally developed for the analysis of
microarrays for data from all of the aforementioned techniques
(see Note 2528).
(b) Batch effect correction (optional). In case the preliminary analysis
using unsupervised clustering or MDS plots performed during
the preprocessing steps provides evidence for a batch effect, this
effect can be corrected before the actual differential expression
analysis. A typical batch effect would cause sequential samples

processed on a single day/array/flow cell, etc., to cluster
together, rather than replicates.
To perform a batch correction, we recommend the
ComBat procedure implemented in the R package sva.
> library(sva)
# read information about batches and the
class difference of interest from the tar-
gets file
> batch <- targets$batch
> target <- targets$target
# build model matrix
> mod <- model.matrix(~as.factor(target),
data=data)
# correct data for batches using ComBat pro-
cedure
> data_combat <- ComBat(dat=data,
batch=batch, mod=mod, numCoves=NULL, par.
prior=TRUE)
(c) Generate a table for the experimental design, using data from
the targets file.
# load the limma library
> library(limma)
> condition <- factor(targets$condition)
> design <- model.matrix(~0 + condition)
> colnames(design) <- levels(condition)
# if the need arises to correct this compar-
ison of condition for a covariate, it can
easily be added to the design (fixed factor)
# design <- model.matrix(~0 + condition +
covariate1)
# colnames(design) <- c(levels(condition),
covariate1)
(d) Introduction of a blocking factor to correct for multiple sam-
ples from the same individual (random factor).
> library(statmod)
> corfit <- duplicateCorrelation(data_combat,
design, block=targets$patient)
(e) Fit a model to the data.
> fit=lmFit(data_combat, de-
sign, block= targets$patient,
correlation=corfit$consensus)
# make contrasts, e.g. to compare healthy
and diseased samples (assuming that the var-
iable condition has the levels diseased
and healthy)
> contr <- makeContrasts(healthDisease =

diseased - control, levels=design)
#fit to model
fit2=contrasts.fit(fit, contrasts=contr)
fit2=eBayes(fit2)
# if needed, add annotation step here
# make lists of differentially expressed
genes between entities, with correction
for multiple testing using the Benjamini
Hochberg False Discovery Rate [24]
> healthDisease <- topTable(fit2, coef=1,
number=Inf, p.value=0.05, sort.by="logFC",
adjust.method="BH", lfc=0.25)
# write list to file
> write.table(healthDisease, file="results_
disease_health.txt", sep="\t")
3.4 Functional There are several open source software packages that, based on a
Analysis differential expression analysis as described above, generate lists
and/or networks of functional groups enriched in the experimen-
tal conditions. Here, we outline how to format the results from the
differential expression analysis to run basic functional analyses in
ermine [19] or GSEA [16] coupled to visualization using the
EnrichmentMap plugin [18] in Cytoscape [17]. An example of the
comparison of enriched functional groups in different clinical con-
ditions is in Fig. 3.
Regulation of Induction of apoptosis

Response to apoptosis
immune other organism Immune System Process by extracellular signals
response Cell development
Signal
Response Regulation of
transduction
to virus Apoptosis programmed
cell death
Defense
response
G protein Programmed
Response to Immune response signaling cell death
biotic stimulus
apoptosis
Response to
oxidative stress Negative regulation
Electron Negative regulation
of cellular process
transport of biological process
Organic acid
metabolic
Ectodermal process Negative regulation Regulation of
Regulation Negative regulation developmental
development
nt of transcription of transcription from RNA
Muscle of transport System process
(DNA dependent) polymerase II promoter
development process
epithelial
integrity Biopolymer
Steroid
Cellular metabolic
metabolic RNA metabolic
lipid process
process process
Epidermis metabolic
process Transcription from RNA
Digestion developmentt polymerase II promoter
Carboxylytic
metabolic Heart
process Regulation of development
Lipid transcription from RNA
metabolic polymerase II
Golgi vesicle
process Nucleic acid
transport
metabolic process enrichment in
metabolism signal transduction &
Functional map of chronic transcription
& aggressive periodontitis aggressive chronic
Fig. 3 Visualization of GSEA results using the Enrichment Map plugin in Cytoscape. Reprinted from [4] with
permission from Sage. Visualization of gene sets significantly enriched in diseased gingival tissues from
patients with chronic or aggressive periodontitis. Gene sets are depicted as nodes in a network. Color describes
the disease entity (red for AP and blue for CP), and the color intensity represents the degree of enrichment. The
size of the node represents the size of the enriched gene set, and the thickness of the connectors stands for
the degree of overlap between the nodes [18]
(a) Prepare a ranked list of features. Rank all genes by t-value.

(b) Use these data for Gene Set Enrichment Analysis (GSEA) using
the GSEAPreRanked function (when using the GSEA graphi-
cal user interface, this function can be found in the tools
pull-down menu).
(c) Import the results into Cytoscape following this tutorial http://
www.baderlab.org/Software/EnrichmentMap/Tutorial.
(d) (Alternatively) use ranked list in ErmineJ.
3.5 Upload Most journals require the submission of the raw and/or processed
to Repositories data from high-throughput experiments to online repositories.
Repositories exist in the US as well as in Europe, with differences
in the accepted data formats.
1. Array repositories.
(a) The Gene Expression Omnibus (GEO, http://ncbi.nlm.
nih.gov/geo) at NIH.
(b) ArrayExpress (http://www.ebi.ac.uk/arrayexpress) at the
European Bioinformatics Institute.
2. Sequencing data repositories.
(a) The Sequence-Read-Archive (SRA, http://ncbi.nlm.nig.
gov/sra) stores raw data and alignment information from
Illumina sequencers and other machines.
(b) In contrast, the Gene Expression Omnibus (GEO, http://
ncbi.nlm.nih.gov/geo) holds processed sequence data files.
(c) The European repository ArrayExpress only accepts sub-
missions that include the raw data plus meta data. Only the
meta data will be stored at ArrayExpress, the raw data will
be deposited at the SRA of the European Nucleotide
Archive (http://ebi.ac.uk/ena).
4 Notes
1. As an alternative to the use of local hardware, cloud computing

providers such as Amazon Web Services, Microsoft Azure, or
Google offer platforms that allow for a very flexible utilization
for bioinformatics workflows.
2. HT-12 arrays feature 12 samples per slide, thereby reducing
batch effects in comparison to array systems that only load a
single sample per chip, e.g., from Affymetrix. On the other
hand, the design with several arrays per slide may well be rea-
son for additional statistical concern. For example, for the
related Sentrix-6 Expression BeadChips (for murine samples),
a separate normalization routine was proposed to address the
specifics of the design [25].
3. The percentage of failed probes on bead arrays with good

quality is by far less than 1 %. Arrays with higher failure rate
should possibly be excluded.
4. The beanplots should show a pronounced U-shape, with a lot
of signal at 0 and 1, indicating unmethylated and hypermethyl-
ated regions. Plots with an inverse behavior should be excluded
from further analysis.
5. The minfiQC function provides a very quick overview of what
sample could be bad and should be scrutinized.
6. FastQC typically flags the Per base sequence content assessment
as a failure with Illumina RNA Seq data. The considerable bias
seen in the first bases (Fig. 2b) is caused by random hexamer
priming [26].
7. In an ideal human RNA Seq experiment, the GC content
should follow a normal distribution with a single peak at the
mean GC content of the human organism. Deviations from
this shape (Fig. 2c) are indicative of a contamination, possibly
by rRNA or other contaminantsleading to peaks on the
right-hand side. Additional peaks on the very left are often
caused by sequencing of poly-A tails. In the case of clinical
samples from human gingiva, a very considerable source of
contamination of the library is oral microorganisms. In this
case, sequences from microbes are often found among the
over-represented sequences, and can be tested by BLASTing at
http://blast.ncbi.nlm.nih.gov/Blast.cgi
8. If contamination with rRNA is suspected, tools such as
SortMeRna can be used to remove it [27]. Some groups gen-
erally recommend filtering of rRNA reads, because varying
proportions of them in the libraries will not be detectable by
measures as the percentage of aligned reads, and thereby intro-
duce a bias.
9. In contrast to the situation with DNA sequencing, sequence
duplicates will inherently be found in RNA Seq studies, because
of obvious different expression levels for different genes. It is
therefore not recommended to remove duplicates.
10. Base quality in FASTQ files is expressed in the Phred scale
that is the log10 of the probability that a base call was wrong
multiplied by -10, e.g., for a one in twenty chance (5 % = 0.05),
the score would be 13, for one in one hundred, 20, for one in
one thousand 30. Phred scores usually range from 0 to 40 and
are encoded, to save disk space, by a single ASCII character.
In recent FASTQ files, the Sanger encoding is used, with the
33rd ASCII character representing a score of zero, while
FASTQ files containing older Illumina data may well be
encoded differently. FastQC can detect the encoding used.
Generally, the quality of base calls decreases toward the end of
the read (see FastQCs Per base sequence quality graph,

Fig. 2a). It is recommended to aim for the majority of reads to
have a mean phred score of 25 or higher (better). Reads with
bad quality base calls can be addressed either by filtering
removing the entire reador trimming, where the lower
quality ends of the reads can be removed. The latter proce-
dure preserves a read that can later be aligned.
11. Note that the order of the options given to trimmomatic in the
command line mattersadapter clipping should be done before
all other steps to avoid disguising adapters by trimming.
12. Alignment of NGS data is computationally intensive, the STAR
aligner uses a lot of RAM to provide considerable speedups in
comparison to older software like Tophat.
13. Consider to limit the number of threads to ~80 % of those
available in the system to allow for other processes to be able
to run efficiently.
14. One big advantage of the STAR aligner is the so called soft
clipping functionality. In contrast to other aligners like bowtie
that try to align a read end-to-end, STAR performs a local
alignment base-by-base, until a threshold of mismatches is
reached. Thereby, adapters, poor quality sequencing tails at the
end of reads can be removed.
15. Using the --genomeLoad option, STAR can share the genome
index data stored in the main memory between several annota-
tion processes (shared memory concept), reducing the footprint
of the aligner when used for several samples in parallel.
16. If you suspect microbial contamination in your samples, the
nonaligned reads can be preserved using the --outReadsUn-
mapped Fastx option and tested for alignment with nontarget
species.
17. As an alternative to the alignment to the genome, the reads
could also theoretically be aligned to a transcriptome, e.g., after
an assembly with Trinity [28] following this protocol [29].
18. A new trend in RNA-Seq analysis is the use of pseudoalign-
ment engines, such as Kallisto, that can quantify abundances of
transcripts without the need for alignment [30], thereby mas-
sively reducing the computational demands of the analysis
workflow.
19. The voom transformation routine assumes that genes with zero
or very low counts were removed after featureCount by filtering.
20. The voom function also allows performing microarray-style
normalization functions, such as quantile normalization. This
is recommended only for very noisy samples.
21. The main fraction of reads in small RNA sequencing should be
around 2024 nucleotides in length (corresponding to the
miRNA fraction). The raw reads, however, include the adapter

sequences that need to be removed before alignmentdue to
the short target sequence, an alignment would not be possible
in most cases. After clipping of the adapters, the length distri-
bution should show a peak at the about 20 nt.
22. There exists a plethora of other aligners for small RNA sequenc-
ing workflows, the most common alternative to STAR is bow-
tie. However, bowtie seems to be very susceptible to not or
not completely trimmed adapters.
23. Aligner will miss the target regions, seed region is only 68 nt.
24. As an alternative to aligning the short RNA reads to the
genome, an alignment to the miRBase database of known
miRNAs is possible [31].
25. It is beyond the possibilities of this chapter to address all features
and possibilities of the limma package. However, the reader is
encouraged to download the latest version of the very through
limma manual from https://www.bioconductor.org/pack-
ages/3.3/bioc/vignettes/limma/inst/doc/usersguide.pdf.
26. Array weights in limma allow to account for microarrays with
varying quality, e.g., from human samples, by assigning differ-
ent weights. It is generally recommended to utilize weights in
situations with difficult samples, sparse source materials, and
the observation of varying quality. Conversely, when using
material from very well controlled cell culture systems, weights
need not be used.
27. In RNA Seq experiments, a combination of the array weights
strategy for individual samples and the weighting method used
by voom is possible to correct outlier samples. This method
is implemented by the voomWithQualityWeights function in
limma.
28. This workflow describes how to perform a differential expres-
sion analysis of RNA Seq data based on gene counts. Still, the
sequencing data also allow for more detailed analyses, e.g., a
differential splicing analysis. This analysis can be performed
with minor changes to the workflow described herein by sim-
ply changing the focus of the featureCounts function from
gene to exon by setting useMetaFeatures=FALSE.
Acknowledgments
Periodontology (DG PARO) and the German Society for Oral
and Maxillo-Facial Sciences (DGZMK) to M.K. and by grants
from NIH/NIDCR (DE015649, DE021820 and DE024735)
and by an unrestricted gift from Colgate-Palmolive Inc. to
author P.N.P.
References
1. Kebschull M, Demmer RT, Grun B, Guarnieri differential expression analyses for RNA-
P, Pavlidis P, Papapanou PN (2014) Gingival sequencing and microarray studies. Nucleic
tissue transcriptomes identify distinct peri- Acids Res 43:e47
odontitis phenotypes. J Dent Res 93:459468 13. Dobin A, Davis CA, Schlesinger F, Drenkow J,
2. Nowak M, Kramer B, Haupt M, Papapanou PN, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras
Kebschull J, Hoffmann P, Schmidt-Wolf IG, TR (2013) STAR: ultrafast universal RNA-seq
Jepsen S, Brossart P, Perner S, Kebschull M aligner. Bioinformatics 29:1521
(2013) Activation of invariant NK T cells in peri- 14. Dobin A, Gingeras TR (2015) Mapping RNA-
odontitis lesions. J Immunol 190:22822291 seq reads with STAR. Curr Protoc
3. Kramer B, Kebschull M, Nowak M, Demmer Bioinformatics 51:11.14.1119.
RT, Haupt M, Korner C, Perner S, Jepsen S, doi:10.1002/0471250953.bi1114s51
Nattermann J, Papapanou PN (2013) Role of 15. Bolger AM, Lohse M, Usadel B (2014)
the NK cell-activating receptor CRACC in Trimmomatic: a flexible trimmer for illumina
periodontitis. Infect Immun 81:690696 sequence data. Bioinformatics 30:21142120
4. Kebschull M, Guarnieri P, Demmer RT, 16. Subramanian A, Tamayo P, Mootha VK,
Boulesteix AL, Pavlidis P, Papapanou PN Mukherjee S, Ebert BL, Gillette MA, Paulovich
(2013) Molecular differences between chronic A, Pomeroy SL, Golub TR, Lander ES,
and aggressive periodontitis. J Dent Res Mesirov JP (2005) Gene set enrichment analy-
92:10811088 sis: a knowledge-based approach for interpret-
5. Stoecklin-Wasmer C, Guarnieri P, Celenti R, ing genome-wide expression profiles. Proc
Demmer RT, Kebschull M, Papapanou PN Natl Acad Sci U S A 102:1554515550
(2012) MicroRNAs and their target genes in 17. Shannon P, Markiel A, Ozier O, Baliga NS,
gingival tissues. J Dent Res 91:934940 Wang JT, Ramage D, Amin N, Schwikowski
6. Kebschull M, Papapanou PN (2010) The use B, Ideker T (2003) Cytoscape: a software
of gene arrays in deciphering the pathobiology environment for integrated models of biomo-
of periodontal diseases. Methods Mol Biol lecular interaction networks. Genome Res
666:385393 13:24982504
7. Papapanou PN, Behle JH, Kebschull M, 18. Merico D, Isserlin R, Bader GD (2011)
Celenti R, Wolf DL, Handfield M, Pavlidis P, Visualizing gene-set enrichment results using
Demmer RT (2009) Subgingival bacterial col- the cytoscape plug-in enrichment map.
onization profiles correlate with gingival tissue Methods Mol Biol 781:257277
gene expression. BMC Microbiol 9:221 19. Gillis J, Mistry M, Pavlidis P (2010) Gene
8. Demmer RT, Behle JH, Wolf DL, Handfield function analysis in complex data sets using
M, Kebschull M, Celenti R, Pavlidis P, ErmineJ. Nat Protoc 5:11481159
Papapanou PN (2008) Transcriptomes in 20. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-
healthy and diseased gingival tissues. Acosta C, Feinberg AP, Hansen KD, Irizarry
J Periodontol 79:21122124 RA (2014) Minfi: a flexible and comprehensive
9. Joensson D, Ramberg P, Demmer RT, bioconductor package for the analysis of infin-
Kebschull M, Dahlen G, Papapanou PN (2011) ium DNA methylation microarrays.
Gingival tissue transcriptomes in experimental Bioinformatics 30:13631369
gingivitis. J Clin Periodontol 38:599611 21. Liao Y, Smyth GK, Shi W (2014) feature-
10. Gentleman RC, Carey VJ, Bates DM, Bolstad Counts: an efficient general purpose program
B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge for assigning sequence reads to genomic fea-
Y, Gentry J, Hornik K, Hothorn T, Huber W, tures. Bioinformatics 30:923930
Iacus S, Irizarry R, Leisch F, Li C, Maechler M, 22. Robinson MD, Oshlack A (2010) A scaling
Rossini AJ, Sawitzki G, Smith C, Smyth G, normalization method for differential expres-
Tierney L, Yang JY, Zhang J (2004) sion analysis of RNA-seq data. Genome Biol
Bioconductor: open software development for 11:R25
computational biology and bioinformatics. 23. Law CW, Chen Y, Shi W, Smyth GK (2014)
Genome Biol 5:R80 Voom: precision weights unlock linear model
11. Ritchie ME, Diyagama D, Neilson J, van Laar analysis tools for RNA-seq read counts.
R, Dobrovic A, Holloway A, Smyth GK (2006) Genome Biol 15:R29
Empirical array quality weights in the analysis 24. Benjamini Y, Hochberg Y (1995) Controlling
of microarray data. BMC Bioinformatics 7:261 the false discovery rate: a practical and power-
12. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, ful approach to multiple testing. J Roy Stat Soc
Shi W, Smyth GK (2015) Limma powers 57:289300
25. Shi W, Banerjee A, Ritchie ME, Gerondakis S, without a reference genome. Nat Biotechnol
Smyth GK (2009) Illumina WG-6 BeadChip 29:644652
strips should be normalized separately. BMC 29. Haas BJ, Papanicolaou A, Yassour M, Grabherr
Bioinformatics 10:372 M, Blood PD, Bowden J, Couger MB, Eccles
26. Hansen KD, Brenner SE, Dudoit S (2010) D, Li B, Lieber M, Macmanes MD, Ott M,
Biases in illumina transcriptome sequencing Orvis J, Pochet N, Strozzi F, Weeks N,
caused by random hexamer priming. Nucleic Westerman R, William T, Dewey CN, Henschel
Acids Res 38, e131 R, Leduc RD, Friedman N, Regev A (2013)
27. Kopylova E, Noe L, Touzet H (2012) De novo transcript sequence reconstruction
SortMeRNA: fast and accurate filtering of ribo- from RNA-seq using the trinity platform for
somal RNAs in metatranscriptomic data. reference generation and analysis. Nat Protoc
Bioinformatics 28:32113217 8:14941512
28. Grabherr MG, Haas BJ, Yassour M, Levin JZ, 30. Bray NL, Pimentel H, Melsted P, Pachter L
Thompson DA, Amit I, Adiconis X, Fan L, (2016) Near-optimal probabilistic RNA-seq
Raychowdhury R, Zeng Q, Chen Z, Mauceli quantification. Nat Biotechnol 34:525527
E, Hacohen N, Gnirke A, Rhind N, di Palma F, 31. Kozomara A, Griffiths-Jones S (2014) miR-
Birren BW, Nusbaum C, Lindblad-Toh K, Base: annotating high confidence microRNAs
Friedman N, Regev A (2011) Full-length using deep sequencing data. Nucleic Acids Res
transcriptome assembly from RNA-Seq data 42(Database issue):D68D73
Chapter 20
Exploring Genome-Wide Expression Profiles Using

Machine Learning Techniques
Abstract
Although contemporary high-throughput omics methods produce high-dimensional data, the resulting
wealth of information is difficult to assess using traditional statistical procedures. Machine learning meth-
ods facilitate the detection of additional patterns, beyond the mere identification of lists of features that
differ between groups.
Here, we demonstrate the utility of (1) supervised classification algorithms in class validation, and (2)
unsupervised clustering in class discovery. We use data from our previous work that described the tran-
scriptional profiles of gingival tissue samples obtained from subjects suffering from chronic or aggressive
periodontitis (1) to test whether the two diagnostic entities were also characterized by differences on the
molecular level, and (2) to search for a novel, alternative classification of periodontitis based on the tissue
transcriptomes.
Using machine learning technology, we provide evidence for diagnostic imprecision in the currently
accepted classification of periodontitis, and demonstrate that a novel, alternative classification based on
differences in gingival tissue transcriptomes is feasible. The outlined procedures allow for the unbiased
interrogation of high-dimensional datasets for characteristic underlying classes, and are applicable to a
broad range of omics data.
Key words Periodontal disease, Aggressive periodontitis, Chronic periodontitis, Gene expression,
Transcriptome, Gingiva, Classification, Machine learning
1 Introduction
The high-dimensional data produced by contemporary omics

methodology (see Chapter 18 by Kebschull et al. of this volume)
provide a wealth of information that is difficult to analyze using
traditional statistical methods. In Chapter 18 of this volume, we
have presented a workflow for the identification of features in the
dataset that differ between a priori-defined subgroups of samples,
e.g., based on clinical diagnosis or experimental treatment alloca-
tion. These analyses produce lists of features, and, subsequently, of
ontology groups that are differentially expressed after correction
for multiple hypothesis testing. Nevertheless, differential
347
348 Moritz Kebschull and Panos N. Papapanou
expression of features between groups does not necessarily imply

that these groups are distinguishable based on characteristic pat-
terns of these features. In addition, differential expression analyses
can only assess dissimilarities between already defined groups,
while novel groupings based on characteristic patterns in the data
are impossible to generate.
To address these problems, omics researchers have ventured
into the field of machine learning, a genre of computer science that
uses artificial intelligence for pattern recognition and computa-
tional learning. Specifically, both supervised and unsupervised
learning approaches proved useful for the analysis of omics data.
Supervised learning encompasses the training of a learning algo-
rithm on labeled samples and the subsequent use of the learned
algorithm to predict the labels for new, unlabeled samples. This
approach for the classification of samples based on patterns recog-
nized by the learner is commonly used for class validation, i.e., the
evaluation of the learnability of a class distinction, e.g., two differ-
ent diagnoses. In contrast, unsupervised learning entails the subdi-
vision of a set of samples with no prior allocation, into two or more
novel classes, based on characteristic similarities of their encom-
passing features.
Our group has used machine learning techniques to study the
classification of periodontal diseases based on the transcriptomes of
240 disease-affected gingival tissue samples from 120 subjects with
chronic or aggressive periodontitis. First, we performed a class
validation analysis to evaluate whether supervised classification
algorithms were able to distinguish chronic from aggressive peri-
odontitis based on the tissue transcriptomes. Indeed, the best-
performing algorithms were able to reach high diagnostic accuracy
in the differentiation between chronic and aggressive periodontitis.
However, to do so, the algorithms had to utilize the expression of
thousands of genes as diagnostic features, rendering the classifier
very computationally demanding, and the results likely less gener-
alizable. In addition, we found a substantial heterogeneity in clas-
sifier performance, despite the use of generally accepted, robust
methods and otherwise identical procedures, which is strongly
suggestive of diagnostic imprecision in the current classification of
periodontitis [1]. We subsequently sought to detect novel classes
of periodontitis patients based on characteristic transcriptomic pat-
terns in their diseased gingival tissues using unsupervised cluster-
ing. Since disease severity at the particular gingival unit was shown
to be a major determinant of local gene expression, and given our
major goal to allocate the patient, rather than the individual tissue
sample, we utilized a model-based clustering approach. Specifically,
we used mixture models implemented in the flexmix package in R
[2], and corrected for both the severity of periodontitis at the par-
ticular gingival tissue sample (i.e., the maximum probing depth
adjacent to the sample) and the interdependency of multiple tissue
Machine Learning Analysis of Omics Data 349
samples obtained from the same patient. Our approach identified

two novel clusters of periodontitis patients that did not only differ
substantially in their defining underlying transcriptomic features,
but also in their whole-mouth clinical and microbiological profiles,
as well as in serological markers of periodontitis. We suggested
that, after appropriate validation steps in independent cohorts and
in longitudinal studies, these findings could support a novel,
pathobiology-based classification of periodontitis [3].
In this chapter, we provide an overview of the practical applica-
tion of machine learning techniques on high-dimensional omics
data. First, we delineate the steps necessary to perform a class vali-
dation analysis of a dataset obtained using the information provided
in our accompanying chapters in this volume (see Chapters 18
and 19 by Kebschull et al.), utilizing the CMA package in R. Then,
we describe the unsupervised, mixture model-based clustering of
the same dataset using the flexmix package.
2 Materials
2.1 Hardware 1. A computer with x86-64 compatible processor(s) running

either Linux, Mac OS X, or Windows (see Note 1).
RAM 16GB.
2.2 Software 1. The R statistical environment, including the Bioconductor

framework, and the following libraries.
CMA [4].
reshape [5].
ConsensusClusterPlus [6].
flexmix [2].
limma [7].
gplots [8].
mclust [9].
2. (Optional, but highly recommended) An integrated program-
ming environment (IDE) for R, e.g., RStudio, or a program-
ming editor, e.g., GNU Emacs/ESS.
3. (Optional, but highly recommended) A version control sys-
tem, e.g., git.
2.3 Data 1. Quality-controlled, preprocessed mRNA expression profiling

data from microarray or RNASeq experiments (see Chapters 18
and 19, both by Kebschull et al. of this volume). For this case
study, we utilize a hypothetical dataset:
mRNA expression profiles generated using microarrays
from clinically diseased gingival tissue biopsies.
200 Subjects with periodontitis, 1 sample per subject = 200

samples in total (see Note 2).
For each subject, a diagnosis of chronic or aggressive peri-
odontitis [10, 11] was assigned by consensus.
For each tissue biopsy, there exist clinical and microbio-
logical data.
Expression data were quality-controlled, normalized, and
batch-corrected (see Note 3).
3 Methods
3.1 Use 1. Preprocessing of data for use with the CMA package.
of Supervised Use batch-corrected, preprocessed, and normalized data
Learning Algorithms from microarray or sequencing experiments (see Chapters 18
for the Distinction and 19, both by Kebschull et al. of this volume). After the steps
of Aggressive described in (see Chapter 18 by Kebschull et al. of this vol-
and Chronic ume), the data are usually in the form of a large array with
Periodontitis Based thousands of rows for the different features (i.e., genes, tran-
on mRNA Expression scripts, CpG islands, etc.) and columns for the individual
samples.
In this example, we assume that only samples associated
with periodontal disease are present (edata_aff, a subset of the
edata expression data matrix generated in Chapter 18 of this
volume, with only affected samples remaining). In R, we for-
mat the data according to the specifications of the CMA R
package we intend to use for the supervised analysis.
# label with diagnosis
> colnames(edata_aff) <- pheno_aff$Diagnosis
# rotate (samples -> rows, features -> col-
umns)
> edata_aff_rot <- as.data.frame(t(edata_
aff))
# generate a factor containing the diagnosis
information with two levels, in this exam-
ple chronic and aggressive
> diseased <- rownames(edata_aff_rot)
> diseasedY <- as.factor(diseased)
# change back into matrix format, label rows
with subject number
> diseasedX <- as.matrix(edata_aff_rot)
> labels <- pheno_aff$Patient
> rownames(diseasedX) <- labels
2. Generate training and evaluation sets.
For the evaluation of the performance of a learning
algorithm in our dataset, we perform an internal validation
procedure (see Note 4).
# load CMA package

> library(CMA)
#set a random seed important to keep constant
for reproducible results
> set.seed(651)
# generate learning sets of the same size
comprising on average about 2/3 of the different
available samples by bootstrapping (sampling
with replacement) or other methods (see Note
5) for a high number of iterations, e.g.
1000 different sets (see Notes 2 and 5).
This step needs to be adjusted in cases of
multiple samples per subject (see Note 7).
> datboot <- GenerateLearningsets(y=disease
dY, method="bootstrap", niter=1000, ntrain=f
loor(0.66*length(diseasedY)), strat=TRUE)
3. Feature selection.
For each learning set, characteristic features are identified
using statistical tests comparing the predefined groups. Here,
we use the moderated t-test implemented in the limma R
package that was introduced in Chapter 18 (see Note 6).
> varsel <-GeneSelection(X=diseasedX,
y=diseasedY, learningsets=datboot,
method="limma")
4. Supervised learning analysis.
# perform classification of the evaluation
sets generated in (b) by different (see Note
8) classifier algorithms, using the best nb-
gene features identified by the feature se-
lection process in (c)
# these procedures produce warnings from the
R system (see Note 9)
> class_svm <- classification(X=diseasedX,
y=as.factor(diseasedY),
learningsets=datboot, genesel=varsel, nb-
gene=250, tuninglist = list(grids = list()),
classifier=svmCMA, probability=TRUE)
> class_lda <- classification(X=diseasedX,
y=as.factor(diseasedY),
learningsets=datboot, classifier = dldaCMA,
genesel=varsel, nbgene=250)
> [] more classifiers
5. Presentation and interpretation of results.
The performance of the classifiers can be assessed by differ-
ent measures (see Note 10), including the Area under the
Receptor Operating Curve (AUC) that plots the false positive
by the true positive rates.
sensitivity specificity AUC

1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0

SCDA
DLDA
PLSLDA
SVM
SCDA
DLDA
PLSLDA
SVM
SCDA
DLDA
PLSLDA
SVM
Fig. 1 Microarray-based classification of AP and CP gingival lesions. Four different microarray classifier algo-
rithms were trained to distinguish gingival lesions from AP or CP patients based on their whole-transcriptome
expression profiles. For each of the 1000 splittings into training/evaluation sets that accounted for multiple
tissue samples per participant, variable selection was performed based on the training set using a mixed-
effects linear model. Subsequently, four different classifier algorithms [diagonal linear discriminant analysis
(DLDA), partial least square analysis combined with linear discriminant analysis (PLS-LDA), shrunken centroids
discriminant analysis (scDA), or a support vector machines (SVM)] were trained on the training set to distin-
guish between AP and CP gingival lesions based on 250 genes (DLDA, PLS-LDA, and SVM). Performance of the
algorithms in the classification of the corresponding evaluation sets was then assessed using the sensitivity
and specificity of AP detection, as well as (ROC) area-und-the-curve. With permission from Sage Publishing,
reprinted from [1]
The performance data can then be plotted, either for all

1000 iterations (Figs. 1 and 2a), or for a random iteration
(Fig. 2b).
> auc_svm<-evaluation(class_svm, measure="auc",
scheme=iter)
> auc_lda<-evaluation(class_lda, measure="auc",
scheme=iter)
# plot the AUC for the different classifiers
and the different iterations
> boxplot(attributes(auc_lda)$score,
attributes(auc_svm)$score, names=c("DLDA","S
VM"),main="AUC")
3.2 Identification As in Subheading 3.1 (1), we use a dataset of expression data from
of Novel Classes periodontally affected subjects (edata_aff, a matrix of >50,000 fea-
of Periodontitis Based tures [rows] 200 samples [columns], with a correspondent data
on mRNA Expression frame with phenotypical information, pheno_aff). In this unsuper-
Profiles Using vised analysis, the data are not labeled.
Unsupervised # set a random seed important to keep constant
Clustering for reproducible results
> set.seed(651)
3.2.1 Preprocessing # take top genes used for clustering and
of Data for Use number of bootstrap iterations (see Note 11)
with the flexmix package > numbertop <- 5000
X- and Y-linked genes can lead to undesired phenomena

during clustering and should be removed (see Note 12).
# get X-/Y-linked genes (example code for
Affymetrix arrays)
> x <- hgu133plus2CHR
> mapped_probes <- mappedkeys(x)
> xx <- as.data.frame(x[mapped_probes])
> is.X <- subset(xx, chromosome=="X")
> is.y <- subset(xx, chromosome=="Y")
> sexChr <- rbind(is.X, is.y)
> sexChr <- sexChr[,1]
> rownames(edata_aff) -> allGenes
> overlap <- allGenes %in% sexChr
> edata_aff <- edata_aff[!overlap,]
# take top genes by median absolute deviation
> mads<-apply(edata_aff,1,mad,na.rm=TRUE)
> edata_aff = edata_aff[order(mads,
decreasing=TRUE) [1:numbertop],]
# scale data
> edata_aff = sweep(edata_aff,1,
apply(edata_aff,1,median,na.rm=TRUE))
# combine top genes with probing depth infor-
mation for each sample
> data <- t(edata_aff)
> data <- cbind(data, patient, ppd)
> data <- as.data.frame(data)
sensitivity specificity AUC

1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0

svm5
svm10
svm50
svm100
svm250
svm500
svm750
svm1000
svm2500
svm5000
svm5
svm10
svm50
svm100
svm250
svm500
svm750
svm1000
svm2500
svm5000
svm5
svm10
svm50
svm100
svm250
svm500
svm750
svm1000
svm2500
svm5000
Fig. 2 Microarray classifier distinction of gingival lesions from AP and CPSVM algorithm using different feature
set sizes. For each of the 1000 splittings into training/evaluation sets, a support vector machine (SVM) classifier
algorithm was trained based on the training set to distinguish AP from CP gingival lesions using either 5, 10,
50, 100, 250, 500, 750, 1000, 2500, or 5000 genes. Performance of the algorithms in the classification of the
corresponding evaluation datasets was then assessed using the sensitivity and specificity of AP detection, as
well as (ROC) area-und-the-curve (a). In addition, for each number of features, a ROC curve was generated for
a representative iteration (b). The SVM algorithm showed improving performance with increasing signature
size. With permission from Sage Publishing, reprinted from [1]
354
SVM 5 genes SVM 10 genes SVM 50 genes SVM 100 genes SVM 250 genes
1.0 1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4 0.4
Sensitivity
Sensitivity
Sensitivity
Sensitivity
Sensitivity
0.2 0.2 0.2 0.2 0.2
AUC=0.784 AUC=0.832 AUC=0.912 AUC=0.938 AUC=0.965
0.0 0.0 0.0 0.0 0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
1specificity 1specificity 1specificity 1specificity 1specificity
SVM 500 genes SVM 750 genes SVM 1,000 genes SVM 2,500 genes SVM 5,000 genes
1.0 1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6 0.6

0.4 0.4 0.4 0.4 0.4
Sensitivity
Sensitivity
Sensitivity
Sensitivity
Sensitivity
0.2 0.2 0.2 0.2 0.2

AUC=0.975 AUC=0.979 AUC=0.982 AUC=0.987 AUC=0.99
0.0 0.0 0.0 0.0 0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
1specificity 1specificity 1specificity 1specificity 1specificity
Fig. 2 (continued)
# convert data into long format

> library(reshape)
> data.long <- melt(data, id=c("patient",
"ppd"), variable_name="probe")
3.2.2 Assess Influence Depending on the nature of the samples, there may be phenotypes
of Phenotypical Variables that primarily drive expression. In our example, the gene expres-
sion of gingival tissue biopsies was found to be strongly related to
the maximum probing depth associated with the biopsy (Fig. 3).
This information is important, because if uncorrected, the
strong influence of probing depth would have led to a separation
of deep from shallow lesions by the clustering algorithm Fig. 4.
> ppd <- pheno_aff$PDMax
# color code samples according to their as-
sociated maximum probing depth
> colors <- as.character(factor(ppd, 4:12,
c(green4, green3, green2, orchid,
orchid, orchid1, orchid2, orchid3,
orchid4))
5 10
6
0.4
8
10 5 5
7 7 8
7 58 7 5 5 5
4 5 8
9 9 7 78 5 5
7 58 7 69
5 5
0.2
11 67 97 5 5 5 5 6
4
7 5 5 5
11 10 108 10 5 6 5 6 45
10 11 4 10 5 5 7
8 98 6 5 5
10 8
6 55 7
89 4 65
6 5 6 7554
10 5 5
5 6 65
10
776 598 5 8 5 10 7 5 5
8 10 5 5
Dimension 2
10 8 5 7
0.0
10 4 68 9 10 7 44
5
55 6
7 8 55 6 8 5 5 6 586 7
10 6 55 7 5 6 5 75
10 8 8 9
5 5
5 1056 511 4 9 5 5 5 4
554 811 5 6
6 5 6
7 89 7 57 5
10 6 10 9 8 4 5
0.2
8 10 10 5
8 7
7 8 10 5 10 55 5
9 7 5
85 5 5
8
6 5
9 8 9 4
4
6
0.4
5
6 8
5 4 5
7
8
7
5
0.6
1.0 0.5 0.0 0.5 1.0

Dimension 1
Fig. 3 MDS plot of gene expression profiles of individual tissue samples according to probing depth.
Multidimensional scaling plot of transcriptomic profiles from 241 tissue samples, based on all autosomal
probes on the array. Each individual tissue sample was labeled with the maximal pocket depth associated with
the particular biopsy. Samples with PD of 46 mm are coded in different shades of green, and samples with
of 812 mm in different shades of purple. Note that most biopsies aggregate according to their local disease
severity, with most shallow pockets on the right side of the plot, and most deep sites on the left. With permission
from Sage Publishing, reprinted from [3]
consensus matrix k=2
1
2
Fig. 4 Consensus Clustering (k = 2) of individual biopsies. Gene expression pro-

files from 241 periodontitis-affected tissue samples from 120 patients with peri-
odontitis were subjected to consensus clustering, based on the 5000 most
variable probes across the entire dataset. Consensus clustering was run for 1000
iterations with a resampling rate of 80 %. This exploratory analysis disregarded
the interdependency of samples from the same donor and was meant to identify
phenotypic features of the individual tissue samples that determined clustering
allocation. With permission from Sage Publishing, reprinted from [3]
# MDS plot of individual biopsies using all

genes, color-coded for probing depth
plotMDS(edata_aff, gene.selection="common",
labels=ppd, col=colors)
# Perform exploratory clustering analysis
without accounting for other factors, e.g.
by consensus clustering [6], a resampling
approach to standard hierarchical cluster-

ing (see Fig. 4). The resulting clusters can
then be assessed for differences in main
phenotypical variables
> library(ConsensusClusterPlus)
> results <- ConsensusClusterPlus(edata_
aff,maxK=10,
reps=1000,pItem=0.8,pFeature=1,
title=exploratory_clustering.pdf,
clusterAlg="hc", distance="pearson",seed=65
1,plot="pdf")
3.2.3 Perform Mixture The flexmix R package performs model-based clustering based on
Model-Based Clustering finite mixtures of regressions that allows for the inclusion of fixed
with Correction for Probing and random factors, e.g., the maximum probing depth associated
Depth with a gingival biopsy as a fixed factor, and the subject as a random
factor which ensures that cluster membership is determined on the
subject level (not relevant in this example, since all subjects only
contributed a single biopsy). The resulting models for several num-
bers of clusters can be assessed for the goodness-of-fit using stan-
dard measures, such as the Akaike Information Criterion (AIC).
> library(flexmix)
> model2_ppd <- flexmix(value ~ probe | pa-
tient, model = FLXMRglmfix(fixed = ~ 0+ppd,
k = 2), k=2, data=data.long, control =
list(verbose = 1, iter.max = 20, minprior
= 0))
# extract cluster assignments for the dif-
ferent samples
> model2_ppd@cluster -> cluster2_ppd
3.2.4 Phenotypic 1. Differences in expression of features and the corresponding

Analysis of the New biological groups between the novel clusters can be identified
Clusters using the methodology introduced in Chapter 18 (this vol-
ume), i.e., limma [7] for a differential expression analysis and
ermineJ [12] for a subsequent ontology analysis.
2. Often, the separation of the obtained clusters is illustrated
using a heatmap (Fig. 5), e.g., utilizing heatmap.2 in the gplots
R package [8].
# extract most informative features from the
model and format
> pars <- parameters(model2_ppd)
> ordering <- order(abs(apply(pars[2:101,],
1, diff)), decreasing = TRUE)
> data_corrected <- as.matrix(data[,
1:100] - data$ppd * pars[1, 1])
Cluster #1 Cluster #2
Probes
Subjects
Fig. 5 Heatmap of cluster-defining genes. The cluster-defining genes were sorted

by their importance for the model, and the preprocessed expression data were
corrected for the probing depth of the individual sample. Data were visualized
using the heatmap.2 function in the gplots package v.2.11.3 [8]. Note the clear
separation of gene expression between the two clusters. With permission from
Sage Publishing, reprinted from [3]
> ordered <- data_corrected[,ordering][,

1:100]
> ordered <- ordered[order(data$cluster),]
# add sideline to the heatmap to illustrate
the different clusters in red and green.
> sideline <- rep(c("red", "green"),
table(data$cluster))
# load the gplots package and plot heatmap
> library(gplots)
> heatmap.2(t(ordered), Colv=FALSE,
dendrogram="none", trace="none",
col=redgreen, ColSideColors=sideline)
Comparison of Cluster Assignments

Adjusted Rand
Extent
Index 0.0095
Localized Periodontitis
Generalized Periodontitis
Adjusted Rand
1999 Classification
Index 0.0143
Chronic Periodontitis
Aggressive Periodontitis
Transcriptome-based
Classification
Cluster #1
Cluster #2
Fig. 6 Stability of cluster assignments from the model-based clustering approach using finite mixtures over a
wide range of feature numbers. Cluster assignments obtained with model-based clustering of 241 transcrip-
tomic profiles of periodontitis-affected gingival tissue samples from 120 patients, using different numbers of
features. All autosomal probes on the microarray were sorted by absolute variance across the whole dataset,
and the top 10053,243 probes were employed by the clustering algorithm. The graph shows robust cluster-
ing, with most patients assigned to the same cluster in all situations. Cluster #1 is represented in blue color,
Cluster #2 in red. When using small (<1000 features) or very large (>10,000 features) sets, some promiscu-
ous patients change clusters. This behavior is expected, since very small sets tend to lack all information
required for correct cluster assignment, and very large sets add a considerable amount of nonspecific noise.
With permission from Sage Publishing, reprinted from [3]
3. To compare cluster assignments, e.g., the similarity of the

novel classes identified by the unsupervised analysis and tradi-
tional groupings, measures like the Hubert-Arabie adjusted
Rand index [13] (ranging from 0 indicating entirely random
overlaps, and 1 indicating perfect agreement) can be used
(Fig. 6).
# load the mclust [9] library and compare
the novel, mixture model clustering based
classes and the 1999 classification
> library(mclust)
> clusterComp <- adjustedRandIndex(cluster2_

ppd, pheno_aff$Diagnosis)
4. Finally, it is of high interest to assess whether the novel classes
identified by mixture model-based clustering would also dis-
play disparate phenotypes. To do so, standard parametrical or
nonparametrical tests can be employed.
# t-tests
pval <- t.test(NoOfDeepSites ~ clus-
ter2_ppd, data=clusters, var.equal = TRUE,
na.action="na.omit")
# contingency tables and Fisher test
> gendertable <- xtabs(~ cluster2_ppd +
Gender, data=clusters)
> genderfisher <- fisher.test(gendertable)
4 Notes
1. When running Windows and/or a graphical user interface,

several methods of script optimization to use multiple comput-
ing cores of the CPU(s) may not be available.
2. Multiple samples from the same subject may deliver more stable
insights into the pathobiology of disease than single samples
[14], but also present with several challenges from a statistical
perspective: Most algorithms for supervised (e.g., CMA [4])
and unsupervised learning (e.g., ConsensusClusterPlus [6]) do
not account for multiple statistically dependent samples and
therefore cannot be utilized for the analysis of such datasets
without modification. Specifically, in the case of classifier algo-
rithms, care must be taken to avoid training an algorithm using
one sample from a particular individual, and then evaluating the
classifier using another sample from the same individual. Since
multiple samples from the same individual are correlated, this
would lead to artificially high performance of the classifier.
Similarly, in unsupervised clustering analyses that aim to identify
classes of subjects based on omics profiles of multiple samples,
methods such as model-based clustering need to be employed to
correct for the correlation of the samples, and possibly also for
their inherent characteristics. By no means should multiple sam-
ples be coerced into a single sample (e.g., by averaging of the
expression values) to facilitate the analysis.
3. Batch effects introduced by different chemistries, operators,
but also by normal day-to-day variation are of particular con-
cern for machine learning applications, because samples within
a particular batch correlate with each other irrespective of their
underlying biology and may lead to bias [15]. In Chapter 18,
we describe a workflow to (a) visually assess data for evidence

of batch effects, e.g., by multidimensional scaling plots, and
(b) correct for these effects using the ComBat procedure in the
sva R package [16].
4. Ideally, the performance of a classifier algorithm trained on a
labeled dataset should be assessed using a second, indepen-
dent evaluation set. For example, this external validation can
involve a different cohort of subjects. This way, the true gen-
eralizability of the learner can be assessed and the inherent
danger of overfitting a classifier that is highly specific only to
one particular dataset can be mitigated. In case independent
datasets are not available, an internal validation is performed
by splitting the dataset into pairs of learning and evaluation
sets by procedures such as bootstrapping or other cross-
validation methods (see Note 5). The learner is then trained
using the learning set, and evaluated using the corresponding
evaluation set. The performance of the algorithm for that par-
ticular pair of learning and evaluation sets is then recorded.
The average of a high number of iterations (e.g., 1000)
reduces the variance of the error estimator, and thereby over-
optimism, a typical problem of not-well controlled machine
learning applications in omics research [17].
5. The GenerateLearningsets procedure offers several different
options for the generation of splits into learning and evaluation
sets, including the bootstrap that was used in our example, but
also k-fold cross-validation, Monte-Carlo cross-validation, and
k-fold cross-validation. The strat = TRUE argument ensures a
similar distribution of the classes in the original dataset and the
obtained splits.
6. For the selection of features, the CMA package offers a variety
of different methods. First, there are filter methods comparing
the predefined classes in the learning set using statistical testing
to rank differentially expressed features. The limma method
[18] used in our examples falls in this category. Second, meth-
ods such as the random forest variable importance measure
rank features according to their discriminatory potential.
Third, methods such as elastic nets or the lasso exist that are
classification algorithms themselves selecting sparse sets of vari-
ables that can also be used by other algorithms.
7. As indicated in Note 2, CMA does not account for multiple
samples. Therefore, if planning to use this package for the
supervised classification analysis of data with multiple, statisti-
cally dependent samples, as our group did in our paper testing
the 1999 classification of periodontitis [1], the datboot object
generated using GenerateLearningsets needs to be modified.
Specifically, we randomly sampled subjects, rather than indi-

vidual samples, and assigned all samples belonging to a drawn
individual to the learning sets. In so doing, we avoided to test
an algorithm on a sample from an individual that had already
had provided a sample for the learning phase, as outlined in
Note 2.
8. In a class validation situation, to obtain reliable information
about the distinction between two or more classes, it is highly
recommended to perform classification using several different
algorithms. To avoid bias, sound scientific practice mandates
the reporting of all results obtained, rather than only the best
one [19].
9. Note that these function calls result in warnings when using
the GenerateLearningsets option bootstrapone warning
about duplicate entries for each iteration of the learning set
generation (which is perfectly normal for a sampling procedure
with replacement).
10. In addition to the area-under-the-ROC-curve (AUC), several
other measures for the classifier performance can be obtained
using the evaluation function. These include sensitivity and
specificity (for K = 2, as in our example). In addition, and also
in the case of more than two classes, the Brier score, the mis-
classification rate, and the average probability of correct clas-
sification can be employed.
11. The optimal choice of features used for the unsupervised clus-
tering is subject to an ongoing debate. Generally, by selecting
a low proportion of available features one may overlook criti-
cal biological information that could have informed the unsu-
pervised analysis. In contrast, when using a very high
proportion or all features present, the considerable noise may
lead to suboptimal results. It is, however, not entirely clear
what numbers or proportions are too few or too many.
Therefore, it is advisable to assess the stability of the clusters
obtained using a range of feature numbers. Relatively stable
cluster assignments, as depicted in Fig. 7, support the notion
of robust novel classes.
12. Often, the features used for clustering are ranked using a mea-
sure of variation within the dataset, with those features show-
ing the highest variation being interpreted as the likely most
informative. However, in a dataset that comprises samples
from both male and female subjects, those highly variable fea-
tures include genes with links to sex chromosomes. Depending
on the issue under investigation, it may be preferable to exclude
these features from the analysis.
100
500
1000
2000
number of features employed

3000
4000
5000
10000
20000
53243
87
118
8
47
54
123
111
18
101
91
10
21
125
84
35
5
24
78
2
59
131
126
124
119
115
112
109
106
104
99
94
93
90
88
86
77
76
71
70
69
68
67
66
63
56
55
50
49
48
46
44
42
39
37
33
31
28
27
26
23
16
15
14
13
9
7
3
4
34
127
100
117
108
1
92
130
129
121
120
116
113
110
107
105
102
98
95
89
85
83
82
81
80
75
74
73
72
65
64
62
61
60
58
53
52
45
43
41
40
38
36
32
30
29
25
22
20
17
11
12
Subjects
Fig. 7 Comparison of cluster assignment. Transcriptomic profile-based cluster assignments of study partici-
pants (panel a) are compared with the current primary classification of periodontitis [chronic (CP) or aggres-
sive (AgP); (panel b)], as well as with an extent-based subdivision based on the 1999 International Workshop
criteria [localized or generalized periodontitis; (panel c)]. The degree of concordance between the three ways
of classifying periodontitis was assessed using the Herbert-Arabie adjusted Rand index (ARI). Perfect align-
ment is indicated by an ARI of 1, random alignment by an ARI of 0. Each column across panels ac represents
a patient. With permission from Sage Publishing, reprinted from [3]
Acknowledgments
Periodontology (DG PARO) and the German Society for Oral and
Maxillo-Facial Sciences (DGZMK) to M.K., and by grants from
NIH/NIDCR (DE015649 and DE024735) and by an unre-
stricted gift from Colgate-Palmolive Inc. to P.N.P. The authors
thank Prof. Anne-Laure Boulesteix (Munich, Germany) and Prof.
Bettina Grn (Linz, Austria) for their support with the CMA and
flexmix packages, respectively.
References
1. Kebschull M, Guarnieri P, Demmer RT, Boulesteix 2. Grn B, Leisch F (2008) FlexMix version 2:
AL, Pavlidis P, Papapanou PN (2013) Molecular finite mixtures with concomitant variables and
differences between chronic and aggressive peri- varying and constant parameters. J Stat Softw
odontitis. J Dent Res 92:10811088 28:135
3. Kebschull M, Demmer RT, Grun B, Guarnieri and aggressive periodontitis. Periodontol 2000
P, Pavlidis P, Papapanou PN (2014) Gingival 53:1227
tissue transcriptomes identify distinct periodon- 12. Gillis J, Mistry M, Pavlidis P (2010) Gene
titis phenotypes. J Dent Res 93:459468 function analysis in complex data sets using
4. Slawski M, Daumer M, Boulesteix AL (2008) ErmineJ. Nat Protoc 5:11481159
CMA: a comprehensive bioconductor package 13. Hubert L, Arabie P (1985) Comparing parti-
for supervised classification with high dimen- tions. J Classif 2:193218
sional data. BMC Bioinformatics 9:439 14. Papapanou PN, Abron A, Verbitsky M, Picolos
5. Wickham H (2007) Reshaping data with the D, Yang J, Qin J, Fine JB, Pavlidis P (2004)
reshape package. J Stat Software 21:120 Gene expression signatures in chronic and
6. Wilkerson MD, Hayes DN (2010) aggressive periodontitis: a pilot study. Eur
ConsensusClusterPlus: a class discovery tool J Oral Sci 112:216223
with confidence assessments and item tracking. 15. Leek JT, Scharpf RB, Bravo HC, Simcha D,
Bioinformatics 26:15721573 Langmead B, Johnson WE, Geman D,
7. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Baggerly K, Irizarry RA (2010) Tackling the
Shi W, Smyth GK (2015) Limma powers dif- widespread and critical impact of batch effects
ferential expression analyses for RNA- in high-throughput data. Nat Rev Genet
sequencing and microarray studies. Nucleic 11:733739
Acids Res 43:e47 16. Leek JT, Johnson WE, Parker HS, Jaffe AE,
8. Warnes GR, Bolker B, Bonebakker L, Gentleman Storey JD (2012) The sva package for removing
R, Huber W, Liaw A, Lumley T, Maechler M, batch effects and other unwanted variation in
Magnusson A, Moeller S, Schwartz M, Venables high-throughput experiments. Bioinformatics
B (2009) gplots: various R programming tools 28:882883
for plotting data. R Package Version 2(4) 17. Boulesteix AL (2010) Over-optimism in bioin-
9. Fraley C, Raftery AE, Murphy TB, Scrucca L formatics research. Bioinformatics
(2012) MCLUST version 4 for R: normal mix- 26:437439
ture modeling for model-based clustering, clas- 18. Smyth GK (2004) Linear models and empirical
sification, and density estimation. Technical bayes methods for assessing differential expres-
Report no. 597, Department of Statistics, sion in microarray experiments. Stat Appl
University of Washington, USA Genet Mol Biol 3:Article3
10. Armitage GC (1999) Development of a classi- 19. Boulesteix AL, Strobl C (2009) Optimal clas-
fication system for periodontal diseases and sifier selection and negative bias in error rate
conditions. Ann Periodontol 4:16 estimation: an empirical study on high-
11. Armitage GC, Cullinan MP (2010) dimensional prediction. BMC Med Res
Comparison of the clinical features of chronic Methodol 9:85
Part III
Cells and Tissues

Chapter 21
Embryonic Explant Culture: Studying Effects

of Regulatory Molecules on Gene Expression
in Craniofacial Tissues
Katja Nrhi
Abstract
The ex vivo culture of embryonic tissue explants permits the continuous monitoring of growth and mor-
phogenesis at specific embryonic stages. The functions of soluble regulatory molecules can be analyzed by
introducing them into culture medium or locally with beads to the tissue. Gene expression in the manipu-
lated tissue explants can be analyzed using in situ hybridization, quantitative PCR, and reporter constructs
combined to organ culture to examine the functions of the signaling molecules.
Key words Mouse, Morphogenesis, Signaling molecule, Organ culture, Tooth, Whisker, Palate,
Calvarial bone, In situ hybridization, Real-time quantitative PCR
1 Introduction
The development of embryonic organs is characterized by dynamic

morphogenetic events such as budding, branching, and tissue
fusions that are accompanied by proliferation, migration, and dif-
ferentiation of cells. These developmental processes are controlled
by interactions between cells and tissues. In the case of ectodermal
organs like teeth and hairs, interactions between the epithelium
and underlying mesenchyme are of particular importance. The for-
mation of the epithelial placode and the subsequent folding of the
epithelium in an organ-specific manner are regulated by the recip-
rocal and sequential interactions between the epithelium and mes-
enchyme. Cell and tissue interactions are mediated by soluble
regulatory molecules that are conserved in evolution and regulate
the development of all organs. The most important and widely
used signals belong to four families: bone morphogenetic protein
(BMP), hedgehog (HH), fibroblast growth factor (FGF), and the
Wnt family. The signals are integrated into complex regulatory
networks in which specific inhibitors of signals also play important
367
368 Katja Nrhi
roles in fine tuning signaling. A good example of an organ where

the functions of signaling networks have been elucidated in detail
is the mouse tooth [1].
To follow the morphogenesis of developing organs, the tis-
sues can be transplanted in vivo or cultured as explants in vitro.
Organ culture is not suitable for long-term follow-up and does
not provide a physiological environment, which are the advan-
tages of the transplantation method. However, organ culture
techniques are superior in many other aspects. These methodolo-
gies allow continuous monitoring, they provide reproducible con-
ditions, known medium composition, and importantly, tissues can
be manipulated by multiple controlled ways. Over the years, sev-
eral types of organ culture systems have been used for the exami-
nation of embryonic morphogenesis. A technique employing a
platform of perforated metal gauze is called Trowell-type organ
culture [2] and it has been applied to study the morphogenesis of
numerous different organs [39]. In this technique, tissue explants
are cultured at the air-liquid interface on thin membrane filters
supported by a metal grid (Fig. 1). We have applied the Trowell
technique modified by Saxn [10] to elucidate the mechanisms of
tissue interactions during embryonic tooth development [1118]
and also for studies on the fusion of palatal shelves, osteogenesis
of calvarial bones, and suture formation, as well as the formation
of whiskers [8, 1922].
a tissue b tissue
grid culture explant culture
explant filter
medium medium
PBS/H2O
c d
Fig. 1 Schematic representations of (a) Trowell-type organ culture and (b) Hanging-drop technique. (c) In the
Trowell-type organ culture dish the metal grid supports six pieces of filters placed on the holes punched in the
grid. (d) Close-up of a cultured tissue explant lying on a filter in the Trowell-type culture dish. (Picture c cour-
tesy of Otso Hr)
Embryonic Explant Culture 369
Recombinant signaling molecules or their inhibitors can be

applied either in the culture medium or locally by beads to examine
their regulatory functions. Gene expression can be examined dur-
ing culture by using tissue from transgenic reporter mice express-
ing fluorescent markers. Alternatively, gene expression can be
localized in fixed tissues by in situ hybridization analysis either
from the whole explants or from tissue sections. This method is
based on labeled (radioactive or nonradioactive) RNA (riboprobe)
that binds to the corresponding mRNA sequences produced by
the tissue [23].
In addition to the Trowell-type organ culture we have also
used the so-called hanging drop culture method (Fig. 1) and com-
bined it with real-time quantitative PCR (qPCR) to accurately
quantify gene expression changes [24, 25]. The hanging-drop
technique requires only a tiny volume of culture medium allowing
smaller consumption of expensive signaling molecules, and the
harvesting of the tissues from the drops is effortless. Exposing tis-
sue explants over very short periods of time to signaling molecules
allows the detection of their direct targets by real-time qPCR on
complementary DNA produced from the tissue. In this article, we
describe two organ culture techniques: the Trowell and the
hanging-drop methods, which we have used for culturing teeth,
palate, whiskers, and calvarial bones. We also present the tech-
niques for tissue processing for gene expression analysis by in situ
hybridization, and describe the qPCR method.
2 Materials
2.1 Solutions All solutions should be sterile.

and Culture Media
1. Phosphate-buffered saline (PBS), pH 7.4. 10 PBS stock solu-
tion: 1.37 M NaCl, 27 mM KCl, 79 mM Na2HPO4 2H2O,
15 mM KH2PO4. Adjust pH with HCl if necessary. Prepare 1
PBS working solution by diluting one part of 10 stock solu-
tion with nine parts of distilled water. Autoclave (121 C,
15 min) the working solution and store at room temperature.
2. Dulbeccos phosphate-buffered saline, modified (D-PBS).
Store at room temperature.
3. Culture medium: Dulbeccos Modified Eagle Medium
(DMEM; store at 4 C) supplemented with 1 % (v/v) gluta-
MAX-1 (store in single use aliquots at 20 C), 10 % (v/v)
heat-inactivated fetal bovine serum (FBS; freeze in single-use
aliquots at 20 C), and 0.2 % (v/v) PenStrep (penicillin
10,000 IU/mL, streptomycin 10,000 g/mL; freeze in single
use aliquots at 20 C). PenStrep is stable in medium for ~4
weeks. Store culture medium at 4 C. Usually, 50 mL of culture
medium is enough for one experiment.
370 Katja Nrhi
4. F-12 medium (Hams nutrient mixture; store at 4 C). Use as

1:1 mixture in culture medium (see Note 1).
5. 10 mg/mL ascorbic acid (Merck, Darmstadt, DE, pro analysi).
Prepare solution in distilled water and freeze in single-use
aliquots at 20 C. Use 100150 g/mL in culture medium
(see Note 1).
2.2 Dissection All glassware and metal instruments should be sterile. We use auto-
and Culture claved glassware. For sterilization of forceps and scissors we use
Steri 250 glass bead sterilizer (Simon Keller Ltd., Burgdorf, CH).
1. A standard incubator with 37 C temperature, 21 % oxygen
and an atmosphere of 5 % CO2 in air and 9095 % humidity for
tissue culture.
2. Dissection of tissues: 10-cm diameter plastic bacteriological
Petri dishes (Bibby Sterilin Ltd., Stone, Staffs, UK) and 10-cm
diameter glass Petri dishes, small scissors (Instrumed 96, DE),
forceps (Medicon, DE), watchmaker forceps (Durmont, CH),
and disposable 20- and 26-gauge needles (Terumo, Neolus,
Leuven, BE) attached to 1-mL plastic syringes (Euromedis)
(see Note 2).
3. Culture dishes: 35 mm/10 mm plastic Petri dishes (bacterio-
logical or cell culture dishes).
4. Metal grids: Prepare from stainless-steel mesh (corrosion-
resistant, size of mesh 0.7 mm) by cutting approx 30-mm
diameter disk and bending the edges to give 3 mm height (the
height of the metal grids can be altered affecting the needed
amount of culture medium). Use nails to make holes in the
grid to allow the analysis and photography of the explants
(Figs. 1 and 2). There are commercially available organ culture
dishes featuring a central well in which a metal grid (even with-
out bent edges) can be placed (Falcon, Becton Dickinson Ltd.,
Oxford, UK).
5. Filters: 25-mm diameter Nuclepore Polycarbonate Track-
Etch Membranes (Whatman, Schleicher & Schuell, DE). The
pore size routinely used is 0.1 m (see Note 3). The filters are
stored in 70 % ethanol at room temperature.
6. Protein-releasing beads: 17150 m diameter Affi-Gel Blue
agarose beads (Bio-Rad Laboratories, Hercules, CA) or
heparin-coated acrylic beads (Sigma-Aldrich, St. Louis, MO)
are divided into aliquots and stored at 4 C.
7. 0.1 % Bovine serum albumin (BSA) in PBS (0.1 % BSA/PBS)
for dilution of the recombinant proteins used with beads.
Typically high concentrations are used, e.g., 2025 ng/L of
FGF4 (R&D Systems, Minneapolis, MN) and 100 ng/L of
BMP4 (R&D Systems, Minneapolis, MN).
8. Glass Pasteur pipettes to transfer beads and tissue explants.
Fig. 2 Use of Trowell-type culture to follow the morphogenesis of a molar tooth and the initiation of whiskers.
(a, b) The bud staged molar was dissected from the mandible of an E13 transgenic mouse embryo expressing
GFP in the Shh locus [30] and placed on a piece of Nuclepore filter (the arrow points to the edge of the filter)
covering the hole of a metal grid. The occlusal side is up, the mesial side on left, and the distal side on right. The
explant was photographed using (a) light and (b) fluorescent microscopy. (b) After 2 days of culture (+2), GFP
expression is localized to a spot in the center of the molar indicating the formation of the primary enamel knot
and development to cap stage. The following day (+3), new GFP expressing spots indicate the formation of sec-
ondary enamel knots for protoconid and metaconid and progress to early bell stage, and on the last culture day
(+4) three additional secondary enamel knots for anteroconid, hypoconid, and entoconid are detected. (c) E12.5
vibrissa pad was dissected, cultured, and photographed daily using stereomicroscope. After 2 days of culture (+2)
all five whisker rows are seen (brace). During days +3 and +4, hair placodes (arrow head) are detected under the
whisker rows. Arrow points to the developing nostril. (Pictures a, b courtesy of Enni Penttil)
9. Stereomicroscope attached to a camera.

10. Prefixation after culture: 100 % methanol (store at 4 C).
11. Fixation: 4 % paraformaldehyde (PFA) in PBS, freeze in ali-
quots (15 mL) at 20 C. It is recommended to use fresh PFA
(after melting, store at 4 C, use within 2 days).
2.3 In Situ 1. Microscope slides for histological paraffin sections cut by a

Hybridization (ISH) microtome (e.g., Leica).
2. Methanol series for dehydration: 25, 50, 75, and 100 %.
Prepare dilutions in PBS and store at 4 C (only for samples
used in whole-mount ISH).
372 Katja Nrhi
35
3. S-UTP- (radioactive, e.g., Perkin Elmer) or digoxigenin
(nonradioactive, e.g., Roche)-labeled riboprobes.
2.4 Complementary 1. RNA isolation: use, e.g., RNeasy Mini Kit (Qiagen, Hilden,
DNA (cDNA) Synthesis DE).
2. Prepare lysis buffer: Supplement lysis buffer provided by the
kit with 1 % -mercaptoethanol.
3. Nanodrop spectrophotometer to measure RNA concentration
and quality.
4. cDNA synthesis: 500 g/mL random Primers (Promega,
Madison, WI), 40 U/L RNase inhibitor RNasin (Promega,
Madison, WI), 10 mM dNTP mix (ThermoFisher Scientific,
FI), 200 U/L SuperscriptTM II Reverse Transcriptase
(Invitrogen, Carlsbad, CA; the kit includes the 5 first strand
buffer and DTT). Store all reagents at 20 C.
2.5 Real-Time 1. DyNAmo TMFlash SYBR Green qPCR Kit (ThermoFisher

Quantitative PCR Scientific, FI).
(RT-qPCR) 2. Primers for sample and control cDNA (design primers with,
e.g., Primer 3 software).
3. Multiwell Plate 96 or 384 with sealing foils (e.g., LightCycler
480, Roche, DE).
4. For running qPCR: a plate-based real-time PCR amplification
and detection instrument (e.g., LightCycler 480, Roche).
5. Software for analysis of qPCR data (e.g., Lightcycler 480
software, Roche).
3 Methods
The preparation of materials and dissection of tissues is carried out

in laminar flow hood. The dissection microscope should be placed
in the hood, as well.
3.1 Treatment 1. Pipette agarose beads or heparin-coated acrylic beads to PBS in a

of Beads glass Petri dish. Count 100200 beads under the microscope, and
transfer to a nonstick Eppendorf tube with mouth-controlled glass
capillary pipette. Before transferring beads, pipettes are drawn by
heating to adjust the mouth diameter of the pipette to the size of the
beads. Ideally, the diameter should be the minimal to allow free pas-
sage of the beads one by one. Spin down the beads, and remove PBS.
1. Add recombinant proteins in a small volume (1050 L) of
0.1 % BSA/PBS. An equal amount of 0.1 % BSA in PBS is
pipetted to the control beads. Incubate for 30 min at 37 C
and store at 4 C. The beads can be used at least for 14 days
(depending on the stability of the protein).
3.2 Preparation 1. Take one sheet of Nuclepore filter from ethanol and rinse three
of Tissue times in PBS in a plastic 10-cm diameter Petri dish.
Culture Dishes 2. Cut the filter in pieces (35 mm2 depending on the size of tis-
sue), using small scissors and watchmaker forceps and leave in
PBS. Filter pieces can be stored in PBS for several days at 4 C.
3. Place metal grids in 35-mm diameter plastic Petri dishes. Add
13 mL culture medium (see Note 1) by pipetting through the
grid. Avoid air bubbles. The surface of the medium should be
flush with the plane of the grid. Excess medium results in float-
ing of the filters and tissues.
3.3 Dissection 1. Place the mouse uterus (E11E18) in a 10-cm diameter plastic
of Tissues Petri dish containing D-PBS, and cut open the uterine wall
using small scissors and forceps. Under stereomicroscope
remove the embryos from fetal membranes, and transfer them
to a fresh plastic Petri dish with D-PBS. Cut off the heads
using disposable needles (or with scissors when dissecting older
embryos) and transfer the heads to a 10-cm diameter glass
Petri dish containing D-PBS.
2. Dissect the tissue piece of interest using needles: mandible,
calvarial bones, vibrissa pads, or tooth buds (see Notes 4 and 5).
If tissues are cultured using the hanging-drop technique con-
tinue as described below (see Subheading 3.6). For the hang-
ing-drop technique it is essential to remove all the extra tissue
surrounding the tissue of interest to avoid skewed data in real-
time qPCR.
3. Pipette warm (37 C) culture medium to a 35-mm plastic
Petri dish with a grid. Transfer the dissected tissue pieces on
the metal gauze by lifting with a filter piece and watchmaker
forceps. Alternatively, the explants can be transferred by the
capillary force between the tips of watchmaker forceps or
with the Pasteur pipette and placed on filters lying on the
metal grid.
4. Add signaling molecules (see Note 6) either directly to culture
medium or introduce them with beads soaked in high concen-
tration of molecules. Under the stereomicroscope, transfer the
beads one at a time to the tissues. Depending on the experi-
ment and tissue, 15 beads can be placed on one explant.
Examples of BMP4- and BMP2-bead experiments are shown
in Fig. 3 (see Note 7).
3.4 Culture 1. Culture the tissues in an incubator. Change the culture medium
and Fixation every other day (see Note 8).
2. Photograph the explants, e.g., daily with a camera attached to
a stereomicroscope (see Notes 9 and 10).
3. Remove the culture medium by sucking, and pipette ice-cold
methanol (prefixation) gently on the tissues to avoid detach-
374 Katja Nrhi
ment of tissues from the filters. Leave for 5 min, and transfer
filters by watchmaker forceps to Eppendorf tubes to fix the
explants in PFA for 1024 h at 4 C. Continue with gene
expression analysis.
3.5 In Situ 1. To perform ISH on tissue sections, process tissue explants to

Hybridization (ISH) paraffin using standard protocols. Cut serial sections (47 m
of paraffin-embedded tissue explants and process for ISH to
analyze the expression of genes of interest by using labeled
(radioactive or nonradioactive) riboprobes. ISH is performed
according to a protocol described in [23] with modifications
[13, 26]. An example of sample processed by radioactive ISH
is shown in Fig. 3.
2. For whole-mount ISH (with digoxigenin-labeled riboprobes)
the tissue explants are rinsed in PBS and dehydrated in metha-
nol series (each step 510 min at room temperature) after fixa-
tion. Wash with 100 % methanol twice and store samples in
100 % methanol at 20 C until use (can be stored for several
months). Process for ISH according to instructions described
in ([14, 27, 28]; see Note 11). Examples of whole-mount ISH
staining are shown in Fig. 3.
3.6 Hanging-Drop 1. Pipette 3040 L drops of warm culture medium on the lid of
Culture a 35-mm plastic Petri dish. Different signaling molecules or
other molecules are added to culture medium, e.g., 0.53 ng/
L BMP4, 2 ng/L SHH, or 0.12 ng/L FGF4. The control
culture medium is supplemented with the solvent used to dissolve
the protein of interest to eliminate any effects caused by sol-
vent or dilution of the medium.
2. Transfer the dissected tissue samples carefully to the drops
using the capillary force between watchmaker forceps.
Fig. 3 Examples of bead experiments combined with in situ hybridization (ISH) analysis. All explants were
cultured for 24 h before fixation for gene expression analysis. (ac) Inhibition of BMP2-induced Msx2 expres-
sion by SOSTDC1 (Ectodin) in E13 tooth buds [28]. (a) BMP2-releasing bead has induced Msx2 in the immedi-
ate surroundings of the bead. (b) A SOSTDC1 (15 ng/mL)-releasing bead placed next to the BMP2-bead has
markedly reduced the expression of Msx2 and (c) several SOSTDC1-beads have completely inhibited the
inductive effect of BMP2. (d) BMP4-releasing beads on E12 palatal mesenchyme. Whole-mount ISH shows
induction of Msx2 expression in the immediate surroundings of the bead. (e, f) Id1 expression is induced in
explants of E15 calvarial mesenchyme between the approximating parietal bones (p) around (e) BMP2- and (f)
BMP4-releasing beads (arrow). (e) Whole-mount ISH, (f) radioactive ISH on histological section. (g, h) E16
mandibular incisors cultured with (g) a control bead soaked in BSA and (h) a BMP4-releasing bead. Whole-
mount ISH indicates stronger and more enlarged Ameloblastin expression in the explant exposed to BMP4
compared to BSA control. The apical side of the incisor is on left, the proximal side on right, and the lingual side
on top. (Pictures ac courtesy of Johanna Laurikkala, df courtesy of David Rice, and g and h courtesy of
Marika Suomalainen)
376 Katja Nrhi
3. Turn the lid quickly and place on the top of the Petri dish
containing 12 mL of sterile liquid (PBS or distilled water) in
the bottom to prevent evaporation from the hanging drop.
4. Culture as described above (see Subheading 3.4). Culture time
is usually not more than 24 h.
3.7 RNA Isolation 1. After culture in hanging drops, lids are turned upside down to
and cDNA Synthesis allow the collection of tissue samples under stereomicroscope
with small forceps into Eppendorf tubes (see Note 12).
2. Lyse the tissues with 350 L of lysis buffer and isolate RNA
immediately according to manufacturers instructions. Store
RNA samples at 80 C and avoid repeated thawing (use
aliquots!).
3. Quantify the total RNA with UV spectroscopy, absorbance at
260 nm.
4. Synthesize complementary DNA (cDNA) from total RNA
according to instructions specified by the manufacturer.
Transcribe 1001000 ng of total RNA with 500 ng of Random
primers and 100 U of Superscript II. Dilute cDNA samples
with distilled water to get final volume of 100 L and freeze in
aliquots at 20 C.
3.8 Real-Time 1. Prepare standard series (see Note 13) and 10 (e.g., 15 M)
Quantitative primer mix of forward and reverse primers for each gene to be
PCR (qPCR) studied.
2. Prepare master mix: 2 L 10 primer mix, 3 L H2O, 10 L 2
LightCycler 480 SYBR Green I Master. This is for one reaction,
so multiply the volumes of each reagent by number of your sam-
ples (sample cDNA and standards). Also, note that a master mix
has to be prepared for each gene to be studied. Mix well the
prepared master mixes and keep them on ice and protected from
light as they contain the SYBRGreen fluorophore.
3. Pipette 5 L of cDNA or standard into the wells.
4. Pipette 15 L of master mix into each well of the plate. The
final volume is 20 L in each well.
5. Seal the plate with a transparent adhesive foil and use the
default PCR conditions for the real-time qPCR instrument.
6. Normalize data against ranbp1 (tooth, mandible) or keratin 14
(skin) and analyze with an appropriate software. Gene expres-
sion is quantified by comparing the sample data against a dilu-
tion series of PCR products (see Note 13) of the gene of
interest. An example of real-time qPCR data is shown in Fig. 4.
7. All PCR products can be separated on a 2 % agarose gel using
electrophoresis to check for the correct size of the PCR prod-
uct and to eliminate the possibility of primer dimers.
Fig. 4 Gene expression analysis of E13.5 dental epithelia by real-time qPCR after
3 h Hanging-drop culture with recombinant BMP4 and SHH. For each treatment
three replicates that all included three tooth explants were prepared. Sostdc1
(Ectodin, Wise) is induced by BMP4 and Patched-1 by SHH. Expression is shown
as number of transcripts
4 Notes
1. Depending on the tissue, the composition of the optimal cul-

ture medium varies. The culture medium in this protocol is
suitable for most tissues at early stages of development, but
during more advanced stages, different tissues may have special
requirements. For cultures of whole tooth buds, we use cul-
ture medium composed of DMEM and F12. For more
advanced stages of tooth or calvarial bone development, ascor-
bic acid is added to allow deposition of collagen [8, 12].
2. During dissection, it is recommended to use disposable needles
instead of other instruments, such as scalpels or iris knives,
because needles need no sharpening or sterilization. The size of
the needles can be chosen. In addition, the syringes do not need
to be changed every time as they do not have to be absolutely
sterile. To preserve the tissue structure, dissecting should be
done by determined cuts avoiding stretching and tearing of tis-
sue. Glass Petri dishes are superior to plastic dishes because
needles easily scrape and loosen pieces from the plastic surface.
3. As supporting material, lens paper may be used for large tissue
pieces. Nuclepore filters with different pore sizes (0.058 m)
may also be used. The maximum thickness of filters is approx
10 m which allows good diffusion of the medium to the tis-
sue. Small pores (0.050.2 m) allow better examination of
the explants in the stereomicroscope using transmitted light,
but the tissues tend to detach from these filters more readily
during treatments after culture. Therefore, a larger pore size
(up to 0.6 m) may be preferable, depending on the
experiment.
378 Katja Nrhi
4. The preparation and dissection of tissues should be done as

quickly as possible to promote survival of the tissues. Embryos
waiting to be dissected should be kept on ice in a Petri-dish
with D-PBS and only one uterus at a time should be prepared.
The dissected tissues should be transferred to the culture dishes
and the incubator within 23 h.
5. To study the reciprocal interactions between epithelium and
mesenchyme the interacting tissues need to be enzymatically
separated from each other. Various manipulations can be per-
formed after which their advancing development is followed.
Separation of epithelium and mesenchyme is described in [29].
6. Signaling molecules include proteins, growth factors, and their
antagonists [30], as well as other molecules for example reti-
noic acid [15].
7. The experimental and control beads can be placed on opposite
halves of the jaw. Antagonistic and synergistic functions of dif-
ferent regulatory molecules can be examined by placing beads
releasing different signals, or a signal and its putative antago-
nist near each other on the same explant (Fig. 3).
8. Although organs are usually cultured for 12 weeks max, it is
possible to culture tooth buds for 36 weeks and examine the
formation of roots [31].
9. Reporter mouse lines carrying green fluorescent protein (GFP)
labeled reporter constructs are useful if one wishes to follow
the development of specific structure or gene expression dur-
ing advancing development in vitro. GFP expression is detected
by fluorescent stereomicroscope. For example, a mouse line
expressing GFP in the Shh locus [32] is useful for tooth devel-
opment studies because it allows the visualization of the
dynamics of the formation of enamel knot signaling centers
that express Shh locally [33, 34]. An example of culture of
Shh-GFP tooth buds is shown in Fig. 2.
10. Reporter gene expression can be used to follow, e.g., cell move-
ments in freshly cut tissue slices (150 m), e.g., of coronally
sectioned embryonic incisors with a vibratome (Leica) [35].
For this, the cut tissue slices are placed in Trowell-type cell cul-
ture conditions for 2 h to recover, live-stained with additional
fluorescent markers if necessary (e.g., Draq5/nuclear marker)
prior to laser scanning confocal microscopy analysis. During
8-h confocal microscope imaging, the tissue slices are kept in an
environmental chamber (5 % CO2 humidified and maintained at
37 C) [35].
11. After whole-mount ISH analysis it is possible to analyze the
gross morphology of the explants (e.g., to differentiate between
epithelial vs. mesenchymal staining) from vibratome sections.
For this purpose explants are fixed for 24 h in PFA, rinsed in
PBS, embedded in gelatin/albumin, and cut to thick sections

by vibratome (30200 m) [18, 21, 28].
12. In hanging drops, several tissue pieces can be placed in one
drop depending on the size of the tissue. For instance, three to
four E13 tooth germs or E11 mandibles fit well into one drop.
13. We use standard series of 106, 105, 104, and 103 amplicons in
5 L. It has to be prepared for all the genes to be studied. To
calculate the number of molecules of a PCR amplicon you need
to know the length (bp) and the concentration (g/L) of the
amplicon. First, calculate the molecular weight (length 650 g/
mol) of one amplicon. Second, calculate the amount of ampli-
cons in 1 L (concentration g/L 106 L) and then divide the
product with the molecular weight to resolve the concentration
of amplicon as mol/L. By using Avogadros number
(6.022 1023), calculate the number of molecules in liter and
further in microliter.
References
1. Tummers M, Thesleff I (2009) The impor- 9. Voutilainen M, Lindfors P, Mikkola ML (2013)
tance of signal pathway modulation in all Protocol: ex vivo culture of mouse embryonic
aspects of tooth development. J Exp Zool mammary buds. J Mammary Gland Biol
312B:309319 Neoplasia 18:239245
2. Trowell OA (1959) The culture of mature 10. Saxn L (1966) The effect of tetracycline on
organs in a synthetic medium. Exp Cell Res osteogenesis in vitro. J Exp Zool
16:118147 162:269294
3. Grobstein C (1953) Inductive epithelio- 11. Thesleff I, Lehtonen E, Wartiovaara J, Saxn L
mesenchymal interaction in cultured organ (1977) Interference of tooth differentiation
rudiments of the mouse. Science 118:5255 with interposed filters. Dev Biol 58:197203
4. Saxn I (1973) Effects of hydrocortisone on 12. Partanen AM, Ekblom P, Thesleff I (1985)
the development in vitro of the secondary pal- Epidermal growth factor inhibits tooth mor-
ate in two inbred strains of mice. Arch Oral phogenesis and differentiation. Dev Biol
Biol 18:14691479 111:8494
5. Saxn L, Lehtonen E, Karkinen-Jskelinen 13. Vainio S, Karavanova I, Jowett A, Thesleff I
M, Nordling S, Wartiovaar J (1976) (1993) Identification of BMP-4 as a signal
Morphogenetic tissue interactions: mediation mediating secondary induction between epi-
by transmissible signal substances or through thelial and mesenchymal tissues during early
cell contacts? Nature 259:662663 tooth development. Cell 75:4558
6. Nogawa H, Takahashi Y (1991) Substitution 14. Jernvall J, Aberg T, Kettunen P, Keranen S,
for mesenchyme by basement-membrane-like Thesleff I (1998) The life history of an embry-
substratum and epidermal growth factor in onic signaling center: BMP-4 induces p21 and
inducing branching morphogenesis of mouse is associated with apoptosis in the mouse tooth
salivary epithelium. Development enamel knot. Development 125:161169
112:855861 15. Mitsiadis T, Muramatsu T, Muramatsu H,
7. Nogawa H, Ito T (1995) Branching morpho- Thesleff I (1995) Midkine (MK), a heparing-
genesis of embryonic mouse lung epithelium in binding growth/differentiation factor, is regu-
mesenchyme-free culture. Development lated by retinoic acid and epithelial-mesenchymal
121:10151022 interactions in the developing mouse tooth,
8. Kim HJ, Rice DPC, Kettunen PJ, Thesleff I and affects cell proliferation and morphogene-
(1998) FGF-, BMP- and Shh-mediated signal- sis. J Cell Biol 129:267281
ing pathways in the regulation of cranial suture 16. Laurikkala J, Mikkola M, Mustonen T, Aberg
morphogenesis and calvarial bone develop- T, Koppinen P, Pispa J, Nieminen P, Galceran
ment. Development 125:12411251 J, Grosschedl R, Thesleff I (2001) TNF signal-
380 Katja Nrhi
ing via the ligand-receptor pair ectodysplasin Eda-A1/Edar pathway reveals an unexpected
and edar controls the function of epithelial sig- role of ectodysplasin as inhibitor of Wnt signal-
naling centers and is regulated by Wnt and ling in ectodermal placodes. Dev Biol
activin during tooth organogenesis. Dev Biol 320:6071
229:443455 26. Vaahtokari A, Vainio S, Thesleff I (1991)
17. Wang XP, Suomalainen M, Felszeghy S, Associations between transforming growth fac-
Zelarayan LC, Alonso MT, Plikus MV, Maas tor 1 RNA expression and epithelial-
RL, Chuong CM, Schimmang T, Thesleff I mesenchymal interactions during tooth
(2007) An integrated gene regulatory network morphogenesis. Development 113:985994
controls stem cell proliferation in teeth. PLoS 27. Kettunen P, Thesleff I (1998) Expression and
Biol 5:13241333 function of FGFs-4, -8, and -9 suggest func-
18. Jrvinen E, Salazar-Ciudad I, Birchmeier W, tional redundancy and repetitive use as epithe-
Taketo MM, Jernvall J, Thesleff I (2006) lial signals during tooth morphogenesis. Dev
Continuous tooth regeneration in mouse in Dyn 211:256268
induced by activated epithelial Wnt/-catenin 28. Mustonen T, Tmmers M, Mikami T, Itoh N,
signaling. Proc Natl Acad Sci U S A 103: Zhang N, Gridley T, Thesleff I (2002) Lunatic
1862718632 fringe, FGF, and BMP regulate the Notch
19. Rice DP, Aberg T, Chan Y, Tang Z, Kettunen pathway during epithelial morphogenesis of
PJ, Pakarinen L, Maxson RE, Thesleff I (2000) teeth. Dev Biol 248:281293
Integration of FGF and TWIST in calvarial 29. Sahlberg C, Mustonen T, Thesleff I (2002)
bone and suture development. Development Explant cultures of embryonic epithelium:
127:18451855 Analysis of mesenchymal signals. Methods Mol
20. Rice R, Spencer-Dene B, Connor EC, Gritli- Biol 188:373382
Linde A, McMahon AP, Dickson C, Thesleff 30. Laurikkala J, Kassai Y, Pakkasjrvi L, Thesleff I,
I, Rice DP (2004) Disruption of Fgf 10/ Itoh N (2003) Identification of a secreted
Fgf2b-coordinated epithelial-mesenchymal BMP antagonist, ectodin, interacting BMP,
interactions causes cleft palate. J Clin Invest FGF, and SHH signals from the tooth enamel
113:16921700 knot. Dev Biol 264:91105
21. Nrhi K, Jrvinen E, Birchmeier W, Taketo 31. Tmmers M, Yamashiro T, Thesleff I (2007)
MM, Mikkola ML, Thesleff I (2008) Sustained Modulation of epithelial cell fate of the root
epithelial -catenin activity induces precocious in vitro. J Dent Res 86:10631067
hair development but disrupts hair follicle 32. Harfe BD, Scherz PJ, Nissim S, Tian H,
down-growth and hair shaft formation. McMahon AP, Tabin CJ (2004) Evidence for
Development 135:10191028 an expansion-based temporal Shh gradient in
22. Nrhi K, Tummers M, Ahtiainen L, Itoh N, specifying vertebrate digit identities. Cell 118:
Thesleff I, Mikkola ML (2012) Sostdc1 defines 517528
the size and number of skin appendage plac- 33. Kavanagh KD, Evans AR, Jernvall J (2007)
odes. Dev Biol 15:149161 Predicting evolutionary patterns of mamma-
23. Wilkinson DG, Green J (1990) In situ hybrid- lian teeth from development. Nature
ization and the three-dimensional reconstruc- 449:427432
tion of serial sections. In: Copp AJ, Cockroft 34. Munne PM, Tummers M, Jrvinen E, Thesleff
DL (eds) Postimplantation mammalian I, Jernvall J (2009) Tinkering with the induc-
embryos: a practical approach. IRL Press, tive mesenchyme: Sostdc1 uncovers the role
Oxford, pp 155171 for dental mesenchyme in limiting tooth induc-
24. James MJ, Jrvinen E, Wang XP, Thesleff I tion. Development 136:393402
(2006) Different roles of runx2 during early 35. Juuri E, Saito K, Ahtiainen L, Seidel K,
neural crestderived bone and tooth develop- Tummers M, Hochedlinger K, Klein OD,
ment. J Bone Miner Res 21:10341044 Thesleff I, Michon F (2012) Sox2+ stem cells
25. Fliniaux I, Mikkola ML, Lefebvre S, Thesleff I contribute to all epithelial lineages of the tooth
(2008) Identification of dkk4 as a target of via Sfrp5+ progenitors. Dev Cell 23:1728
Chapter 22
Oral Epithelial Cell Culture Model forStudying

thePathogenesis ofChronic Inflammatory Disease
MikeR.Milward, MartinR.Ling, MelissaM.Grant, andIainL.C.Chapple
Abstract
The interactions between bacteria, epithelium, and neutrophilic polymorphonuclear leukocytes (neutrophils)
are the key to the initiation and progression of many chronic inflammatory-immune diseases. In addition,
all can be influenced by external factors, such as micronutrients, thereby providing potentially novel
approaches to therapy. This chapter will therefore provide detailed methods for core techniques involved
in studying cellular and molecular epithelial responses to a bacterial challenge in relation to chronic inflam-
matory disease pathogenesis and therapy.
Key words Epithelium, Immunocytochemistry, Gene expression, Polymerase chain reaction,

Electrophoresis, H400 oral epithelial cells, NF-kappa B, DNA, RNA
1 Introduction
Epithelial tissues were traditionally believed to act purely as a phys-

ical barrier designed to protect organisms from the external envi-
ronment, an example of this being protection from microbial
ingress into the underlying connective tissues. However, in recent
years, it has been shown that epithelial cells rather than just pos-
sessing an innate barrier function, play an active role in the host
response to bacterial infections [1]. The epithelium responds to
bacterial challenge by facilitating the initiation of an innate
inflammatory-immune response. The ability of bacteria to adhere
to host cells is important in cell stimulation. In vivo studies have
demonstrated a correlation between the number of bacteria
attached to the periodontal epithelium and the magnitude of the
inflammatory response [2]. Porphyromonas gingivalis has been
shown to attach to a range immortalized epithelial cell lines, as well
as to primary gingival epithelial cells [3, 4]. A variety of mecha-
nisms exist which allow bacterial adhesion to epithelial cells; these
include fimbriae, proteases, hemagglutinins, and lipopolysaccha-
rides (LPS) [5]. Adhesion and invasion by periodontal pathogens
381
382 Mike R. Milward et al.
is thought to be crucial in the pathogenesis of periodontitis especially

in the initial stages; however, some studies have suggested that
adhesion is more important than invasion in terms of activation of
a pro-inflammatory response by epithelial cells [6], through pat-
tern recognition receptor binding.
In order to characterize epithelial responses to bacterial chal-
lenge invitro epithelial models may be employed. While primary
explant cell cultures are closer to invivo situations, they show sig-
nificant heterogeneity and limit repeatability studies due to their
mortality. Adopting well-characterized cell lines permits the estab-
lishment of core and reproducible experimental conditions, in the
absence of interference from antimicrobial use, which may then be
evaluated against primary explants. This chapter describes an
example of such by utilizing an immortal oral epithelial cell line
(H400), chosen because it is currently one of the best character-
ized oral epithelial cell lines available [7] and has a phenotype
closely matching that of gingival epithelium.
H400 cells may be exposed to a variety of bacterial stimulants
and subsequent downstream cellular responses characterized at the
gene and protein expression level. Specifically, activation of NF-kB
(a key pro-inflammatory gene transcription factor), which induces
gene expression changes and subsequently downstream cytokine
production (biologically active proteins responsible for initiating
the inflammatory/immune response); these stages will be described
in this chapter (Fig.1).
Fig. 1 The key biological stages following epithelial cell stimulation and activation of the NF-kB pathway.
(a) Translocation of NF-kB to the nucleus, (b) gene expression changes, (c) production of cytokines
Oral Epithelial Cell Culture Model forStudying thePathogenesis ofChronic 383
Cell culture is a technique-sensitive discipline in which careful,

aseptic manipulation of the cells is required to produce consistent
results. All cell culture-based experimental work should be con-
ducted in a positive flow biological hood to help minimize poten-
tial contamination. Supplementing cell culture growth media with
antibiotics to suppress bacterial contamination is in widespread
use, and is required when using contaminated biological samples
(explants); however, this is an unwanted confounder with the
potential to influence experimental results. It may also give a false
sense of security and mask poor aseptic technique. It is therefore
recommended that all experiments are conducted using this model
system without antibiotic supplementation.
One limitation of mono-bacterial stimulation of oral epithelial
cells is that invivo, periodontal bacteria exist within a plaque bio-
film. The biofilm comprises a wide variety of bacterial species, con-
stantly changing according to local environmental conditions.
However, invitro investigation of this interaction is difficult due to
the complexity and dynamic nature of the biofilm. The technique
described in this chapter focuses on utilizing single whole dead
bacteria in an attempt to characterize how different bacteria stimu-
late the pocket/sulcular epithelium. Techniques for living bacteria
are beyond the scope of this chapter. Porphyromonas gingivalis, and
Fusobacterium nucleatum will be discussed as example species as
they are key bacteria that have been implicated in periodontal dis-
ease pathogenesis.
This chapter will present methods for immunocytochemical
analysis of NF-kB. Immunocytochemistry is a technique based on
the interaction of an antibody with an insoluble immunoreactive
tissue or cellular antigen, then localization of the site using a label
(e.g., an enzyme or fluorescent marker). Detection is dependent
upon the label used, and can be by ultra-violet, visible light, or elec-
tron microscopy. A number of methods can be employed with anti-
gens being localized using direct, indirect, and multilayer techniques.
These differ in the way in which the label is located at the site of the
tissue antigen/antibody reaction) [8]. In order to ensure that this
staining technique is specific for the target protein (in this case
NF-B), a number of controls must be utilized, such as the use of a
known positive section, replacing primary antibody with (1) non-
immune serum, (2) antibody known to have a different localization
pattern, (3) buffer, as well as preabsorbed primary antibody with (a)
purified homologous antigen, (b) non-cross reacting antigen (neg-
ative controls). These controls are run during development of the
NF-B staining technique, once specificity has been ascertained,
subsequent staining of samples can be performed using a positive
staining control (Ki-67) and a negative control (primary antibody
replaced by diluents). Following staining, visual localization of
NF-B can be determined where translocation of NF-B into
the nucleus indicates NF-B activation. The mechanism of
Fig. 2 Diagrammatic representation of immune-cytochemical methodology. (1)

Cell substrate, (2) primary monoclonal antibody, (3) biotinylated secondary anti-
body (multi-link), (4) peroxidase linked to streptavidin (label), (5) diaminoben-
zidene (DAB)
immunocytochemical staining is illustrated in Fig.2. The practical

steps involved are summarized in Fig.3 as described [9, 10].
In addition to detection of nuclear translocation of NF-kB by
standard microscope slide-based immuno-cytochemistry, high-
throughput technology can also be used to determine levels of
NF-kB activation. This technique allows quantitative analysis of
NF-kB translocation, with the added benefit of high sample
throughput. The basic principal of this method is similar to the
staining techniques described for slide-based analysis, but uses a
fluorescent label. However, the main difference is the way in which
stained images are analyzed. This technology utilizes an automated
fluorescence microscopic imaging system that is computer driven
allowing rapid cell identification as well as quantification of a vast
array of parameters that include cell size, number, shape, nuclear
intensity, cytoplasmic intensity. Its use in this context allows determi-
nation of levels of NF-B staining in the nuclei and the surrounding
Wash slides quickly in PBS x3

Fix in dry Acetone for 15mins
Fix & Dry Air dry for 15 minutes
Monoclonal Apply appropriate monoclonal antibody

Antibodies (PBS with bovine serum albumin (10mg/ml)
60 mins
Wash in PBS (3x2 min)
Multi link Overlay with multilink (1:50 dil in buffer)

60 mins
Label Overlay with label (1:50 dil in buffer)

60 mins

Overlay fresh DAB (10mg in 20ml PBS
DAB filtered and 25microl of 30% H2O2 prior to
use 5 mins
Wash in tap water(2 min)
Stain Overlay with copper sulphate (0.5%
development w/v in NaCl) 5 min
Wash in tap water

Counter Counter stain with Meyers Haematoxylin
stain 30 sec
Wash in tap water
Dry and mount Pass slide through graded alcohols

and xylene (30 seconds in each) and
mount in XAM
Fig. 3 Flow diagram summarizing the immune-cytochemical staining methodology
cytoplasmic area. Use of high-throughput technology requires cells

to be grown in black flat bottom 96-well tissue culture grade plates
as described within this chapter, with cells then treated dependent
on the particular experimental protocol being used.
Following methods to determine NF-kB activation in an epi-
thelial cell model, techniques to determine gene expression
changes are included within this chapter. Once activated NF-kB
translocates and binds to nuclear DNA causing a wide range of
gene transcription changes resulting in mRNA production and
subsequent downstream cytokine production. This chapter briefly
describes isolation of RNA from the model system along with the
steps required to analyze mRNA levels. The latter includes the use
of polymerase chain reaction (PCR), a technique that allows rep-
lication of a portion of the DNA molecule to produce multiple
copies of this DNA region that can then be detected using agarose
gel electrophoresis.
2 Materials
2.1 Epithelial Cell 1. Dulbeccos modification of Eagles medium (DMEM) supple-

Culture Media mented aseptically with 2mML-glutamine, hydrocortisone
andReagents (0.25mg per 500mL medium), and 10% (v/v) fetal calf serum.
2. 5% CO2 Incubator.
3. Microbiological fridge (5C).
4. Cell culture flasks.
5. Passage reagent: Hanks balanced salt solution containing
0.5g/L trypsin (1:250) and 0.2g/L EDTA.4Na.
6. Buffered saline solution: Dulbeccos phosphate buffered saline
(D-PBS) without calcium and magnesium, containing 0.2g/L
KCl, 0.2g/L KH2PO4, 8g/L NaCl, 1.15g/L anhydrous
Na2HPO4.
7. H400 oral epithelial cell (OEC) line is derived from a squa-
mous cell carcinoma (2040mm, stage 2, moderately differen-
tiated, node negative tumor) of the alveolar process in a
55-year-old female patient, and has been well characterized
(e.g., cytokeratins) indicating close similarities to gingival epi-
thelium [7]. It is a suitable cell line for use in the described
model system.
8. Ensure sterility of all reagents by incubating 10mL of supple-
mented media at 37C in 5% CO2 overnight and checked for
evidence of bacterial/fungal contamination (indicated by
turbidity and pH change).
2.2 Cell Culture 1. Heat-inactivated foetal calf serum.

Methodology (Fig.4) 2. Dimethyl sulphoxide (DMSO).
4. Liquid nitrogen.
5. Water bath.
6. Pre-warmed DMEM cell culture media.
7. Centrifuge.
8. Pasteur pipette.
9. Cell culture flask.
10. Microbiological incubator.
11. Inverted microscope.
12. Warm Dulbeccos phosphate buffered saline (D-PBS) without
calcium and magnesium, containing 0.2g/L KCl, 0.2g/L
KH2PO4, 8g/L NaCl, 1.15g/L anhydrous Na2HPO4.
13. Trypsin-EDTA: Hanks balanced salt solution containing
0.5g/L trypsin (1:250) and 0.2g/L EDTA.4Na.
Fig. 4 Substrates for H400 cell growth; (a) Petri dish with multi-well side, (b) printed
four-well slide, (c) plastic cell culture flask, (d) 96-well flat bottomed plate
14. Glass slides printed with four wells.

15. Hot air oven.
16. Sterile Petri dishes.
17. Warmed DMEM (37C).
18. Square plastic dish.
19. Flat bottomed 96-well plates.
2.3 Bacterial Growth 1. Blood agar: 40g trypticase soy base powder and 15g agar base
(See Note 1) to 1L dH2O and mix. Autoclave at 121C for 15min. While
just above setting temperature (to prevent blood lysis), add 5%
defibrinated sheep blood and gently mix.
2. Trypticase soy broth: Dissolve 40g of trypticase soy base powder
in 1L dH2O.
3. Petri dishes.
4. Bijoux bottles.
5. Reconstituted bacteria.
7. Anaerobic incubation chamber (80% nitrogen, 10% carbon
dioxide, 10% hydrogen).
8. Centrifuge.
9. PBS.
10. Water bath.

11. Spectrophotometer.
2.4 Immunocyto- 1. PBS.

chemistry ofCells 2. Dry acetone.
Grown onMulti-Well
3. Darkened humidified box.
Slides
4. Monoclonal antibody to NF-kB p65 subunit (clone F-6).
5. Positive staining control Ki-67 (clone MM1).
6. Negative staining control (replacement of the primary antibody
with PBS).
7. Incubator.
8. Multilink (biotin-labeled goat anti-mouse/rabbit Ig).
9. Label (peroxidase linked to avidin, Biogenex).
10. Diaminobenzidene reagent (filter and H2O2).
11. 0.5% copper sulfate in 0.15M NaCl.
12. Mayers haematoxylin.
13. Graded alcohols.
14. Xylene.
15. XAM mounting medium.
2.5 High-Throughput 1. Formaldehyde.

Immuno- 2. PBS.
Cytochemistry
3. Primary antibody (anti-NF-kB p65 monoclonal antibody F-6).
4. 0.1M phosphate buffer.
5. Bovine serum albumin.
6. Secondary antibody (anti-STAT1 rabbot polyclonal immuno-
globulin conjugated to a fluorescent label).
7. Detergent buffer.
8. ArrayScan HCS Imaging Cytometer.
9. ArrayScan II Data Acquisition and Data Viewer 3.0.
2.6 mRNA forGene 1. Commercially available RNA extraction kit.

Expression 2. Lysis buffer containing a reducing agent that induces cell lysis
while protecting DNA and RNA from degradation.
3. Microscope.
4. Vortex mixer.
5. Ethanol.
6. DNase (DNase kit).
7. Centrifuge.
8. RNase-free water.
9. Omniscript reverse transcriptase kit.
10. Oligo (dT) primer.
11. RNase inhibitor.
12. Microcon YM-30 centrifugal filter unit.
13. dH2O.
14. Spectrophotometer (Biophotometer).
2.7 Polymerase 1. Ice.

Chain Reaction(PCR) 2. RedTaq PCR system containing RedTaq ready reaction mix,
dH2O, and forward and reverse primers.
3. Thermal Cycler (Eppendorf Mastercycler Gradient).
2.8 Agarose Gel 1. 1.5% Agarose gel (0.6g agarose) to 40mL 1 Tris-acetate-EDTA
Electrophoresis (TAE) buffer.
2. Bunsen burner.
3. Ethidium bromide.
4. Electrophoresis plate and sample well-forming comb.
5. Electrophoresis tank.
6. Hyperladder IV (100-bp ladder).
7. Tracking dye (in RedTaq mix).
8. UV light source.
9. Image-capturing equipment and image analysis software.
3 Methods
3.1 Production 1. Seed cells into cell culture flasks and allow to grow until cell
andMaintenance confluence is achieved.
ofEpithelial Cell Model 2. At (or just before) confluence, perform cell passage (disruption
System of monolayer, and reseeding of cells into a new flask so the
growth cycle can continue).
3. Passage using commercially available reagent containing
0.5g/L trypsin (1:250) and 0.2g/L EDTA4Na in Hanks
balanced salt solution.
3.2 Bacterial 1. Pour blood agar into petri dishes, allow to set, and store plates
Culture/Growth at 4C prior to use.
2. Dispense the trypticase soy broth suspension into bijou bottles
and autoclaved at 121C for 15min.
3. Allow broth to cool and store at 4C prior to use.
4. Reconstitute bacteria and inoculate onto blood agar plates.

5. Incubate anaerobically (80% nitrogen, 10% carbon dioxide,
10% hydrogen; anaerobic chamber) at 37C for 5 days.
6. Examine plates and check to ensure purity by Gram staining
and colony morphology.
7. Colonies are harvested and seeded into 10mL trypticase soy
broth (for P. gingivalis growth additional supplementation of
hemin, N-acetyl muramic acid and vitamin K) and incubated at
37C in an anaerobically for 4 days.
8. Centrifuge bacterial suspension to produce a pellet.
9. Wash pellet three times in sterile phosphate buffered saline (PBS).
10. Heat-inactivate at 100C for 10min.
11. Measure optical density at 600nm (OD600) of the resultant
suspension using a spectrophotometer, and use standard curve
to determine cell number.
12. A suspension containing 4108 bacterial cells per mL is pro-
duced which is stored in aliquots at 20C prior to use.
3.3 Cell Culture 1. Preserve cells for future use by preparing 1mL suspension of
1106 cells containing 20% (v/v) heat inactivated fetal calf
serum and 10% (v/v) dimethyl sulphoxide (DMSO).
2. Freeze suspensions slowly to 80C overnight before long-
term storage under liquid nitrogen.
3. When required, remove cell aliquots from liquid nitrogen and
rapidly defrost in water bath at 37C.
4. Once defrosted, transfer suspension to 9mL of pre-warmed
(37C) DMEM cell culture media and gently mix.
5. Centrifuge at between 600 and 800g for 10min.
6. Remove resulting supernatant gently using a Pasteur pipette to
prevent disruption of the cell pellet.
7. Resuspend pellet in 1mL fresh pre-warmed (37C) DMEM.
8. Mix resultant suspension gently and use to seed 75cm2 cell
culture flask containing 9mL of pre-warmed (37C) DMEM.
9. Incubate flask at 37C in 5% CO2 for 4 days.
10. On day 4, inspect flask for bacterial/fungal contamination
(turbidity/pH change).
11. Also examine microscopically (with the inverted microscope)
to visualize cell growth and typical epithelial cell morphology.
12. Remove old growth media and feed monolayer with 10mL
fresh preheated (37C) DMEM.
13. Re-incubate flask at 37C for a further 23 days (time depen-
dent on cell growth rate).
14. Passage when cells reach (near) confluence and pH is neutral

(as indicated by the phenol read indicator present in DMEM).
Over-incubation results in exhaustion of media nutrients and
creates an acidic pH leading to cell death.
3.3.1 Cell Passage (See 1. Remove old media from the cell monolayer.
Notes 2 and3) 2. Wash with warm (37C) D-PBS and remove.
3. Add 4mL of T-EDTA and incubate for 10min (37C) with
occasional mixing.
4. Confirm dissociation of the monolayer microscopically using
an inverted microscope.
5. Transfer cell suspension to a sterile universal container contain-
ing 4mL of warm (37C) DMEM, and mix.
6. Centrifuge at 150g for 10min.
7. Remove supernatant and discarded.
8. Resuspend cell pellet in 5mL warm (37C) DMEM.
9. Count cells and seed flasks (cell number dependent on sub-
strate being used) (Table1).
3.3.2 Cell Growth 1. Growth on multi-well slides is required for immuno-

onGlass Multi WellSlides cytochemical analysis.
2. Use glass slides printed with four wells, sterilize using a hot air
oven (150C for 2h), allow to cool, aseptically transfer to
sterile plastic Petri dishes.
3. Add 15mL warmed (37C) DMEM.
4. The Petri dish is placed inside a square plastic dish to help pre-
vent contamination and evaporation of DMEM.
5. Following passage, seed 4105 cells into Petri dish mix well by
agitating petri dish and incubate at 37C in 5% CO2.
3.3.3 Growing Cells 1. Growth on 96-well plates is required for high-throughput

in96-Well Plates immuno-cytochemical analysis.
2. Following passage seed cells (8103 cells in 150L DMEM
per well) into flat bottomed 96-well plates.
Table 1
Cell counts for seeding a range of cell culture receptacles
Receptacle Volume DMEM Cell number

25cm2 cell culture flask 5mL 2105
75cm2 cell culture flask 10mL 5105
Petri dish 15mL 4105
96 Well plate 150L (per well) 8103
3. Grow cells for between 24 and 48h until a semi-confluent

monolayer is achieved.
4. Treat monolayer depending upon the experimental protocol
being used (e.g., staining for NF-B which is discussed later in
this chapter).
3.4 Immunocyto- 1. Quickly remove multi-well glass slides from the growth media,
chemical Analyses and wash in PBS for 2min3 (3 changes of PBS).
ofNF-kB 2. Remove excess PBS and fix slides in dry acetone (15min at
3.4.1 Staining Procedure room temperature).
(See Note 4) 3. Air dry slides (10min) and place in darkened humidified box
prior to staining.
4. Prepare and apply monoclonal antibody to NF-kB p65 sub-
unit (clone F-6; 1:100) to appropriate slide wells, a positive
staining control Ki-67 (clone MM1; 1:100) as well as a nega-
tive staining control (replacement of the primary antibody
with PBS).
5. Incubate for 60min.
6. Wash in PBS (32min).
7. Dry slides to remove excess moisture from around the wells
then overlay with Multilink (biotin-labelled goat anti-mouse/
rabbit Ig, 1:50 dilution) and incubate for 60min.
8. Wash slides in PBS (32min) and overlay with label (peroxi-
dase linked to avidin. 1:50 dilution) and incubate for 60min.
9. Wash slides in PBS (32min), remove excess PBS, and overlay
with freshly prepared diaminobenzidene (DAB) reagent
(10mg in 20mL PBSfiltered then 25L H2O2 added
used immediately 5min), incubate for 5min.
10. Wash in water for 2min, then apply copper sulphate solution
(0.5%w/v in 0.15M NaCl) for 5min (this step darkens the
brown staining allowing easier visualization).
11. Wash and counterstain for 30s using Meyers haematoxylin.
12. Wash again, dehydrate in graded alcohols, clear in xylene, and
mount in XAM mounting medium.
3.4.2 Quantification In order to determine levels of NF-B activation, cell counts are
ofCell Translocation performed on stained cell mono layers. A binocular microscope
fitted with an eyepiece graticule divided into 1010 squares is
used, with slides viewed at 100 magnification. In order for cells to
be classified as demonstrating NF-B activation (i.e., nuclear
translocation), the presence of nuclear staining for p65 is
required with no visible staining present within the c ytoplasm,
where any ambiguity exists cells are recorded as negative.
Cell counts were performed on random fields and numbers of
1 2 3 4 5 6 7 8 9 10
11 12 13 14 15 16 17 18 19 20
21 22 23 24 25 26 27 28 29 30
31 32 33 34 35 36 37 38 39 40
41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60
61 62 63 64 65 66 67 68 69 70
71 72 73 74 75 76 77 78 79 80
81 82 83 84 85 86 87 88 89 90
91 92 93 94 95 96 97 98 99 100
b
Hunting Curve
18
17
Cell number
16
15
14
13
12
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Field number
Fig. 5 (a) Represents a microscopic field using a graticule to divide into 0.1mm by 0.1mm squares, (b)
example of a hunting curve
cells showing nuclear NF-B translocation determined. A graph

(Hunting curves) was produced by plotting field number vs. cell
number (Fig.5); this provides the number of fields that should be
counted to be representative of the sample (i.e., the number of
fields required to show reduced graph variability). The total and
mean number of positive cells for each specimen is determined and
used to calculate the percentage of cells demonstrating NF-B
activation.
3.5 High-Throughput Use of high-throughput technology requires cells to be grown in

Immuno- black 96-well tissue culture grade flat bottom plates as previously
Cytochemistry described, with cells then treated dependent upon the particular
experimental protocol being used. Once experiments are com-

pleted the cells are fixed and stained.
3.5.1 Fixing ofCell 1. Remove growth media.

Monolayer 2. Fix cells using 1% formaldehyde solution for 20min.
3. Remove and wash twice with 100L PBS.
4. Fill wells with PBS, seal wells, and store at 4C prior to
analysis.
3.5.2 Staining 1. Wash plates with PBS.

2. Incubate with 100L (per well) primary antibody (anti-NF-
kB p65 monoclonal antibody (F-6) diluted 1:100in 0.1M
phosphate buffer pH 7.8 with 0.1% bovine serum albumin) for
1h at room temperature.
3. Wash cells three times in wash buffer and incubate with sec-
ondary antibody (anti-STAT1 rabbit polyclonal immunoglob-
ulinconjugated to a fluorescent label) for a further hour.
4. Add detergent buffer to wells for 15min, wash with PBS, and
add 200L of detergent buffer to each well prior to analysis.
3.5.3 Data Acquisition Following staining, data acquisition utilizes an ArrayScan HCS
andAnalysis imaging cytometer (Fig.6), consisting of an automated fluorescent
imaging microscope that allows collection of data on the spatial
distribution of the fluorescent marker. The scanner reads multiple
fields to build up a picture of each individual well. Software allows
identification of individual cells that are then subdivided into
Fig. 6 Cellomics ArrayScan II imaging cytometer, used for high-throughput

immunocytochemical analysis (Imagen Biotech, UK)
Fig. 7 Quantification of nuclear and cytoplasmic intensity; (a) shows a diagrammatic representation of cell
compartmentalization, (b) shows how this is related to an actual scanned image
nuclear and cytoplasmic compartments (Fig.7). The software can

identify the edge of the nucleus and then creates areas two pixels
either side of this line, therefore producing separate nuclear and
cytoplasmic compartments. The fluorescent intensity is deter-
mined for both nuclear and cellular compartments. A software
package (ArrayScan II Data Acquisition and Data Viewer) is used
for data acquisition and analysis. The localization of NF-kB is
determined by relative analysis of nuclear and cytoplasmic com-
partment fluorescence.
3.6 mRNA forGene This section briefly describes isolation of RNA from the model
Expression system along with the steps required to analyze levels of mRNA
production using polymerase chain reaction (PCR).
3.6.1 RNA Isolation RNA is extracted from cells grown in culture flasks using commer-
cially available extraction kits. The key stages in this process are:
1. Remove growth media and replace with lysis buffer; this
contains a reducing agent that induces cell lysis while protecting
DNA and RNA from degradation.
2. Examine the cell layer microscopically to ensure complete
cell lysis.
3. Homogenize the lysate, and vortex for 30s, to help shear

genomic DNA as well as reducing the viscosity.
4. Use ethanol to adjust binding conditions, allowing binding of
RNA to the membrane on the spin column.
5. Wash membrane (three times).
6. Use DNase to digest and remove any DNA contamination that
would interfere with downstream analysis.
7. Following further washing with buffer the membrane is dried
by centrifugation.
8. RNA attached to the dried membrane is eluted by the addition
of RNase-free water to the membrane followed by centrifuga-
tion. This step results in RNA being removed from the mem-
brane and present in the water.
3.6.2 Reverse Reverse transcription is performed to generate a DNA strand

Transcription (RT) complementary to the mRNA present within the sample of total
RNA. RT reactions are carried out using the Omniscript reverse
transcriptase kit coupled with an oligo (dT) primer and an
RNase inhibitor. An oligo (dT) primer is used to ensure that
only mRNA species within the total RNA sample were reverse
transcribed.
Carry out reverse transcription in accordance with manufac-
turers instructions, the key stages are:
1. Incubate template RNA and oligo (dT) primers in a heated
block to 80C for 10min (removes any secondary structure
from the RNA sample).
2. Place sample on ice for 5min.
3. Add remaining components of the RT reaction, gently mix and
incubate at 37C for 60min. This step allowed extension of
the transcription product.
4. Inactivate the reverse transcriptase enzyme by heating at 93C
for 5min.
5. Rapidly cool samples on ice.
3.6.3 Concentration This stage has two key aims (1) to remove unincorporated RT
andPurification ofcDNA reaction components, which could interfere with downstream
analysis, and (2) concentration of cDNA in the samples.
1. Dilute samples with 500L dH2O and centrifuged using a
Microcon YM-30 centrifugal filter unit. The filter unit retains
cDNA molecules while allowing contaminants to pass through
the filter.
2. Wash samples with dH2O and centrifuge at 10,000g for
8min to concentrate cDNA.
3. Collect cDNA by inverting the filter into a fresh collection

tube and centrifuging for 3min at 1000g.
4. Quantify the amount of cDNA in the resulting solution using
a spectrophotometer.
3.6.4 Quantification The amount of cDNA is quantified spectrophotometrically using

ofRNA andDNA a Biophotometer. The solution is diluted and the absorbance at
260nm determined. An absorbance value of 1unit at 260nm
corresponds to 40g/mL single stranded DNA.In addition to
quantifying the amount of DNA, purity of the sample is deter-
mined by assessing the ratio between readings taken at 260nm
and 280nm (A260/A280). A highly pure sample is signified by an
A260/A280 of 1.82.1.
3.7 Polymerase This technique allows replication of a portion of the DNA mole-
Chain Reaction cule to produce multiple copies of this DNA region which can then
be detected using electrophoresis of the sample loaded into an
agarose gel.
All PCRs are set up on ice to limit primer dimer formation and
avoid premature reactions. Assays are performed using the RedTaq
PCR system (Sigma-Aldrich, UK) in reaction volumes containing
RedTaq ready reaction mix, dH2O, forward primer, and reverse
primers.
Reactions are amplified in a Thermal Cycler (Mastercycler
Gradient, Eppendorf, UK). A standard PCR program is shown in
Table2, with a schematic diagram of the PCR process in Fig.8.
Initial experiments are performed to determine (1) the annealing
temperature required, and (2) the cycle number required for each
specific primer, which is dependent on the relative abundance of
the target sequence and efficiency of the specific reaction, by opti-
mizing the conditions for each primer this allows better visualiza-
tion of the resulting product on agarose gel.
Samples and controls are normalized to a house-keeping gene
so that meaningful comparisons can be made of gene expression
between control and experimental samples. The gene used for this
purpose is glyceraldehyde-3-phosphate-dehydrogenase (GAPDH)
that is used to normalize levels in all samples. PCR is performed
using GAPDH primers with the products being visualized using
agarose gel electrophoresis. The resulting images are digitally
captured and band intensity determined using image analysis soft-
ware (Aida, Fuji, UK). The calculated differences in band intensities
seen are used to determine the amount of cDNA required to produce
bands of equal intensity for each sample for GAPDH.The process of
normalization is repeated until all bands are of similar intensity on
agarose gel; this ensures the same amount of cDNA is added to each
subsequent reaction, therefore allowing investigation of gene expres-
sion differences between test and control samples.
Table 2
Cycle program for PCR product production
Reaction Temperature
Time
stage (oC)
5 min
Denaturing 94
30 secs Cycle number
dependent on
Annealing Primer dependent 30 secs
primer used
30 secs
Polymerisation 72
10 min
Fig. 8 Schematic diagram showing the stages in the PCR reaction

3.8 Agarose Gel 1. Add 1.5% agarose gel (0.6g agarose) to 40mL 1 Tris-acetate-
Electrophoresis EDTA (TAE) buffer.
3.8.1 Preparation 2. Mix and heat to boiling point, to allow the agarose to
ofAgarose Gel dissolve.
3. Briefly cool, and add ethidium bromide (10mg/mL) (this allows
visualization of nucleic acids when exposed to ultra-violet light).
4. Pour the molten gel into an electrophoresis plate and insert the
sample well forming comb (Fig.9).
5. Allow gel to set at room temperature.
3.8.2 Gel Electrophoresis 1. Place gel into an electrophoresis tank containing sufficient 1
TAE buffer to cover the gel and wells.
2. Remove the well forming comb, thus flooding the resulting
wells with TAE buffer.
3. Load products from the PCR reaction into the appropriate
sample well.
4. Include one well with Hyperladder IV, (100-bp ladder) to
enable size determination of DNA products.
5. Perform electrophoresis at 80120V, with the tracking dye (in
RedTaq mix) allowing visualization of the extent of sample
movement.
6. The gel is run for sufficient time to allow good separation of
sample.
Following electrophoresis, gel images are captured digitally
under ultra violet light and analyzed with image analysis software.
Fig. 9 Agarose gel electrophoresis plate used for visualization of PCR

4 Notes
1. Lyophilized stocks of bacteria are obtained from the American

Type Culture Collection (ATCC; Rockville, MD, USA). This
section describes the growth of two periodontal pathogens
Fusobacterium nucleatum ATCC 10953 and Porphyromonas
gingivalis ATCC 33277.
2. Maintenance of a cell line requires cells to undergo passage as
they approach confluence. This technique disrupts the cell
monolayer to produce a single cell suspension, which is subse-
quently diluted and reseeded into a new flask, starting the pro-
cess of monolayer growth again.
3. After initial seeding (day 0), feed cells on day 4 by replacing
old DMEM with fresh, warmed (37C) DMEM and con-
tinue incubation. Perform experiments between cell passage
420.
4. All procedures are performed at room temperature with PBS
(pH 7.6) used for washing stages and reagent dilution.
Acknowledgments
The methods and diagrams are adapted and updated from Oral
Epithelium in the Pathogenesis of Periodontitis Ph.D. by Michael
R Milward (2010). High content analysis was performed in con-
junction with Imagen Biotech, UK
References
1. Dale BA (2002) Periodontal epithelium: a 5. Lamont RJ, Jenkinson HF (1998) Life below
newly recognized role in health and disease. the gum line: pathogenic mechanisms of
Periodontol 2000 30:7078 Porphyromonas gingivalis. Microbiol Mol Biol
2. Vaahtoniemi LH, Raisanen S, Stenfors LE Rev 62:12441263
(1993) Attachment of bacteria to oral epithelial 6. Eaves-Pyles T, Szabo C, Salzman AL (1999)
cells invivo: a possible correlation to gingival Bacterial invasion is not required for activation
health status. JPeriodontal Res 28:308311 of NF-kappaB in enterocytes. Infect Immun
3. Isogai H, Isogai E, Yoshimura F, Suzuki T, 67:800804
Kagota W, Takano K (1988) Specific inhibition 7. Prime SS, Nixon SV, Crane IJ, Stone A,
of adherence of an oral strain of Bacteroides Matthews JB, Maitland NJ, Remnant L, Powell
gingivalis 381 to epithelial cells by monoclonal SK, Game SM, Scully C (1990) The behaviour
antibodies against the bacterial fimbriae. Arch of human oral squamous cell carcinoma in cell
Oral Biol 33:479485 culture. JPathol 160:259269
4. Duncan MJ, Nakao S, Skobe Z, Xie H (1993) 8. Van Noorden S, Stuart MC, Cheung A,
Interactions of Porphyromonas gingivalis with Adams EF, Polak JM (1986) Localization of
epithelial cells. Infect Immun 61:22602265 human pituitary hormones by multiple
immunoenzyme staining procedures using odontal pathogens. Clin Exp Immunol 148:
monoclonal and polyclonal antibodies. 307324
JHistochem Cytochem 34:287292 Milward MR, Chapple ILC, Carter K,
10.
9. Milward MR, Chapple ILC, Wright HJ, Matthews JB, Cooper PR (2013) Micronutrient
Millard JL, Matthews JB, Cooper PR (2007) modulation of NF-kB in oral keratinocytes
Differential activation of NF-kB and gene exposed to periodontal bacteria. Innate Immun
expression in oral epithelial cells by peri- 19:140151
Chapter 23
Fabrication and Characterization of Decellularized

Periodontal Ligament Cell Sheet Constructs
Amro Farag, Cdryck Vaquette, Dietmar W. Hutmacher,
P. Mark Bartold, and Saso Ivanovski
Abstract
Decellularized tissue-engineered constructs have the potential to promote regeneration by providing a
biomimetic extracellular matrix that directs tissue-specific regeneration when implanted in situ. Recently,
the use of cell sheets has shown promising results in promoting periodontal regeneration. Here, we
describe the fabrication of decellularized periodontal cell sheets with intact extracellular matrix structural
and biological properties. Melt electro spun polycaprolactone (PCL) scaffolds are used as a carrier for the
inherently fragile cell sheets, to provide support during the processes of decellularization. An optimized
decellularization method is outlined using perfusion with a combination of NH4OH and Triton X-100
together with a DNase treatment step for DNA removal. The maintenance of extracellular matrix structural
and biological integrity is important, and here we describe the assessment of these properties using immu-
nostaining for extracellular matrix proteins and ELISA for growth factor quantification.
Key words Cell sheet, Decellularization, Tissue engineering, Scaffolds, Constructs, Perfusion,
Polycaprolactone, Periodontal regeneration
1 Introduction
A novel approach to tissue engineering using cell sheet technol-

ogy has been proposed, whereby contiguous cell monolayers com-
plete with extracellular matrix could be produced from various cell
types including periodontal ligament cells [1]. Periodontal ligament
cell sheets have been shown to promote periodontal tissue regenera-
tion in vivo [2, 3]. Recently, periodontal ligament cell sheets pre-
pared in vitro and delivered using a novel biphasic polycaprolactone
(PCL) scaffold were shown to be capable of simultaneous regenera-
tion of bone and periodontal ligament [4]. Furthermore, periodon-
tal ligament cells were shown to be superior to other periodontal
tissue sources (alveolar bone, gingiva) in promoting new functional
periodontal ligament, alveolar bone, and new cementum formation
in a surgically created rat periodontal defect model [5].
403
404 Amro Farag et al.
While a tissue-engineered cell sheet approach is very promising,

there are several underlying limitations hindering this technology
from becoming translated to clinical practice. One such limitation
is the reliance on the patients own cells, resulting in problems
associated with reproducibility and safety. Another limitation is
that cell sheet technology requires cell culture facilities, technical
expertise, and thus considerable expense to bring this technology
into daily use in the clinic.
To overcome these limitations that are common to many cell-
based tissue engineering strategies, the use of decellularized matrices
is gaining attention for tissue engineering applications. Indeed, cell-
derived tissue-engineered decellularized matrices prepared in vitro
have also been shown to retain their structural integrity and maintain
their molecular functionality [6, 7]. We have recently reported that
decellularized periodontal ligament cell sheets can retain their struc-
tural and biological integrity, and be readily recellularized by allo-
genic cells [8]. The aim of this chapter is to describe the method of
fabrication and characterization of these tissue-engineered decellular-
ized periodontal ligament cell sheet constructs.
2 Materials
All fluids used in the cell culture and decellularization processes

have to be sterile and warmed to 37 C. Aseptic techniques should
be used at times.
All cell culture incubation must be performed in a 37 C, 10 %
CO2 humidified incubator.
2.1 Primary Human 1. Human periodontal ligament tissue harvested from freshly
Periodontal Ligament extracted teeth (see details in Subheading 3.1).
Cell (hPDLC) 2. 25- and 175-cm2 culture flasks.
Harvesting
3. Dulbeccos Modification of Eagles medium (DMEM).
and Expansion
4. 10 % fetal calf serum (FCS).
5. Penicillin (50 U/mL, Invitrogen).
6. Streptomycin (50 g/mL, Invitrogen).
7. 24-Well culture plates.
8. Ascorbic acid (100 and 1000 g/mL).
9. Cell culture incubator.
10. Centrifuge.
11. 50-mL falcon tubes (sterile).
2.2 Melt Electrospun 1. Direct writing melt electrospinner with suitable G-coding soft-
PCL Carrier Membrane ware (see Note 1).
Fabrication 2. Medical grade polycaprolactone (PCL, 80kDa Corbion).
Decellularized Periodontal Cell Sheets 405
2.3 Cell Sheet 1. Fine pointed curved tweezers (sterile).

Fabrication 2. 5-mm sterile dermal Biopsy punch (Kai Medical).
and Harvesting
3. PCL scaffolds with 090 fiber orientations.
2.4 Perfusion 1. Phosphate-buffered saline (PBS).

Decellularization 2. 0.05 % trypsin.
3. 20 mM NH4OH solution.
4. 0.5 % (v/v) Triton X-100.
5. Bidirectional perfusion system (explained in Subheading 3.4,
Fig. 1).
6. DNase I solution (100 U/mL, Invitrogen).
7. CaCl2 (0.9 mM) and MgCl2 (0.5 mM) in sterile PBS.
8. Sterile double distilled water.
9. Petri dishes (sterile).
2.5 Immunostaining 1. 4 % Paraformaldehyde solution in PBS at pH 7.4

for Extracellular (Sigma-Aldrich).
Matrix Components 2. Triton X-100 (0.2 %) in PBS.
Fig. 1 Perfusion bioreactor system for cell decellularization. (a) Perfusion pump with attached chambers. (b)
3D printed chambers used to house the cell sheet constructs during the decellularization procedure
3. Bovine serum albumin 1 % (Sigma-Aldrich) in PBS.

4. Antibodies against human Collagen I and Fibronectin (Life
Technologies, Invitrogen).
5. 4,6-Diamidino-2-phenylindole (DAPI, 5 g/mL).
6. Phalloidin-tetramethylrhodamine B isothiocyanate conjugate
(Phalloidin-TRITC, 0.8 U/mL, Life Technologies).
7. Fluorescently labeled secondary antibody Alexa 633 goat anti-
mouse antibody (5 g/mL, Alexa Fluor, Invitrogen Catalog
#A-21126).
2.6 Growth Factor 1. 2 M NaCl.

Extraction 2. 20 mM HEPES.
3. EDTA protease inhibitor cocktail (Roche Complete Mini,
Roche Applied Science, Indianapolis, IN).
3 Methods
All procedures should take place within a cell culture room in a

biosafety cabinet and under aseptic conditions.
3.1 Primary Human 1. Place redundant freshly extracted teeth immediately in warm
Periodontal Ligament DMEM.
Cell (hPDLC) 2. Hold the extracted tooth by the crown to avoid any damage to
Harvesting the periodontal tissues.
and Expansion 3. Obtain periodontal ligament (PDL) tissues by gently scraping
the tissues attached to the middle third of the roots.
4. Dice the tissues into smaller portions (approximately
1 1 1 mm3).
5. Using a plastic pipette, transport the diced PDL tissues to a
25-cm2 flask, and wet them with one or two drops of
DMEM. Place the flask in the incubator upright for 5 min,
then rewet the diced tissues again, and incubate for further
5 min until the PDL tissues are firmly attached to the flask
inner surface. Add 5 mL of DMEM supplemented with 10 %
foetal calf serum (FCS), penicillin (50 U/mL), and streptomy-
cin (50 g/mL), with culture medium changes every 48 h
until the cells reach 80 % confluency (see Note 2, Fig. 2a).
6. Passage the cells by discarding the culture medium, rinsing
with warm PBS twice, and then adding sufficient 0.25 % tryp-
sin to just cover the cell layer. Place the flask in the incubator
for 1 min, and then check under an inverted microscope for
cell detachment. Collect the cells in 10 mL of DMEM and
then split the cells in a 1:3 ratio.
7. Use cells between the 3rd and 4th passages for optimal cell
growth and expansion (see Note 3).
Fig. 2 Periodontal ligament cell cultures as observed under inverted microscopy. (a) Establishment of primary
periodontal ligament cell cultures from explanted tissues. (b) Confluent periodontal ligament cultures forming
a cell sheet. (c) Decellularized cell sheet constructs
3.2 Melt Electrospun 1. Load medical grade polycaprolactone granules into a syringe,
PCL Carrier Membrane set the temperature of the water tank to 100 C until the PCL
Fabrication melts completely (see Note 4).
2. Adjust the spinneret collector distance to 2 cm.
3. Set the feed rate to 20 L/h and the voltage to 10 kV.
4. Set the transitional speed of the collector at 250 mm/min to
obtain straight fibers.
5. Set the spinner to deposit alternating series of layers with 90
orientation.
6. Collect the deposited scaffolds using clean fine tipped tweezers
into a clean Petri dish and seal the dish with parafilm.
7. PCL scaffolds must be cut into 5 mm diameter using a biopsy
punch and sterilized before cell sheet harvesting (see Note 5).
Use UV in a biosafety cabinet to perform scaffold sterilization
overnight where scaffolds should be embedded in ethanol
100 % (Fig. 3a, b).
3.3 Cell Sheet 1. Discard culture medium from the cell culture flask/s, rinse
Fabrication gently twice with warm PBS, then detach the cells by adding
and Harvesting trypsin as described in Subheading 3.1, step 6.
2. Resuspend the cells using a master mixture of 50 mL into a
24-well cell culture plate, with seeding density of 50,000
cells/well.
3. Leave the cells in high glucose DMEM for the first 24 h.
4. After 1 day post-seeding, discard the old medium and add
200 L of DMEM supplemented with 1000 mg/mL ascorbic
acid for a further 72 h.
5. Discard the medium and add the same volume of DMEM but
only with 100 mg/mL ascorbic acid, and change this medium
Fig. 3 Fabrication of decellularized cell sheet constructs. (a, b) Melt electrospun polycaprolactone (PCL)
scaffolds. (c, d) Periodontal ligament cell sheets placed on PCL scaffolds. (e) Inverted microscope image of
periphery of PCL scaffold with attached cell sheet. (f) Scanning electron microscopy (SEM) image of decel-
lularized cell sheet construct
every 48 h for the following 15 days until the cell sheets can be
visibly detected (Fig. 2b).
6. Wet the 5-mm diameter PCL scaffolds with DMEM.
7. Place the PCL scaffold exactly in the center of each well after
discarding most of the medium from each well (see Note 6).
8. Make sure that the scaffold is in contact with the cell sheet and
avoid excessive forces that may damage the cell sheet.
9. Using fine curved tweezers, start detaching the cell sheets from
the boundaries of the wells in a circumferential manner (see
Note 7).
10. Pull the cell sheet edges upward and toward the center of each
well, fold it over the scaffold.
11. Gently use the tweezers to flip the scaffold with the cell sheet
facing upward.
12. Wet the detached cell sheet with 1020 L of medium every
5 min for 20 min total, with the culture plate placed back into
the incubator after each wetting cycle.
13. Transfer the cell sheet constructs (scaffold + cell sheet/ CSC) to a
new 6-well cell culture plate, then add 300 L of DMEM/well.
The cell sheet constructs are shown in Fig. 3ce (see Note 8).
14. Leave the constructs overnight so the cell sheets adhere to the
PCL scaffolds.
3.4 Perfusion 1. Use a perfusion system for the cell sheet construct (CSCs)
Decellularization decellularization.
2. We designed a perfusion bioreactor system composed of an
infusion/withdrawal syringe pump, 30-mL plastic syringes,
silicone tubes, and decellularization chambers fabricated from
photo-curable material. The chambers and its components
were designed with CAD software and additive manufactured
using an inkjet 3D printer (Objet30 Pro Desktop, Stratasys)
using an acrylic resin (Verowhite Plus 835, Stratasys), as shown
in Fig. 3.
3. All components must be placed in a biosafety cabinet and
exposed to UV radiation overnight for sterilization.
4. Rinse the CSCs once with warm PBS at 37 C, and place them
in the decellularization chambers with a maximum of 11 con-
structs per chamber (see Notes 9 and 10).
5. Perfuse the CSCs in 30 mL of 20 mM NH4OH solution with
0.5 % v/v Triton X-100 at room temperature (see Note 11).
6. Bidirectional perfusion of the constructs needs to be per-
formed for 60 min at a rate of 1000 mL/h, with a flow inver-
sion every 50 s (see Note 12).
7. Discard the liquids from the decellularization chambers by
simply detaching the silicone tubes from the perfusion cham-
bers, then immediately connect another loaded syringe for the
DNase perfusion step to the perfusion chambers.
8. Perfuse with 30 mL of DNase I solution (100 U/mL,
Invitrogen) in CaCl2 (0.9 mM) and MgCl2 (0.5 mM) in sterile
PBS at 37 C for 60 min (see Note 13).
9. Finally, perfuse the constructs with sterile water at 37 C for
another 60 min.
10. Carefully disconnect the perfusion chambers from the syringe
pump after discarding all fluids.
11. Collect the CSCs by opening the chambers inside the cabinet
using sterile tweezers. Then place them in a culture Petri dish,
add sterile PBS to completely cover the scaffolds, seal the
dishes completely with parafilm, then place them in a refrigera-
tor overnight at 4 C. The appearance of the cell sheet con-
structs using an inverted microscope and SEM is shown in
Figs. 1c and 2f respectively.
3.5 Immunostaining 1. Place constructs in a clean multiwell culture plate and rinse
for Extracellular samples carefully twice with PBS at room temperature.
Matrix Components 2. Fix constructs in 4 % paraformaldehyde (PFA) for 2030 min.
3. Discard PFA and then wash once with PBS.
4. If necessary, permeabilize cells for 5 min in 0.2 % Triton
X-100 in PBS.
5. Discard solutions and then wash samples twice gently with

PBS.
6. Transfer samples into 1 % BSA in PBS and incubate for 10 min.
7. Prepare solutions of antibodies diluted in 1 % BSA in PBS,
1:200 for collagen type I, and 1:300 for fibronectin.
8. Add 100200 L of antibody solution for each well and then
incubate for 45 min.
9. Rinse constructs three times with PBS.
10. Incubate samples in 1 % BSA in PBS containing fluorescently
labeled secondary antibody (Alexa 633 goat anti-mouse, 5 g/
mL), 0.8 U/mL TRITC-conjugated phalloidin, and 5 g/mL
DAPI for 45 min protected from light.
11. Rinse twice with PBS at room temperature and then proceed
with confocal imaging.
3.6 Growth Factor 1. Dissolve 1 tablet of the protease inhibitor in 10 mL of the

Extraction extraction buffer solution (2 M NaCl in 20 mM HEPES).
2. Rinse constructs twice with PBS at room temperature.
3. Add 300 L of extraction buffer solution with protease inhibi-
tor for each well/construct.
4. Seal the plate with parafilm and incubate for 60 min on an
orbital shaker at room temperature.
5. Collect the supernatant for each construct into a sterile eppen-
dorf tube and store at 80 C.
4 Notes
1. We used a custom-made machine fabricated in-house at the

Institute of Health and Biomedical Innovation, Queensland
University of Technology [8].
2. Avoid cell over-confluency during the cell propagation (expan-
sion) phase as it affects their future survival, growth, and ability
to form extracellular matrix.
3. When seeding in large 175 cm2, start with at least 500,000
cells as lower seeding densities are much slower to reach
semi-confluency.
4. Multiple biomaterial options exist for the carrier component of
the construct. We utilized electro-spun polycaprolactone
(PCL) because it is a medical grade material with excellent bio-
compatibility. It has the key properties of being able to be fab-
ricated in a highly porous structure that allows perfusion of the
decellularization reagents, while still retaining appropriate
mechanical properties to provide sufficient support during the
perfusion process. It is also a chemically stable polymer that

retains its structural integrity during the decellularization
process. Other biomaterials with similar properties could also
be utilized as the cell sheet carrier component of the
construct.
5. If using PCL scaffolds as a carrier for the cell sheets, it is advis-
able to enhance the scaffold hydrophilicity by immersion into
2 M NaOH for 30 min at 37 C followed by five rinses of
ultrapure water.
6. Avoid excessive pressure when placing a scaffold on the top of
a cell sheet during the cell sheet harvesting step.
7. When using fine tweezers to detach a cell sheet, always aim the
tip to the boundaries of the well, to be able to harvest an intact
cell sheet.
8. When relocating or transferring the constructs from one plate
to another or to perfusion chambers, always handle the con-
structs with extra care holding each one separately at a time.
Use broad tipped tweezers for this procedure while holding
the constructs from the sides, also avoid squeezing the con-
struct in order not to damage the cell sheet and/or the
scaffold.
9. Use spacers between the scaffolds inside the perfusion
chamber.
10. Arrange the scaffolds with enough distance between each, so
they do not adhere to each other. Avoid overfilling of the
decellularization chambers with fluids. However, please note
that scaffolds must be immersed completely in fluids all the
time.
11. Avoid rapid fluid perfusion as this will disrupt the constructs.
12. CaCl2 (0.9 mM) and MgCl2 (0.5 mM) concentrations in PBS
are essential to activate the DNase enzyme. However, if these
concentrations are exceeded, calcium ions will chelate with
phosphate forming a turbid white solution. If this occurs dur-
ing the DNase perfusion step, the solution should be discarded
and a fresh one prepared.
13. DNase enzyme concentration and the period of perfusion
should not exceed what is outlined in Subheading 3.4, step 8,
as this will damage the extracellular matrix components.
Acknowledgments
This work was supported by Australias National Health and

Medical Research Council project grant 1086181.
References
1. Ishikawa I, Iwata T, Washio K, Okano T, 5. Dan H, Vaquette C, Fisher AG, Hamlet SM,
Nagasawa T, Iwasaki K, Ando T (2009) Cell Xiao Y, Hutmacher DW, Ivanovski S (2014)
sheet engineering and other novel cell-based The influence of cellular source on periodontal
approaches to periodontal regeneration. regeneration using calcium phosphate coated
Periodontol 2000 51:220238 polycaprolactone scaffold supported cell sheets.
2. Flores MG, Yashiro R, Washio K, Yamato M, Biomaterials 35:113122
Okano T, Ishikawa I (2008) Periodontal ligament 6. Gawlitta D, Benders KE, Visser J, van der Sar
cell sheet promotes periodontal regeneration in AS, Kempen DH, Theyse LF, Malda J, Dhert
athymic rats. J Clin Periodontol 35:10661072 WJ (2015) Decellularized cartilage-derived
3. Akizuki T, Oda S, Komaki M, Tsuchioka H, matrix as substrate for endochondral bone
Kawakatsu N, Kikuchi A, Yamato M, Okano T, regeneration. Tissue Eng Part A 21:
Ishikawa I (2005) Application of periodontal 694703
ligament cell sheet for periodontal regenera- 7. Sadr N, Pippenger BE, Scherberich A, Wendt
tion: a pilot study in beagle dogs. J Periodontal D, Mantero S, Martin I, Papadimitropoulos A
Res 40:245251 (2012) Enhancing the biological performance
4. Vaquette C, Fan W, Xiao Y, Hamlet S, of synthetic polymeric materials by decoration
Hutmacher DW, Ivanovski S (2012) A bipha- with engineered, decellularized extracellular
sic scaffold design combined with cell sheet matrix. Biomaterials 33:50855093
technology for simultaneous regeneration of 8. Brown TD, Dalton PD, Hutmacher DW
alveolar bone/periodontal ligament complex. (2011) Direct writing by way of melt electros-
Biomaterials 33:55605573 pinning. Adv Mater 23:56515657
Chapter 24
A Method to Isolate, Purify, and Characterize Human

Periodontal Ligament Stem Cells
Krzysztof Mrozik, Stan Gronthos, Songtao Shi, and P. Mark Bartold
Abstract
Human periodontal ligament stem cells (PDLSCs) are a unique population of mesenchymal stem cells
(MSCs) that demonstrate the capacity to generate cementum- and periodontal ligament-like structures
in vivo. As such, PDLSCs represent a promising cell-based therapy in reconstructive dentistry for the treat-
ment of periodontal disease. The present chapter describes two methods for isolating PDLSCs from human
PDL tissue including traditional plastic adherence, and immunomagnetic selection based on the expres-
sion of MSC-associated surface markers STRO-1 antigen, CD146 (MUC-18), CD29 (Integrin -1),
CD44, and CD106 (VCAM-1). Although no single antibody demonstrates specificity for MSCs, isolation
based on expression of individual markers results in homogenous preparations of PDLSCs. Methods to
further characterize the immunophenotype and multipotent capacity of PDLSCs to differentiate into
adipocytes, osteoblast-, and cementoblast-like cells in vitro, and cementum- and periodontal ligament-like
tissues in vivo, are also described.
Key words Periodontal ligament stem cells, Mesenchymal stem cells, Adherence isolation,
Immunomagnetic isolation, Differentiation potential
1 Introduction
The presence of multiple cell types (fibroblasts, cementoblasts, and

osteoblasts) within the postnatal periodontal ligament suggests
that these cells may share common ancestors. The possibility that
progenitor cells might exist in the postnatal periodontal ligament
has been recognized for some time but until recently had never
been formally proven [1]. These cells are believed to provide a
renewable cell source for normal tissue homeostasis and periodon-
tal wound healing [2, 3].
Recently, multipotent stem cell populations, termed periodon-
tal ligament stem cells (PDLSCs), have been isolated from the
periodontal ligament of extracted human third molar teeth [4].
These PDLSCs give rise to adherent clonogenic clusters that
resemble fibroblasts and are capable of developing into adipocytes,
413
414 Krzysztof Mrozik et al.
osteoblast-, and cementoblast-like cells in vitro, and demonstrate

the capacity to produce cementum- and periodontal ligament-like
tissues in vivo [46]. PDLSCs express an array of cementoblast and
osteoblast markers as well as several bone marrow-derived mesen-
chymal stem/stromal cell (MSC) associated markers [7, 8]. The
similarity between PDLSCs and bone marrow MSCs suggests that
PDLSCs represent another MSC-like population. Further work is
now focusing on identifying markers uniquely expressed by
PDLSCs to discriminate these cells from other types of MSC-like
cells identified in dental tissues [9].
The first reported isolation and identification of MSCs in
human periodontal ligament was in 2004 [4]. Since then, there has
been considerable activity trying to understand the function of
these cell populations and their interactions with each other with a
view to laying the fundamental groundwork for clinical applica-
tions in regenerative periodontics. A number of studies have now
been carried out to confirm the presence of MSC-like cells in the
periodontal ligament. These have not been limited to human but
also include mouse, rat, and sheep [5, 8, 1014]. All of these stud-
ies have confirmed the multipotent nature of PDLSCs, and while
the initial studies indicated this to include an ability to differentiate
into osteoblast, cementoblast, or adipogenic phenotypes, at least
one recent study has indicated an ability of these cells to also dif-
ferentiate into neuronal precursors [14]. Importantly, cryopreser-
vation does not seem to alter the functional properties of PDLSCs
[15]. This will have significant relevance should banking of these
cells become a clinical necessity.
Identification of stem cells in postnatal dental tissues has pre-
sented exciting possibilities for the application of tissue engineer-
ing, as well as gene and cell-based therapies in reconstructive
dentistry. The use of stem cells with these technologies may consti-
tute novel strategies for regenerative periodontal therapy.
Periodontitis is a disease of the periodontium characterized by irre-
versible loss of connective tissue attachment and supporting alveo-
lar bone [16]. These changes often lead to an aesthetically and
functionally compromised dentition. For many decades, periodon-
tists have been interested in regenerating tissues destroyed by peri-
odontitis. Periodontal regeneration can be defined as the complete
restoration of the lost tissues to their original architecture and func-
tion by recapitulating the crucial wound healing events associated
with their development [17, 18]. The isolation of adult stem cells
from human periodontal ligament has presented new opportunities
for tissue engineering [5, 10]. Clearly, in order for such therapies to
be successful, suitable means of procuring these cells and expanding
them in vitro are essential. This chapter describes methods for the
isolation, ex vivo expansion, and characterization of PDLSCs from
human periodontal ligament.
Human Periodontal Ligament Stem Cells 415
2 Materials
2.1 Processing 1. Wash buffer (Hanks balanced salt solution [HBSS; JRH
of Periodontal Biosciences, Lenexa, KS, USA] supplemented with 5 % (v/v)
Ligament fetal bovine serum [FBS; JRH Biosciences] and 50 U/mL
penicillin and 50 g/mL streptomycin).
2. Type I Collagenase (6 mg/mL stock solution in PBS
[phosphate buffered saline]).
3. Dispase II (8 mg/mL stock solution in PBS).
4. White Cell Fluid; 2 % acetic acid in distilled H2O.
5. 70-m cell strainer.
6. 10-cm tissue culture dish.
7. 14-mL polypropylene round bottom tube.
8. Forceps.
9. Scalpel handle Size 3.
10. Surgical blade Size 11.
2.2 Dynal 1. Blocking buffer (HBSS supplemented with 5 % FBS, 5 %

Immunomagnetic Cell normal human serum [NHS; AB+, see Note 1], 1 % bovine
Isolation serum albumin [BSA], 50 U/mL penicillin, and 50 g/mL
and Fluorescence streptomycin)
Activated Cell Sorting 2. Wash buffer (see above).
3. Murine monoclonal primary antibodies (anti-human): (a)
STRO-1 (IgM, anti-human stromal cell; Developmental
Studies Hybridoma Bank, University of Iowa, Iowa City, IA,
USA), (b) anti-CD29 (IgG1, BD Biosciences, San Jose, CA,
USA), (c) H9H11 (IgG1, anti-CD44; A/Prof. Andrew
Zannettino, Division of Hematology, IMVS, Adelaide, SA,
Australia [19]), (d) 6G10 (IgG1, anti-CD106, VCAM-1;
American Type Culture Collection Manassas, VA. ATCC No.
HB 10519), (e) CC9 (IgG2a, anti-human CD146, MUC-18;
A/Prof. Stan Gronthos, Division of Hematology, IMVS,
Adelaide, SA, Australia [20, 21], (f) CD166 (ALCAM; BD
Biosciences), (g) anti-CD105 (IgG1, BD Biosciences), (h)
anti-CD166 (IgG1, BD Biosciences), (i) anti-CD14
Fluorescein isothiocyanate (FITC) conjugated (IgG2a;
Beckman Coulter, Fullerton, CA, USA), (j) anti-CD31 FITC
conjugated (IgG1; Beckman Coulter), (k) anti-CD45 FITC
conjugated (IgG1; Beckman Coulter).
4. Isotype-matched control (a) 1B5 (IgG1), (b) 1A6.11 (IgG2),
(c) 1A6.12 (IgM) (Prof. L.K. Ashman; The University of
Newcastle, Newcastle, NSW, Australia). Isotype-matched con-
trols are also commercially available.
5. Dynabead-conjugated Rat anti-Mouse IgM and Goat anti-

Mouse IgG secondary antibodies (Dynal Biotech ASA, Oslo,
Norway).
6. Goat anti-mouse IgM-FITC conjugated antibody (Southern
Biotechnology Associates, Inc., Birmingham, AL) and Goat
anti-mouse IgG-FITC conjugated antibody Southern
Biotechnology Associates, Inc.).
7. Dynal MPC-2 Magnetic Particle Concentrator (Dynal
Biotech ASA).
8. FACS Fix: 1 % (v/v) formalin, 0.1 M D-glucose, 0.02 % sodium
azide in PBS.
2.3 Cell Culture 1. -MEM growth medium (Alpha-Modification of Eagles

of Human PDLSC Medium [-MEM; JRH Biosciences, Lenexa, KS, USA] sup-
plemented with 20 % (v/v) FBS, 2 mM L-glutamine,
100 M L-ascorbate-2-phosphate, 1 mM sodium pyruvate,
50 U/mL penicillin and 50 g/mL streptomycin).
2. Hanks balanced salt solution (HBSS; JRH Biosciences, Lenexa,
KS, USA).
3. Osteogenic inductive medium (-MEM supplemented with
10 % (v/v) FBS, 2 mM L-glutamine, 100 M L-ascorbate-2-
phosphate, 107 M dexamethasone, 1.8 mM inorganic phos-
phate (KH2PO4), 50 U/mL penicillin and 50 g/mL
streptomycin.
4. Adipogenic inductive medium (-MEM supplemented with
10 % (v/v) FBS, 0.5 mM isobutylmethylxanthine, 60 M indo-
methacin, 0.5 M hydrocortisone, 15 mM Hepes buffer,
2 mM L-glutamine, 100 M L-ascorbate-2-phosphate, 1 mM
sodium pyruvate, 50 U/mL penicillin and 50 g/mL
streptomycin.
5. Phosphate buffered saline solution, pH 7.4.
6. 0.5 % Trypsin/0.2 % EDTA solution (10x stock solution).
7. 0.4 % Trypan blue/PBS.
8. 0.1 % (w/v) toluidine blue.
9. T-25 and T-75 culture flasks.
10. 6-Well culture plates.
11. 14-mL polypropylene round bottom tubes.
12. 1.8-mL Cryo-tubes.
13. Freeze mix (20 % Dimethyl sulphoxide [DMSO] in FBS).
14. Cryo 1 C freezing container Mr. Frosty (Nalge, Nunc
International).
15. Alizarin red stain (1 % Alizarin Red [Sigma-Aldrich, St. Louis,

MO], 2 % ethanol in distilled water).
16. Oil Red O stain (0.5 g Oil Red O [ICN Biomedicals, Inc.,
Aurora, Ohio] stain dissolved in 100 mL isopropanol and
mixed in 3:2 ratio with distilled water).
17. 1 % (w/v) paraformaldehyde (PFA) in PBS.
18. 10 % Neutral buffered formalin.
2.4 Attachment 1. Hydroxyapatite/tricalcium phosphate (HA/TCP) ceramic

of PDLSCs to HA/TCP particles (Zimmer Corporation, Warsaw, IN).
Particles 2. Fibrinogen (Sigma-Aldrich).
and Subcutaneous
3. Thrombin (Sigma-Aldrich).
Implantation
4. Surgical scissors.
5. AUTOCLIP Wound Clips 9 mm (BD Biosciences).
6. AUTOCLIP Applier 9 mm (BD Biosciences).
2.5 Recovery 1. 4 % Paraformaldehyde (4 g paraformaldehyde in PBS).

of Transplant, 2. 10 % EDTA (ethylenediaminetetraacetic acid in deionized
Processing, water).
and Immunohisto-
3. Ethanol.
chemistry
4. Xylene.
5. Paraffin wax.
6. Forceps.
7. Scalpel handle Size 3.
8. Surgical blade Size 11.
9. Mayers Hematoxylin (Lillies Modification).
10. Acid Alcohol: 0.3 % concentrated hydrochloric acid (HCl),
70 % ethanol in distilled water.
11. Bicarbonate solution (1 % in distilled water).
12. Eosin: 0.114 % eosin Y, 0.0114 % aqueous phloxine, 0.46 %
glacial acetic acid [v/v], 84.1 % ethanol in distilled water.
13. Gurrs DePeX mounting medium.
14. 30 % Hydrogen Peroxide.
15. Sodium Azide.
16. Goat serum.
17. Rabbit polyclonal primary antibodies: (a) Bone sialoprotein
(BSP) LF-120 (Dr. Larry Fisher, Craniofacial and Skeletal
Diseases Branch, NIDCR/NIH, Bethesda, MD, USA [22]),
(b) Osteocalcin (OSC) LF-32 (Dr. Larry Fisher, Craniofacial
and Skeletal Diseases Branch, NIDCR/NIH, Bethesda, MD,
USA [23]).
18. Rabbit Ig control (Caltag Laboratories, Burlingame, CA; Code

No. 10500).
19. Secondary antibody: Biotinylated Goat anti-rabbit IgG (H + L)
antibody (Vector Laboratories, Inc., Burlingame, CA).
20. Vectastain ABC Kit (Streptavidin-horseradish peroxidase kit;
Vector Laboratories, Inc.).
21. AEC kit (Horseradish peroxidase substrate; Vector Laboratories,
Inc.).
3 Methods
3.1 Processing Following informed consent, teeth with healthy periodontal

of Human Periodontal ligament can be obtained as a result of tooth extraction for orth-
Ligament (PDL) odontic purposes or removal of third molars. PDL tissue can be
pooled from multiple teeth and from different donors provided
3.1.1 Collection
they are all processed within 2 h of extraction.
of Periodontal Ligament
Cells 1. Gently separate the periodontal ligament (PDL) from the
middle third of the tooth root surface using forceps and a size
11 surgical blade in a 10-cm tissue culture dish containing
10 mL wash buffer.
2. Transfer wash buffer containing PDL tissue into a 14-mL
round bottom tube and centrifuge at 400 g for 10 min at
4 C.
3. Resuspend and digest PDL in a solution of 1 mL Type I col-
lagenase (3 mg/mL final) and 1 mL dispase II (4 mg/mL
final) for 1 h at 37 C (see Note 2).
4. Add an excess volume of wash buffer to the digested tissue to
neutralize enzyme activity and then strain through a 70-m
cell strainer to remove undigested tissue from the liberated
periodontal ligament cells.
5. Pellet PDL cells by centrifugation at 400 g for 10 min at
4 C, resuspend in 2 mL wash buffer, and keep on ice.
6. Remove a 10 L aliquot and dilute 1:10 into White Cell Fluid
(WCF) and enumerate nucleated cells using a hemocytometer
(see Note 3). Typically, the PDL cell count ranges from 0.5 to
2.5 104 per one to three teeth processed.
3.2 Isolation Single-cell suspensions generated from digested human PDL tissue
of Periodontal have the capacity to form adherent clonogenic cell clusters with a
Ligament Stem Cells fibroblast-like morphology in an in vitro culture setting. Each col-
ony originates from a single progenitor cell (colony-forming-unit-
fibroblast, CFU-F), similarly to colonies formed by dental pulp
stem cells (DPSCs) and bone marrow-derived MSCs (BM MSCs)
[4, 5, 7, 20, 24]. The colony-forming cell population residing in
PDL tissue has been termed periodontal ligament stem cells

(PDLSCs) [4]. However, traditional methods of isolating bone
marrow derived CFU-F based on plastic adherence result in a het-
erogeneous population, whereby some individually expanded col-
onies display a multipotent capacity for lineage differentiation
whereas others demonstrate a restricted differentiation potential
[2427].
Homogeneous populations of PDLSCs and other MSCs can
be immunomagnetically isolated according to the particular mark-
ers expressed on the cell surface. Although no single antigenic
marker demonstrates specificity for MSCs, ex vivo expanded
PDLSCs express an unidentified early mesenchymal stem cell-
associated surface antigen reactive to the antibody STRO-1.
Similarly to DPSCs and BM MSCs, STRO-1 antibody-based isola-
tion of PDLSCs released from freshly digested PDL tissue demon-
strates that the majority of colony-forming units are contained
within the STRO-1+ fraction. Thus, the reactive antigen to the
STRO-1 antibody is considered an important MSC-associated
marker expressed by PDLSCs [4]. Further discrimination of the
STRO-1+ DPSC and BM MSC fractions has been achieved by puri-
fication based on their expression of endothelial cell markers
CD146 (MUC18) and CD106 (VCAM-1), demonstrating their
localization within the perivascular niche [7, 20, 26, 27]. Although
PDLSCs also express CD146 and CD106 implying a perivascular
origin [4, 5, 9], the relatively small number of cells released from
PDL tissue compared to dental pulp and bone marrow is a signifi-
cant limitation in applying a dual-antibody isolation approach.
However, positive isolation of PDLSCs can also be achieved using
antibodies to various other MSC-associated markers including
CD106 [5], CD29 (Integrin 1; [7], CD44 [7, 9], and CD146
[4, 9] as stand alone reagents. Similarly to STRO-1, these antibod-
ies positively select the majority of colony-forming units within
PDL tissue with varying efficiencies.
3.2.1 Adherence 1. Single-cell suspensions generated from digested PDL are ini-
Isolation of Periodontal tially plated in T-25 culture flasks or 6-well plates in -MEM
Ligament Stem Cells growth medium. Cultures are incubated at 37 C in 5 % CO2
and Ex Vivo Culture and >90 % humidity (see Note 4).
2. Adherent primary PDLSC colonies (CFU-F) are passaged
when 7080 % confluency is achieved after approximately 2
weeks. At this point in time, PDLSC cultures are washed once
with HBSS and liberated by enzymatic digestion by the addi-
tion of 1 mL 0.05 % trypsin/0.02 % EDTA solution per T-25
culture flask or a 6-well plate for 510 min at 37 C. The single
cell suspension is then washed twice in wash buffer using a
14-mL tube (see Note 5).
3. Assess cell viability by removing a 10 L aliquot of the single
cell suspension and diluting 1:10 in 0.4 % trypan blue/PBS.
The number of viable cells can be enumerated using a hemocy-

tometer (nonviable cells take up the blue dye).
4. To expand cultures, PDLSCs are reseeded (passaged) into
T-75 culture flasks at 58 103 per cm2 in -MEM growth
medium. Cultures are refed twice weekly by aspirating the
growth medium and replacing with an equal volume of fresh
growth medium warmed to 37 C (see Note 5).
3.2.2 Immunomagnetic 1. Resuspend single-cell suspensions generated from digested

Isolation of Periodontal PDL in 10 mL blocking buffer using a 14-mL round bottom
Ligament Stem Cells tube and incubate on ice for 30 min to reduce the possibility of
and Ex Vivo Culture Fc receptor-mediated binding of antibodies.
2. Pellet cells by centrifugation at 400 g for 10 min at 4 C and
resuspend all cells (usually 0.52.5 104) in 1 mL of desired
primary antibody (STRO-1 [IgM] and anti-CD146 [IgG2a]
monoclonal supernatants [1/2 dilution] or purified IgG1
monoclonal antibodies anti-CD29, -CD44, and -CD106
[10 g/mL]) using a 5-mL tube and incubate on ice for 1 h
with occasional, gentle mixing.
3. Cells are washed twice in wash buffer by centrifugation at
400 g for 10 min at 4 C and resuspended in 1 mL of appro-
priate Dynabead-conjugated secondary antibody (see Note 6),
mixed and incubated on ice for 10 min. Add 2.5 mL wash buf-
fer and incubate for 2 h at 4 C on a rotator. Check for beads
bound to cells microscopically.
4. Mix the cell suspension and place the tube in a Dynal MPC-2
Magnetic Particle Concentrator for 2 min. Aspirate off cells
that are not bound to the magnet (negative fraction), add
3 mL cold wash buffer to bead-bound cells, mix the cell sus-
pension, again place the tube in the magnetic particle concen-
trator, and aspirate unbound cells. Repeat this process until
there are no unbound cells remaining in the suspension. This
can be checked microscopically.
5. At this point, purified PDLSCs can be culture expanded similarly
to PDL cells following digestion as described for adherence
isolation of PDLSCs (see Subheading 3.2.1) or used in colony-
forming unit-fibroblastic (CFU-F) efficiency assays as described
in Subheading 3.3.
3.2.3 Cryopreservation 1. Single cell suspensions of culture expanded PDLSCs are pre-
of Ex Vivo Expanded pared by 0.05 % trypsin/0.02 % EDTA digestion and cells enu-
PDLSCs merated and viability assessed using 0.4 % trypan blue/PBS as
described above.
2. Cells are resuspended at a concentration of 4 106 per mL of
FBS and kept on ice. An equal volume of ice-cold freeze mix is
then added drop-wise while gently agitating the cells to give a
final concentration of 2 106 per mL in 10 % DMSO/

FBS. Aliquots of 1 mL are distributed into 1.8-mL cryo-vials
precooled on ice and then frozen at a rate of approximately
1 C per minute using a Cryo 1 C freezing container Mr.
Frosty precooled to 4 C. Place the container holding the
cryo-vials at 80 C overnight before transferring the cryo-
vials into liquid nitrogen for long-term storage.
3. Recovery of the cryopreserved stock is achieved by rapidly
thawing the cells in a 37 C water bath (see Note 7). Resuspend
the cells in 20 mL cold wash buffer and spin at 280 g for
10 min.
4. Assess viability of cells using 0.4 % trypan blue/PBS as described
above. Typically, this procedure results in cell viabilities
between 80 and 90 %.
3.3 Assessment 1. Single-cell suspensions harvested from freshly digested PDL or

of Colony Forming- immunomagnetically isolated PDLSCs are seeded in triplicate
Efficiency into 6-well culture plates at 0.251.0 104 per well in -MEM
(CFU-F Assay) growth medium. Cultures are incubated at 37 C in 5 % CO2
and >90 % humidity for 12 days (see Note 8).
2. Cultures are washed twice with PBS and then fix for 20 min in
1 % (w/v) PFA in PBS.
3. Stain fixed cultures with 0.1 % (w/v) toluidine blue (in 1 %
PFA) for 1 h, then rinse with tap water and allow to dry.
4. Aggregates of greater than 50 cells are scored as CFU-F using
either an inverted or dissecting light microscope.
3.4 Flow-Cytometric To characterize the immunophenotype of ex vivo expanded

Analysis of PDLSCs PDLSCs, flow-cytometric analysis can be used to measure the
expression of mesenchymal and non-mesenchymal stem-cell associ-
ated surface markers at early passages. The relatively low number of
cells initially harvested from the digestion of PDL tissue (<2.5 104)
is usually inadequate to perform flow-cytometric analysis without
ex vivo expansion.
1. Adherent ex vivo expanded PDLSCs are washed once with
HBSS and liberated by enzymatic digestion by the addition of
3 mL 0.05 % trypsin/0.02 % EDTA solution per T-75 culture
flask for 510 min at 37 C. The single cell suspension is then
washed twice in wash buffer.
2. Cell count and assessment of viability is performed as described
above.
3. Resuspend PDLSCs for immunolabeling in 0.5 mL blocking
buffer and incubate on ice for approximately 30 min to
reduce the possibility of Fc receptor-mediated binding of
antibodies.
4. Individual 5-mL polypropylene round bottom tubes contain-

ing 1 105 PDLSCs are incubated with appropriate primary
murine monoclonal antibodies; STRO-1, anti-CD29, anti-
CD44, anti-CD105, anti-CD106, anti-CD146, anti-CD166
(mesenchymal stem cell-associated markers), anti-CD14, anti-
CD31, anti-CD45 (hemopoietic, non-mesenchymal stem cell-
associated markers), or isotype-matched control at a
concentration of 20 g/mL for 1 h on ice. Wash the cells twice
in 1 mL wash buffer.
5. Incubate cells with secondary detection reagent; goat anti-
mouse IgG- or goat anti-mouse IgM-FITC (fluorescein iso-
thiocyanate) conjugated antibody (1:50, Southern
Biotechnology, Birmingham, AL) for 45 min on ice (see Note
9). Wash the cells twice in 1 mL wash buffer.
6. Add 0.5 mL FACS Fix solution to each tube to fix the cells.
7. Analyze the cells on any fluorescence-activated cell sorter fitted
with a 250 MW argon laser emitting light at a wavelength of
488 nm able to detect FITC (or other fluorophores).
3.5 Differentiation The capacity of mesenchymal stem cells to generate stromal tissues
Potential of PDLSCs including those similar to which they were derived from is recog-
In Vitro nized as a hallmark feature of these cells. The ability of PDLSCs to
differentiate into various stromal cell lineages in vitro can be inves-
tigated by culturing under inductive conditions.
3.5.1 In Vitro Formation 1. Seed 1 105 in vitro expanded PDLSCs per T-25 culture flask
of Bone Mineral in 5 mL -MEM growth medium and incubate at 37 C in 5 %
CO2 and >90 % humidity.
2. After 24 h, aspirate the -MEM growth medium and add an
equivalent volume of osteogenic inductive medium. Replace
the osteogenic inductive medium twice a week.
3. After 4 weeks, aspirate the medium and gently rinse the
osteogenic-induced culture with PBS five times, fix with 10 %
neutral buffered formalin for 1 h at room temperature (RT),
and then rinse three times with distilled H2O.
4. Stain the osteogenic-induced culture with 1 % Alizarin Red,
2 % ethanol in distilled H2O for 1 h at RT. Mineralized deposits
of calcium will appear red (see Note 10).
3.5.2 In Vitro 1. Seed 5 104 in vitro expanded PDLSCs per well using a 24-well
Differentiation plate in 500 L -MEM growth medium and incubate at
into Adipocytes 37 C in 5 % CO2 and >90 % humidity.
2. After 24 h, aspirate the -MEM growth medium and add an
equivalent volume of adipogenic inductive medium. Replace
the adipogenic inductive medium twice a week.
3. After 4 weeks, aspirate the medium and gently rinse the

adipogenic-induced culture once with PBS and fix with 10 %
neutral buffered formalin for 10 min at RT.
4. Aspirate the formalin and stain the adipogenic-induced culture
with Oil Red O stain for at least 2 h at RT. Lipid-laden vacu-
oles within adipocytes will appear red (see Note 11).
3.6 Differentiation In order to demonstrate that ex vivo expanded PDLSCs can dif-
Potential of PDLSCs ferentiate into functional cementoblast- or osteoblast-like cells,
In Vivo cells attached to osteogenic-conductive hydroxyapatite/tricalcium
phosphate (HA/TCP) ceramic carrier particles can be subcutane-
ously transplanted into immunocompromised mice (see Note 12).
3.6.1 Attachment 1. Prepare single cell suspensions of ex vivo expanded PDLSCs

of PDLSCs to HA/TCP following 0.5 % trypsin/EDTA digestion and cell viability
Particles assessed using 0.4 % trypan blue/PBS as described above.
2. Resuspend 5 106 ex vivo expanded PDLSCs in 1 mL -MEM
growth medium and transfer to a 1.8-mL cryo-vial containing
40 mg HA/TCP ceramic carrier particles (Zimmer, Warsaw, IN)
(see Note 13).
3. Gently mix the cell suspension and HA/TCP particles using a
rotator while incubating at 37 C for 1 h to enhance cell attach-
ment to the particles.
4. Gently pellet the mix at 300 g for 2 min and discard the
supernatant.
5. Approximately 10 min prior to implantation, add 20 L mouse
fibrinogen (30 mg/mL in PBS) and 20 L mouse thrombin
(100 U/mL in 2 % CaCl2) to the cells attached to HA/TCP
ceramic carrier particles, and gently mix in using a pipette tip
to form a plug.
3.6.2 Subcutaneous 1. In 610 week old immunocompromised NOD/SCID mice,

Implantation Procedure perform a 1 cm mid-longitudinal skin incision on the dorsal
surface and create a subcutaneous pocket by blunt dissection.
2. Place the polymerized transplant into the subcutaneous pocket
and close the incision with AUTOCLIP 9 mm wound clips
using AUTOCLIP 9 mm Applier (see Note 14).
3.6.3 Recovery 1. Recover the transplants 8 weeks after transplantation, cut into
of Transplants, Processing, two pieces using a surgical blade, and fix in 4 % paraformalde-
and Hematoxylin and Eosin hyde for 2 days.
Staining 2. Decalcify transplant for 10 days in 12 mL 10 % EDTA solution
using a 14-mL round bottom tube while rotating (see Note 15).
3. Transplants are processed by dehydration through an increasing
gradient of ethanol concentrations (50 %, 70 %, 90 %, and several
changes in 100 %) and then three changes in xylene. Transplants

are then washed twice in molten paraffin wax before embed-
ding in molten paraffin wax. Allow to cool to form a block and
prepare 5 m sections.
4. Sections are deparaffinized in xylene (2 5 min) and then rehy-
drated through a decreasing gradient of ethanol concentra-
tions (5 min each; 2 100 %, 1 90 %, 1 70 %, 1 50 %, and
2 distilled water).
5. Stain with Mayers hematoxylin (Lillies Modification) for 5 min,
wash off hematoxylin in running tap water, and then rinse in dis-
tilled water for 10 s. Immerse in bicarbonate solution for 10 s,
wash in running tap water, decolorize in 0.3 % acid alcohol for 5 s,
and again wash in running tap water. Blue sections in lithium
carbonate and wash in running tap water. Counterstain sections in
eosin for 2 min and dehydrate in three changes of 100 % ethanol
(30 s each). Immerse in two changes of xylene for 30 s each and
mount in Gurrs DePeX mounting medium.
3.6.4 Immunohisto- 1. Sections are deparaffinized in xylene (2 5 min) and then rehy-
chemistry drated through a decreasing gradient of ethanol concentra-
tions (5 min each; 2 100 %, 1 90 %, 1 70 %, 1 50 %, and
2 distilled water).
2. Endogenous peroxidase activity is blocked using 1 % hydrogen
peroxide diluted in 0.1 % sodium azide and PBS for 20 min.
3. Sections are rinsed three times in PBS for 5 min and blocked in
5 % goat serum for 1 h at RT.
4. Primary rabbit polyclonal antibodies are diluted in 5 % goat
serum (1:500; bone sialoprotein [BSP, LF-120], osteocalcin
[OSC, LF-32], and equivalent concentration of rabbit Ig con-
trol), added to each slide, and incubated for 2 h at RT.
5. Sections are washed three times in PBS (5 min per wash), then
incubated with the secondary antibody goat anti-rabbit Ig-
biotinylated antibody (1/100 dilution) for 60 min, before
washing three times in PBS.
6. Streptavidin peroxidase conjugate (Vectastain ABC Kit) is
prepared as recommended by the manufacturer and added to
the sections for 30 min at room temperature.
7. After three washes in PBS, horseradish peroxidase substrate
(AEC [3-Amino 9-ethylcarbazole] kit) is added to the sections
according to the manufacturers protocol and incubated until
color development has occurred (see Note 16).
8. Wash sections three times with distilled water, counterstain
with Mayers hematoxylin (Lillies Modification) for 2 min,
dehydrate in three changes of 100 % ethanol (30 s each),
immerse in two changes of xylene for 30 s each, and mount in
Gurrs DePeX mounting medium.
4 Notes
1. Prior to use, normal human serum should be heat inactivated

at 56 C for 30 min in a shaking water bath, then centrifuged
at 1000 g for 10 min, and supernatant collected.
2. The stated volume of Type I collagenase and dispase II is ade-
quate for processing up to four teeth.
3. Wipe any excess cell suspension on pipette tip with tissue
before mixing with White Cell Fluid or 0.4 % trypan blue/PBS
to ensure the cell count number is not overestimated.
4. Store growth medium at 4 C. If greater than 1 week old at
4 C, add fresh 2 mM L-glutamine prior to use.
5. If adherent cell cultures are overconfluent, wash cells once
with HBSS and incubate with equal volume of Type I collage-
nase (3 mg/mL final) and dispase II (4 mg/mL final) (1 mL
total per 25 cm2 surface area) for 1 h at 37 C. Wash the liber-
ated single cell suspension twice in wash buffer. If cells appear
to be clumped, pass them through a 70-m cell strainer prior
to subculture.
6. Prior to incubation, the Dynabead-conjugated secondary anti-
body should be washed to remove immunoglobulins not
attached to beads. Add the required volume of Dynabeads
(calculate volume based on four beads per cell) to 3 mL wash
buffer, mix the suspension, and place the tube in a Dynal
MPC-2 Magnetic Particle Concentrator for 2 min. Aspirate
off wash buffer containing unbound immunoglobulins, resus-
pend beads in fresh wash buffer, and leave on ice until required.
7. Do not heat thawing cells to 37 C. Remove the cryo-tube
from the water bath as soon as sample is thawed.
8. Check for over-growth of CFU-F at day 10 to prevent colonies
growing into each other.
9. Phycoerythrin (PE) conjugated secondary antibodies can also
be used.
10. Rinse flasks with distilled H2O until excess Alizarin Red stain is
removed.
11. Rinse flasks with distilled H2O until excess Oil Red O stain is
removed and store in distilled H2O at 4 C (do not allow to air
dry).
12. This procedure requires animal ethics approval from the appro-
priate body and should be performed in accordance with speci-
fications of an approved small-animal protocol.
13. Prior to cell attachment, prewash HA/TCP particles in 2 mL
wash buffer on a rotator at 37 C for 1 h. Remove wash buffer
before addition of cell suspension.
14. Up to four transplants can be performed per animal (one

transplant per subcutaneous pocket created).
15. Change 10 % EDTA solution daily and confirm decalcification
is complete by x-ray analysis.
16. Steps 5 and 6 can be substituted using a broad spectrum
immunoperoxidase AEC (Rabbit) staining kit (Invitrogen,
Carlsbad, CA, USA).
References
1. Melcher AH (1985) Cells of periodontium: 11. Jo YY, Lee HJ, Kook SY, Choung HW, Park
their role in the healing of wounds. Ann R Coll JY, Chung JH, Choung YH, Kim ES, Yang
Surg Engl 67:130131 HC, Choung PH (2007) Isolation and charac-
2. Pitaru S, McCulloch CA, Narayanan SA (1994) terization of postnatal stem cells from human
Cellular origins and differentiation control dental tissues. Tissue Eng 13:767773
mechanisms during periodontal development 12. Ivanovski S, Haase HR, Bartold PM (2001)
and wound healing. J Periodont Res 29: Isolation and characterization of fibroblasts
8194 derived from regenerating human periodontal
3. Gould TR, Melcher AH, Brunette DM (1980) defects. Arch Oral Biol 46:679688
Migration and division of progenitor cell popu- 13. Luan X, Ito Y, Dangaria S, Diekwisch TG
lations in periodontal ligament after wounding. (2006) Dental follicle progenitor cell hetero-
J Periodont Res 15:2042 geneity in the developing mouse periodon-
4. Seo BM, Miura M, Gronthos S, Bartold PM, tium. Stem Cells Dev 15:595608
Batouli S, Brahim J, Young M, Robey PG, 14. Techawattanawisal W, Nakahama K, Komaki
Wang CY, Shi S (2004) Investigation of multi- M, Abe M, Takagi Y, Morita I (2007) Isolation
potent postnatal stem cells from human peri- of multipotent stem cells from adult rat peri-
odontal ligament. Lancet 364:149155 odontal ligament by neurosphere-forming cul-
5. Gronthos S, Mrozik K, Shi S, Bartold PM ture system. Biochem Biophys Res Commun
(2006) Ovine periodontal ligament stem cells: 357:917923
isolation, characterization, and differentiation 15. Seo BM, Miura M, Sonoyama W, Coppe C,
potential. Calcif Tissue Int 79:310317 Stanyon R, Shi S (2005) Recovery of stem cells
6. Shi S, Bartold PM, Miura M, Seo BM, Robey from cryopreserved periodontal ligament.
PG, Gronthos S (2005) The efficacy of mesen- J Dent Res 84:907912
chymal stem cells to regenerate and repair dental 16. Pihlstrom BL, Michalowicz BS, Johnson NW
structures. Orthod Craniofac Res 8:191199 (2005) Periodontal diseases. Lancet 366:
7. Gronthos S, Mankani M, Brahim J, Robey PG, 18091820
Shi S (2000) Postnatal human dental pulp stem 17. Polimeni G, Xiropaidis AV, Wikesjo UM
cells (DPSCs) in vitro and in vivo. Proc Natl (2006) Biology and principles of periodontal
Acad Sci U S A 97:1362513630 wound healing/regeneration. Periodontol
8. Trubiani O, Di Primio R, Traini T, Pizzicannella 2000 41:3047
J, Scarano A, Piattelli A, Caputi S (2005) 18. Lin NH, Gronthos S, Bartold PM (2008) Stem
Morphological and cytofluorimetric analysis of cells and periodontal regeneration. Aust Dent
adult mesenchymal stem cells expanded ex vivo J 53:108121
from periodontal ligament. Int J Immunopathol 19. Zannettino AC, Paton S, Arthur A, Khor F,
Pharmacol 18:213221 Itescu S, Gimble JM, Gronthos S (2008)
9. Chen SC, Marino V, Gronthos S, Bartold PM Multipotential human adipose-derived stromal
(2006) Location of putative stem cells in stem cells exhibit a perivascular phenotype in vitro
human periodontal ligament. J Periodont Res and in vivo. J Cell Physiol 214:413421
41:547553 20. Shi S, Gronthos S (2003) Perivascular niche of
10. Nagatomo K, Komaki M, Sekiya I, Sakaguchi postnatal mesenchymal stem cells in human
Y, Noguchi K, Oda S, Muneta T, Ishikawa I bone marrow and dental pulp. J Bone Miner
(2006) Stem cell properties of human peri- Res 18:696704
odontal ligament cells. J Periodont Res 41: 21. Filshie RJ, Zannettino AC, Makrynikola V,
303310 Gronthos S, Henniker AJ, Bendall LJ, Gottlieb
DJ, Simmons PJ, Bradstock KF (1998) marrow stromal fibroblasts form bone after
MUC18, a member of the immunoglobulin transplantation in vivo. J Bone Miner Res
superfamily, is expressed on bone marrow 12:13351347
fibroblasts and a subset of hematological malig- 25. Muraglia A, Cancedda R, Quarto R (2000)
nancies. Leukemia 12:414421 Clonal mesenchymal progenitors from human
22. Fedarko NS, Fohr B, Robey PG, Young MF, bone marrow differentiate in vitro according
Fisher LW (2000) Factor H binding to bone to a hierarchical model. J Cell Sci 113:
sialoprotein and osteopontin enables tumor 11611166
cell evasion of complement-mediated attack. 26. Gronthos S, Brahim J, Li W, Fisher LW,
J Biol Chem 275:1666616672 Cherman N, Boyde A, DenBesten P, Robey PG,
23. Ingram RT, Clarke BL, Fisher LW, Fitzpatrick Shi S (2002) Stem cell properties of human den-
LA (1993) Distribution of noncollagenous tal pulp stem cells. J Dent Res 81:531535
proteins in the matrix of adult human bone: 27. Gronthos S, Zannettino AC, Hay SJ, Shi S,
evidence of anatomic and functional heteroge- Graves SE, Kortesidis A, Simmons PJ (2003)
neity. J Bone Miner Res 8:10191029 Molecular and cellular characterisation of
24. Kuznetsov SA, Krebsbach PH, Satomura K, highly purified stromal stem cells derived from
Kerr J, Riminucci M, Benayahu D, Robey PG human bone marrow. J Cell Sci 116:
(1997) Single-colony derived strains of human 18271835
Chapter 25
Constructing Tissue Microarrays: Protocols and Methods

Considering Potential Advantages and Disadvantages
for Downstream Use
Lynne Bingle, Felipe P. Fonseca, and Paula M. Farthing
Abstract
Tissue microarrays were first constructed in the 1980s but were used by only a limited number of researchers
for a considerable period of time. In the last 10 years there has been a dramatic increase in the number of
publications describing the successful use of tissue microarrays in studies aimed at discovering and validating
biomarkers. This, along with the increased availability of both manual and automated microarray builders
on the market, has encouraged even greater use of this novel and powerful tool. This chapter describes the
basic techniques required to build a tissue microarray using a manual method in order that the theory behind
the practical steps can be fully explained. Guidance is given to ensure potential disadvantages of the
technique are fully considered.
Key words Tissue, Microarray, Immunohistochemistry, Candidate, Biomarker, Validation
1 Introduction
In 1986, Battifora [1] first described the technique of assembling

multiple tissues in a block or multitumor (sausage) tissue blocks
now referred to as tissue microarrays (TMA). Individual research-
ers further refined the technique [24] before automated microar-
rayers appeared on the market. These have allowed an explosion of
publications citing the importance of tissue microarrays in bio-
marker validation, particularly with regard to candidate cancer
genes. The recent increased interest in the use of tissue arrays has
followed the increased output of potential biomarkers from high-
throughput assays such as cDNA microarrays. Advances in a num-
ber of technologies related to biomedical research mean that
parallel in situ detection of DNA, RNA and protein targets, using
microarrays, is a now a real possibility, with true validation of
candidate markers. Tissue microarrays are also no longer limited to
formalin-fixed paraffin embedded tissues as frozen tissue
429
430 Lynne Bingle et al.
microarrays have been used successfully and are being marketed

alongside more traditional tissue arrays.
The immediate benefits of this research tool include the abil-
ity to screen large numbers (hundred plus) of tissue samples on
one slide; this not only reduces costs in terms of both labor and
consumables, but it also optimizes the use of tissue specimens
and allows greater uniformity in the detection of the target and
thus allows for more consistent analysis and greater confidence in
results obtained. Selection of material (cores) used to build the
array needs to be made with great care as not all tissue sections
are homogenous and bias may be introduced. Furthermore, it
might be the aim of the study to investigate changes in specific
areas of diseased tissues, for example leading edge of tumors, and
again without careful selection of tissue cores bias could be intro-
duced to a study. To avoid redundancy of tissue, to fully validate
a candidate gene, and to allow a microarray to be of value to
multiple researchers with multiple interests, a broad range of tis-
sues needs to be included. At first inspection the collection of
multiple tissue cores from a limited resource in a tissue bank
might be considered a negative aspect of the method, as when
multiple cores of tissue are removed from a tissue block that
block/sample is no longer available to the tissue bank. However,
as a large number of studies can be done with one sample, includ-
ing many immunohistochemical reactions, as long as the con-
struction of the TMA is well planned then the negative aspects
can be overcome. Technical problems have been highlighted by a
number of researchers, including the use of adhesive-coated tape
to assist cutting sections from the array, which has led to increased
nonspecific staining with immunohistochemistry. Sections should
be used as soon after sectioning as possible to ensure optimal
results but over a period of time this can mean cutting into a
block on numerous occasions and each time this results in tissue
loss and wasting of vital resources. This can be further exacer-
bated, as cutting arrays is more technically demanding than cut-
ting normal sections. In order to overcome this problem all
sections should be cut at the same time but to ensure any unused
sections are preserved for future use, commercial companies coat
arrays with a protective layer of wax.
Including more than one core per case is also important
because, dependant upon the tissue type being used, and also on
the necessity for antigen retrieval in the staining method used,
cores of tissue may be lost during the detection procedure. This
can not only reduce the number of samples available for analysis
but also make identification of remaining cores technically chal-
lenging. These points will be addressed in greater detail in the
following sections.
Constructing Tissue Microarrays 431
2 Materials
1. Tissue Arrayer: Manual or automated: there are many companies

now that supply equipment needed to manufacture tissue
microarrays ranging from manual arrayers at the lower end of
the price range to very complex, and expensive, automated
arrayers.
2. Silicon reusable molds for recipient blocks, bought alongside
punch needles (with plungers), are the most inexpensive option
for constructing a microarray. These molds can also be used to
construct frozen tissue microarrays.
3. Simple manual tissue microarrayers usually consist of premade
recipient blocks (melt when heated at 60 C for 30 min) and
biopsy punch needles (available in 1, 1.5, 2, 3, and 5 mm
diameter) with plunger (Fig. 1).
4. Plastic cassettes.
5. Embedding material (paraffin or OCT depending on source of
tissue).
6. Glass slides.
7. Cold plate.
8. Water bath.
9. Laboratory oven.
10. 20 C Freezer.
11. Ice tray.
12. Microtome.
13. Cryostat.
14. Metal chuck: this will be attached to the OCT array mold to
enable cutting of sections.
Fig. 1 Biopsy punch needle with plunger used to collect tissue core and transfer
to recipient block (also shown) for manual tissue microarray assembly
15. Copper plate: to be used when slightly melting OCT to ensure

the surface of the array is completely flat and thus allows
sections to be cut easily and efficiently.
16. H&E sections of tissues to be used to build TMA.
17. Marker pens.
3 Methods
The basics of constructing a tissue microarray are to prepare the

embedding media, to prepare the recipient block, to extract cylin-
ders of donor tissue using a sharp punch, and then to assemble
these cores into a recipient block. Initially, the only embedding
material considered for tissue arrays was paraffin but more recently
it has been demonstrated that OCT also provides an ideal medium
for the construction of frozen tissue arrays. Once the array has
been constructed the real benefits of this resource become imme-
diately apparent as many sections (up to 200) from up to 600 indi-
vidual donors can be produced saving labor, reagents, and, in some
cases, valuable resources (tissue). However, once again the user
must reflect and consider the source of the material, as the cases
may have been fixed/processed in many different ways, even if
they all come from the same institution. To collect 600 cases for an
array the user may have collected material from a number of differ-
ent institutions where the tissues will almost certainly have been
processed differently. This must be taken into consideration when
analyzing results.
For the purposes of this chapter it is easier, and of greater
benefit to the reader, to describe the manual construction of a tis-
sue array as this can very easily be done in almost all laboratories
with a tissue bank. The automated methods of creating a tissue
array are based on this manual method and will have specific steps
dependent on the source of the instrument. It is not within the
remit of this chapter, therefore, to describe all of the automated
arrayers currently available on the market but an understanding of
the manual process will still assist those using automated arrayers.
3.1 Paraffin 1. The mold needs to be submerged in paraffin along with a plas-
Embedded Tissue tic cassette for around 15 min to heat them up. Molds, shown
Microarrays in Fig. 2, demonstrate the variation in the number of core
holes that can be created.
2. Place the mold onto a cold plate and after filling the mold with
paraffin (embedding infiltration type) set the plastic cassette in
the center of the mold. Ensure the mold and cassette are com-
pletely filled with paraffin. .
3. Allow to cool for around 60 min and then it should be relatively
easy to separate the cassette and the mold. This should have
Fig. 2 Image depicting Arraymold (www.arraymold.com)
produced a cassette with a number of holes (determined by

your mold and the design of your array) ready to receive donor
tissue cores.
4. Using an H&E section determine the selected area of tissue for
each core and mark the slide accordingly.
5. Use this as a guide and the biopsy punch needle to collect the
first tissue core.
6. Transfer the first tissue core to the correct place in your recipi-
ent block by placing the biopsy needle over the whole and
pushing the stylet. As the stylet is flush with the needle tip the
core can continue to be inserted until it is flush with the top of
the array block. To ensure the core is flush with the top of the
array, it is good practice to turn the block over onto a hard, flat
surface after each core has been added and press down hard.
7. Repeat this process until all cores have been transferred to the
recipient block and then place the block, set face down on a
glass slide, into an oven at 40 C to soften both the paraffin
and the tissue so that they stick to each other.
8. When the paraffin and tissue are both soft gently press the slide
and the block together to set the cores into the array block. Do
not separate the slide and the block yet.
9. A second, hotter slide (7080 C) is needed at this stage. Place
the hot slide on the top of the first slide, the surface of the
paraffin quickly softens and becomes malleable so both slides
can move easily across the surface of the block. This sets the
surface of the array block and gets rid of any air pockets.
10. Remove the hot slide and turn the original slide/block onto a
flat surface, allow to cool to room temperature. Transfer slide/
block to an ice tray to cool further so that the slide can be

removed from the block easily.
11. The block is ready to be cut in the same way as a normal tissue
block.
3.2 Frozen Tissue 1. The same molds and biopsy punch needles used for manual
Microarrays preparation of paraffin tissue arrays can be used to construct
arrays from frozen tissue samples.
2. Fill the mold with OCT taking care not to introduce air bub-
bles as these can result in air pockets and interfere with down-
stream steps.
3. Place the mold either in a 20 C freezer or in a cryostat
(20 C) until the OCT is completely frozen at which time the
OCT can be pulled away from the mold. Attach a metal chuck
to the array mold to enable cutting of sections.
4. Frozen or fresh tissue can be used in these arrays but the same
considerations as for Formalin-fixed paraffin embedded (FFPE)
tissue selection should be made.
5. The cores need to be set as they will be loose in the OCT mold
at this stage. Place the array mold carefully onto a room tem-
perature copper plate to melt the OCT, ensuring you do not
dislodge any cores. When you can see melted OCT around the
edge of the array mold move the mold backward and forward
on the plate to help flatten the surface and set the cores. Place
the plate and mold into a freezer or a cryostat and allow the
plate to freeze to the mold.
6. Separate the plate and mold; the array will now have a flat sur-
face and the cores will be set.
7. Trim the melted OCT so that the array block can clear the
cryostat blade and cut sections in the same manner as normal
frozen sections.
4 Notes
1. Aside from the manual arrayers described in the methods

section of this chapter, there are alternative array builders now
available on the market.
(a) Semiautomated tissue arrayers are available allowing auto-
mated movement of the recipient block so that up to four
recipient blocks can be created at one time.
(b) The Automated Tissue Microarrayer varies considerably
in price but is essentially a computer-aided instrument
that selects sample tissue from donor blocks and auto-
matically inserts the selected tissue into a premade recipient
block.
2. The most important consideration in building a tissue array is

the source of the donor tissue and also the final use of the tissue
array. Once tissue cores have been removed from a block, the
block is no longer available to other members of a tissue bank/
department/research group. Thus, the array builder must be
confident that the tissue is not so valuable that it is completely
irreplaceable; the final array must be fit for purpose and ideally
fit for use by a number of researchers.
3. A further consideration when building a tissue array is how
representative a core might be of the tissue and previous stud-
ies have been conducted to demonstrate validity of TMAs in
the microscopic analysis of various human diseases. The main
purpose of many tissue arrays is to allow researchers to com-
pare expression of genes and/or proteins in diseased tissues
(quite often cancers) with respect to expression in normal
tissues. The selection of cores needs to be made to answer
the particular question of the researcher. For example if the
study aims to determine changes in protein expression across a
tumor, into the leading edge and relating to the metastatic
potential of the tumor, then the array must contain representa-
tive regions of all of these areas of a tumor but also related
normal tissues.
4. A common criticism of the use of small tissue cores is that they
do not truly represent the parent tissue and in answer to this
criticism the size of punches used to collect cores has both
increased and decreased. The downside of using larger cores is
that fewer samples can be assembled into an array and more
damage is caused to the donor block. Overall, it must be
remembered that cores are taken from previously selected
areas; they are not randomly collected, a feature of arrays that
is quite often forgotten.
5. The most important steps in building an FFPE array, as alluded
to above, are Subheading 3.1, steps 4 and 5, the initial choice
of the tissue to be used.
(a) It is vital that the full use of the microarray is considered
before any tissue cores are selected. Use of tissue cores
renders a tissue block redundant and thus to ensure maxi-
mum use is made of potentially limited resources construc-
tion of a tissue microarray must be made with great care.
(b) Selection of appropriate and representative areas of a sec-
tion is more likely to be successful if the help of a patholo-
gist is sought.
(c) Areas of a section to be used are marked on an H&E slide
that should have been recently prepared to ensure vital fea-
tures have not been lost through their use in previous
studies.
6. Subheading 3.1, steps 5 and 6 describe the collection of cores

using biopsy punch needles. If the arrays are to be used only
for protein expression, the same needle can be used for all
cores in the array. If, however, the array is going to be used for
gene, particularly RNA expression, the needle should be
changed for each core.
7. Loss of tissue cores can present major difficulties, can under-
mine the power of the planned experiment, and can make
interpretation of results difficult as orientation becomes near
impossible. Orientation, correct identification of all cores in an
array, is usually facilitated by careful mapping of the position-
ing of all tissue cores when first placing them into a block. The
use of control tissues, especially normal samples within an array
of cancers or diseased samples, will also assist correct orienta-
tion as will negative controls in specific sites and asymmetric
patterning. Some researchers also suggest using nonrelevant
tissue cores around the periphery of the TMA block to avoid
immunohistochemical precipitation in the most peripherally
located cores.
8. Subheading 3.1, steps 710 describe the softening of both the
paraffin and tissue so that they stick to each other and are set in
the array. The amount of time needed for this step is variable
and dependent upon the tissue. The time needed could be as
short as 15 min, is quite often standardized to 12 h but in
some instances it may be best to leave the block in the oven at
40 C overnight to ensure the tissue and paraffin are fully
mixed.
9. If you do not have more than one oven (for Subheading 3.1,
step 9) you need to find somewhere to keep the block and
slide warm until your second slide has been warmed. The side
of a water bath is often a convenient place for this step but do
ensure that the slide/block does not get too warm as they will
start to melt. It may be prudent to use a blank paraffin block
and slide to test whether the side of your water bath is, in fact,
the ideal place to keep them at around 40 C while warming
up the second slide to 70 C.
10. When cutting sections from the blocks it might first seem that
some punch cores are missing when in fact they are more likely
to just be curled. The cores will flatten out when the ribbon is
laid onto a water bath.
11. Frozen tissue arrays can be prepared in a very similar manner
to that of FFPE arrays. The important differences have been
outlined in Subheading 3.2 with a further variation on the col-
lection and use of frozen tissue in TMAs being described by
Torata et al. [5].
12. Ideal frozen tissue thickness is around 1 cm and cores should

be transferred, using a biopsy punch, to the array mold that is
kept in the cryostat to ensure that neither the OCT mold nor
the tissue warms. It is important that the tissue is inserted into
the mold as quickly as possible; otherwise, it will freeze as it is
being inserted and the final structure of the array will be
uneven.
13. It is also important to be aware that although the methods sec-
tion describes setting of cores in the OCT it is almost certain
this will need to be repeated as you trim the block and also as
you cut deeper into the block. The use of a room temperature
copper plate, exactly as described in the methods section, will
ensure the cores remain set into the OCT and you will continue
to be able to cut your block without damaging any cores.
14. Using tissue microarrays for immunohistochemistry has many
benefits, as described previously, and particularly in terms of
labor, precious and potentially expensive reagents (antibodies),
and valuable tissue resources. There are, however, a number of
problems that have been highlighted through the increased
use of tissue microarrays. A number of antibodies used in
immunohistochemistry rely upon heat-induced epitope
retrieval and this can result in the loss of a significant number
of small tissue cores. This tissue loss is, to a certain extent,
dependent upon the source of the tissue and can be amelio-
rated somewhat if the arrays are built on high quality glass
slides, specifically prepared to improve tissue adhesion.
15. The efficiency of staining appears to be very much dependent
upon the use of fresh sections cut from tissue arrays. If sections
are cut and stored for any period of time before use, then the
staining pattern is not consistent and the use of positive con-
trol antibodies has suggested that the tissues do not react with
the antibodies in the manner expected. This could lead to erro-
neous results. In order to alleviate this problem major manu-
facturers producing multiple tissue microarrays for commercial
purposes coat the array sections with a protective layer of paraf-
fin that can easily be removed by heating the slide in an oven
immediately prior to use. Individual users who do not have
access to facilities that will enable this coating procedure should
take care to use the array immediately after the section has
been cut.
16. Analysis of results reveals the true power of this method but
must be undertaken with great care and rigor so as not to
introduce bias. There are many scanning and image analysis
software packages available on the market. These are not only
less time consuming for the researcher, they also allow greater
uniformity and reduce user bias.
References
1. Battifora H (1986) The multitumor (sausage) 4. Kononen J, Bubendorf L, Kallioniemi A,
tissue block: novel method for immunohisto- Barlund M, Schraml P, Leighton S, Torhorst J,
chemical antibody testing. Lab Invest 55: Mihatsch MJ, Sauter G, Kallioniemi OP (1998)
244248 Tissue microarrays for high-throughput molecu-
2. Wan WH, Fortuna MB, Furmanski P (1987) A lar profiling of tumor specimens. Mat Med
rapid and efficient method for testing immuno- 4:844847
histochemical reactivity of monoclonal antibod- 5. Torata N, Ohuchida K, Akagawa S, Cui L,
ies against multiple tissue samples simultaneously. Kozono S, Mizumoto K, Aishima S, Oda Y,
J Immunol Methods 103:121129 Tanaka M (2014) Tissue tablet method: an effi-
3. Battifora H, Mehta P (1990) The checkerboard cient tissue banking procedure applicable to
tissue block. An improved multitissue control both molecular analysis and frozen tissue micro-
block. Lab Invest 63:722724 array. Hum Pathol 45:143152
Chapter 26
Growing Adipose-Derived Stem Cells Under

Serum-Free Conditions
DiogoGodoyZanicotti andDawnE.Coates
Abstract
Growing adipose-derived stem cells (ADSC) in serum-free conditions is important as it represents a way of
expanding multipotent cells in a clinical grade medium. Most cultured ADSC are expanded and tested in
serum-containing media, which can pose significant health risks if these cells were used in clinical applica-
tions. Moreover, cells grown in serum-free conditions behave significantly different than those cultured in
serum-containing media. Here, we present a technique to culture adipose-derived stem cells in serum-free
conditions. The methods described in this chapter were optimized for ovine ADSC.The appropriate
optimization should be done for other cell lines.
Key words Adipose-derived stem cells (ADSC), Mesenchymal stem cells (MSC), Serum-free media,
Ovine, Fetal bovine serum (FBS)
1 Introduction
The use of mesenchymal progenitor cells, also called mesenchymal

stem cells (MSC), has arisen as an important tool for regenerative
procedures. MSC are progenitor cells with regenerative potential liv-
ing in virtually every tissue of mesenchymal origin. These cells have
been isolated from multiple tissues, including but not limited to,
bone marrow, fat, periosteum, gingival connective tissue, dental
pulp, and periodontal ligament [16]. MSC have shown the capacity
to differentiate into multiple cell types (e.g., osteoblasts, chondro-
blasts, and adipocytes) and to collaborate with the regeneration
process of tissues through different mechanisms such as microenvi-
ronment modification, stimulation of native stem cells, direct dif-
ferentiation, neovascularization, and immunoregulation [7].
One of the major issues regarding the culture of ADSC for
clinical application is the routine usage of serum of animal origin.
Serum of animal origin (i.e., fetal bovine serum-FBS) has been
traditionally used to provide nutrients and adhesion proteins to
MSC. However, there is lot-to-lot variation between FBS batches,
439
440 DiogoGodoyZanicotti andDawnE.Coates
which affects the proliferation and differentiation potential of

ADSC.Also, unknown serum proteins may have unexpected and
uncontrolled effects on ADSC.Moreover, there is an inherent risk
of FBS being contaminated with mycoplasma, prions, and viruses,
which pose significant safety concerns for clinical use [811].
For the reasons described above, it is of utmost importance
that ADSC be cultured in serum-free media. This chapter presents
a method to culture ovine ADSC (oADSC) in serum-free condi-
tions and can be optimized to MSC derived from tissue locations
other than adipose.
2 Materials
Prepare all materials at room temperature (RT) unless stated

otherwise. All reagents were made up with ultrapure water unless
otherwise stated. Prepare all cell culture reagents inside a Class II
Biosafety Cabinet. Please dispose of all biological material and
waste products following good laboratory practice and waste
disposal regulations.
2.1 Cell Culture 1. Preparing sheep serum (SS): The final volume of serum
Components obtained will be approximately 50% of the whole blood after
clotting. Retrieve 100mL of whole blood and place into BD
vacutainer tubes containing gel for serum separation (Becton,
Dickinson and Company 0268396). Let the blood clot at 4C
for 2030min and then spin the vacutainers at 2000g for
10min at 4C.Filter the serum with a Millex-GP 45m
syringe filter and then with a 22m filter (Merck-Millipore
Corporation, Germany). Transfer the serum to a 50mL sterile
universal flask. Perform heat inactivation for 30min at 56C
(see Subheading2.1, item 2).
2. Heat inactivation of SS (protocol from Serum Source
International [12]: Allow the SS to acclimate at RT for a mini-
mum of 10min. Prepare the water bath, with sufficient water
to submerge the serum, to a controlled temperature of 56C
for the heat inactivation process. Prepare a control bottle filled
with water to monitor the water bath temperature. Place the
control and SS bottles into the 56C water bath. Use a sus-
pended glass thermometer (not touching the sides or bottom
of the bottle) to monitor the temperature inside the control
bottle. Set the timer to 30min once the temperature of the
control bottle reaches 56C.Gently swirl the bottles every
4min to make certain the SS will remain uniform throughout
the heating process. Remove the heat-inactivated SS after
30min and gently swirl once again. Allow the bottle to cool
down to RT.Aliquot and store the sheep serum at -20C.
Growing Adipose-Derived Stem Cells Under Serum-Free Conditions 441
3. Krebs-Ringer buffer (prepared as per manufacturers recom-

mendations, pH 7.4; Sigma-Aldrich K4002) containing
4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES):
add 100L of 1M HEPES to 9.9mL of Krebs-Ringer buffer.
This will result in 10mM HEPES in Krebs-Ringer buffer.
4. Phosphate-buffered saline (PBS) containing 20mg/mL
bovine serum albumin (BSA also referred to as Fraction V;
Sigma-Aldrich) and 1 % antibiotic-antimycotic (Gibco,
ThermoFisher, Auckland, New Zealand): gently mix 600mg
of BSA in 30mL of PBS (see Note 1). Add 0.3mL of antibiotic-
antimycotic and mix gently.
5. Make a 0.6mg/mL collagenase solution (Collagenase type II
for adipocyte isolation, derived from Clostridium histolyticum;
Sigma-Aldrich C6885) diluted with KrebsRinger buffer con-
taining HEPES (see Subheading2.1, item 3): mix 6mg of
collagenase type II in 10mL of Krebs-Ringer buffer with
HEPES.Filter sterilize using a 0.22m membrane filter. This
solution should be used immediately.
6. Complete serum-free medium (SFM): StemPro MSC SFM
Basal Medium (Life Technologies A13829-01) requires sup-
plementation with StemPro MSC SFM XenoFree Supplement
(Life Technologies A11577-01) and GlutaMAX-CTS (Life
Technologies A12860) or 200mM l-Glutamine (Life
Technologies). To make 500mL of complete medium, asepti-
cally add 5mL of StemPro MSC SFM XenoFree supplement
to 500mL of StemPro MSC SFM basal medium. Add 5mL
of GlutaMAX and 250L of 10mg/mL of gentamicin reagent
solution (Life Technologies), to the medium before use. In
addition, add 5mL of antibiotic-antimycotic (Life
Technologies) to the medium to a final concentration of 1%.
The use of antibiotics and antimycotic is optional. The use of
antibiotics and antimycotic does not replace good aseptic
technique.
7. Coat a T75 tissue culture flask (MediRay, New Zealand) with
1mg/cm2 of bovine plasma fibronectin (cell culture grade
fibronectin from bovine plasma; Sigma-Aldrich) or use sheep
fibronectin if available, and incubate for 1h at 37C.The fibro-
nectin solution was prepared in PBS but HBSS (Hanks
Balanced Salt Solution; Life Technologies) can also be used
instead. A volume of 3mL of fibronectin solution should be
used for a T75 flask. Make sure to coat the whole surface of
the T75. The fibronectin solution should be used immediately.
The number of T75 flasks will depend on the number of cells
and or cell passages required.
8. The PBS, pH 7.4, buffer used in all experiments is without
calcium or magnesium unless otherwise stated. 1M stock
EDTA (Sigma-Aldrich) was made in deionized H2O and

added to the PBS for a final concentration of 1mM for
Subheading3.1, step 7.
9. 70m nylon mesh, gamma irradiated cell strainer in a polypro-
pylene frame (MediRay, New Zealand).
10. TrypLE reagent (Life Technologies). Used for the dissociation
of cells from the tissue culture plastic during cell culture.
11. Dimethyl sulfoxide (DMSO; Sigma Aldrich).
3 Methods
3.1 Isolating ADSC This method was adapted from that described previously by
Niemeyer etal. [13].
1. Start with approximately 10g of adipose tissue collected into
PBS containing 20mg/mL BSA and antibiotic-antimycotic
kept at RT (see Note 2).
2. Wash the tissue twice with 30mL PBS containing 20mg/mL
BSA and 1% antibiotic-antimycotic. Discard the solution and
keep the adipose tissue (use tweezers or any other sterile instru-
ment as necessary).
3. Place the tissue in a Petri dish and reduce to portions of
approximately 12mm2 (see Note 3). Remove as much of the
fibrous material and blood vessels as possible (see Note 4).
4. Incubate the adipose tissue with an equivalent volume of col-
lagenase solution (1:1) and agitate lightly at 37C for 90min.
5. Prepare the SFM containing 2.5% of SS.Add 12.5mL of SS to
487.5mL of SFM.
6. Centrifuge the cells at 600g for 10min at RT, discard the
supernatant together with the lipid layer, and keep only the cell
pellet.
7. To inactivate the collagenase briefly wash and resuspend the
cell pellet with 10mL PBS containing EDTA (1mM) heated
to 37C.
8. Filter the cells through a 70m cell strainer to remove the cell
clumps and endothelial cell aggregates (see Note 5).
supernatant, and keep the cell pellet.
10. Wash the cell pellet with 10mL of PBS pre-warmed to 37C.
supernatant, and keep the cell pellet.
12. Resuspend the cells in 10mL of SFM containing autologous
SS and then culture and adapt to serum-free conditions as fol-
lows in Subheading3.2.
Table 1
Serum-free adaptation
Passage Sheep seruma (%) Sheep serum (%)

P0 20 2.5
P1 2.5 0
P3 0 0
a
Use this column only if poor cell attachment was achieved, otherwise use the second
column
3.2 Cell Culture 1. Centrifuge the cells at 600g for 10min at RT and discard the
andSerum-Free supernatant. Resuspend the cells in 12mL of SFM with 2.5%
Adaptation SS (see Table1, Notes 6, and 7).
2. Plate the 12mL SFM with 2.5% SS containing the oADSC
(ovine adipose-derived stem cells) into a T75 tissue culture
flask in a cell culture incubator with a humidified atmosphere
at 37C and 5% CO2.
3. After 1h, remove the medium and gently wash the cells once
with 15mL of PBS pre-warmed to 37C to remove nonadher-
ent cells [14]. Add 12mL of fresh SFM containing 2.5% SS
(see Note 8).
4. Replace the medium after 24h and then every 48h subse-
quently until reaching 90% confluence. At this stage cells are
ready to be used unless higher number of cells are required.
If necessary passage cells as follows.
3.3 Cell Passage 1. Coat three T75 with 1mg/cm2 of bovine plasma fibronectin
(use sheep fibronectin if available) and incubate for 1h at 37C.
2. Observe the oADSC under an inverted microscope and confirm
that cells are ready to be sub-passaged (8090% confluent).
3. Pre-warm TrypLE reagent and SFM to 37C before use and
add 10mL of pre-warmed SFM to a 50mL sterile universal
flask in preparation for the cells.
4. Discard the medium from the T75 flask.
5. Briefly wash the cells with 10mL of sterile PBS and then
discard.
6. Add 3mL of TrypLE to the T75 flask. Tilt the flask in all direc-
tions to evenly distribute the reagent. Incubate the cells in
TrypLE for 3min at 37C.
7. After incubation, check the flask under the microscope for cell
detachment. Tap the flask firmly (more than once if necessary)
to facilitate complete cell detachment.
8. Add 7mL of pre-warmed SFM to the T75 flask. Collect the
cell suspension and transfer to the 50mL sterile universal
containing 10mL of SFM.Tap the T75 firmly, rewash with

10mL SFM, and collect the suspension to the 50mL sterile
universal containing the remaining of the cell suspension.
9. Centrifuge the cell suspension at 150g for 5min at RT.
10. Discard the supernatant and resuspend the cells in a minimal
volume (usually 1mL) of pre-warmed SFM for cell counting.
11. Stain cells 1:1 with trypan blue solution and count using a
haemocytometer.
12. Add 10mL of pre-warmed SFM (with or without SS, see
Table1) to each fibronectin coated T75.
13. Add enough cell suspension to each T75 to provide 5104
cells/cm2 (i.e., 3.75106 cells per T75 flask). Gently swirl the
cell suspension to ensure even distribution.
14. Place the culture flask in the cell culture incubator at 37C
with a humidified atmosphere of 5% CO2.
15. Replace the medium after 24h and then every 48h subse-
quently until reaching 90% confluence. At this stage cells are
ready to be used unless a higher number of cells are required.
If so, repeat the passage steps above.
3.4 Cryopreservation If the cells need to be stored, follow the steps below after
Subheading3.3, step 10 of the passage protocol described above.
1. Prepare a cryopreservation solution (2) by supplementing
pre-warmed SFM with 20% DMSO.A fresh solution should
be used.
2. Resuspend the harvested cell pellet to twice the desired final
cell concentration (i.e., enough to seed a T25 flask; 2106cells/
mL) in pre-warmed SFM.
3. Add 1:1 the cryopreservation solution slowly to the cell sus-
pension, and mix gently to ensure homogeneity.
4. Add immediately the desired volume of cell suspension (i.e.,
1mL) to the sterile cryovials at RT.
5. Place the cryovials in a cryogenic freezing container (i.e., Mr
Frosty (-1C/min) Freezing Container) and then place in a
freezer at -70C.
6. After 24h, transfer the frozen cells to liquid nitrogen (vapor
phase) for long-term storage.
4 Notes
1. Complete mixing of the BSA will take some time and is volume-
dependent. Use a rotating platform at low speed with the BSA
solution on ice during mixing. The liquid will be clear when
mixing is complete. Do not shake to avoid bubbles.
2. The quantity of ADSC will depend on the method used for

retrieving fat from the animal. We believe the easiest method is
to perform a surgical exposure and dissection of the fat. The
best location to collect the fat is between the shoulder blades
of the sheep. Good quantities can also be obtained from the
hip region. This method allows for a clean and abundant quan-
tity of fat. Other methods such as liposuction are also efficient
but the tissue obtained is less clean. However, the quantity and
quality of fat will depend on the animal. We recommend col-
lecting fat from animals weighing at least 65kg and during
late spring or summer when the animal condition is optimal
(in temperate countries like New Zealand).
3. Use two new sterile scalpel (No. 15) blades to chop the tissue.
Use a cross action to finely chop the tissue.
4. You do not have to remove 100% of the fibrous material and
blood vessels as this would also remove a good part of the tis-
sue containing ADSC.Only remove the large blood vessels
and large pieces of fibrous material.
5. The liquid will take a while to pass through the filter. Also,
large cellular clumps and fibrous tissue might clog the filter. If
necessary, use a sterile instrument such as tweezers to swirl the
cell suspension and speed up the filtration process.
6. Most cells will cope well with 2.5% SS for the initial isolation
of ADSC.However, when poor attachment is noted, it is
important to collect and resuspend the cells in 20% SS.
7. The presence of collagenase may compromise the attachment
of ADSC.This step is important to remove the remaining col-
lagenase from the cell suspension.
8. This method aims to select a purer population of ADSC and also
to ensure that cells are adapting well to the medium offered.
Acknowledgments
The financial support provided by grants from the New Zealand

Lottery Board (Lottery Health Research Grant) and the Otago
Medical Research Foundation (Jack Thompson Arthritis Grant),
and a University of Otago Doctoral Scholarship (awarded to
D.G.Z.) are gratefully acknowledged. We also thank the Molecular
Biosciences Laboratory (Faculty of Dentistry) personnel.
References
1. Pittenger MF, Mackay AM, Beck SC, Jaiswal esenchymal stem cells. Science 284:
m
RK, Douglas R, Mosca JD, Moorman MA, 143147
Simonetti DW, Craig S, Marshak DR (1999) 2. Zuk PA, Zhu M, Mizuno H, Huang J, Futrell
Multilineage potential of adult human JW, Katz AJ, Benhaim P, Lorenz HP, Hedrick
MH (2001) Multilineage cells from human culture system under xeno-free conditions. Tissue
adipose tissue: implications for cell-based Eng Part C Methods 17:12011210.
therapies. Tissue Eng 7:211228. doi:10.1089/ten.tec.2011.0255
doi:10.1089/107632701300062859 9. Wang Y, Han ZB, Song YP, Han ZC (2012)
3. Roberts SJ, Owen HC, Tam WL, Solie L, Van Safety of mesenchymal stem cells for clinical
Cromphaut SJ, Van den Berghe G, Luyten FP application. Stem Cells Int 2012:652034.
(2014) Humanized culture of periosteal pro- doi:10.1155/2012/652034
genitors in allogeneic serum enhances osteo- 10. Brunner D, Frank J, Appl H, Schoffl H, Pfaller
genic differentiation and invivo bone W, Gstraunthaler G (2010) Serum-free cell cul-
formation. Stem Cells Transl Med 3:218228. ture: the serum-free media interactive online
doi:10.5966/sctm.2012-0137 database. Altex 27:5362
4. Gronthos S, Mankani M, Brahim J, Robey PG, 11. Shahdadfar A, Fronsdal K, Haug T, Reinholt
Shi S (2000) Postnatal human dental pulp stem FP, Brinchmann JE (2005) In vitro expansion
cells (DPSCs) invitro and invivo. Proc Natl of human mesenchymal stem cells: choice of
Acad Sci U S A 97:1362513630. serum is a determinant of cell proliferation, dif-
doi:10.1073/pnas.240309797 ferentiation, gene expression, and transcrip-
5. Seo BM, Miura M, Gronthos S, Bartold PM, tome stability. Stem Cells 23:13571366.
Batouli S, Brahim J, Young M, Robey PG, doi:10.1634/stemcells.2005-0094
Wang CY, Shi S (2004) Investigation of multi- 12. International SS (2014) Heat inactivation of
potent postnatal stem cells from human peri- fetal bovine serum (FBS). Serum Source Int.
odontal ligament. Lancet 364:149155. http://www.serumsourceintl.com/pdf/heat_
doi:10.1016/S0140-6736(04)16627-0 inactivation.pdf. Accessed 10 Aug 2014
6. Jin SH, Lee JE, Yun JH, Kim I, Ko Y, Park JB 13. Niemeyer P, Fechner K, Milz S, Richter W,
(2014) Isolation and characterization of human Suedkamp NP, Mehlhorn AT, Pearce S, Kasten
mesenchymal stem cells from gingival connec- P (2010) Comparison of mesenchymal stem
tive tissue. JPeriodont Res 50:461467. cells from bone marrow and adipose tissue for
doi:10.1111/jre.12228 bone regeneration in a critical size defect of the
7. Raynaud CM, Rafii A (2013) The necessity of sheep tibia and the influence of platelet-rich
a systematic approach for the use of MSCs in plasma. Biomaterials 31:35723579.
the clinical setting. Stem Cells Int 2013:892340. doi:10.1016/j.biomaterials.2010.01.085
doi:10.1155/2013/892340 14. Griesche N, Luttmann W, Luttmann A,
8. Santos F, Andrade PZ, Abecasis MM, Gimble Stammermann T, Geiger H, Baer PC (2010) A
JM, Chase LG, Campbell AM, Boucher S, simple modification of the separation method
Vemuri MC, Silva CL, Cabral JM (2011) Toward reduces heterogeneity of adipose-derived stem
a clinical-grade expansion of mesenchymal stem cells. Cells Tissues Organs 192:106115.
cells from human sources: a microcarrier-based doi:10.1159/000289586
Chapter 27
Quantitative Real-Time Gene Profiling ofHuman

Alveolar Osteoblasts
DawnE.Coates, SobiaZafar, andTrudyJ.Milne
Abstract
The use of quantitative real-time reverse transcriptase PCR (qRT2-PCR) for the identification of differentially
regulated genes is a powerful technology. The protocol presented here uses qRT2-PCR gene arrays to
investigate the regulation of 84 angiogenic related genes in human primary alveolar osteoblasts following
treatment with the bisphosphonate, zoledronic acid (ZA), and geranylgeraniol (GGOH). GGOH has
potential as a therapeutic agent for Bisphosphate-Related Osteonecrosis of the Jaw (BRONJ), a serious
side-effect resulting from the treatment for metastatic cancer (Zafar etal., J Oral Pathol Med 43:711721,
2014; Ruggiero, Ann NY Acad Sci 1218:3846, 2011). The isolation of the primary osteoblast cells fol-
lows the methods previously described (Dillon etal., Methods Mol Biol 816:318, 2012) with a new RNA
extraction technique described fully. The method highlights the importance of obtaining high-quality
RNA which is DNA-free. Relative levels of gene expression are normalized against selected housekeeping
genes (HKG) and a number of examples of how fold regulation (2Cq) and gene expression level (2Cq)
data can be presented are given.
Key words Quantitative real-time reverse transcriptase PCR (qRT2-PCR), Bisphosphonate-related

osteonecrosis of the jaw (BRONJ), Zoledronic acid, Geranylgeraniol, PCR arrays
1 Introduction
Human primary osteoblasts are the critical building block for bone
growth and remodeling. Understanding the activation and
response of osteoblasts to different stimuli or medicinal drug
therapies is important in multiple clinical conditions [13]. The
importance of autocrine and paracrine signaling from angiogenic
factors for bone growth is now well recognized [4, 5]. Research is
beginning to understand better the coupling of angiogenesis and
osteogenesis during bone growth and remodeling [6, 7].
The method presented here describes the isolation of primary
human osteoblasts, which were phenotyped elsewhere, and shown
to produce mineralized nodules that stained positively for osteocal-
cin (immunofluorescence) and produced calcium (alizarin red S),
447
448 Dawn E. Coates et al.
alkaline phosphatase activity, and phosphate (von kossa). There are

multiple immortalized osteoblast cell lines commercially available;
however, evidence suggests that they are phenotypically distinct
from primary human osteoblast cells invitro and thus this method
uses primary cells derived from the jaw bone to most closely align
with the cells affected by BRONJ [8]. A method for the recovery
of total RNA following treatment with of ZA and ZA+GGOH is
described. Once transcribed to cDNA the determination of the
expression levels of a panel of 84 angiogenic genes is outlined. A
number of ways to present fold regulation (2Cq) and gene expres-
sion level (2Cq) data in a meaningful way are also discussed.
2 Materials
2.1 Primary Human 1. All the procedures were conducted in a Class II biosafety cabi-
Alveolar Osteoblast net (AIRPURE) with sterile solutions and laboratory
Isolation andCulture consumables.
2. Dulbeccos Modified Eagle Medium with GlutaMAX (DMEM;
ThermoFisher, NZ).
3. Fetal Bovine Serum, qualified, New Zealand origin (FBS;
Gibco, ThermoFisher, NZ).
4. Antibiotic-antimycotic 100 (Gibco, ThermoFisher, NZ).
5. Gentamicin at 10mg/mL (ThermoFisher, NZ).
6. 2-Phospho-l-ascorbic acid trisodium salt (Cat. No. 49752,
Sigma-Aldrich, USA). To make a 100mM stock solution
161mg of the 2-Phospho-l-ascorbic acid trisodium salt is
diluted in 5mL of deionized Milli-Q water (MQH2O), mixing
thoroughly. The solution is filtered through a 0.22m filter,
aliquoted and frozen at -20C.
7. Dexamethasone Water-soluble and suitable for cell culture
(Sigma-Aldrich, USA). A concentrated stock solution of
10mM is made by dissolving 39.25mg of the dexametha-
sone in 10mL of MQH2O, with thorough mixing. The solu-
tion is filtered through a 0.22m filter, aliquoted and frozen
at -20C. A working stock of 10M is made with a 1:1000
dilution in sterile MQH2O, which is also aliquoted and fro-
zen at -20C.
8. -Glycerophosphate disodium salt hydrate (Cat. No. G9422,
Sigma-Aldrich, USA). A 500mM stock solution (1:100 con-
centrated) is made by thoroughly mixing 1.08g of
-glycerophosphate disodium salt hydrate in 10mL of MQH2O.
The solution is filtered through a 0.22m filter, aliquoted and
frozen at -20C.
9. Phosphate Buffer Saline, pH7.4without calcium, magne-
sium, or phenol red (PBS; ThermoFisher, NZ).
Quantitative Real-Time Gene Profiling ofHuman Alveolar Osteoblasts 449

10.
0.25
% Trypsin/EDTA with phenol red (Gibco,
ThermoFisher, NZ).
11. Dimethyl sulfoxidecell culture tested and >99% pure (Sigma-
Aldrich, USA).
12. Bone Ronguer 140165mm (Falcon Medical, UK).
13. 50mL sterile universalsCellStar blue screw cap tubes (Greiner
Bio-One, Germany).
14. Millex-GP 0.22m syringe filter (Merck Millipore
Corporation, Germany).
15. Multiwell sterile plates, 6-well, polystyrene, clear with lid
(Greiner Bio-One, Germany).
16. CELLSTAR Filter cap culture flask (T-75; Greiner Bio-One,
Germany).
17. Sterile surgical bladesNo 10 (Swann Morton Ltd, UK).
18. CO2 incubatorUV (MCO-19AIC, Sanyo Electric Biomedical).

19.
Counting chamber. Neubauerdouble cell clear sight
(Hawksley, UK).
20. Cryovials 2mL external thread (Greiner Bio-One, Germany).
21. Cell freezing deviceMr Frosty (Nalgene, USA).
2.2 Experimental 1. Zoledronic acid (ZA; Cat. No. 118072-93-8, Zometa,

Treatment Novartis, Switzerland). The ZA is aliquoted and stored at
ofOsteoblasts 4C.Prior to use a stock solution of 1mM is made in PBS.
2. Geranylgeraniol (GGOH; Cat. No. G3278, Sigma-Aldrich,
USA). A 10mM stock solution of GGOH is made in 100%
ethanol, which is then aliquoted and frozen at -20C.
2.3 Total RNA 1. TRIzol Reagent (ThermoFisher, NZ).

Extraction andDNase 2. PureLink RNA Mini Kit (Ambion, ThermoFisher, NZ).
1 Treatment
3. PureLink DNase I (ThermoFisher, NZ). For on-column
DNase treatment of the RNA.
4. Nuclease-free water (not DEPC-treated; ThermoFisher, NZ).
5. 70% ethanol (ENSURE Absolute for analysis, Merck, NZ).
Made with nuclease-free water.
6. Nikon Eclipse Ti-S microscope with a TS-TCC5.0ICE cooled
camera (Coherent Scientific, Australia).
2.4 Reverse 1. RT2 First Strand Kitkit includes genomic DNA elimination,
Transcription random hexamers and oligo-dT prime reverse transcription,
reverse transcriptase and built-in external RNA control to
monitor reverse transcription efficiency (Qiagen, Australia).
2. 0.2mL flat cap PCR tubes (Neptune Scientific, USA).
3. PTC-100 Programmable Thermal Controller (PCR; MJ
Research, USA).
2.5 SYBR Green 1. 96-Well RT2 Profiler PCR Array System (Qiagen, Australia)
qPCR Arrays using the Human Angiogenic Growth Factors Array (Cat No.
PAHS-072; Qiagen, Australia). Contains primers for 84 genes
involved in angiogenesis plus housekeeping genes and control
wells within a 96-well plate. Also included are the MicroAmp
Optical adhesive films.
2. RT2 SYBR Green qPCR Mastermixcontains real-time PCR
buffer, HotStart DNA polymerase, nucleotides, and ROX
(Qiagen, Australia).
3. 50mL sterile reagent reservoir (Corning Corporation, USA).
4. 8-channel precision pipette50L maximum (Hamilton,
USA).
5. Applied Biosystems 7500 Real-Time Fast PCR instrument
(Applied Biosystems, USA).
2.6 Data Analysis 1. qBASE software (Biogazelle, Belgium).

2. Microsoft Excel-based PCR Gene Data Analysis Template
(Qiagen, Australia).
3. GraphPad PRISM software (Version 6.0, GraphPad, USA).
3 Methods
3.1 Primary Human Osteoblast cells are recovered using methods similar to those
Alveolar Osteoblast described by Dillon etal. [3]. All procedures are conducted in a
Isolation andCulture Class II biosafety cabinet with sterile solutions and plastic labora-
tory consumables.
1. Collect bone tissue of 55mm during third molar extraction
surgery into a sterile universal containing 30mL of sterilized
phosphate buffered saline (PBS) without Ca2+ or Mg2+ and
immediately transport to the laboratory for processing.
2. The excised bone is rinsed and transferred to a petri dish
containing sufficient PBS to cover the tissue. Tissue is then
thoroughly scrapped with a No 10 scalpel blade to remove as
much soft tissue as possible.
3. The bone tissue is transferred to a fresh petri dish containing
PBS and divided into 35mm pieces using a bone rongeur.
4. The PBS is gently decanted and discarded while the bone
fragments are transferred to a 50mL universal containing
20mL of fresh PBS.
5. The tube is then vortexed vigorously for three pulses of 10s
and the bone fragments allowed to settle for 30s. The super-
natant containing the hematopoietic cells and soft tissue is
decanted and discarded.
6. PBS (20mL) is again added to the bone fragments and the

vortexing process (as above) repeated three to four times until
the bone fragments are white in appearance with no soft tissue
evident on close inspection.
7. The bone fragments are then seeded as explants into 6-well cul-
ture plates (two to three wells with 1015 fragments/well) with
7mL of pre-warmed osteoblast growth media added to each
well. The basal osteoblast growth medium contains DMEM,
10% FBS, 100 units/mL penicillin, 100g/mL streptomycin,
250ng/mL Fungizone, and 50g/mL gentamicin (make in
advance and store at 4C). To complete the osteoblast growth
medium supplements are added just prior to use, to give a final
concentration of 100M 2-phospho-l-ascorbic acid and 10nM
dexamethasone (see Notes 1 and 2).
8. The bone explant cultures are placed in a cell culture incubator
at 37C in 5% CO2 /95% air and left undisturbed for 7 days.
The medium is then replaced twice weekly taking care not to
dislodge the explants (see Note 3).
9. Cells are subplated when approaching 80% confluence using
0.25% Trypsin/EDTA and the bone fragments removed at
this point.
10. Trypsinization is conducted by briefly washing the cells in PBS
pre-warmed to 37C.After the removal of the PBS; 600L of
0.25% Trypsin/EDTA is added to each well of the 6-well plate
(1800 L to a T-75). The plate is incubated at 37C for
510min (see Note 4). Cells are observed under an inverted
microscope and when rounded the plate is given a firm tap to
dislodge the cells. The trypsin is inactivated by the addition of
osteoblast growth media containing 10% FBS from step 7.
11. Cells are split 1:3 until 3 T-75 flasks at 80% confluence are
obtained at which point the cells are cryopreserved ready for
experimentation.
12. Freezing media contains 90% FBS and 10% DMSO.For each
cryovial a T-25 equivalent of cells is resuspended in 500L of
freezing medium and placed in a 2mL cryovial.
13. The cells are placed in a cell freezing device (Mr Frosty) over-
night at -80C to allow slow cooling and then are transferred
to liquid nitrogen for long-term storage.
3.2 Experimental 1. Cells are recovered from cryopreservation and placed in a

Treatment water bath at 37C until nearly thawed.
ofOsteoblasts 2. Osteoblast growth medium (37C; 500L) is slowly added to
the cryovial.
3. The cell suspension is added to a 50mL universal and 10mL
pre-warmed (37C) osteoblast culture medium added and
then cells centrifuged at 250g for 5min at RT to obtain

a pellet.
4. The supernatant is decanted and the cell pellet resuspended in
10mL pre-warmed (37C) culture medium prior to transferring
to a T-25 cell culture flask. The cells are then placed in a cell
culture incubator at 37C in 5% CO2 /95% air.
5. Cells are passaged until 80% confluent in a T-75 flask.
6. Counting is conducted with a haemocytometer and cells plated
at a density of 8000cells/cm2 and then incubated at 37C for
6h in 6-well cell culture plates. One plate is used for each
treatment/control group for each time point and incubated
for 6h to allow cells to adhere (see Note 5).
7. The cells are synchronized in osteogenic growth medium sup-
plemented with 0.5% FBS for 12h.
8. Media is then replaced with fresh osteogenic growth medium
containing 10% FBS.Equal volumes of PBS (as the carrier con-
trol), or the treatments (30M ZA) or (30M ZA+50 M
GGOH) are added for each time point investigated. Plates are
placed in cell culture incubator at 37C in 5% CO2 /95% air.
3.3 Total RNA 1. RNA is extracted using TRIzol with a sequential harvesting
Recovery inTRIzol method (see Fig.1) to give a final volume of 500L.The cell
culture medium is removed from one well and TRIzol is
added and incubated for 5min while being pipetted repeatedly
Fig. 1 Schematic representation of TRIzol extraction technique

(30, drawing the fluid in and out of the pipette tip) to assist
complete dissociation of nucleoprotein complexes. The
medium from the second well is then removed and the TRIzol
from the first well transferred to the second well. The process
is repeated until all six wells of each sample are processed as
shown in Fig.1. The plate is observed after each step using a
phase contrast microscope to ensure all the cells have been col-
lected into the TRIzol.
2. Samples are frozen at -80C until required.
3.4 Total RNA A PureLink RNA Mini Kit is used to purify the RNA following
Extraction andDNase I the manufacturers instructions and using RNase-free
Treatment consumables.
3.4.1 Total RNA 1. The 1.5mL microfuge tubes containing the samples in 500L
Extraction of TRIzol are thawed on ice and the following procedures
conducted in a fume hood.
2. Chloroform (100L) is added to each sample and the tube
shaken vigorously by hand for 15s and then incubated for
3min at RT.
3. RNA purification is conducted by centrifugation at 12,000g
for 15min at 4C.Following centrifugation the aqueous
upper phase is carefully removed to a clean microfuge tube tak-
ing care not to disrupt the phenol-chloroform interphase.
4. An equal volume of 70% ethanol (made in nuclease-free water)
is added and vortexed.
5. The sample is transferred to a Spin Cartridge, which contains a
clear silica-based membrane to which the RNA binds.
3.4.2 DNase I Treatment 1. The removal of DNA is essential for accurate qRT2-PCR. On-
column DNase I (30 units/column) treatment is conducted
for 15min at RT with wash steps before and after the
treatment.
3.4.3 Elution andQuality 1. RNA is eluted in 50L of nuclease-free water, which is care-
Assessment fully pipetted onto the membrane and allowed to sit for 1min.
The column is then centrifuged at 12,000g for 2min.
2. RNA quality and quantity is assessed using 2L of sample on
a NanoVue Plus Spectrometer after calibration with RNase-
free water. RNA should have an A260/A280 ratio >1.8 and a
yield >50ng/L.
3.5 Reverse Total RNA purified from the samples is reverse transcribed into
Transcription cDNA using the RT2 First Strand Kit following the manufacturers
instructions. Reverse transcription of RNA is carried out in a final
volume of 25L using 500ng total RNA (we also use 400ng).
The procedure involves a second gDNA elimination step followed

by reverse transcription.
1. RNA is made up to a final concentration of 500ng in 8L
nuclease-free water in a 0.2mL PCR tube. 5 genomic DNA
elimination buffer (2L) is added and the reaction is incu-
bated at 42C for 5min followed by cooling at 4C for 1min
using a PTC-100 PCR machine.
2. A reverse transcription cocktail is made for each reaction con-
taining 4L of 5 RT Buffer, 1L of primer and external
control mix, 2L of RT enzyme mix, and 3L or nuclease-
free water.
3. The RT cocktail (10L) is added to 10L of the RNA and the
contents mixed well in 0.2mL PCR tubes and then briefly
centrifuged.
4. Thermal cycling is conducted at 42C for 15min, 95C for
5min, and held at 4C.
5. Nuclease-free water (92L) is added to each tube before stor-
ing the cDNA at 20C.
3.6 SYBR Green Angiogenic Growth Factors Array plates (Cat No. PAHS-072;
RT2-qPCR Arrays Qiagen, Australia) and RT2 SYBR Green qPCR Mastermix are
used on a 7500 Real-Time Fast PCR instrument.
1. For each plate, an experimental cocktail is prepared in a pipette
reservoir. The cocktail contains 102L of diluted cDNA syn-
thesis reaction, 1,350L of 2 RT2 SYBR green master mix,
and 1248L of nuclease-free water to give a final volume of
2700L.This needs thorough mixing with a pipette prior to
dispensing into the wells.
2. Twenty-five microliters of the experimental cocktail is trans-
ferred to each well of the PCR array plate (sitting on a holder)
using an 8-channel pipette.
3. The plate is sealed with MicroAmp Optical adhesive film and
placed in an Applied Biosystems 7500 Real-Time Fast PCR
instrument for qRT2-PCR thermal cycling and detection. The
melting temperature (Tm) is also collected for each assay. The
PCR cycling parameters are 1 cycle at 95C for 10min, 40
cycles of 95C for 15s with 60C for 1min where fluores-
cence data collection occurs.
3.7 Data Analysis The data analysis is conducted using the Microsoft Excel-based
PCR Gene Data Analysis Template in combination with GraphPad
PRISM software. Genes with mean overall Cq values of 34 or
greater were considered beyond the detection limit of the system
and were not included for further analysis. qBASE software is used
for the selection of housekeeping genes. All selected HKGs have
M-values <1.0.
Fig. 2 Volcano plot illustrating the fold regulation of 70 angiogenic genes with
Cq-values <34in response to 30M ZA treated HOBs (n=3) as compared with
the control cells at the 72-h time point
There are multiple ways to present the qRT2-PCR data when

investigating relative gene regulation. In the graphs/tables below,
we present some of the graphs we have found helpful when pre-
senting our research findings. All graphs were generated using
GraphPad PRISM software.
3.7.1 Volcano Plot A volcano plot (see Fig.2) demonstrates a way of presenting the
ofRelative Gene fold regulations and statistical significance for all genes of interest.
Expression
1. The Y-axis is a log10 scale with the p-values for ZA treatment
versus the control group.
2. The X-axis shows the gene expression as fold regulation
(2Cq) values on a log2 scale.
3. The vertical black dotted lines represent a fold regulation of
2.0.
4. The horizontal dotted line represents a p-value of 0.05.
5. Blue dots represent the significantly up-regulated genes.
6. Green dots represent the significantly down-regulated genes.
7. Black dots are those genes not significantly regulated.
8. Genes that are significantly regulated >2.0-fold are labeled.
3.7.2 Volcano Plot A modified volcano plot (see Fig.3) can also be used to focus on
ofSelected Regulation specific genes of interest. Here, eight genes that were significantly
down-regulated in the presence of 30M ZA as compared to
Fig. 3 qRT2-PCR expression of the eight genes significantly down-regulated genes

with 30M ZA treatment (Genes 1-8a) and their response to 30M ZA+50 M
GGOH (Genes 1-8b). RNA was isolated from the HOBs (n=3) at 72h. Results show
the significantly dis-regulated Genes 1a-8a (p-value <0.05 and Fold Regulation
>2.0) exerted some reversal in the presence of GGOH (Genes 1b-8b)
c ontrols are plotted. The effect of the addition of 30M ZA+50M

GGOH on the expression of these genes is also shown.
1. The Y-axis is a log10 scale with the p-values.
2. The X-axis shows the gene expression as fold regulation
(2Cq) values on a log2 scale.
3. The vertical black dotted lines represent a fold regulation of
2.0.
4. The horizontal dotted line represents a p-value of 0.05.
5. Only those genes significantly regulated >2.0-fold in the
presence of 30M ZA as compared to controls are presented
as Genes 1-8a.
6. The correction when 30M ZA+50 M GGOH is compared
to 30M ZA is give by Genes 1-8b.
7. Individual genes are colored and linked for clarity.
3.7.3 Graph A scatter plot (see Fig.4) demonstrates a way of presenting the
Demonstrating Expression gene expression levels of one gene of interest after it has been nor-
Levels ofanIndividual malized with a selected HKG.
Gene (2Cq)
1. The Y-axis is the relative gene expression and plots the 2Cq
values.
2. The X-axis is the different groups.
Fig. 4 Relative qRT2-PCR expression of the CCL2 gene with 30M ZA alone and
in combination with 50M GGOH or control conditions. HOBs (n=3) at 72h of
treatment/control conditions. Results expressed as meanSD, D1=ZA/Control;
D2=(ZA+GGOH)/ZA; *p-value0.05; **p-value 0.005
3. The standard deviation (SD) is presented for the data with the
lines drawn in later to present the significantly different groups.
4. Neither fold change nor fold regulation is presented; however,
this is a clear way of presenting the relative expression levels.
3.7.4 Graph This graph (see Fig.5) presents the fold regulation of an individual
Demonstrating theFold gene under different treatments as compared to control levels.
Regulation ofanIndividual Asterisks denote the statistical significance.
Gene (2Cq)
1. The Y-axis is fold regulation with the 2Cq values.
2. The X-axis is the different groups.
3. The SD is presented for the data with asterisks to denote statis-
tical significance.
4. The horizontal solid line is no change in the fold regulation
and the 2 thresholds are given as dotted lines.
4 Notes
1. Stock solutions (1000) are made in sterile deionized H2O

and then filtered through a 0.22m filter, aliquoted and fro-
zen at 20C.
2. Only in phenotyping experiments where matrix deposition is
desired is 5mM -Glycerophosphate added to the osteoblast
medium.
Fig. 5 Relative qRT2-PCR expression, of the CCL2 gene, by HOB cells (n=3) after
treatment with 30M ZA alone and in combination with 50M GGOH as com-
pared to control. The horizontal solid line is no change in fold regulation and the
2-fold regulation thresholds are given as dotted lines. Results expressed as
meanSD; *p-value0.05; **p-value0.005
3. After 1020 days of culture, cells are evident migrating out

from the explants and after 34 weeks the cultures reached
80% confluence ready for subplating.
4. The longer time, and incubation at 37C, is necessary to
detach osteoblasts.
5. Cells are seeded at 8,000 cells/cm2 in 66-well cell culture
plates with a surface area of 9.6cm2/well. The 6-well plates
make recovery with TRIzol easier than in the T-25/T-75 cell
culture flasks. In an experiment to assess the number of cells
required (see Table1), RNA was extracted with no treatment
or after treatment with 30M ZA for 96h. RNA was eluted
from the PureLink RNA Mini Kit columns in 50L of
nuclease-free water. Results indicated that seeding in 6-wells
with a total of 4.6105 cells was ideal in this experiment to
gain high-quality RNA (A260/A280 ratio >1.8) with good yield
(>50ng/L). It is, however, important to assess the effects of
your treatment(s) on RNA recovery, prior to initiation of the
experiments.
Table 1
Determination of the number of cells required to yield adequate amounts of RNA for qRT2-PCR
experiments
Concentration Total amount of

Surface area Sample A260/A280 (ng/L) RNA extracted (ng)
9.6cm2 (1 well) Control 1.8 36.8 1840
30M ZA 1.7 46.0 2300
28.8cm2 (3 wells) Control 1.9 40.0 2000
30M ZA 2.0 84.8 4240
57.6cm2 (6 wells) Control 1.9 218.8 10940
30M ZA 2.0 177.6 8880
References
1. Zafar S, Coates DE, Cullinan MP, Drummond Nakamura T (2003) Vascular endothelial
BK, Milne T, Seymour GJ (2014) Zoledronic acid growth factor is expressed along with its recep-
and geranylgeraniol regulate cellular behaviour and tors during the healing process of bone and
angiogenic gene expression in human gingival bone marrow after drill-hole injury in rats.
fibroblasts. JOral Pathol Med 43:711721 Bone 32:491501
2. Ruggiero SL (2011) Bisphosphonate-related 6. Ramasamy SK, Kusumbe AP, Wang L, Adams
osteonecrosis of the jaw: an overview. Ann NY RH (2014) Endothelial Notch activity promotes
Acad Sci 1218:3846 angiogenesis and osteogenesis in bone. Nature
3. Dillon JP, Waring-Green VJ, Taylor AM, Wilson 507:376380
PJM, Birch M, Gartland A, Gallagher JA (2012) 7. Kusumbe AP, Ramasamy SK, Adams RH (2014)
Primary human osteoblast cultures. Methods Coupling of angiogenesis and osteogenesis by a
Mol Biol 816:318 specific vessel subtype in bone. Nature
4. Chim SM, Tickner J, Chow ST, Kuek V, Guo B, 507:323328
Zhang G, Rosen V, Erber W, Xu J(2013) 8. Czekanska EM, Stoddart MJ, Ralphs JR,
Angiogenic factors in bone local environment. Richards RG, Hayes JS (2014) A phenotypic
Cytokine Growth Factor Rev 24:297310 comparison of osteoblast cell lines versus human
5. Uchida S, Sakai A, Kudo H, Otomo H, primary osteoblasts for biomaterials testing.
Watanuki M, Tanaka M, Nagashima M, JBiomed Mat Res 102:26362643
Chapter 28
Proteomic Analysis of Dental Tissue Microsamples

Jonathan E. Mangum, Jew C. Kon, and Michael J. Hubbard
Abstract
Improved understanding of dental enamel development will benefit not only dentistry but also biomedicine
more generally. Rat and mouse models of enamel development are relatively well characterized and experi-
mentally powerful. However, the diminutive size of murine teeth makes them difficult to study using
standard proteomics approaches. Here, we describe gel-based proteomic methods that enable parallel
quantification, identification, and functional characterization of proteins from developing rat and mouse
teeth. These refined methods are applicable to other scarce samples including human enamel defects.
Key words Microsample proteomics, Dental development, Enamel defects, Rat and mouse models,
Ameloblast, Sample preparation, Gel electrophoresis, Functional proteomics
1 Introduction
Improved understanding of dental enamel development (amelo-

genesis) will not only stimulate advances in dental health but also
benefit biomedical research more generally. Dentally, elucidating
the causes of enamel malformations should help with their preven-
tion in many cases, thereby saving major costs at individual and
societal levels. Biomedically, better appreciation of enamel devel-
opment will benefit allied topics including the cellular mechanisms
of handling calcium in bulk, of avoiding calcium cytotoxicity, and
of biomineralization. In pursuit of these widespread benefits, we
have established proteomic approaches to query molecular and cel-
lular aspects of amelogenesis in animal models.
We have used gel-based proteomic strategies to investigate
amelogenesis in developing teeth from rats and mice. Murine teeth
provide a well-characterized and powerful model of dental devel-
opment, particularly given their accessibility to genetic and phar-
macological manipulations. Gel-based proteomics enables
hundreds of proteins to be quantified, identified, and functionally
characterized in parallel. However, the standard approaches were
not well suited to analyzing small amounts of tissue, as found in
461
462 Jonathan E. Mangum et al.
developing molars from rats and mice. This sample limitation

prompted us to tailor procedures for proteomic analysis of dental
tissue microsamples.
In this chapter, we describe refined methods for microsample
proteomics and their application to murine enamel epithelium.
Specifically, the preparation of proteins from enamel epithelia
using a serial extraction approach, customized mini-gel 2-dimen-
sional electrophoresis (2DGE), and various downstream modes
of proteome analysis is outlined. These approaches have helped
us elucidate mechanisms of transcellular calcium transport [14],
functions of cytosolic calcium-binding proteins [510], and char-
acteristics of a new type of molecular chaperone that we discov-
ered in rat enamel epithelium [1115]. Many of the proteomics
data generated with these methods are available on ToothPrint, a
freely available online database (http://tooth-print.mdhs.
unimelb.edu.au) [16, 17]. In addition to being useful for murine
enamel epithelium, these methods have also proven adaptable to
other scarce samples including human enamel defects [18, 19],
suggesting a broader utility.
2 Materials
2.1 Microdissection 1. Dissection buffer (see Note 1): 10 mM HEPES pH 7.4,

of Murine Enamel 129 mM NaCl, 5 mM NaHCO3, 4.7 mM KCl, 1.2 mM
Epithelium KH2PO4, 1 mM CaCl2, 1.2 mM MgSO4, 2.8 mM glucose,
and Enamel Matrix stored 20 C in 50 mL aliquots.
2.2 Sequential 1. TBS extraction buffer: 10 mM TrisHCl pH 7.2, 120 mM

Protein Extraction NaCl, 10 mM ethylene glycol-bis(b-aminoethylether)n,n,n,n-
tetraacetic acid (EGTA), 5 mM dithiothreitol (DTT), plus the
following protease inhibitors added just before use (see Note 2):
1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride
(PMSF), 5 g/mL leupeptin, 5 g/mL pepstatin.
2. SDS denaturant: 2 % sodium dodecyl sulfate (SDS), 10 mM
Tris pH 7.2, 2 mM DTT, 10 mM EGTA, plus protease inhibi-
tors as for TBS above.
3. Benzonase: working stock (10 units/L) made in storage
buffer (50 % glycerol, 20 mM TrisHCl pH 8, 2 mM MgCl2,
20 mM NaCl).
4. SDS/DTT: 10 % SDS, 10 mM DTT.
5. Isoelectric focussing solubilization buffer (IEF-SoB): 9 M
urea, 4 % 3-[(3-cholamidopropyl)dimethylammonio]-1-
propanesulfonate (CHAPS), 50 mM DTT, 5 % carrier
ampholytes pH 3.510.
Dental Proteomics 463
2.3 Gel Preparation 1. Glass tubes: inner diameter 1.5 mm, outer diameter 3 mm,
length 7.5 cm, from Sigma. Glass tubes are washed by over-
2.3.1 First-Dimension
night soaking in 20 % HCl, sonicated for 10 min in a water
Carrier Ampholyte Gels
bath, then rinsed with water until pH neutralizes. Tubes are air
dried and stored in a dust-free environment.
2. Acrylamide solution (see Note 3 for safety information):
premade 40 % acrylamide solution with 2.6 % cross-linker.
3. NP40/CHAPS: Nonidet P40 (NP40), and CHAPS are
combined as a 10 %/0.49 M solution ready for 1:18 dilution.
Care should be taken when dispensing NP40, which clings to
the surface of pipette tips due to its viscosity.
4. Carrier ampholytes: pH 310 (from GE healthcare), 36.5
(from BDH), pH 35 & 46 (from BioRad).
5. APS: ammonium persulfate made as a fresh 10 % solution in
water (just before use).
6. Tube-gel solution: 9.25 M urea, 5 % acrylamide, 1.1 mM
EGTA, 0.56 % NP40, 27 mM CHAPS, 2.2 % carrier ampho-
lytes (Table 1), 0.22 % APS. To dissolve urea this solution
should be vortexed vigorously and sonicated in <10-s bursts
(to avoid heating the solution, see Note 4). Just before pour-
ing the gel, N,N,N',N'-tetramethyl-ethane-1,2-diamine
(TEMED) is added to 0.22 %, the solution is mixed and used
immediately.
2.3.2 Second-Dimension 1. Customized gel-casting plates (see Note 5 and Fig. 1).
SDS-PAGE 2. Resolving gel mixture: 375 mM TrisHCl pH 8.8, 0.1 % SDS,
0.1 % APS, 0.1 % TEMED, acrylamide. Note acrylamide con-
centration will depend on mass range of interest. We routinely
use 12.5 % acrylamide for resolving 15250 kDa proteins.
3. Gel combs: 1.5 mm preparative gel combs from BioRad
(creates one narrow lane for mass ladder, and one wide lane for
tube gel).
Table 1
Carrier ampholyte mixtures used for making IEF tube gels with various
resolving ranges
Ampholyte premixes (pH range) Ratio Useful resolving range (pH)

36.5 and 310 9:1 3.54.5
36.5 and 310 1:1 46
35 and 46 and 310 1:1:2 47
36.5 and 310 1:3 5.57
a b
76 mm
45 15 mm
bevel
c d
1.5 mm
tube gel
0.75 mm
slab gel
e f
Parafilm
Fig. 1 Customized thick/thin gel apparatus. Customization for the Hoefer

Mighty Small II system (distributed by GE Healthcare) is illustrated, but the prin-
ciples should apply to any mini-gel setup. (a) Assembly of customized glass
plate, using a standard Hoefer glass sheet, and a modified 0.75-mm-thick glass
sheet with notch cut as specified in (b). The two pieces are bonded with epoxy
cement to produce a single unit as shown in (b). (c) The modified glass plate is
assembled with a Hoefer alumina backing plate, 0.75 mm T-spacers, and a
1.5-mm-thick preparative gel comb. (d) Cross-section showing relationship
between the thick 1.5 mm tube gel and the thin 0.75 mm slab gel. The role
of the customized glass plate in channelling proteins from the thick tube gel into
the thinner slab gel is highlighted in the magnified view. (e) Plastic transfer
pipette used for casting tube gels, showing positions to be cut. (f) Assembly of
gel casting apparatus with glass tubes protruding ~0.7 cm above the top, and
Parafilm sealing the base
2.4 Gel 1. Catholyte (upper) running solution: 20 mM sodium hydrox-

Electrophoresis ide, diluted from 0.2 M stock that is stored at 4 C and replaced
every 4 weeks. Just before IEF, the catholyte running solution
is degassed by bath sonication for 5 min at 30 C (see Note 6).
2. Anolyte (lower) running solution: 0.03 % (v/v) phosphoric
acid, diluted from 85 % aqueous solution (AR grade) just
before use.
2.4.2 Second-Dimension 1. Laemmli running buffer: 25 mM Tris, pH 8.3, 192 mM gly-

SDS-PAGE cine, 0.1 % SDS.
2. Transfer buffer: 70 mM TrisHCl, pH 6.8, 3 % SDS, 0.002 %
bromophenol blue.
2.5 In-Gel 1. Destain Solution: 50 % acetonitrile, 50 mM ammonium bicar-

Trypsinolysis bonate, pH 8.
2. Reducing Solution: 10 mM dithiothreitol (DTT), 25 mM
ammonium bicarbonate, pH 8.
3. Alkylating Solution: 55 mM iodoacetamide, 25 mM ammo-
nium bicarbonate, pH 8.
4. Trypsin Stock Solution: dissolve trypsin at 125 ng/L in
1 mM HCl (pH ~4). The low pH keeps the enzyme inactive
and so reduces autolysis, allowing storage at 4 C for up to 1
month.
5. Trypsin Digest Solution: for trypsinolysis, the 10 stock is
diluted to 12.5 ng/L trypsin in 25 mM ammonium bicar-
bonate pH 8 just before use.
3 Methods
Murine molars provide limited amounts of tissue (e.g., two man-

dibular first-molars from rat yield ~6 L enamel epithelium, and
mouse gives half this amount). Such limitations pose a challenge
for standard immobilized pH gradient (IPG) 2DGE. To maximize
experimental yields and reduce animal usage, we developed a
higher sensitivity approach based on mini-format, carrier
ampholyte-based 2DGE in combination with sequential protein
extraction. Carrier ampholyte isoelectric focussing (CA-IEF) pro-
vides superior protein loading capacity and recovery over IPG-
based IEF [17]. Customized mini-gels are used to interface thick
CA-IEF tube gels with thin slab gels for the second dimension,
increasing sensitivity by concentrating individual proteins into
smaller gel volumes. Sequential protein extraction reduces sample
complexity by producing, in our case, three fractions that each
contain distinct protein populations associated with cellular compo-
nents (i.e., cytosol, membranes, or nucleus/cytoskeleton; Fig. 2).
a
Coomassie
kDa: Myosin
Nuclear Mitotic
220 Apparatus Protein
Desmoplakin
Collagen VI
100
70
Unc-5
60
50
Actin
40
Histone
30
Ribosomal proteins
S27, L10, L13, L15
Ribosomal protein S5
20
TBS TX-100 SDS
b
Coomassie
kDa:
100
70
60
50
40
30
20
4.5 5.5 6.0 6.5
p/
Fig. 2 Gel-based identification of enamel cell proteins. (a) SDS-PAGE analysis of fractions obtained from rat
enamel epithelium following sequential extractions with Tris-buffered saline (TBS), Triton X-100 and SDS as
indicated. Distinct banding patterns are evident after Coomassie staining, consistent with sampling of different
cellular compartments as intended (nominally cytosol, organelles, and cytoskeleton/nucleus, respectively).
Four bands from the SDS-soluble lane subjected to gel-LC/MS2 analysis and the proteins identified therein are
boxed. Actin and histone bands, identified previously, are indicated. (b) 2DGE analysis of the TBS-soluble frac-
tion from (a), illustrating the effective isolation of numerous proteins. Only the acidic region of this Coomassie-
stained gel is shown. Protein identifications made to establish the limit of sensitivity are indicated, and other
abundant proteins have been reported elsewhere [8, 17]. This figure was taken from ref. [20], and reproduced
with permission from Blackwell Publishing
This fractionation boosts detection sensitivity, provides links to

subcellular location, and effectively compensates for the limited
spatial resolution of mini-gels. Using these methods we have
resolved ~500 Coomassie-stained gel spots using enamel epithe-
lium from a single rat [20]. In combination with blot-based analy-
ses, 2DGE can be married with a variety of downstream procedures
that enable the quantitation, identification, and functional charac-
terization of many proteins in parallel.
3.1 Microdissection 1. Mandibular first molars are isolated during the secretion or
of Enamel Epithelium maturation phase of amelogenesis (i.e., from 45 to 910 day-
and Enamel Matrix old pups, respectively) [6, 21]. Enamel epithelium is subse-
quently microdissected under ice-cold dissection buffer as
rapidly as practicable (generally completed within 57 min
after killing the pup) to minimize postmortem modification of
proteins (e.g., proteolysis). Epithelia from each animal are
transferred to a fresh 1.5 mL centrifuge tube (see Note 7),
excess liquid is removed using a gel-loading pipette tip, and
then the tissue is frozen over dry ice and stored at 80 C.
2. Enamel proteins are isolated from the extracellular matrices
that remain after epithelial isolation (step 1) [1]. The exposed
enamel matrix surface is gently scraped with a micro-knife to
release the soft layer of partially mineralized enamel proteins
from underlying dentine (which is firmer and distinctly col-
ored). This white particulate material is then pooled and sedi-
mented by centrifugation (1000 g for 2 min). The enamel
matrix pellet is dissolved in 10 volumes of 4 % trifluoroacetic
acid, assisted by bath sonication. Next, SDS and EGTA are
added (1 % and 100 mM, respectively) along with a trace of
bromophenol blue, then the mixture is dried in a vacuum cen-
trifuge. The dried pellet is dissolved in the original volume of
water and the pH neutralized with ammonia vapor as required
(hold a small droplet at the end of a pipette tip close to the
sample surface until the solution turns from yellow to blue).
Next, TrisHCl pH 7.2 and dithiothreitol are added to final
concentrations of 25 and 10 mM, respectively. EGTA is added
incrementally (1 L additions from a 1 M, pH 7 stock) until
any remaining calcium dodecyl sulfate (visible as a white pre-
cipitate) is dissolved. Finally, the sample is boiled for 5 min,
centrifuged (17,000 g for 2 min) and stored at 80 C.
3.2 Sequential 1. Two frozen enamel epithelia are thawed in 30 L of ice-cold

Protein Extraction TBS extraction buffer, then vortexed and bath sonicated for
5 min to homogenize the tissue. Homogenates are then frozen
over dry ice for 5 min. After thawing on ice, this vortex/sonicate/
freeze cycle is repeated three times (see Note 8). After the
fourth cycle is completed, the homogenate is centrifuged at
22,000 g for 5 min (4 C). The supernatant (nominal cytosolic

fraction, see Note 9) is collected. The pellet is washed with
30 L of TBS extraction buffer, centrifuged again, and the
ensuing supernatant is pooled with the cytosolic fraction.
2. The pellet is then taken up in 30 L of TBS extraction buffer
containing 2 % Triton X-100, vortexed and sonicated for
1 min, then incubated for 15 min at 4 C with intermittent
vortexing/sonication. The extract is centrifuged as above
and the supernatant (nominal organellar fraction) is collected.
The pellet is washed with 30 L TBS extraction buffer contain-
ing 2 % Triton X-100, centrifuged again, and the supernatant
pooled as above.
3. The pellet is then taken up in 30 L of SDS denaturant,
vortexed and sonicated for 1 min, then boiled for 2 min. The
extract is centrifuged 22,000 g for 5 min at 20 C, and then
the supernatant (nominal cytoskeletal/nuclear fraction) is col-
lected and stored on ice. The pellet is washed with 30 L of
SDS denaturant, centrifuged again, and the supernatant is
pooled as above.
4. Before the extracts are used for 2DGE or SDS-PAGE (see
Note 10), they should be treated to degrade any contaminat-
ing nucleic acid, and subsequently ethanol precipitated to
remove salts/small molecules and concentrate the sample.
5. To degrade nucleic acids, 1 L of Benzonase is added (10 units,
see Note 11) per 60-L extract and incubated on ice for 30 min
(see Note 12). To assist protein solubilization after ethanol
precipitation, 3 L of SDS/DTT is added and the extracts are
incubated at 100 C for 2 min. After cooling on ice for 1 min,
9-volumes of cold (20 C) ethanol is added. Samples are
incubated at 20 C for at least 30 min to maximize protein
precipitation, and then centrifuged at 14,000 g for 5 min
(4 C). The supernatant is discarded and residual ethanol com-
pletely removed under a gentle stream of dry nitrogen or with
a vacuum centrifuge. Next, 3 L of SDS/DTT is added directly
onto the dried pellet and the tube is vortexed for 1 min before
adding 20 L of IEF-SoB (see Notes 13 and 14). The pellet is
completely dissolved by repeated pipetting (avoid creating air
bubbles), and then the samples are centrifuged (20,000 g for
5 min at 25 C) to sediment particulates that might occlude
the loading surface of the tube gel. Samples can be stored at
80 C before 2DGE, but if so should be thoroughly mixed
and centrifuged again before loading.
3.3 Gel Preparation 1. A disposable 3-mL plastic transfer pipette is adapted to produce
a tube-gel casting chamber. Four glass tubes are placed inside
the chamber and Parafilm is used to seal the base (see Fig. 1 and
Note 15). Tube gels should be positioned as close to vertical
as possible (e.g., place casting chamber inside a 15-mL Falcon

tube that is standing upright in a storage rack).
2. Note: rapidly complete this step (<2 min) to ensure even
polymerization of acrylamide. The tube-gel solution is pipet-
ted into the casting chamber between the gaps of adjacent
tubes, making ~0.2-mL additions and moving around the
perimeter of the chamber with each addition. Water containing
0.002 % Bromophenol Blue (overlay solution) is added in the
same manner to displace the tube-gel solution up into the glass
tubes. Addition of the blue overlay solution is continued until
the colorless tube-gel solution is ~ 7 mm from the top of the
glass tubes (see Note 16). To ensure a flat surface for sample
loading and prevent dehydration of gels during polymeriza-
tion, 5 L of overlay solution is gently added on the top of the
acrylamide in each glass tube using a gel-loading tip. If sample
volumes greater than 20 L are to be loaded, the amount of
overlay solution can be increased as appropriate.
3. After initial polymerization has proceeded for 15 min at room
temperature, the top of the casting tube is gently covered with
stretched Parafilm to prevent dust contamination and reduce
evaporation of overlay buffer. Complete polymerization of
acrylamide minimally requires incubation for a further 45 min
at room temperature.
4. Once the tube gels have completely polymerized, the Parafilm
is gently peeled from both ends of the casting chamber and the
glass tubes are ejected. The lower polyacrylamide plug is
carefully excised from the tubes with a clean razor blade. It is
important at this step that the base of the tube gel is not
stretched or otherwise disturbed, because resulting disconti-
nuities between the gel and glass tube interior (i.e., bubbles)
are a common cause of current leak during IEF. Residual poly-
acrylamide is removed from the glass tube exterior with a lint-
free tissue. Liquid is removed from the top of the gel using a
gel-loading tip while taking care to avoid disturbing the load-
ing surface (see Notes 17 and 18). Once all tubes are clean and
checked for integrity (reject any tubes with bubbles) their ends
are sealed with Parafilm, again taking particular care with the
base of the tubes. Finally, the tubes are wrapped in cling-film,
labeled (date, ampholyte mix) and then stored at 4 C for up
to 2 weeks.
3.3.2 Second-Dimension 1. Gel plates are assembled inside the caster with 0.75-mm spac-
SDS-PAGE ers according to the manufacturers instructions.
2. Resolving gel mixture is poured into the caster until the gel
cassettes are full, and then 1.5-mm thick preparative combs are
inserted (see Note 19).
3. Gels are incubated at room temperature for at least 1 h before

the caster is disassembled. The gels are cleaned then wrapped
in clingfilm together with a moistened paper towel (to maintain
gel hydration) and stored at 4 C for up to 2 weeks.
3.4 Gel 1. Tubes are assembled into a focussing rack according to the
Electrophoresis manufacturer's instructions (care should be taken to align the
tube tops as closely as possible). Up to 12 tube gels can be
focussed concurrently without having to change the standard
power settings (see below). Samples in IEF-SoB are loaded
into each tube using a gel-loading pipette tip and, if the tube is
not full after loading, it is topped up with IEF-SoB. Once all
samples have been loaded, the rack is mounted onto the gel-
running rig. The catholyte running solution is poured slowly
into the upper chamber until all tubes are just submerged (by
12 mm). The anolyte running solution is poured into the
lower chamber, fully submerging the lower ends of the tubes.
2. Isoelectric focussing is initially performed with power settings of
200 V, 5 mA, 1.5 W, and without cooling. Once the current has
reduced to ~0.8 mA (typically this takes 3045 min), the power
settings are changed to 1000 V, 2.5 mA, 1.5 W, and water-cool-
ing of the gel rig is started. Normally, it takes 3060 min to reach
1000 V, depending on the number of tubes being focussed
(more tubes take longer). Progress of this IEF step is tracked by
recording the time, voltage, current, watts, and cumulative volt-
hours (Vh) every 1530 min. If the voltage fails to reach 1000 V
after 2 h of focussing there is likely to be a current leak, which
requires intervention as described in Note 20. For tissue extracts,
IEF is typically optimal once a total of 32003600 Vh has been
reached (note that somewhat shorter focussing times may be
beneficial for less-complex samples). The focussing rack is
removed and immediately placed on ice to reduce defocusing
(diffusion) of proteins while awaiting extrusion.
3. To eject each gel, the glass tube is connected to a water-filled
syringe using an extrusion adapter (available from Sigma-
Aldrich). The basic (lower) end of the tube is placed on a piece
of Parafilm and then the syringe plunger is pressed gently until
the gel begins to emerge. Once ~1 cm of gel has emerged, a
yellow pipette tip is placed on it to prevent flipping should the
subsequent extrusion be poorly controlled (e.g., due to over-
pressurization). Once the gel is fully extruded, excess water is
removed with a lint-free tissue and the Parafilm is labeled to
indicate gel orientation and sample identity. The Parafilm-
supported gel is then placed inside a 15-mL screw-cap Falcon
tube and snap-frozen on dry ice. Tube gels are stored at 80 C
until ready for second-dimensional separation. Used glass
tubes are stored submerged in water before cleaning as
described (Subheading 2.3.1).
3.4.2 Second-Dimension 1. Resolving gels are assembled into the gel-running rig and
SDS-PAGE Laemmli running buffer is added to the upper chamber until
it almost reaches the level of the loading well. The comb is
carefully removed and any liquid remaining in the loading
well is removed with a gel-loader tip (this helps prevent the
tube gel from floating when running buffer is added in step 3,
below).
2. Tube gels (on Parafilm) are thawed by adding 200 L of
transfer buffer and incubated for 2 min at room temperature
(see Notes 21 and 22). When completely thawed the gel
appears optically clear, with no sign of bubbles.
3. Transfer buffer is drained and the tube gel is loaded by slowly
lowering one end into the loading well with the help of a thin
metal spatula. The rest of the tube-gel is gradually placed into
the well, working from one end to the other, taking care to
avoid trapping air bubbles against the resolving gel (see Note 23).
Any trapped bubbles should be removed carefully by tapping
gently on the tube gel with a spatula. Running buffer is then
carefully topped up to immerse the tube gel and any further
bubbles are removed as before. Molecular weight markers are
also loaded if desired (see Note 24). Second-dimension sepa-
ration is performed at 200 V and 1215 mA per gel, and
continued until the dye front reaches the bottom of the
resolving gel.
3.5 Protein Analysis To quantify proteins we routinely use densitometric analysis after
staining gels with Coomassie Brilliant Blue, which provides a good
3.5.1 Protein
dynamic range (unlike silver staining) and interfaces well with pro-
Quantitation
tein identification technologies including mass spectrometry
(Fig. 2 and [20]). Although higher detection sensitivities can be
achieved with fluorescent stains, these methods might compromise
mass spectrometry [22]. Quantitative immunoblotting is a power-
ful adjunct approach for proteins of particular interest, when suit-
able antibodies are available [23].
3.5.2 Protein
Identification by Mass For identification by mass spectrometry of in-gel tryptic digests,
Spectrometry proteins stained with Coomassie Blue are preferred for reasons of
sensitivity and sequence coverage. With 2DGE protein spots,
MALDI-TOF-based peptide mass fingerprinting is generally suffi-
cient for identification when combined with the electrophoretic
properties (Mr and pI). For SDS-PAGE bands, LC-MS/MS
sequence tags (n 2) can provide identification of several proteins
from a single band, albeit with reduced quantitative information
(Fig. 2 and [20]). In all cases, however, it is imperative to substanti-
ate functionally important identifications with additional measures
(e.g., immunoblotting or functional characterization as described
in the following section) [11].
Given the diversity of mass spectrometry equipment and

methodology currently available, it would be inappropriate to
provide a comprehensive account here (for an excellent overview
of the subject, see [24, 25]). Instead, we exemplify a relatively
straightforward method for identifying proteins in dental tissue
microsamples using in-gel trypsinolysis and LC-MS/MS, which
has worked well for us.
1. Coomassie Blue stained bands of interest are excised from the
gel using a clean razor blade (see Note 25), diced into 1 mm
cubes and transferred to a clean 1.5 mL microcentrifuge tube.
It is highly recommended that a positive and negative control
be conducted in parallel to all analyses. The positive control
can be a marker protein (we routinely use 1 g bovine serum
albumin) while the negative control should be a blank region
of the gel.
2. NB: all incubation steps are done with orbital mixing at a rate
sufficient to keep the gel pieces suspended (~700 rpm on a
vortex mixer). Wash the gel pieces twice with 300 L of water
for 15 min and then repeatedly with 50 L Destain Solution
for 15 min until gel pieces are colorless. Two washes may be
sufficient for moderate intensity bands.
3. Dehydrate gel with 100 L of neat acetonitrile for 20 min (gel
pieces should shrink and become opaque white). Discard ace-
tonitrile and dry in a vacuum centrifuge for 5 min (NB: if a
vacuum centrifuge is not available gel pieces can be dried
30 min at room temperature with tube lids openhowever a
clean container should be placed over the tubes to prevent
dust/particulates from entering the tubes).
4. Rehydrate samples in 50 L Reducing Solution and incubate
at 56 C for 1 h. Decant Reducing Solution then add Alkylating
Solution and incubate at room temperature in the dark for
45 min.
5. Wash gel pieces with 100 L 25 mM ammonium bicarbonate
pH 8 for 10 min, then dehydrate again as done in step 3.
Rehydrate with 25 mM ammonium bicarbonate solution for
15 min and then dehydrate again. Prepare the Trypsin Digest
Solution at this stage.
6. Rehydrate gel in 20 L Trypsin Digest Solution for 20 min.
This allows the trypsin to infuse into the gel plug for in-gel
digestion. Remove excess Trypsin Digest Solution and then
add 1020 L of 25 mM ammonium bicarbonate (pH 8) until
the gel pieces are just covered to ensure proper hydration during
digestion.
7. Digest at 37 C overnight in an incubator (NB: heating blocks
are not recommended as they do not provide uniform heat and
can cause condensation to form on the tube lid).
8. Stop digestion by adding 25 L of 10 % formic acid and incu-

bate for 15 min at room temperature. Recover supernatant,
which can be used directly for LC-MS/MS (see Note 26).
9. Apply tryptic peptides (15 L) to an electrospray ion trap
mass spectrometer that has a reversed-phase C18 chromatog-
raphy front end. Elute peptides over an acetonitrile gradient
(dependent on the apparatus) and analyze the eluted peptides
in positive ion mode with automated MS/MS.
10. Generated mass peak lists are searched against sequence databases
using publicly available software (e.g., the Mascot MS/MS Ions
search algorithm available at www.matrixscience.com). Note that
for high-confidence identifications more than one algorithm
should be used and the raw MS/MS data should be scrutinized for
quality wherever possible to reduce false discovery rates.
3.5.3 Functional After electrophoresis, proteins can be electroblotted onto a mem-

Characterization brane (nitrocellulose or PVDF) and functionally probed with
potential ligands. For example, protein-binding or calcium-binding
functionality may be determined in this way [6, 7, 26].
Immunoblotting can also be used for targeted analyses of protein
heterogeneities, such as isoforms or posttranslational modifications
(e.g., phospho/glyco-forms or amidation states [6, 12]).
3.6 Extension While the preceding methods were optimized for murine enamel
of These Methods tissues, they have also been successfully used on human enamel
to Human Enamel samples. For example, Molar Hypomineralization (MH) is a wide-
Defects spread and often severe developmental defect that presents as
demarcated patches of chalky enamel on permanent first molars
[19, 27, 28]. Seeking pathogenic clues to cause, we investigated
the protein content of MH-affected enamel using the microsample
methods described here. Using samples as small as 2 mg we were
able to determine the protein content of MH enamel and use those
profiles to infer pathological significance by comparing broken and
intact lesions as shown in Fig. 3 [19].
4 Notes
1. All solutions are made in ultrapure deionized water (e.g., MilliQ,

18.2 M cm). All reagents are electrophoresis or analytical
grade and purchased from Sigma or BDH unless stated other-
wise. Buffers are diluted from 1 M stocks, the pH of which is
measured at room temperature. The pH of working solutions
should not be adjusted after mixing buffer stocks with other
ion-containing solutions.
2. A cocktail of protease inhibitors is used in an effort to protect
against multiple classes of proteases. Benzamidine (trypsin
Complement C3 fragment (7)

-1-antitrypsin (7)
Lactotransferrin (11)
Albumin (1,3,4,5,6) Albumin (10)
Albumin fragment (3,5,6) Alpha amylase (8)
Albumin (10)
lg alpha chain 2 (10)
-1-antitrypsin (7)
Leukocyte elastase inhibitor (7)
Antithrombin-III (7)
Albumin (10)
-1-antitrypsin (7)
lg-kappa chain VIII (7,8,11)
Protein S100-A9 (7)
Serpin B3 (10)
Uncharacterized c6orf58 (10)
Albumin (10)
Hemoglobin (7,10,11) & (7,11)

Protein S100 A9 (7,11)
Albumin (7)
Prolactin-inducible protein (11)
Albumin (7)
)
)
(7
(6
en
ct
ta
ok
In
Br
Fig. 3 Proteomic analysis reveals numerous body fluid proteins in iMH enamel.
The indicated major gel bands from intact and broken lesions were subjected to
proteomic identification. The figure depicts the proteins identified in each band,
and the specimens in which these identifications were made (specimen
numbers in parentheses). Gel lanes for specimens 6 and 7 illustrate intact and
broken lesions, respectively. This figure was taken from ref. [19], and reproduced
with permission from Sage Journals
inhibitor), pepstatin (aspartic acid protease inhibitor), and

leupeptin (serine & cysteine protease inhibitor) are all stable in
water. However, PMSF (serine protease inhibitor) undergoes
autohydrolysis in aqueous solution (t1/2 < 2 h at pH 7 [29])
and so must be stored in anhydrous organic solvent (we use a
0.1 M stock in neat ethanol). It is primarily due to the instability
of PMSF that protease inhibitors are added just before homog-
enizing enamel epithelia.
3. Although acrylamide premixed in aqueous solution is safer to
use than the powder form, protective equipment (lab coat,
gloves, safety glasses) still must be employed when handling
this potent neurotoxin.
4. When handling urea in aqueous solution, it is important to
avoid low and high temperatures. Temperatures below 15 C
will cause urea to precipitate, decreasing its protein-solubilizing

capacity. Temperatures above 37 C can result in degradation
of urea to reactive cyanate species that in turn can carbamylate
nucleophilic amino acids (lysine, arginine, cysteine, and initia-
tor methionine). Such carbamylation is to be avoided because
it introduces charge heterogeneity to proteins, which manifests
as acidic spot trains on 2DGE (see [30] and refs therein).
5. Glass plates are modified to allow 1.5-mm diameter tube gels
to be interfaced with 0.75-mm-thick resolving gels. This is
beneficial because thin resolving gels provide higher protein
concentration in each gel spot, thereby improving detection
sensitivity and the reliability of protein identification from
in-gel digests.
6. Thoroughly degassing the catholyte buffer just prior to use is
widely thought to be important for achieving good IEF results.
Failure to do this properly can lead to gas bubble formation
inside the tube (above the gel surface) as the apparatus warms,
resulting in current disruption and under-focussing.
7. All procedures use low-protein-binding microcentrifuge tubes
(e.g., Eppendorf LoBind). Such tubes minimize the amount of
protein that is lost to the tube wall, obviating the need for sur-
face pretreatments such as siliconization. Besides providing
good protein recovery, these tubes have a low content of plasti-
cizer, which is important for downstream mass spectrometry.
8. We have found this freeze/thaw homogenization approach
gives greater protein recovery than traditional micro-pestle
homogenization approaches, presumably due to reduced sam-
ple handling and loss to surfaces.
9. The freeze-thaw procedure produces a soluble fraction (nomi-
nally cytosol) due to rupture of the plasma membrane by vol-
ume expansion during ice formation. Insoluble complexes of
cytosolic proteins (e.g., cytoskeleton) and unbroken organelles
(e.g., ER, mitochondria, nuclei) remain in the pellet predomi-
nantly. Subsequent Triton-solubilization of the pellet produces
an organelle-enriched fraction by releasing membranous and
lumenal proteins, while insoluble cytoskeletal proteins and
DNA-bound proteins remain in the pellet. SDS solubilizes
many of the latter, and subsequent DNAse treatment releases
any remaining DNA-bound proteins.
10. All three extracts can be applied to 2DGE; however, it should
be kept in mind that some proteins in the Triton and SDS frac-
tions may not be amenable to IEF. For example, when several
proteins from the SDS fraction were analyzed by SDS-PAGE
and LCMS (gel-LCMS), we found the majority were too
hydrophobic, too large, or too highly charged to be visualized
by 2DGE [20]. Hence, a combination of 2DGE and gel-
LCMS can be used to maximize coverage of the enamel epi-

thelium proteome.
11. Benzonase is a DNA/RNA nuclease that is particularly useful
due to its resilience against harsh conditions (e.g., 7 M urea,
1 % SDS, pH 610, 040 C) [31].
12. Two epithelia from rat provide sufficient protein for Coomassie-
stained 2DGE, as determined empirically [20]. SDS-PAGE
requires approximately 30 % this amount, reflecting the broader
range of proteins resolved by the gel.
13. SDS/DTT is added to pellets to maximize protein solubility in
IEF-SoB and so to reduce subsequent protein precipitation
during initial stages of IEF. The amount of SDS/DTT was
determined empirically to improve protein solubility and yet
not impair resolution.
14. Carrier ampholyte IEF gels are able to accommodate substantial
amounts of SDS/DTT due to the open nature of the system.
Protein is initially electrophoresed into the tube gel before the
carrier ampholytes migrate to their pI positions. A current is gen-
erated during this initial electrophoresis phase, which is sufficient
to facilitate rapid migration of charged SDS, DTT, and other
small ions out of the tube gel into the running solution.
15. A larger number of tube gels (~25) can be cast by using a
modified 15-mL disposable syringe. The needle outlet is
removed to create an open-ended cylinder, and the plunger is
used to seal the base and later provide a gentle means for eject-
ing the cast tubes.
16. It is important to empirically determine the volume of overlay
buffer needed to cast tube gels of a desired length. A substan-
tial plug of polyacrylamide gel is required at the bottom of the
caster (2 cm from base) to maintain the intended acrylamide
concentration inside the tubes.
17. It is important to remove the overlay solution to prevent dilu-
tion of urea at the gel surface during storage. A high concentra-
tion of urea is needed to maximize sample entry to the gel.
18. After polymerization, acrylamide properties can be assessed
using a gel-loading pipette tip; properly prepared gels feel
quite firm. If the gel is able to be drawn into the pipette tip
and appears viscous, the polyacrylamide concentration at the
loading surface is too lowthe batch of tube gels should be
rejected.
19. We routinely omit stacking gels for simplicity and reproduc-
ibility, as suggested by others [32].
20. Maximum voltage is considered to be more important for spot
resolution than length of focussing time, so ensure that 1000 V
is reached (by minimizing current leaks).
21. Generally, we do not reduce and alkylate tube gels before sec-
ond dimension electrophoresis because samples are heavily
reduced before loading and our transfer procedure is rapid
enough to avoid cysteine oxidation. By avoiding this step,
sensitivity and resolution are improved both through reducing
sample loss from the gel (i.e., washout during transfer) and
minimizing protein diffusion in the tube gel.
22. Time spent equilibrating the tube gel in transfer buffer can be
critical. Too long and proteins can be lost, too short and the
CHAPS/NP40 in the tube gel may not be displaced by SDS
(resulting in loss of small acidic proteins particularly).
23. It is important to completely thaw the gels before loading.
Incompletely thawed gels produce minute gas bubbles, which
can cause vertical streaking of 2DGE maps.
24. Regular mass standards (in solution) will give approximate
mass (Mr) comparisons only. For accurate mass calibration the
Mr markers need to be cast inside a tube gel, which is then cut
to size and placed into the marker lane.
25. MS-based protein identification is sensitive to contamination
from skin, hair, dust, or unclean gel-handling equipment.
Common protein identifications from contamination include
keratins, collagens, and caseins. To minimize these problems
always use clean staining trays, tubes, etc., wear gloves, avoid
leaning over gels, and if possible work in a laminar flow
hood.
26. NB: for scarce or hydrophobic proteins further extraction can
be done on the gel pieces as follows: Add 50 % acetoni-
trile/50 mM ammonium bicarbonate for 15 min to recover
additional peptides; repeat this step and pool supernatants.
Add 50 L of 100 % acetonitrile to gel pieces for 15 min, col-
lect and pool with supernatant from the previous step.
Concentrate to 5 L or dryness in a vacuum centrifuge.
Concentration to dryness is more convenient but may result in
some loss of peptides that will not resolubilize. Dried peptides
can be stored at 20 C or 80 C for long periods of time
(1 year). Immediately prior to mass analysis dissolve dried
peptides in 20 L 1 % formic acid.
Acknowledgments
We thank Nicola McHugh for skilfully assisting with development

of the 2DGE procedures described here. This work was supported
by the Melbourne Research Unit for Facial Disorders, the National
Health and Medical Research Council of Australia, and the Health
Research Council of New Zealand.
References
1. Hubbard MJ (1996) Abundant calcium reticulum protein, from rat enamel cells evi-
homeostasis machinery in rat dental enamel dence for a unique role in secretory-protein
cells. Up-regulation of calcium store proteins synthesis. Eur J Biochem 267:19451957
during enamel mineralization implicates the 14. Hubbard MJ, Mangum JE, McHugh NJ
endoplasmic reticulum in calcium transcytosis. (2004) Purification and biochemical charac-
Eur J Biochem 239:611623 terisation of native ERp29 from rat liver.
2. Hubbard MJ (2000) Calcium transport across Biochem J 383:589598
the dental enamel epithelium. Crit Rev Oral 15. Hermann VM, Cutfield JF, Hubbard MJ
Biol Med 11:437466 (2005) Biophysical characterization of ERp29.
3. Franklin IK, Winz RA, Hubbard MJ (2001) Evidence for a key structural role of cysteine
Endoplasmic reticulum Ca2 + -ATPase pump is 125. J Biol Chem 280:1352913537
up-regulated in calcium-transporting dental 16. Hubbard MJ, Faught MJ, Carlisle BH,
enamel cells: a non-housekeeping role for Stockwell PA (2001) ToothPrint, a proteomic
SERCA2b. Biochem J 358:217224 database for dental tissues. Proteomics
4. Turnbull CI, Looi K, Mangum JE, Meyer M, 1:132135
Sayer RJ, Hubbard MJ (2004) Calbindin inde- 17. Hubbard MJ, Kon JC (2002) Proteomic analy-
pendence of calcium transport in developing sis of dental tissues. J Chromatogr B Analyt
teeth contradicts the calcium ferry dogma. Technol Biomed Life Sci 771:211220
J Biol Chem 279:5585055854
18. Mangum JE, Farlie PG, Hubbard MJ (2005)
5. Hubbard MJ, McHugh NJ (1995) Proteomic profiling of facial development in
Calbindin28kDa and calbindin30kDa (cal- chick embryos. Proteomics 5:25422550
retinin) are substantially localised in the par-
ticulate fraction of rat brain. FEBS Lett 19. Mangum JE, Crombie FA, Kilpatrick N,
374:333337 Manton DJ, Hubbard MJ (2010) Surface
integrity governs the proteome of hypominer-
6. Hubbard MJ (1995) Calbindin28kDa and
alized enamel. J Dent Res 89:11601165
calmodulin are hyperabundant in rat dental
enamel cells. Identification of the protein phos- 20. Mangum JE, Veith PD, Reynolds EC, Hubbard
phatase calcineurin as a principal calmodulin MJ (2006) Towards second-generation pro-
target and of a secretion-related role for calbi- teome analysis of murine enamel-forming cells.
ndin28kDa. Eur J Biochem 230:6879 Eur J Oral Sci 114(Suppl 1):259265
7. Hubbard MJ, McHugh NJ (1996) 21. Kardos TB, Hubbard MJ (1981) Rapid dissec-
Mitochondrial ATP synthase F1-beta-subunit tion of rodent molar-tooth germs. Lab Anim
is a calcium-binding protein. FEBS Lett 15:371373
391:323329 22. Lanne B, Panfilov O (2005) Protein staining
8. Hubbard MJ (1998) Enamel cell biology. influences the quality of mass spectra obtained
Towards a comprehensive biochemical under- by peptide mass fingerprinting after separation
standing. Connect Tissue Res 38:1732 on 2-d gels. A comparison of staining with
9. Hubbard MJ (1998) Proteomic analysis of coomassie brilliant blue and sypro ruby.
enamel cells from developing rat teeth: big J Proteome Res 4:175179
returns from a small tissue. Electrophoresis 23. Shnyder SD, Mangum JE, Hubbard MJ (2008)
19:18911900 Triplex profiling of functionally distinct
10. Sayer RJ, Turnbull CI, Hubbard MJ (2000) chaperones (ERp29/PDI/BiP) reveals marked
Calbindin28kDa is specifically associated with heterogeneity of the endoplasmic reticulum
extranuclear constituents of the dense particu- proteome in cancer. J Proteome Res 7:
late fraction. Cell Tissue Res 302:171180 33643372
11. Demmer J, Zhou C, Hubbard MJ (1997) 24. Simpson RJ (2003) Proteins and proteomics: a
Molecular cloning of ERp29, a novel and laboroatory manual. Cold Spring Harbour
widely expressed resident of the endoplasmic Laboratory Press, Cold Spring Harbour, NY
reticulum. FEBS Lett 402:145150 25. Aebersold R, Mann M (2003) Mass
12. Hubbard MJ, McHugh NJ (2000) Human spectrometry-based proteomics. Nature 422:
ERp29: isolation, primary structural character- 198207
isation and two-dimensional gel mapping. 26. Hubbard MJ, Klee CB (1987) Calmodulin
Electrophoresis 21:37853796 binding by calcineurin. Ligand-induced rena-
13. Hubbard MJ, McHugh NJ, Carne DL (2000) turation of protein immobilized on nitrocel-
Isolation of ERp29, a novel endoplasmic lulose. J Biol Chem 262:1506215070
27. Weerheijm KL (2004) Molar incisor hypo- electrophoresismyth or reality? J Proteome

mineralization (MIH): clinical presentation, Res 2:239242
aetiology and management. Dent Update 31. Biedermann K, Jepsen PK, Riise E, Svendsen I
31:912 (1989) Purification and characterization of a
28. http://www.thed3group.org. Accessed Sept Serratia marcescens nuclease produced by
2015 Escherichia coli. Carlsberg Res Commun
29. James GT (1978) Inactivation of the protease 54:1727
inhibitor phenylmethylsulfonyl fluoride in buf- 32. Hochstrasser DF, Harrington MG, Hochstrasser
fers. Anal Biochem 86:574579 AC, Miller MJ, Merril CR (1988) Methods for
30. McCarthy J, Hopwood F, Oxley D, Laver M, increasing the resolution of two-dimensional
Castagna A, Righetti PG, Williams K, Herbert protein electrophoresis. Anal Biochem 173:
B (2003) Carbamylation of proteins in 2-D 424435
Chapter 29
Characterization, Quantification, and Visualization

of Neutrophil Extracellular Traps
Phillipa C. White, Ilaria J. Chicca, Martin R. Ling, Helen J. Wright,
Paul R. Cooper, Mike R. Milward, and Iain L.C. Chapple
Abstract
Following the discovery of neutrophil extracellular traps (NETs) in 2004 by Brinkmann and colleagues,
there has been extensive research into the role of NETs in a number of inflammatory diseases, including
periodontitis. This chapter describes the current methods for the isolation of peripheral blood neutrophils
for subsequent NET experiments, including approaches to quantify and visualize NET production, the
ability of NETs to entrap and kill bacteria, and the removal of NETs by nuclease-containing plasma.
Key words Neutrophil extracellular traps, Reactive oxygen species, Fluorescence microscopy, SEM,
Chemiluminescence, DNA, Elastase, Myeloperoxidase, Cathepsin G, Immunostaining
1 Introduction
Neutrophils are terminally differentiated effector cells of the

haematopoietic myeloid lineage, which are critical to both the
innate and acquired (humoral) immune response. In order to elim-
inate invading pathogens, neutrophils have a comprehensive cyto-
toxic arsenal at their disposal, which includes reactive oxygen
species (ROS) and granule-derived proteins [1]. In 2004, a new
paradigm in neutrophil biology was described by Brinkmann and
colleagues, referred to as neutrophil extracellular traps (NETs) [2].
NETs are highly conserved extracellular mesh-like structures of
decondensed nuclear chromatin and associated antimicrobial pep-
tides (AMPs), whose putative role is to immobilize and kill invad-
ing pathogens. The expulsion of nuclear NETs is believed to arise
during an active form of programmed cell death, termed NETosis,
that is dependent upon ROS production and the citrullination of
arginine and methylarginine residues of the nuclear chromatin by
peptidyl arginine deiminase 4 (PAD4) [3, 4]. However, there is
emerging evidence to suggest that NETs are also induced by other
481
482 Phillipa C. White et al.
intricate molecular pathways, such as the release of mitochondrial

derived NETs from cells that remain viable following NET expulsion
[5] and a rapid form of ROS-independent NET release [6].
To determine the role of NETs in various disease states,
neutrophils are often isolated from peripheral venous blood and
investigated ex vivo. Neutrophils isolation requires an efficient,
aseptic, and reproducible method to obtain nonactivated and viable
cells. The most frequently used approaches utilize density gradient
centrifugation and other techniques that are widely employed
include Percoll gradients, Histopaque gradients, and Dextran sedi-
mentation. Our own comparison of neutrophil isolation techniques
indicates that neutrophils isolated using discontinuous Percoll
gradients produce nonactivated cells that are more responsive to
exogenous stimulation [7].
To induce NET release, in vitro stimuli such as phorbol
12-myristate 13-acetate (PMA) [8] and hypochlorous acid (HOCl)
[9] are routinely used. Gram-positive and gram-negative bacteria
[10, 11], protozoan parasites [12], fungi (cellular and hyphal
forms) [13], and viruses [14] have all been reported to elicit NET
release. Interestingly, host-derived inflammatory mediators, such
as pro-inflammatory cytokines [15], can also stimulate NET
generation.
The quantification of NET release frequently utilizes fluoro-
metric detection of DNA [16]. To ensure NET-DNA is being
measured, and not DNA produced by other cellular processes,
such as necrosis, the AMPs associated with the NET-DNA back-
bone are also routinely co-labeled to demonstrate their close asso-
ciation with NET-derived DNA. Other methods to determine
NET release include the quantification of citrullinated histones, as
well as the use of polymerase chain reaction (PCR) to analyze
genomic NET-DNA sequences [5]. In addition, fluorescence
microscopy, confocal microscopy, and scanning electron micros-
copy (SEM) are all routinely employed to visualize NET struc-
tures, but quantification of NET release by microscopy alone is
problematic.
To study the interactions between NETs and bacterial species,
NET entrapment and microbial killing can be determined using
ex vivo assays. Microscopy provides qualitative results that enable
the visualization of bacteria associating with NET lattices. The
quantification of bacteria entrapped within NET structures can be
determined by the fluorometric detection of labeled bacteria within
NET structures [17]. Following the incubation of bacteria with
NETs, the bacteria-NET suspension can be diluted and inoculated
onto agar media and bacterial colonies can be enumerated to deter-
mine bacterial NET killing after incubation [1820].
Despite the discovery of NETs in 2004, interest in the process
of NET degradation and NET removal has only more recently
arisen. The ability of individuals to degrade NETs and the nature
Characterization, Quantification, and Visualization of Neutrophil Extracellular Traps 483
of the mechanisms involved has been linked to disease pathogene-

sis and is of particular interest in diseases characterized by a destruc-
tive host response, such as chronic periodontitis. NETs are believed
to be disassembled by nucleases (e.g., DNases) transported in cir-
culating plasma [21, 22]; therefore, assays of DNase-containing
plasma to measure NET degradation may provide useful insights
into disease mechanisms and identify novel therapeutic approaches.
Detailed below are the materials, methods, and notes used for
the isolation of neutrophils from human peripheral venous blood,
neutrophil activation, ROS quantification, NET quantification,
NET visualization, the quantification of NET entrapment and
NET killing of bacteria, and NET degradation by human plasma.
2 Materials
2.1 Isolation 1. 1.079 g/mL Percoll solution: Combine 19.708 mL Percoll,

of Neutrophils 11.792 mL distilled water (dH2O), and 3.5 mL of 1.5 M
from Human NaCl. This will equate to 35 mL of 1.079 g/mL Percoll solu-
Peripheral Blood tion, which is sufficient for four gradients.
2. 1.098 g/mL Percoll solution: Combine 24.823 mL Percoll,
6.677 mL dH2O, and 3.5 mL of 1.5 M NaCl. This equates to
35 mL of 1.098 g/mL Percoll solution, which is sufficient for
four gradients.
3. Erythrocyte lysis buffer: Dissolve 8.3 g NH4Cl, 1 g KHCO3,
0.04 g Na2EDTA2H2O, and 2.5 g bovine serum albumin
(BSA) in 1 L dH2O. Store at 4 C.
4. Phosphate buffered saline (PBS): Dissolve 7.75 g NaCl, 0.2 g
KH2PO4, and 1.5 g K2HPO4 in 1 L dH2O. Store at 4 C.
5. Sterile 3 mL Pasteur pipettes.
6. Centrifuge (preferably with ramp/brake and temperature
settings).
7. 25 and 50 mL sterile centrifuge tubes.
8. Haemocytometer, coverslips, and a light microscope (with 20
objective).
2.2 Neutrophil 1. PBS supplemented with glucose and cations (gPBS): Weigh
ROS Assays 7.75 g NaCl, 0.2 g KH2PO4, 1.5 g K2HPO4, 1.8 g glucose,
0.15 g CaCl2 and dissolve in 1 L of dH2O, then add 1.5 mL of
MgCl2. Add and dissolve in order, filter-sterilize (pore size
0.22 m) and store at 4 C.
2. 1 % BSA: Add 10 g of BSA to 1 L of previously made
PBS. Syringe-filter (pore size 0.22 m), aliquot (20 mL), and
store at 20 C. Dilute 1 % BSA to 0.1 % for use in NET
immunostaining.
3. Luminol: Dissolve 0.5 g luminol in 94.05 mL 0.1 M NaOH to

obtain a 30 mM stock solution and store at 4 C (foil wrapped)
until needed. Dilute 1 mL of stock solution with 9 mL PBS for
a 3 mM working solution and adjust pH to 7.3.
4. Isoluminol: Dissolve 0.5 g isoluminol in 94.05 mL 0.1 M
NaOH to obtain a 30 mM stock solution and store at 4 C (foil
wrapped) until needed. Dilute 1 mL of stock solution with
9 mL PBS for a 3 mM working solution and adjust pH to 7.3.
5. Horseradish peroxidase (HRP): Dissolve 5000 units of HRP in
5 mL PBS. Dilute 105 L of stock solution with 945 L PBS
to obtain a 1.5 units/15 L solution.
6. Lucigenin: Dissolve 0.01 g lucigenin in 10 mL PBS to produce
a 1 mg/mL stock solution, store at 4 C, and foil wrapped.
Dilute 1 in 3 in PBS immediately prior to use.
7. 96-well reading luminometer set to 37 C.
8. White tissue culture grade 96-well plates.
2.3 NET 1. RPMI-1640 medium: Containing sodium bicarbonate, no

Functional Assays L-glutamine or phenol red.
2. Dimethyl sulfoxide (DMSO).

3. Black tissue culture grade 96-microwell plates.
4. Clear tissue culture grade 6-, 24-, and 96-microwell plates.
5. Micrococcal nuclease (MNase): Add 1 mL dH2O to MNase
(lyophilized powder, 21,363 units) to produce a stock solu-
tion. Dilute 311.3 L of the stock solution in 50 mL PBS for
a working solution of 13.3 units/mL and a final well concen-
tration of 1 unit/mL. Aliquot (1.5 mL) and store at 20 C.
6. Sytox green: Dilute vial (5 mM DMSO) 1:500 in RPMI to
obtain a 10 M working solution. Foil wrap and store at
20 C.
7. 96-well reading fluorometer.
8. 4-(2-Hydroxyethyl) piperazine-1-ethanesulfonic acid, N-(2-
Hydroxyethyl)piperazine-N-(2-ethanesulfonic acid)
(HEPES): Prepare 0.1 M HEPES by adding 2.38 g HEPES to
100 mL dH2O, adjust to pH 7.5, and store at 4 C.
9. Cytochalasin B: Use at a final concentration of 10 g/mL by
adding 166.5 L of cytochalasin B (2 mg/mL stock in DMSO)
to 4.833 mL of PBS, store at 20 C.
10. 1 M sodium phosphate.
11. N-Methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide: Add
5.64 mL of DMSO to 50 mg vial (Sigma) to obtain a 15 mM
stock. Use at 0.5 mM by diluting stock solution in PBS and
store at 20 C.
12. 3,3,5,5-Tetramethylbenzidine (TMB): Store at 4 C.

13. N-Succinyl-Ala-Ala-Pro-Phe p-nitroanilide: Add 2 mL DMSO
to 25 mg vial to obtain a 20 mM stock and use at 1 mM by
diluting stock solution in 0.1 M HEPES. Store at 20 C.
14. Neutrophil elastase (NE): To produce a standard curve, serially
dilute 9 unit/mL human NE in PBS and store at 20 C.
15. Myeloperoxidase (MPO): To produce a standard curve, seri-
ally dilute 1 unit/mL MPO derived from human leukocytes
and store at 20 C.
16. Cathepsin G (CG): To produce a standard curve, serially dilute
1 unit/mL cathepsin G from human leukocytes and store at
20 C.
17. Fluorescein-5-isothiocyanate (FITC): Use at a final concentra-
tion of 0.3 mg/mL by adding 50 L 6 mg/mL FITC to 1 mL
of bacteria. Foil wrap and store at 20 C.
18. 96-well reading optical density plate reader capable of reading
at wavelengths of 405 and 450 nm.
2.4 Visualization 1. 2.5 % Glutaraldehyde in 0.1 M sodium cacodylate buffer:

of NETs Prepare 0.1 M sodium cacodylate by adding 4.28 g sodium
by Fluorescence cacodylate trihydrate to 100 mL of dH2O, and adjust to pH
Microscopy and SEM 7.4. Make 2.5 % glutaraldehyde by combining 7.5 mL of the
sodium cacodylate buffer, 6 mL of dH2O, and 1.5 mL of glu-
taraldehyde solution.
2. 11-mm round cover slips.
3. Graded ethanol series (20, 30, 40, 50, 60, 70, 90, and 100 %)
produced by diluting 100 % ethanol in dH2O accordingly.
4. Sputter coater to cover specimen with layer of gold.
5. Hexamethyldisilazane (HMDS).
6. 25 mm aluminum stubs with carbon conductive tabs.
7. 20 % Paraformaldehyde (PFA): Dissolve 4 g PFA in 20 mL
PBS on a mixer plate heated to 60 C. Allow the solution to
cool down prior to filter sterilizing (pore size 0.22 m).
8. 0.05 % Digitonin PBS: Dissolve 15 mg Digitonin in 30 mL
PBS to produce a 500 g/mL solution, heat the solution for a
maximum of 20 s, and leave to cool at room temperature for
1 h.
9. PBS/Hoechst: Dilute 40 L of Hoechst solution in 20 mL
PBS to obtain a 20 g/mL working solution.
10. Fluorescence microscope.
11. Scanning electron microscope.
2.5 Neutrophil 1. Phorbol 12-myristate 13-acetate (PMA): Add 1 mL DMSO to

Activation 1 mg vial to obtain a 1.62 mM PMA stock solution. Dilute
stock solution (4000-fold dilution) in PBS to a 405 nM working
solution. Aliquot the solution and store at 20 C. A final well
concentration of 25 and 50 nM is used for ROS and NETs
respectively by diluting further in PBS.
2. Hypochlorous acid (HOCl): Dilute 10 L of sodium hypo-
chlorite stock solution in 1 mL PBS (foil wrapped 1.5-mL
centrifuge tube) and use immediately.
3. Periodontal bacteria American Type Culture Collection
(ATCC).
4. Disposable presterilized loops.
5. Aerobic and anaerobic chamber.
6. Tryptone soya broth (TSB).
8. Fastidious plates.
9. Brain heart infusion (BHI) broth.
10. Fastidious broth.
11. Cling film.
12. Microbiology oven to heat-kill bacteria.
3 Methods
3.1 Isolation 1. To prepare Percoll gradients, layer 8 mL of 1.098 g/mL

of Neutrophils Percoll solution underneath 8 mL of 1.079 g/mL Percoll in a
from Human 25-mL centrifuge tube using a Pasteur pipette. Carefully layer
Peripheral Venous peripheral blood onto discontinuous Percoll density gradients
Blood (see Notes 13).
2. Centrifuge for 8 min at 150 g, followed by 10 min at 1200 g
(both set to 4 C with a brake and ramp of 1).
3. Remove the plasma, lymphocyte, and monocyte layers at the
top of the centrifuge tube using a Pasteur pipette.
4. Carefully collect the neutrophil cell layer located at the top of
the erythrocyte cell layer (see Note 4) and transfer to a 50 mL
centrifuge tube containing ~20 mL erythrocyte lysis buffer. Add
additional erythrocyte lysis buffer to a final volume of ~50 mL
per centrifuge tube and invert the centrifuge tube to mix.
5. Incubate at room temperature for 510 min or until the eryth-
rocytes have lysed (see Notes 5 and 6).
6. Centrifuge the lysed cell suspension for 6 min at 500 g to pellet
the neutrophils (see Note 7).
7. Carefully remove the supernatant and discard.
8. Resuspend neutrophils in 3 mL erythrocyte lysis buffer,

incubate for 5 min at room temperature, and recentrifuge for
6 min at 500 g.
9. Carefully remove the supernatant (i.e., repeat step no. 7) and
resuspend neutrophil cell pellet in 3 mL sterile PBS.
10. Count the cells on a haemocytometer by light microscopy (see
Notes 8 and 9).
3.2 Neutrophil 1. Add 200 L syringe-filtered 1 % BSA to each well of a white

Chemiluminescence 96-well plate and store at 4 C overnight. Following neutro-
for the Measure phil isolation the next day, remove the 1 % BSA from the plate
of Neutrophil ROS and rinse with PBS using a plate washer (see Note 10).
2. Add 1 105 neutrophils in 100 L gPBS to each well. To mea-
sure neutrophil ROS:
(a) Total ROS: 30 L luminol, 45 L gPBS.
(b) Extracellular ROS: 60 L isoluminol, 15 L of 1.5 units
HRP, 30 L gPBS.
(c) Extracellular superoxide: 30 L lucigenin, 45 L gPBS.
3. Place the plate in the luminometer (37 C) and incubate for a
30 min baseline period (see Note 11).
4. Next, pause the luminometer to stimulate the cells with 25 L
aliquots of PBS (negative control), PMA (25 nM) or bacteria
(5 107, multiplicity of infection [MOI] of 500).
5. Place the plate back in the luminometer and continue to mea-
sure light output for a further 120 min.
3.3 Quantification 1. Add 200 L syringe-filtered 1 % BSA to each well of a clear
of NET-DNA 96-well plate and store at 4 C overnight. Following neutro-
phil isolation the next day, remove the 1 % BSA from the plate
using an aspirator (see Note 10).
2. Add 1 105 neutrophils in 175 L RPMI-1640 to each well
and incubate for a 30 min baseline period (37 C, 5 % CO2) (see
Note 11).
3. Stimulate neutrophils in 25 L aliquots with PBS (negative
control), PMA (50 nM), HOCl (0.75 mM), or bacteria
(1 108, MOI of 1000), cover with foil and incubate for 4 h
(37 C, 5 % CO2) (See Notes 1214).
4. Post-incubation add 15 L of MNase at 1 unit/mL to each
well and incubate for 15 min at room temperature.
5. Centrifuge the plate for 10 min at 1800 g. Post-centrifugation,
transfer 150 L of the supernatant to a black 96-well plate (see
Notes 15 and 16).
6. Add 15 L of 10 M Sytox green to quantify any free DNA
within the supernatant and read the fluorescence immediately
on a fluorometer (excitation: 485 nm, emission: 535 nm) (See
Notes 17 and 18).
3.4 Quantification 1. Add 1 mL syringe-filtered 1 % BSA to each well of a clear

of NET-Bound 24-well plate and store at 4 C. Following neutrophil isolation
Components the next day, remove the 1 % BSA from the plate using an aspi-
rator (see Note 10).
3.4.1 Production
and Storage of NETs 2. Add 1 106 neutrophils in 875 L RPMI-1640 to each well
Note 11).
3. Stimulate neutrophils in 125 L aliquots with PBS (negative
control), PMA (50 nM), HOCl (0.75 mM), or bacteria
(1 108, MOI of 1000), cover with foil and incubate for 4 h
(37 C, 5 % CO2) (See Notes 1214).
4. Post-incubation wash the NETs by gently aspirating the super-
natant and adding 1 mL of pre-warmed RPMI-1640. Repeat
these steps once more (See Notes 19 and 20).
5. Add 75 L of MNase at 1 unit/mL to each well and incubate
for 15 min at room temperature.
6. Centrifuge the plate for 10 min at 1800 g (see Note 15).
Post-centrifugation, NET bound components can be quanti-
fied immediately or at a later date (see Note 21).
3.4.2 Measuring 1. Add 100 L of NET supernatant to each well of a clear

NET-Bound Neutrophil 96-microwell plate in duplicate.
Elastase (NE) 2. Add 100 L of 0.5 M N-Methoxysuccinyl-Ala-Ala-Pro-Val
p-nitroanilide to each well, cover the plate with foil, and incu-
bate for 2 h (37 C, 5 % CO2).
3. Post-incubation, measure the absorbance at 405 nm.
4. Generate a standard curve by serially diluting human NE in
100 L aliquots and running in duplicate on the same plate.
3.4.3 Measuring 1. Add 50 L of NET supernatant to each well of a clear 96-well

NET-Bound plate in duplicate.
Myeloperoxidase (MPO) 2. Add 50 L of TMB substrate solution to each well, cover the
plate with foil, and incubate for 20 min at room temperature.
3. Post-incubation, add 50 L of 1 M sodium phosphate to stop
the reaction.
4. Measure the absorbance at 450 nm.
5. Generate a standard curve by serially diluting human MPO in
3.4.4 Measuring 1. Add 50 L of NET supernatant to each well of a clear 96-well

NET-bound cathepsin G plate in duplicate.
(CG) 2. Add 50 L of 1 mM N-Succinyl-Ala-Ala-Pro-Phe p-nitroani-
lide in 0.1 M HEPES to each well, cover the plate with foil,
and incubate for 2 h (37 C, 5 % CO2).
3. Post-incubation, measure the absorbance at 405 nm.

4. Generate a standard curve by serially diluting human CG in
3.5 Quantification 1. Add 200 L syringe-filtered 1 % BSA to each well of a black

of NET Entrapment 96-well plate and store at 4 C overnight. Following neutro-
of Bacteria phil isolation the next day, remove the 1 % BSA from the plate
Note 11).
3. Stimulate neutrophils with 25 L of HOCl (0.75 mM); also
include a neutrophil-free well and cells treated with PBS as
negative controls. Cover with foil and incubate for 3 h (37 C,
5 % CO2) (see Note 14).
4. Following the culture of planktonic bacteria (see Table 1), mea-
sure the turbidity by spectroscopy and dilute to 1 107 per
50 L (MOI of 100) with PBS in 1 mL aliquots. Fluorescently
stain bacteria by incubating bacteria with 0.3 mg/mL FITC
for 30 min on ice with continuous agitation. Next, centrifuge
the bacteria (10 min at 5000 rpm) to precipitate FITC-stained
bacteria, and resuspend in 1 mL PBS.
5. Prior to adding FITC-stained bacteria to the well, add 1 unit/
mL MNase to selected wells containing NETs and incubate for
15 min at room temperature (see Note 22).
6. Add the live FITC-stained bacteria to the NETs plate in 50 L
aliquots (bacteria added to neutrophil-free wells, PBS-treated
cells, HOCl-treated [NETs] cells, and degraded HOCl-NETs)
and incubate for 1 h (37 C, 5 % CO2).
7. Post-incubation, carefully remove the supernatant and replace
with 200 L fresh, pre-warmed RPMI-1640. Repeat wash step
(see Note 23).
8. Quantify entrapped bacteria by measuring the fluorescence
using a fluorometer (excitation: 485 nm, emission: 535 nm).
3.6 Quantification 1. Add 500 L syringe-filtered 1 % BSA to each well of a clear

of NET Killing 24-well plate and store at 4 C overnight. Following neutro-
of Bacteria phil isolation the next day, remove the 1 % BSA from the plate
Note 11).
3. Stimulate neutrophils with 60 L of HOCl (0.75 mM), cover
with foil, and incubate for 3 h (37 C, 5 % CO2) (see Note 14).
490
Table 1
Culture conditions for periodontal bacterial species
Bacteria per mL
Bacteria strain ATCC number Culture medium Growing conditions (37 C) if OD600nm = 1
Actinomyces viscosus (naeslundii genospecies 2) 43146 Solid media: horse blood agar Anaerobic 8.3 108
Liquid media: brain heart infusion
Aggregatibacter actinomycetemcomitans 43718 Solid media: horse blood agar Anaerobic 6.8 109
Phillipa C. White et al.
serotype b Liquid media: brain heart infusion

Propionybacterium acnes 11827 Solid media: horse blood agar Anaerobic 1.69 109
Selenomonas noxia 43541 Solid media: horse blood agar Anaerobic 1.69 109
Veillonella parvula 10790 Solid media: horse blood agar Anaerobic 6.8 109
Streptococcus sanguinis 10556 Solid media: horse blood agar 5 % CO2 1.69 109
Streptococcus oralis 35037 Solid media: horse blood agar 5 % CO2 1.69 109
Streptococcus intermedius 27335 Solid media: horse blood agar 5 % CO2 1.69 109
Streptococcus anginosus 33397 Solid media: horse blood agar 5 % CO2 1.69 109
Streptococcus gordonii 10558 Solid media: horse blood agar 5 % CO2 1.69 109
Capnocytophagia gingivalis 33624 (27) Solid media: horse blood agar Anaerobic 1.62 109
Eikenella corrodens 23834 Solid media: horse blood agar Anaerobic 6.8 109
Aggregatibacter actinomycetemcomitans 29523 Solid media: horse blood agar Anaerobic 6.8 109
serotype a Liquid media: brain heart infusion
Capnocytophaga sputigena 33612(4) Solid media: horse blood agar Anaerobic 1.62 109
Streptococcus constellatus 27823 (M32b) Solid media: horse blood agar 5 % CO2 1.69 109
Camylobacter rectus 33238 (371) Solid media: horse blood agar Anaerobic 6.8 109
Campylobacter showae 51146 Solid media: horse blood agar Anaerobic 6.8 109
Fusobacterium nucleatum sp. Nucleatum 25586 Solid media: horse blood agar Anaerobic 1.62 109
Fusobacterium nucleatum sp. Polymorphum 10953 Solid media: horse blood agar Anaerobic 1.62 109
Porphyromonas gingivalis W83 Solid media: fastidious blood with Anaerobic 1.69 109
neomycin
Liquid media: fastidious
Bacteria we have previously used in the above assays are listed alongside their respective ATCC number. The agar media, broth, and incubation culturing conditions are outlined
below. Following planktonic growth, bacterial concentrations were determined by spectrophotometry (OD600nm) and our own experiments determined the number of bacteria
per mL of broth if the OD is 1.0. ATCC American Type Culture Collection
Characterization, Quantification, and Visualization of Neutrophil Extracellular Traps
491
4. Next, carefully remove the supernatant from each well and

replace with 500 L pre-warmed RPMI-1640, with or without
10 g/mL cytochalasin B or 1 unit/mL MNase, incubate for
15 min at room temperature (see Note 24).
5. Following the culture of planktonic bacteria (see Table 1), mea-
sure the turbidity by spectroscopy and dilute to 1 107 per
50 L (MOI of 100) with PBS in 1 mL aliquots.
6. Add the live bacteria in 50-L aliquots to the selected wells
and centrifuge the plate for 10 min at 700 g, then incubate
the plate for 1 h (37 C, 5 % CO2) (see Note 15).
7. Post-incubation, add MNase in 25 L aliquots (100 unit/mL)
and incubate for 15 min at room temperature (see Note 25).
8. Transfer the well contents to 1.5-mL centrifuge tubes and
dilute with broth. To determine bacteria viability, add 50 L of
this suspension and inoculate an agar plate to enumerate bacte-
rial colonies in 24 h (see Note 26).
3.7 Quantification 1. Stimulate for NETs with 0.75 mM HOCl in a 96-well plate, as
of NET Degradation previously described.
by Human Plasma 2. Defrost previously isolated plasma samples at room temperature
and dilute to 10 % in PBS. Add 10 % plasma to the wells in 50 L
aliquots and incubate for 3 h (37 C, 5 % CO2) (see Note 27).
3. Treat selected wells with 1 unit/mL MNase for 15 min at
room temperature, instead of plasma (see Note 28).
4. Following incubation with plasma, centrifuge the plate (10 min
at 1800 g), and then transfer 150 L of the supernatant to a
black 96-well plate (see Notes 15 and 16).
5. Add 15 L of 10 M Sytox green to quantify any DNA within
the supernatant on a fluorometer (excitation: 485 nm, emis-
sion: 535 nm).
3.8 Fluorescence 1. Add 300 L of syringe-filtered 1 % BSA to each well of a clear

Visualization of NETs 24-well plate and store at 4 C overnight. Following neutro-
phil isolation the next day, remove the 1 % BSA from the plate
Note 11).
3. Stimulate neutrophils in 40-L aliquots with PBS (negative
control), PMA (50 nM) or HOCl (0.75 mM), cover with foil,
and incubate for 4 h (37 C, 5 % CO2) (see Notes 14 and 17).
4. Add 30 L of 10 M Sytox green and visualize immediately
under a fluorescence microscope (excitation: 485 nm, emission:
535 nm).
3.9 Scanning 1. Sterilize round 11-mm glass coverslips (see Note 29) in 0.2 M
Electron Microscopy HCl, followed by two wash steps in dH2O. Once dry, add
(SEM) of NETs 100 L of syringe-filtered 1 % BSA and keep at room tempera-
ture for 1 h prior to use (see Note 10).
2. Add 1 105 neutrophils in 100 L RPMI-1640 to each cover-
slip and after a 30-min baseline incubation period (37 C, 5 %
CO2) (see Note 11), stimulate cells with PBS (negative con-
trol), PMA (50 nM), or live bacteria (1 107 MOI of 100) (see
Note 14).
3. Following a 4-h incubation (37 C, 5 % CO2), fix samples with
2.5 % glutaraldehyde in 0.1 M sodium cacodylate buffer (pH
7.3) for 30 min at room temperature.
4. Dehydrate samples by immersing in a graded ethanol series
(20, 30, 40, 50, 60, 70, 90, 100, and 100 % for 10 min each).
5. Dry samples by adding 100 L of HMDS, and leave to evapo-
rate in a fume hood overnight.
6. Mount samples onto 25-mm aluminum stubs with carbon
conductive tabs and coat in gold for 90 s, prior to analysis by
SEM.
3.10 Immunos 1. Following the production of NETs, centrifuge the plate

taining of NETs (10 min at 1800 g).
2. Postcentrifugation, add 50 L 20 % PFA to each well and incu-
bate for 10 min at room temperature. Next, gently remove the
media and wash twice with PBS.
3. Remove PBS and add 50 L of 0.1 % PBS-BSA to each well.
Incubate for 30 min at room temperature and wash with PBS.
4. Dilute mouse antihuman MPO and rabbit antihuman NE pri-
mary antibodies 1/150 in 0.05 % Digitonin PBS.
5. Add 50 L 0.05 % Digitonin PBS plus antibodies to each well
and incubate for 1 h at room temperature, then wash with PBS.
6. Dilute secondary antibodies goat anti-rabbit IgG Alexa Fluor
594 (Excitation: 590 nm, Emission: 617 nm) and goat anti-
mouse IgG Alexa Fluor 488 (Excitation: 495 nm, Emission:
519 nm) 1/500 in PBS/Hoechst.
7. Add 50 L secondary antibodies in PBS/Hoechst to each well
and incubate for 30 min at room temperature and then wash
with PBS.
8. Visualize NETs using a fluorescence microscope.
3.11 Bacterial 1. Store bacterial suspensions at 80 C in cryotubes containing

Culture TSB and 10 % DMSO.
2. When required, defrost at room temperature and inoculate
agar plates by adding 100 L of bacterial suspension and
spreading with a disposable presterilized loop (see Table 1).
3. Loosely wrap plates in cling film, invert and incubate in the

appropriate conditions for at least 3 days (see Note 30).
4. Grow cultures planktonically by inoculating a single colony
into broth, and incubate for at least 3 days on a shaker (see
Notes 31 and 32).
5. Confirm planktonic growth by measuring the optical density in
a cuvette (600 nm) using noninoculated media to calibrate the
spectrophotometer.
6. Centrifuge the bacterial suspension in broth (15 min, 1800 g,
4 C). Discard the supernatant and resuspend the bacterial pel-
let in PBS. Repeat centrifugation and PBS wash steps twice
more (see Note 33).
7. To heat-kill bacteria, incubate the bacterial suspension in PBS
at 80 C in a microbiology oven for 30 min (see Note 34).
4 Notes
1. Collect blood in lithium-heparin anti-coagulated tubes.

Experiments from our own laboratory reveal greater cell
retrieval and less inadvertent cell activation compared with
other anticoagulants. Isolate cells as soon as possible. We typically
isolate approximately 1 106 cells per mL of blood.
2. If possible, prepare gradients immediately prior to neutrophil
isolation to minimize chance of gradient disruption; however,
gradients may be carefully stored at 4 C for a day if needed.
3. It is more reliable to layer the 1.098 g/mL Percoll underneath
the 1.079 g/mL Percoll solution.
4. Collect as few erythrocytes as possible to minimize the risk of
erythrocyte contamination.
5. Erythrocyte lysis is indicated by the increased transparency of
the blood/lysis buffer solution. Do not incubate neutrophils
in erythrocyte lysis buffer for longer than recommended due
to the potential for cell activation.
6. If erythrocytes are taking time to lyse (solution to go clear) this
could be due to the age of the lysis buffer, which should be
stored for a maximum of 3 weeks at 4 C.
7. During the lysis buffer/PBS washing stages and removal of
supernatant, the operator must take care not to disturb the cell
pellet to maximize cell retrieval.
8. We recommend duplicate cell counting using a haemocytom-
eter and calculation of the mean.
9. We routinely ensure neutrophil viability by trypan blue exclu-
sion and CellTiter-Glo assays, as well as neutrophil purity by
flow cytometry [23].
10. Adding 1 % BSA to plasticware prior to the addition of neutrophils

coats the surface, which has been found to reduce inadvertent
neutrophil activation.
11. A 30-min baseline period prior to neutrophil activation allows
the cells to settle in the plastic-ware and neutrophils are more
responsive to exogenous stimulation.
12. HOCl and PMA stimuli are light-sensitive. Make up/defrost
at room temperature immediately prior to use and protect
from light by foil wrapping.
13. Following their release, NETs are attached to the neutrophil
they are derived from. The addition of MNase degrades the
NET structures, which following centrifugation and the sedi-
mentation of neutrophils, allows for the fluorometric quantifi-
cation of NET-DNA in the supernatants.
14. NETs are produced in response to HOCl in as little as 30 min
[9], PMA and bacteria typically induce NET production in 3
and 4 h, respectively.
15. Following MNase digestion, neutrophils are centrifuged in a
plate spinner. If you do not have a plate spinner then the well
contents can be transferred to 1.5-mL centrifuge tubes for
centrifugation.
16. Care needs to be taken when transferring NET supernatants to
a new plate to ensure the operator is not disturbing the neutro-
phils at the bottom of the well; tipping the plate toward you
helps.
17. Bacteria produce DNA [24] that can interfere with NET-DNA
assays. If quantifying NET production in response to bacteria,
control wells containing bacteria in PBS (and no neutrophils)
need to be measured and subtracted from the final NET pro-
duction values. We do not employ a bacterial NET stimulus
during Sytox green fluorescence microscopy as it is very diffi-
cult to distinguish between NET-DNA and bacterial DNA;
therefore, use PMA or HOCl for NET fluorescence
visualization.
18. Sytox green does not permeate viable cells; therefore, very
high readings (DNA) in control wells are indicative of high
numbers of nonviable neutrophils.
19. Neutrophils and the attached NET structures settle to the bot-
tom of plastic-ware; however, they are not adherent cells and
washing needs to be careful to minimize the disruption of
NET structures.
20. Washing removes components that are not NET-bound but
are concurrently released during neutrophil activation.
21. NET-bound components can be quantified immediately, or
transferred to cryotubes and stored at 4 C (for up to 1 week) or
at 20 C (for up to 6 months) for quantification at a later date.
22. The addition of DNase to NETs in selected wells produces

degraded NET structures to serve as an additional control.
23. The duplicate wash steps ensure bacteria not entrapped within
NET structures are removed and therefore only the fluores-
cence of entrapped bacteria is measured.
24. The addition of cytochalasin B to selected wells following NET
release inhibits neutrophil phagocytosis and the addition of
MNase degrades NETs.
25. The addition of MNase postincubation disassembles the NETs
and frees the bacteria prior to inoculation on agar plates.
26. The dilution factor required prior to inoculating agar plates
with bacteria needs to be determined in preliminary experi-
ments for each bacterial species.
27. Plasma is isolated from whole blood in 6 mL lithium heparin
anti-coagulant tubes by centrifuging for 30 min at 1000 g
(4 C). Following centrifugation, plasma is transferred to cryo-
tubes in 500-L aliquots and stored at 80 C.
28. The treatment of some wells with MNase will serve as a posi-
tive control and the MNase treatment is considered 100 %
NET degradation.
29. It is easier to handle coverslips for SEM use in 6-well plates.
30. Wrapping agar plates in clingfilm helps to prevent the desicca-
tion of agar following prolonged incubation.
31. Periodontal bacteria typically take at least 3 days to grow plank-
tonically. However, growth can be determined at any time
point following inoculation of broth by assessing the turbidity,
as determined by measuring the optical density at 600 nm.
32. Confirm the bacterial species by gram staining and/or PCR.
33. Planktonic bacteria are washed in PBS multiple times to remove
broth prior to neutrophil assays.
34. Killing of bacteria can be confirmed by inoculating a fresh agar
plate with the heated bacteria suspension.
References
1. Borregaard N, Cowland JB (1997) Granules of traps in human chronic granulomatous disease
the human neutrophilic polymorphonuclear after gene therapy is calprotectin-dependent.
leukocyte. Blood 89:35033521 J Allergy Clin Immunol 127:12431252
2. Brinkmann V, Reichard U, Goosmann C, 4. Wang Y, Li M, Stadler S, Correll S, Li P, Wang
Fauler B, Uhlemann Y, Weiss DS, Weinrauch D, Hayama R, Leonelli L, Han H, Grigoryev
Y, Zychlinsky A (2004) Neutrophil extracel- SA, Allis CD, Coonrod SA (2009) Histone
lular traps kill bacteria. Science 303: hypercitrullination mediates chromatin decon-
15321535 densation and neutrophil extracellular trap for-
3. Bianchi M, Niemiec MJ, Siler U, Urban CF, mation. J Cell Biol 184:205213
Reichenbach J (2011) Restoration of anti- 5. Yousefi S, Mihalache C, Kozlowski E, Schmid I,
Aspergillus defense by neutrophil extracellular Simon HU (2009) Viable neutrophils release
mitochondrial DNA to form neutrophil extra- 15. Keshari RS, Jyoti A, Dubey M, Kothari N,
cellular traps. Cell Death Differ 16:14381444 Kohli M, Bogra J, Barthwal MK, Dikshit M
6. Pilsczek FH, Salina D, Poon KK, Fahey C, (2012) Cytokines induced neutrophil extracel-
Yipp BG, Sibley CD, Robbins SM, Green FH, lular traps formation: implication for the
Surette MG, Sugai M, Bowden MG, Hussain inflammatory disease condition. PLoS One 7,
M, Zhang K, Kubes P (2010) A novel mecha- e48111
nism of rapid nuclear neutrophil extracellular 16. Palmer LJ (2010) Neutrophil extracellular
trap formation in response to Staphylococcus traps in periodontitis. Ph.D. thesis, University
aureus. J Immunol 185:74137425 of Birmingham
7. Harris PC (2012) Effect of density gradient 17. Berends ET, Horswill AR, Haste NM,
material upon ex-vivo neutrophil behaviour, Monestier M, Nizet V, von Kckritz-Blickwede
and effect of neutrophil extracellular traps M (2010) Nuclease expression by Staphylococcus
upon the growth and survival of periodonto- aureus facilitates escape from neutrophil extra-
pathogenic bacteria. MRes thesis, University of cellular traps. J Innate Immun 2:576586
Birmingham 18. Wartha F, Beiter K, Albiger B, Fernebro J,
8. Fuchs TA, Abed U, Goosmann C, Hurwitz R, Zychlinsky A, Normark S, Henriques-Normark
Schulze I, Wahn V, Weinrauch Y, Brinkmann B (2007) Capsule and Dalanylated lipotei-
V, Zychlinsky A (2007) Novel cell death pro- choic acids protect Streptococcus pneumoniae
gram leads to neutrophil extracellular traps. against neutrophil extracellular traps. Cell
J Cell Biol 176:231241 Microbiol 9:11621171
9. Palmer L, Cooper PR, Ling MR, Wright HJ, 19. Lauth X, von Kckritz-Blickwede M, McNamara
Huissoon A, Chapple IL (2012) Hypochlorous CW, Myskowski S, Zinkernagel AS, Beall B,
acid regulates neutrophil extracellular trap release Ghosh P, Gallo RL, Nizet V (2009) M1 protein
in humans. Clin Exp Immunol 167:261268 allows Group A streptococcal survival in phago-
10. Beiter K, Wartha F, Albiger B, Normark S, cyte extracellular traps through cathelicidin inhi-
Zychlinsky A, Henriques-Normark B (2006) bition. J Innate Immun 1:202214
An endonuclease allows Streptococcus pneu- 20. Urban CF, Reichard U, Brinkmann V,
moniae to escape from neutrophil extracellular Zychlinsky A (2006) Neutrophil extracellular
traps. Curr Biol 16:401407 traps capture and kill Candida albicans yeast
11. Delbosc S, Alsac JM, Journe C, Louedec L, and hyphal forms. Cell Microbiol 8:668676
Castier Y, Bonnaure-Mallet M, Ruimy R, 21. Hakkim A, Frnrohr BG, Amann K, Laube B,
Rossignol P, Bouchard P, Michel JB, Meilhac Abed UA, Brinkmann V, Herrmann M, Voll
O (2011) Porphyromonas gingivalis participates RE, Zychlinsky A (2010) Impairment of neu-
in pathogenesis of human abdominal aortic trophil extracellular trap degradation is associ-
aneurysm by neutrophil activation. Proof of ated with lupus nephritis. Proc Natl Acad Sci U
concept in rats. PLoS One 6, e18679 S A 107:98139818
12. Behrendt JH, Ruiz A, Zahner H, Taubert A, 22. Leffler J, Martin M, Gullstrand B, Tydn H,
Hermosilla C (2010) Neutrophil extracellu- Lood C, Truedsson L, Bengtsson AA, Blom
lar trap formation as innate immune reac- AM (2012) Neutrophil extracellular traps that
tions against the apicomplexan parasite are not degraded in systemic lupus erythema-
Eimeria bovis. Vet Immunol Immunopathol tosus activate complement exacerbating the
133:18 disease. J Immunol 188:35223531
13. Byrd AS, OBrien XM, Johnson CM, Lavigne 23. Zhou L, Somasundaram R, Nederhof RF,
LM, Reichner JS (2013) An extracellular Dijkstra G, Faber KN, Peppelenbosch MP,
matrixbased mechanism of rapid neutrophil Fuhler GM (2012) Impact of human granulo-
extracellular trap formation in response to cyte and monocyte isolation procedures on
Candida albicans. J Immunol 190:41364148 functional studies. Clin Vaccine Immunol
14. Saitoh T, Komano J, Saitoh Y, Misawa T, 19:10651074
Takahama M, Kozaki T, Uehata T, Iwasaki H, 24. Das T, Sehar S, Manefield M (2013) The roles
Omori H, Yamaoka S, Yamamoto N, Akira S of extracellular DNA in the structural integrity
(2012) Neutrophil extracellular traps mediate a of extracellular polymeric substance and bacte-
host defense response to human immunodefi- rial biofilm development. Environ Microbiol
ciency virus-1. Cell Host Microbe 12:109116 Rep 5:778786
Index
A isolation of structural genes211213

N-terminal sequencing 53, 210, 211, 216
Adeno-associated virus108 purification 206, 210
Adenovirus 109, 110 Biofilm
Adherence isolation419420 bacteriocin competition assay207208
Adhesion confocal laser scanning microscopy205
of bacteria to saliva-coated dentures167 Bioinformatics154, 160, 266, 270, 340
of bacteria to saliva-coated Biomarkers4, 17, 18, 37, 38, 6176, 80, 98, 100, 429
hydroxyapatite 167, 169170, 177 Bisphosphonate-related osteonecrosis of the jaw
of bacteria to saliva-coated medical-grade (BRONJ)448
silicone 167, 171, 179180 Bisulfite sequencing 250, 252, 254,
of saliva-coated yeast to epithelial cells167 260262, 266, 267, 270, 279294
of yeast to immobilized proteins 168169, 174177
of yeast to saliva-coated denture prostheses167 C
of yeast to saliva-coated
Calvarial bone 368, 369, 373, 377
hydroxyapatite 167, 169170, 177
Campylobacter rectus 193, 198
of yeast to saliva-coated medical-grade
Candida albicans
silicone 167, 171, 179180
adhesion to epithelial cells381
Adipose-derived stem cells (ADSC)439444
adhesion to saliva-coated dentures167
Aggregatibacter actinomycetemcomitans 133, 191,
adhesion to saliva-coated
198, 490, 491
hydroxyapatite 167, 169170, 177
AllPrep DNA/RNA/Protein kit300302
adhesion to saliva-coated medical-grade
Allprotect tissue reagent 300, 301, 305
silicone179180
Ameloblastin375
genomic dna purification154
Animal models
Cathepsin G 485, 488489
mouse115
Cell sheet403411
rat113
Chemiluminescence487
Antibody6, 38, 69, 251, 254,
Cloning in Escherichia coli 188, 268
383, 384, 388, 392, 394, 406, 410, 415422, 424,
Colonization 127, 133, 165, 166, 192, 204
425, 437, 471, 493
Comet assay 62, 66, 6971, 75
Antioxidant6176
Competence (for genetic transformation)233
Ascorbic acid62, 64, 67, 74, 370,
Competence-stimulating peptide (CSP)
377, 404, 407, 448, 451
of anginosus group streptococci 219221,
227, 228, 490
B
of mitis group streptococci 130, 219221,
Bacterial DNA 223, 225228
cloning 128, 206, 212 of mutans group streptococci 221, 223,
expression vector215 224, 230, 234
preparation 197, 495 Complementary DNA (cDNA) synthesis 369, 372, 376
Bacteriocin Complementary RNA (cRNA)17
activity assay209 Computational genomics 253, 348
competition assay (agar plate)207 Confocal laser scanning microscopy (CLSM)205
competition assay (biofilm)207208 CpG islands 252, 279, 280, 291, 299305, 333, 350
derivatization 206, 211 CSP. See Competence-stimulating peptide (CSP)
DOI10.1007/978-1-4939-6685-1, Springer Science+Business Media LLC 2017
499
Oral Biology: Molecular Techniques and Applications
500 Index

D Fluorescence microscopy482, 485, 492, 493, 495
Functional proteomics414
Decellularization 404, 405, 409411 Fusobacterium spp.131134, 145, 193195, 198
Delivery devices. See Viral vectors
Denaturing gradient gel electrophoresis G
DNA extraction 62, 140143,
Gel electrophoresis46, 49, 50, 122,
194, 196197, 200, 282
235, 238, 239, 292, 304, 385, 389, 397, 399, 465,
GC clamp 140, 141
470471
Dental development461
Gel preparation (protein) 187, 463, 468470
DGGE. See Denaturing gradient gel electrophoresis
Gene expression analysis215, 269, 369, 375, 377
Differential expression analysis 327328,
Luciferase assay215
337339, 343, 348, 357
Microarrays 308, 309
Differential methylation 257, 260263, 291292
qPCR269, 324, 369, 372,
Differential methylation analysis package
376377, 450, 454
(DMAP) 255, 257, 261, 262, 291, 292
Reporter gene 215, 378
DNA purification
Gene therapy107122
from Candida albicans (and related species)229,
Genomic DNA5, 6, 154, 226, 229,
282, 315
251, 252, 268, 280, 282, 294, 301304, 306, 318,
from Porphyromonas gingivalis 191, 193,
320, 396, 449, 454
197, 198, 381
Geranylgeraniol449
from Streptococcus spp.132, 133, 135, 225, 227
Gingiva
E cells64, 72
tissue 82, 299325, 330,
Elastase 485, 488 339, 348, 349, 359
Embryonic tissue explants Gingival crevicular fluid (GCF) 38, 40, 42,
culture367379 43, 5055, 64, 67, 7273, 80, 83, 84, 191, 194
dissection 370371, 373 Gingival tissue 82, 299305, 308,
fixation 166, 371, 373375 309, 317, 324, 330, 339, 348, 349, 359
In situ hybridization (ISH) 369, 371372, 375
mouse 368, 371, 373, 378 H
Enamel
High-performance liquid chromatography
epithelium462, 465, 467, 474, 476
(HPLC)64, 67
matrix 462, 467
High-throughput DNA sequencing (HTS) 4, 135,
microdissection 462, 467
153, 268269, 292, 311312, 318320, 327343,
protein extraction 462, 465, 467468
384, 385, 391, 393, 394, 429
Enamel defects 462, 473
Histology 371, 375
Epidemiology4
Homologous recombination 213, 215, 233
Epigenetics 249, 260, 263, 265, 270, 279, 299
H400 oral epithelial cells386
Epithelial cells5, 107, 108, 167, 170,
HPLC. See High-performance liquid chromatography
177179, 381383, 385392, 394397, 400, 401
(HPLC)
Epithelium and mesenchyme interactions 367, 378
HumanMethylation450 BeadChip 252, 258
Exosomes 56, 911, 13
Hydrogen peroxide (H2O2) assay 65, 69, 204, 424
Extracellular RNAs (exRNAs) 17, 18, 2033
Hydroxyapatite167, 169170, 177, 417, 423
F
I
FACS
Immunocytochemistry 383, 388
sorting 416, 422
Immunohistochemistry417418, 424, 430, 437
Fetal bovine serum (FBS) 114, 310, 321,
Immunohistology
369, 415, 439, 448
Immunomagnetic bead separation415416
Flow cytometry421422
Immunostaining405406, 409410, 483, 493
Fluorescence 34, 64, 67, 128, 144, 179,
In situ hybridization 369, 371372, 375
192, 193, 200, 208, 269, 325, 384, 395, 415416,
Interspecies competition 207, 208, 213
454, 487, 489
501
Index

K Oral microbiota
amplicon sequencing156
450K. See HumanMethylation450 BeadChip taxonomy 128, 168
Oral streptococci 204, 221
L
Organ culture
Luciferase reporter215 dissection370371
fixation373375
M hanging drop culture 369, 375377
Markerless mutagenesis 223, 229, 230, mouse 314, 368
238239 Trowell-type organ culture 368, 369, 371, 378
Mass spectrometry (MS) 4, 3746, 4858, Oxidative stress6176
67, 79, 100, 102, 210, 471, 472, 475 8-Oxo-2'-deoxyguanosine (8-OHdG) 6466,
Mesenchymal stem cells (MSC) 414, 418, 6970, 74
419, 422, 439, 441
P
Metabolomics83
Microarrays PCR. See Polymerase chain reaction
Affymetrix 340, 353 PCR array 301, 306, 450, 454
data analysis 18, 97, 102, 160, 198, 250, Periodontal diseases 3739, 4145,
256, 259, 260, 266, 267, 270, 296, 305, 450, 454 4750, 5256, 58, 133134, 309, 348, 350, 383
real-time PCR, 128, 129, 300 Periodontal ligament
Microbial community profiling139151 cell culture 309, 321, 407
Microbial ecology153 fibroblasts413
Microdissection252 processing 415, 418
enamel epithelium 462, 467 stem cells 414, 418421
enamel matrix 462, 467 Periodontal pathogens191201, 314, 321, 381, 400
Micronutrients6176 Polycaprolactone (PCL)403405, 407, 408, 410, 411
MicroRNA 23, 316 Polymerase chain reaction
Microsample proteomics462 quantitative real-time PCR128
Morphogenesis 368, 371 Standard PCR protocol397
Mouse model65, 69, 112, 115116, Polymethyl methacrylate 171, 173, 180, 182
121, 252, 265, 266, 280, 314, 368, 371, 373, 378, Porphyromonas gingivalis130, 133, 191195,
388, 392, 406, 410, 414, 416, 422, 423, 465, 493 197, 198, 383, 390, 400, 491
Multiplex qPCR (m-qPCR) 193, 195, 197 Protein 3739, 49, 53, 55, 57, 58, 311,
Myeloperoxidase (MPO) 485, 488, 493 315317, 320, 462, 465468, 471473, 475, 477
analysis
N functional characterization 467, 473
Natural transformation 213, 237 identification37, 38, 49, 53, 55,
Neutrophil extracellular traps (NETs) 481489, 57, 58, 466, 471473, 475, 477
492496 quantitation 39, 58, 471
Next generation DNA sequencing 128, 129, 217 extraction311, 315317, 320,
NMR spectroscopy7983 462, 465, 467468, 477
Nucleic acid techniques (for microbial taxonomy) purification38, 300302, 311, 315317
broad-range PCR129 Protein-releasing beads370
checkerboard DNA-DNA hybridization129 Proteomics
DNA-DNA hybridization128 microsample 461463, 465, 467473, 475, 476
DNA microarray technology129
Q
fluorescence in situ hybridization (FISH)128
multiplex PCR128 qAMP300
nested PCR128 QIIME. See Quantitative Insights Into Microbial Ecology
qPCR. See real-time quantitative PCR
O Qualitative and quantitative proteomics 3746, 4858
Operational taxonomic unit (OTU) 155, 157162 Quantitative insights into microbial ecology153162
Oral diseases 38, 132135, 165 Quantitative real-time reverse transcriptase
Oral fluids 4, 3739, 4145, 4750, 5256, 58 PCR (qRT2-PCR)453459
502 Index

R SDS-PAGE. See Sodium dodecyl sulfate polyacrylamide
gel electrophoresis
Radiolabeling SEM. See Scanning electron microscopy (SEM)
Of bacteria (3H-thymidine)167 Serum-free media (SFM)440444
Of cultured cells, 310311314 Signaling molecules 369, 373, 375, 378
Of yeast (35S-methionine)167169 SigX-inducing peptide (XIP) 219225,
Reactive oxygen species (ROS)62, 481484, 486, 487 227229, 234, 236, 238239, 245
Real-time quantitative PCR (RT-qPCR) 17, 369, Silicone167, 171, 172, 179180,
372, 376 187, 188, 409
Recombinant proteins 188, 370, 372 16S rDNA gene sequencing 140, 141,
Reduced representation bisulfite sequencing 143145, 153162, 195
(RRBS)251257, 261, 262, 279294 Small and long ncRNA profiling 18, 34, 35
Regional DNA methylation 267, 269 Sodium dodecyl sulfate polyacrylamide gel
RNA electrophoresis39, 4549, 51, 53,
cRNA17 54, 57, 58, 174175, 463464, 468471, 475, 476
extraction18, 34, 320, 323, 388, 395 Stable-isotope labeling chemistries5054
gingival tissue300 Staphylococcus epidermidis 166168, 171,
isolation 5, 9, 12, 18, 2021, 385, 395 172, 174, 179183
mRNA5, 17, 27, 263, 300, adhesion to saliva-coated dentures167
349360, 369, 385, 388389, 395397 adhesion to saliva-coated medical-grade
Peripheral blood309 silicone179180
purification 9, 300302, 315 Statistical analysis 98, 264
saliva 5, 89, 1735 Statistical model 98, 260
RNA sequencing12, 1735, 318320, Stem cells 415416, 419420
324, 325, 337, 342 cryopreservation414, 420421, 444, 451
R software differentiation367
2-sample T test351 isolation
gplots 349, 357 adherence419420
immunomagnetic beads 415416, 419
S
mesenchymal 414, 419, 421, 422, 439, 440
Saliva20, 49, 115, 396, 471, 473 periodontal ligament 403, 404, 407
collection Streptococcus/Streptococci
human 20, 115 S. mitis 130, 221223, 225227, 229, 230
mouse 112, 115116 S. mutans 130, 133, 204, 207, 208, 215,
diagnostics 312, 17 217, 219226, 228231, 233235, 238244
microarrays17 S. pneumoniae220, 233235, 240,
processing5, 6, 10, 18, 20 241, 243, 245
proteomics Streptococcus mutans
2-D gel electrophoresis471 bacteriocin assays205
Mascot database searching473 competence220, 224, 229, 230, 234235
reverse transcriptase PCR (RT-PCR)396 mutagenesis233
SEQUEST database searching49 transformation220, 221, 223225,
tandem mass spectrometry (LC-MS/MS)48 229, 230, 238239
Reverse transcriptase PCR (RT-PCR)23, 24 Streptococcus sanguinis
RNA isolation 12, 18, 2021 Hydrogen peroxide production assay204
storage4 SYBR green114, 118, 192, 269,
submandibular 115, 116, 120121 281, 284, 285, 305, 372, 376, 454
transcriptome5 Synthetic peptides 220222, 234
Salivary gland 3, 5, 107, 108, 111114, 121 Systems biology79
Salivary hypofunction 108, 109, 113
Scanning electron microscopy (SEM) 224, 236, T
408, 409, 482, 485, 493, 496 Tannerella forsythia130, 133, 191195, 198
Scintillation counter167, 169171, 180, 185, 188 TaqMan (probe)192
503
Index

Taxonomy 128135, 159, 160 V
Terminal restriction fragment length polymorphism128
DNA extraction140144 Viral vectors
restriction enzymes 140, 143 delivery 111, 115, 116,
Tissue culture170, 177, 188, 370, 120121
373, 385, 393, 415, 418, 441443, 484 recombinant serotype 5 adenoviral (rAd5) 108110,
Tissue engineering 403, 404, 414 114118, 122
Total RNA22, 302, 312, 316318, serotype 2 adeno-associated viral (rAAV2) 108111,
320, 323, 376, 396, 448, 449, 452454 114115, 118120, 122
Transcriptome 12, 308, 309, 348 W
Transformation
of anginosus group streptococci 227, 228 Whole-genome bisulfite sequencing (WGBS) 251255,
of Streptococcus mitis 223, 225227, 257, 261, 262, 280, 291, 292
229, 230
X
of Streptococcus mutans208, 220, 223226,
228230, 233, 234, 238239, 244 XIP. See SigX-inducing peptide
of Streptococcus pneumoniae 220, 233, 234, 240
using synthetic CSPs225227 Z
T-RFLP. See Terminal restriction fragment length Zoledronic acid449
polymorphism

Oral Biology - Molecular Techniques and Applications PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Oral Biology - Molecular Techniques and Applications PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Methods in

Molecular Biology 1537

For further volumes:

Molecular Techniques and Applications

Nicholas C.K. Heng

Nicholas C.K. Heng

ISSN 1064-3745 ISSN 1940-6029(electronic)

Library of Congress Control Number: 9781493967384

Springer Science+Business Media LLC 2017

Printed on acid-free paper

This Humana Press imprint is published by Springer Nature

It is widely accepted that evidence-based dentistry is fundamental to clinical practice and

Dunedin, New Zealand GregoryJ.Seymour

Part I Saliva and Other Oral Fluids

Part II Molecular Biosciences

12 Methods toStudy Antagonistic Activities Among Oral Bacteria . . . . . . . . . . . . 203

Part IIICells and Tissues

24 A Method toIsolate, Purify, andCharacterize Human Periodontal

SandraAfione Molecular Physiology and Therapeutics Branch, National Institute

PaulaM.Farthing Academic Unit of Oral and Maxillofacial Pathology, School

JewC.Kon Department of Pharmacology and Therapeutics, University of Melbourne,

EuanJ.Rodger Department of Pathology, Dunedin School of Medicine, University

Saliva and Other Oral Fluids

Salivary Diagnostics Using Purified Nucleic Acids

Key words Saliva, RNA, DNA, Nucleic acids, Stabilization, Exosomes

especially be taken with respect to processing and stabilizing saliva

preferable biofluid to urine, which is currently the method of

A report by Gallo etal. in 2012 [27] confirming that miRNAs

2. The DNASAL device (Oasis Diagnostics, Vancouver, USA)

the bottom of the Collection Vial containing a proprietary

tion followed by pelleting step to obtain purified RNA for

2.3 Exosomes 1. PureSAL Oral Specimen Collection Device (Catalog

2.4 Cell-Free DNA 1. PureSAL Oral Specimen Collection Device (Catalog

Recently, the PureSAL device has been compared to whole saliva

3.2 Isolation 1. Combine 1.7g of collected sample with 340L of ExoQuick-TC

sufficient centrifuge speeds, non-exosomal materials remaining in

3.3 Cell-Free DNA 1. Sample collection.

1. DNA from samples collected using one of the above commer-

6. The method used in this chapter for isolation of exosomes is

The author would like to acknowledge the support of Dr David T

(2001) Reliability assessment of soluble 30. See https://norgenbiotek.com/product/

RNA Sequencing Analysis ofSalivary Extracellular RNA

Extracellular RNA (exRNA) in human saliva is an emerging field for

studies characterizing ncRNAs in saliva have used RNA-sequencing

2.1 Saliva Collection 1. 50mL sterile tube.

2.2 RNA Sequencing 1. Qiazol.

2.2.2 DNase Treatment 1. DNase.

2.2.3 RNA Quantification 1. QuanTi RiboGreen RNA assay kit.

2.2.5 cDNA Library 1. Qubit dsDNA BR assay kit.

2.2.6 RNA Sequencing 1. EB buffer.

2.2.7 De-Multiplexing 1. Cutadapt.

2. Add 40L DNase Mix/sample (100L final volume=60 L

3.2.3 RNA Quantification 1. Prepare serial dilution of rRNA standards (12.5200ng/mL).

RNA 6000 Pico Chip, Bioanalyzer: detection of intact

Saliva RNA 5.5L

2. Incubate in a preheated thermal cycler for 2min at

RNA+ 3 SR adaptor mix 7L

4. Incubate for 1h at 25C in a thermal cycler.

3 Ligation reaction mix from step 1 20L

2. Heat samples for 5min at 75C.Transfer to 37C for 15min,

Reaction mix from step 2 25.5L

5. Incubate for 1h at 25C in a thermal cycler.

Perform Reverse Mix the following components in a sterile, nuclease-free tube:

Dunedin, New Zealand GregoryJ.Seymour