Volume 142

Number 5

September 3, 2010

www.cell.com

Reorienting
Planar Polarity
Minireview
Histone Crosstalk
bioreprogramming

Biopotential.
Unlock extraordinary potential with stem cell technologies from Sigma . ®

Stem cell biology offers astonishing

research potential; Sigma® Life Science has
the innovations you need to discover the
promise it holds. Access a world of RNAi with
the MISSION® RNAi Library, efficiently edit
genes of interest using advanced CompoZr®
ZFN technology, and characterize your stem
cells with our Prestige Antibodies®, powered
by Atlas Antibodies. The applications are
endless—and so are the possibilities.

wherebiobegins.com/bioreprogramming

Sigma, MISSION, CompoZr and Prestige Antibodies are registered trademarks of

Sigma-Aldrich and its affiliate Sigma-Aldrich Biotechnology, L.P.
 .
E W Intelligent design

N +
inCuSaFe™ copper enriched
stainless steel interior

+ Single-beam, dual capture

infrared CO2 sensor

+ SafeCell UV
protection in situ

+ Hydrogen peroxide
vapor sterilization in situ

1
Good laboratory technique

Sterisonic™ GxP
Performance and Productivity
Delivers Best Efficiency Value:

2 Hours 14 Hours

SANYO Brand X
Sterisonic™

H2O2 sterilization vs. high heat sterilization

= Uptime (Hours) = Downtime (Hours)

The rapid in situ H2O2 sequence returns the

fully sterilized Sterisonic™ GxP to normal use quicker

Spot on results.
than any competitive incubator in the world.

The industry’s first in situ H2O2 sterilization with the fastest turnaround.
For maximum productivity in clinical, general purpose or the most highly compliant GMP applications,
the new SANYO Sterisonic™ GxP CO2 incubator offers an impressive return on investment. With multiple
contamination control safeguards, exclusive on-board H2O2 sterilization, FDA-21CFR data capture and
graphical LCD display, the Sterisonic™ GxP rewards good laboratory technique with performance
you can trust. Learn more, visit www.sterisonic.com or call 800-858-8442.

pictured above: Sterisonic™ GxP MCO-19AICUVH with rapid H2O2 vapor sterilization system.

FREEER
OFF
[ D E TA IL
S
!
O N L IN E ]
FREE!
FREE!
H2O2 sterilization system acessory, plus BD labware consumables kit.
Limited time upgrade offer with purchase of the Sterisonic™ GxP MCO 19AICUVH. ($900 Value)
BD stem cell starter kit with Sterisonic™ GxP quote.
No purchase necessary! Act now. Supplies are limited. ($150 Value)
www.sterisonic.com
©2009 Sanyo Biomedical OWS 1015 05/09
 015.A1.0101.F.US © 2009 Eppendorf AG.
Background: electroporated L-mouse cells.
 Electroporation of
mammalian cells

 Electroporation
of bacteria

 Electroporation
of yeasts

 Cell fusion

Outstanding
transfection
rates!

One electroporator– many applications.

Our Multiporator ® is a compact, flexible system
designed for the electroporation of eukaryotic cells,
bacteria and other microorganisms, as well as the
electrofusion of cells.
Unique features include Soft Pulse microsecond pulses,
patented* electronically regulated pulses, a gentle
hypoosmolar buffer system and a choice of
electroporation and electrofusion options.
A safe and easy method for performing cell fusion:
 Microfusion chamber for viewing and optimization
of fusion parameters
 Helix chamber for larger cell volumes
 Applications include: monoclonal antibodies, plant
protoplast fusion, tumor research, cancer vaccines,
cloning mammals
For more information visit www.eppendorf.com

*U.S. Pat. 6,008,038

www.eppendorf.com • Email: info@eppendorf.com • Application hotline: 516-515-2258

In the U.S.: Eppendorf North America, Inc. 800-645-3050 • In Canada: Eppendorf Canada Ltd. 800-263-8715
 If your standards and your samples are the same size,
they must be the same mass, right?
That’s the problem with calibration-based measuring techniques. You don’t know—you can only assume. Which is why
every major pharmaceutical and biotechnology company, as well as most federal regulatory agencies are switching from relative
methods to Wyatt Technology’s absolute measurements. Our DAWN® multi-angle light scattering (MALS) instruments
allow you to determine absolute molecular weights and sizes without relying on so-called standards, or measurements made in
someone else’s lab. Wyatt instruments measure all of the quantities required for determining absolute molar masses directly.
If you’re ready to abandon the habit and hope of reference standards, call 805-681-9009 or visit wyatt.com to request our
free 28-page Ultimate Guide to Light Scattering. Our booklet will show you how
to make your reliance on standards a thing of the past. CORPORA
CORPORATION

DAWN HELEOS®. The most advanced Optilab rEX®. The refractometer with ViscoStar™. The viscometer with Eclipse. The ultimate system for the DynaPro Plate Reader. Automated
multi-angle light scattering instrument the greatest sensitivity and range. unparalleled signal-to-noise, stable separation of macromolecules and dynamic light scattering for 96 or
for macromolecular characterization. baselines and a 21st-century interface. nanoparticles in solution. 384 or 1536 well plates.

©2008 Wyatt Technology. DAWN, Optilab, DynaPro and the Wyatt Technology logo are registered trademarks, and ViscoStar and Eclipse are trademarks of Wyatt Technology Corporation.
BE THE FIRSTto read the latest issue of any Cell Press journal.

Register for Cell Press Email Alerts and get the complete table of contents as soon as
the issue publishes online — FREE!

Cell Press Email Alerts deliver the news, research, and commentaries featured in each
journal’s latest issue, including the full title of every article, direct links to the articles,
and the complete author list. Plus, to save you time, each research article has a brief
summary highlighting its significant findings.

You don’t have to be a subscriber to sign up for Cell Press Email Alerts. While subscribers have
instant access to the full text of all articles listed in the Email Alerts, non-subscribers can read
the abstracts of all articles as well as the full text of the issue’s Featured Article.

www.cellpress.com
Editor Editorial Board Douglas Green Dana Pe’er
Emilie Marcus Abul Abbas Leonard Guarente Kathrin Plath
C. David Allis Taekjip Ha Carol Prives
Senior Deputy Editor
Geneviève Almouzni Daniel Haber Klaus Rajewsky
Elena Porro
Uri Alon Ulrike Heberlein Venki Ramakrishnan
Scientific Editors Angelika Amon Nobutaka Hirokawa Rama Ranganathan
Karen Carniol Johan Auwerx Mark Hochstrasser Anne Ridley
Connie M. Lee Richard Axel Arthur Horwich James Roberts
Fabiola Rivas Cori Bargmann Tony Hunter Alexander Rudensky
Lara Szewczak Konrad Basler James Hurley Helen Saibil
Bonnie Bassler Richard Hynes Joshua Sanes
Leading Edge Editor
David Baulcombe Thomas Jessell Randy Schekman
Orla M. Smith
Jeffrey Benovic Narry Kim Ueli Schibler
Associate Editor Carolyn Bertozzi Mary-Claire King Joseph Schlessinger
Robert Kruger Wendy Bickmore David Kingsley Hans Schöler
Elizabeth Blackburn Frank Kirchhoff Trina Schroer
Assistant Editor Joan Brugge Richard Kolodner Geraldine Seydoux
Michaeleen Doucleff Lewis Cantley John Kuriyan Kevan Shokat
Senior Managing Editor Joanne Chory Robert Lamb Pamela Sklar
Meredith Adinolfi David Clapham Mark Lemmon Nahum Sonenberg
Andrew Clark Beth Levine James Spudich
Deputy Managing Editor Hans Clevers Wendell Lim Paul Sternberg
Jackie M. Smith Stephen Cohen Jennifer Lippincott-Schwartz Bruce Stillman
Illustration and Pascale Cossart Dan Littman Azim Surani
Design Supervisor Jeff Dangl Richard Losick Keiji Tanaka
Nathan E. Bress Ted Dawson Scott Lowe Craig Thompson
Pier Paolo di Fiore Tom Maniatis Robert Tjian
Lead Illustrator Marileen Dogterom Matthias Mann Jürg Tschopp
Andrew A. Tang Julian Downward Kelsey Martin Ulrich von Andrian
Bruce Edgar Joan Massagué Gerhard Wagner
Illustrators
Steve Elledge Iain Mattaj Detlef Weigel
Yvonne Blanco
Anne Ephrussi Satyajit Mayor Alan Weiner
Kate Mahan
Ronald Evans Ruslan Medzhitov Jonathan Weissman
Production Staff Witold Filipowicz Craig Mello Matthew Welch
Reyna Clancy Marco Foiani Tom Misteli Tian Xu
Elaine Fuchs Tim Mitchison Shinya Yamanaka
Editorial Assistant
Yukiko Goda Alex Mogilner Marino Zerial
Mary Beth O’Leary
Stephen Goff Paul Nurse Xiaowei Zhuang
Joe Goldstein Roy Parker Huda Zoghbi

Cell Office
Cell, Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139
Phone: (+1) 617 661 7057, Fax: (+1) 617 661 7061, E-mail: celleditor@cell.com
Online Publication: http://www.cell.com

Cell (ISSN 0092-8674) is published biweekly by Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139. The institutional subscription rate for
2010 is $1,360 (US and Canada) or $1,532 (elsewhere). The individual subscription rate is $202 (US and Canada) or $305 (elsewhere). The individual copy price is $50.
Periodicals postage paid at Boston, Massachusetts and additional mailing offices. Postmaster: send address changes to Elsevier Customer Service Americas,
Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA.
The paper used in this publication meets the requirments of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed by Dartmouth Printing Company, Hanover, NH.
 FINE SURGICAL INSTRUMENTS FOR RESEARCH TM

The Elements of Discovery

Scissors • Needle Holders • Forceps • Retractors • Clamps • Rongeurs • Probes
Scalpels • Surgical & Laboratory Accessories • Student Quality Instruments

R e q u e s t a c a t a l o g a t f i n e s c i e n c e. c o m o r c a l l 1 - 8 0 0 - 5 2 1 - 2 1 0 9

Scopus is the largest abstract and citation

database of peer-reviewed literature and
quality web sources with smart tools to
track, analyze and visualize research.

enrich
your
experience
www.scopus.com
Cell Press
President & CEO
Lynne Herndon
Editor in Chief, Vice President of Content Development
Emilie Marcus
Vice President of Marketing and Publishing
Els Bosma
Vice President of Web Development and Operations
Keith Wollman
Director of Marketing
Jonathan Atkinson
Production Supervisor
Meredith Adinolfi
Operations Supervisor
Daniel Fauxsmith
Press Officer
Cathleen Genova
ª2010 Elsevier Inc. All rights reserved.
This journal and the individual contributions contained in it are protected
under copyright by Elsevier Inc., and the following terms and conditions
apply to their use:
Photocopying:
Single photocopies of single articles may be made for personal use as al-
lowed by national copyright laws. Permission of the Publisher and payment
of a fee are required for all other photocopying, including multiple or system-
atic copying, copying for advertising or promotional purposes, resale, and
all forms of document delivery. Special rates are available for educational
institutions that wish to make photocopies for nonprofit educational class-
room use. For information on how to seek permission, visit www.elsevier.
com/permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US).
Permissions:
For information on how to seek permission, visit www.elsevier.com/
permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US).
Derivative Works:
Subscribers may reproduce tables of contents or prepare lists of articles
including summaries for internal circulation within their institutions.
Permission of the Publisher is required for resale or distribution outside
the institution. Permission of the Publisher is required for all other deriv-
ative works, including compilations and translations (please consult
www.elsevier.com/permissions).
Electronic Storage or Usage:
Permission of the Publisher is required to store or use electronically any
material contained in this journal, including any article or part of an article
(please consult www.elsevier.com/permissions). Except as outlined
above, no part of this publication may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, electronic, mechan-
ical, photocopying, recording, or otherwise, without prior written permis-
sion of the Publisher.
Notice:
No responsibility is assumed by the Publisher for any injury and/or damage
to persons or property as a matter of products liability, negligence, or other-
wise, or from any use or operation of any methods, products, instructions,
Display Advertising
Midwest/Southeast/Eastern Canada:
Inez Herrero, ph: 585 678 4395; fax: 585 678 4722;
e-mail: i.herrero@elsevier.com
Mid-Atlantic/New England: Vikki Macomber, ph: 508 928
1255; fax: 508 928 1256; e-mail: v.macomber@elsevier.com
West Coast/Western Canada/Asia: Lynne Stickrod, ph: 415
931 9782; fax: 415 520 6940; e-mail: l.stickrod@elsevier.com
UK/Europe: James Kenney, ph: +44 20 7424 4216; fax: +44 18
6585 3136; e-mail: j.kenney@elsevier.com
Classified Advertising
United States and Canada:
Gordon Sheffield, Key Account Manager, ph: 617 386 2189; fax:
617 397 2805; e-mail: g.sheffield@elsevier.com
UK, Europe, and Asia:
Sabrina Dodge, Key Account Manager, ph: +44 20 7424 4997;
fax: +44 18 6585 3136; e-mail: s.dodge@elsevier.com
or ideas contained in the material herein. Because of rapid advances in
the medical sciences, in particular, independent verification of diagnoses
and drug dosages should be made. Although all advertising material is
expected to conform to ethical (medical) standards, inclusion in this
publication does not constitute a guarantee or endorsement of the quality
or value of such product or of the claims made of it by its manufacturer.
Reprints:
Article reprints are available through Cell’s reprint service; for informa-
tion, contact Nicholas Pavlow (e-mail: n.pavlow@elsevier.com; ph: (+1)
212 633 3960).
Subscription Orders and Inquiries:
Mail, fax, or e-mail address changes to Elsevier Customer Service Amer-
icas, allowing 4–6 weeks for processing. Lost or damaged issues will be
replaced, subject to availability, if Cell Press is notified within the claim
period (US and airmail delivery: 3 months from issue date; surface deliv-
ery: 4 months from issue date). Periodical delivery in the US can take up
to 3 weeks. Airmail delivery can take 2–4 weeks.
The price of a single copy of Cell is $50 (excluding special issues). All
orders must be prepaid and in writing. Please include the volume and is-
sue number, payment (check or credit card, MasterCard, Visa, or Amer-
ican Express only), and a delivery address. Allow 4–6 weeks for delivery.
Mailing address: Elsevier Customer Service Americas, Cell Press
Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146,
USA. Toll-free phone within USA/Canada: 866 314 2355; phone for
outside US/Canada: (+1) 314 453 7038; fax: (+1) 314 523 5170; e-mail:
subs@cell.com; internet: www.cellpress.com or <www.cell.com>.
Funding Body Agreements and Policies:
Elsevier has established agreements and developed policies to allow au-
thors whose articles appear in journals published by Elsevier to comply
with potential manuscript archiving requirements as specified as condi-
tions of their grant awards. To learn more about existing agreements and
policies, visit www.cellpress.com.
Guide for Authors:
For a full and complete guide for authors, please go to www.cell.com/
authors.
 REMARKABLY FAST

Phusion® High-Fidelity DNA Polymerase

With Phusion® High-Fidelity DNA Polymerase, there is no need to compromise any aspect of your PCR performance.
A superior choice for cloning, this recombinant polymerase has an error rate 50-fold lower than Taq DNA Polymerase,
combining extreme precision with unparalleled speed and robustness.

Advantages:
Not all PCR polymerases are created equal
• Extreme fidelity • High yield
• High speed • Specificity
• Robustness

Go to www.neb.com/phusion to find out

how Phusion High-Fidelity DNA Polymerase can
improve your PCR performance.

Produced by Distributed by
Amplification of a 3.8 kb fragment from the human beta globin gene clearly illustrates the
extreme speed and robustness offered by using Phusion DNA Polymerase. Reactions were
performed according to the suppliers’ recommendations using varying extension times
(shown above gel).

Phusion is a registered trademark of Finnzymes Oy.

DNA AMPLIFICATION PROTEIN EXPRESSION GENE EXPRESSION

CLONING & MAPPING RNA ANALYSIS
& PCR & ANALYSIS & CELLULAR ANALYSIS
www.neb.com
Leading Edge
Cell Volume 142 Number 5, September 3, 2010

IN THIS ISSUE
MOLECULAR BIOLOGY SELECT

ESSAY
661 Phosphotyrosine Signaling: Evolving a New W.A. Lim and T. Pawson
Cellular Communication System

CORRESPONDENCE

668 What Controls T Cell Receptor R.A. Fernandes, C. Yu, A.M. Carmo, E.J. Evans,
Phosphorylation? P.A. van der Merwe, and S.J. Davis
669 Response: Multilayered Control of T Cell E. Gagnon, C. Xu, W. Yang, H.H. Chu, M.E. Call,
Receptor Phosphorylation J.J. Chou, and K.W. Wucherpfennig

PREVIEWS
672 Fishing Out a Sensor for Anti-inflammatory Oils A.R. Saltiel

674 A New Spin on Planar Cell Polarity P. Olguin and M. Mlodzik

676 Viable Rat-Mouse Chimeras: D. Solter
Where Do We Go from Here?

679 ‘Fore Brain: L.B. Sweeney and L. Luo

A Hint of the Ancestral Cortex

MINIREVIEW

682 The Language of Histone Crosstalk J.-S. Lee, E. Smith, and A. Shilatifard

SNAPSHOT

822 Nuclear Receptors I N.J. McKenna and B.W. O’Malley

 www.phosphosite.org

A Comprehensive Online
Protein Modification Resource
provided by Cell Signaling Technology with grant support from the NIH

PhosphoSitePlus® v3.0 (PSP) is an open web
resource that integrates encyclopedic information
on experimentally determined protein modification
sites, upstream and downstream regulation of
these modifications, and powerful analytical tools
for investigating the structural and biological
significance of protein modifications.
Cell Signaling Technology® and PhosphoSitePlus® are trademarks of Cell Signaling Technology, Inc.
© 2010 Cell Signaling Technology, Inc.
• Expansive and continuously curated content
• Molecular rendering to visualize the location of modification sites
• On-the-fly generation of kinase substrate sequence logos
• Browsing of high-throughput content by disease, cell line, and tissue
• New search interfaces that retrieve modification sites and proteins by
subcellular locations, sequence and motifs, domains, responsiveness
to treatments, disease, tissue, and cell type

Unparalleled product quality, validation and technical support.

www.cellsignal.com
Orders (toll-free) 1-877-616-2355 | Technical support (toll-free) 1-877-678-8324 support@cellsignal.com | Inquiries info@cellsignal.com | Environmental Commitment eco.cellsignal.com
Articles
Cell Volume 142 Number 5, September 3, 2010
687 GPR120 Is an Omega-3 Fatty Acid Receptor
Mediating Potent Anti-inflammatory
and Insulin-Sensitizing Effects
699 Anti-CD47 Antibody Synergizes with
Rituximab to Promote Phagocytosis and
Eradicate Non-Hodgkin Lymphoma
714 A C-Type Lectin Collaborates with a
CD45 Phosphatase Homolog to Facilitate
West Nile Virus Infection of Mosquitoes
726 The ATAC Acetyltransferase Complex
Coordinates MAP Kinases
to Regulate JNK Target Genes
737 A Bacterial mRNA Leader that Employs
Different Mechanisms to Sense
Disparate Intracellular Signals
749 Structural Basis of Semaphorin-Plexin
Recognition and Viral Mimicry from
Sema7A and A39R Complexes with PlexinC1
762 Photoadaptation in Neurospora by
Competitive Interaction of Activating
and Inhibitory LOV Domains
773 Cell Flow Reorients the Axis
of Planar Polarity in the
Wing Epithelium of Drosophila
787 Generation of Rat Pancreas in Mouse
by Interspecific Blastocyst Injection
of Pluripotent Stem Cells
D.Y. Oh, S. Talukdar, E.J. Bae, T. Imamura,
H. Morinaga, W. Fan, P. Li, W.J. Lu,
S.M. Watkins, and J.M. Olefsky
M.P. Chao, A.A. Alizadeh, C. Tang, J.H. Myklebust,
B. Varghese, S. Gill, M. Jan, A.C. Cha, C.K. Chan,
B.T. Tan, C.Y. Park, F. Zhao, H.E. Kohrt, R. Malumbres,
J. Briones, R.D. Gascoyne, I.S. Lossos, R. Levy,
I.L. Weissman, and R. Majeti
G. Cheng, J. Cox, P. Wang, M.N. Krishnan,
J. Dai, F. Qian, J.F. Anderson, and E. Fikrig
T. Suganuma, A. Mushegian, S.K. Swanson,
S.M. Abmayr, L. Florens, M.P. Washburn,
and J.L. Workman
S.-Y. Park, M.J. Cromie, E.-J. Lee,
and E.A. Groisman
H. Liu, Z.S. Juo, A.H.-R. Shim, P.J. Focia, X. Chen,
K.C. Garcia, and X. He
E. Malzahn, S. Ciprianidis, K. Kaldi, T. Schafmeier,
and M. Brunner
B. Aigouy, R. Farhadifar, D.B. Staple, A. Sagner,
J.-C. Ro€per, F. Ju€licher, and S. Eaton
T. Kobayashi, T. Yamaguchi, S. Hamanaka,
M. Kato-Itoh, Y. Yamazaki, M. Ibata, H. Sato,
Y.-S. Lee, J.-i. Usui, A.S. Knisely, M. Hirabayashi,
and H. Nakauchi
(continued)
Let the Countess™ do the counting.
Wow! Consistent, fast, and easy.
The Countess™ Automated Cell Counter takes the subjectivity and tedium out
of one of the fundamental steps of cell culture—counting live and dead cells.
The Countess™ automated technology reduces user error and speeds up cell
counting with less sample, less hassle, and a lower cost than any automated cell
counter on the market. Start counting today at www.invitrogen.com/countcells.

THE MOST

For research use only. Not intended for any animal or human therapeutic or diagnostic use, unless otherwise stated.
© 2009 Life Technologies Corporation. All rights reserved. The trademarks mentioned herein are the property of Life Technologies Corporation or their respective owners. These products may be covered by
one or more Limited Use Label Licenses (see Invitrogen catalog or www.invitrogen.com). By use of these products you accept the terms and conditions of all applicable Limited Use Label Licenses.
800 Profiling by Image Registration R. Tomer, A.S. Denes, K. Tessmar-Raible,
Reveals Common Origin of Annelid and D. Arendt
Mushroom Bodies and Vertebrate Pallium

RESOURCE

810 The Protein Composition of Mitotic S. Ohta, J.-C. Bukowski-Wills, L. Sanchez-Pulido,

Chromosomes Determined Using F. de Lima Alves, L. Wood, Z.A. Chen, M. Platani,
Multiclassifier Combinatorial Proteomics L. Fischer, D.F. Hudson, C.P. Ponting, T. Fukagawa,
W.C. Earnshaw, and J. Rappsilber

POSITIONS AVAILABLE

On the cover: Many epithelial tissues orient external structures such as cilia and hairs, and to
do so they use the planar cell polarity (PCP) proteins, which form polarized cortical domains
at epithelial cell junctions. The mechanisms that globally orient planar polarity have been
mysterious. In this issue, Aigouy et al. (pp. 773–786) show that the pattern of PCP domain
orientation in the Drosophila wing is a consequence of the oriented cell rearrangements and
divisions that shape this tissue.
 R&D Systems Tools for Cell Biology Research™

Proteins
R&D Systems Quantikine® ELISAs
Antibodies

ELISAs
The Most Referenced Immunoassays
A direct measure of product quality is the frequency of citations in the scientific literature. R&D Systems has more
Assay Services
than 20 years of experience designing, testing, and optimizing the most cited ELISA kits in the world. Find out why
MultiAnalyte Profiling scientists trust R&D Systems ELISAs more than any other brand.

Activity Assays R&D Systems is the Most Referenced ELISA Manufacturer NEW Quantikine ELISA Kits
 α1-Acid Glycoprotein
Stem Cells  Angiopoietin-like 3
 Cathepsin V
ELISpot Kits  Clusterin
17.3%  Dkk-1
R&D Systems  EGF R/ErbB1
Flow Cytometry 41.8% 7.9%  EG-VEGF/PK1
 Fetuin A
Cell Selection 3.7%  FGF-21
 Galectin-3
2.5%  Gas 6
2.1%  GDF-15
1.6%  IL-17A/F Heterodimer
1.4%
 IL-19
 Lipocalin-2/NGAL
 MBL
 Proprotein Convertase 9/PCSK9
Approximately 42% of Referenced Immunoassays are Developed and Manufactured by  Periostin/OSF-2
R&D Systems. A survey of 860 manuscripts from 44 journals was conducted to compare the  Progranulin
number of citations specifying the use of R&D Systems ELISAs to the number citing ELISAs from  ST2/IL-1 R4
other commercial sources. A total of 433 ELISA citations referencing immunoassays from 66  Thrombomodulin/CD141
different vendors were identified in the survey.  Tie-1
 TIM-1/KIM-1

For more information visit our website at www.RnDSystems.com/go/ELISA

For research use only. Not for use in diagnostic procedures.

R&D Systems, Inc. www.RnDSystems.com

R&D Systems Europe, Ltd. www.RnDSystems.co.uk

R&D Systems China Co., Ltd. www.RnDSystemsChina.com.cn

Leading Edge

In This Issue

Fish Oil Turns the Tide on Insulin Resistance

PAGE 687
Inflammation mediated by macrophages promotes insulin resistance in obesity. Oh et al. now identify the G protein-coupled
receptor 120 (GPR120) on macrophages and fat cells as a receptor for omega-3 fatty acids (u-3 FAs). The authors show that
GPR120 activation by u-3 FAs inhibits inflammation pathways in macrophages and can reverse insulin resistance in mice.
These results provide a molecular basis for the anti-inflammatory effects of u-3 FAs and suggest that anti-inflammatory
treatments may ameliorate insulin resistance in obesity.

West Nile Virus Stopped at Its Source

PAGE 714
West Nile virus, a potentially deadly virus in humans, propagates in mosquitoes.
Cheng et al. now find that infection of West Nile virus triggers mosquito cells to
produce the lectin protein mosGCTL-1. This C-type lectin enhances entry of the
virus into additional mosquito cells through its interaction with a protein tyrosine
phosphatase receptor, which is homologous to human CD45. Blocking
mosGTCL-1 with an antibody disrupts the infective cycle of West Nile virus in
mosquitoes, suggesting a new strategy for controlling viral dissemination.

Antibodies Are Double-Trouble for Cancer

PAGE 699
Monoclonal antibodies are standard therapeutics for several cancers, including the
anti-CD20 antibody rituximab for B cell non-Hodgkin lymphoma (NHL). However,
antibodies are not curative and must be combined with cytotoxic chemotherapy
for clinical benefits. Now, Chao et al. identify CD47 as an antibody target in NHL
and demonstrate that combining anti-CD47 antibody with the rituximab antibody eradicates human NHL in mice. The synergistic
mechanism used by these two antibodies may be applicable to combined antibody treatments for many types of cancers.

ATAC Wears Two HATs in MAPK Signaling

PAGE 726
Extracellular cues often trigger MAP kinase (MAPK) signaling pathways, which
then activate downstream transcription factors like c-Jun. Here, Suganuma
et al. demonstrate that the ATAC histone acetyltransferase (HAT) governs the
response to MAPK signaling by serving as both a coactivator of transcription
and a suppressor of upstream signaling in the MAPK pathway. The authors
show that ATAC acetylates histone H4 at JNK target genes, which then serves
as a positive cofactor for basal transcription. In addition, ATAC directs upstream
MAPKs to the site of c-Jun binding and restricts the levels of JNK activation.

A Multitasking Leader mRNA

PAGE 737
Bacterial mRNAs often contain leader sequences that regulate transcription
of the adjacent coding region by binding metabolites and ions. For example,
the leader of the mRNA for the Salmonella Mg2+ transporter gene mgtA responds
to Mg2+. Now, Park et al. demonstrate that this leader also contains a short open reading frame with many proline codons, trans-
lation of which places mgtA expression under the control of cytoplasmic proline concentrations as well as Mg2+. Thus, leader
mRNAs can use distinct mechanisms to sense multiple intracellular signals.

Vivid Memories of Days Gone By

PAGE 762
Light responses and photoadaptation in the fungus Neurospora depend on the circadian transcription factor White Collar
Complex (WCC) and its negative regulator Vivid (VVD). Mazhan et al. now demonstrate how WCC and VVD cooperate to
discriminate light intensities over more than five orders of magnitude during the day. At night, previously synthesized VVD
serves as a molecular memory of the sun’s brightness during the preceding day and suppresses responses to light cues
of lower intensity, such as moonlight.

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 649

Cancer Research
Detect and Characterize Genomic Variations
Use powerful Roche instruments and consumables for
454-Ultra Deep Sequencing
high-throughput analysis of copy number variation using
NimbleGen CGH microarrays. Rely on 454 Sequencing
Systems, with their long read lengths (400 to 500 bp). Or
combine both technologies using a Sequence Capture workflow.

 Detect chromosomal imbalances down to approximately

5 kb, commonly found in tumor progression, using high-
qPCR Analysis using the LightCycler® System
density NimbleGen CGH microarrays.
V617F Wild type  Identify all important genomic and transcriptomic
and LOH variants by ultra-deep sequencing of PCR products using
the 454 Sequencing System.
Mutation

Visit www.cancer-research.roche.com for application notes

and more detailed information about using Roche products in
cancer research.
Verification of molecular mutations. The JAK2 V617F mutation is characteristic
in hematopoietic malignancies. Ultra-deep 454 Sequencing detected the
mutation in just 1.16% of reads (top panel). Sensitive Carousel-Based LightCycler®
melting curve analysis confirmed the mutation in approximately
1% of cells (lower panel).

Data provided by A. Kohlmann, MLL Munich Leukemia Laboratory GmbH, Germany.

View the entire Application Note No. 7 at www.cancer-research.roche.com.

For life science research only. Not for use in diagnostic procedures.
454, 454 SEQUENCING, LIGHTCYCLER, and NIMBLEGEN are trademarks of Roche. Roche Diagnostics GmbH
Other brands and product names are trademarks of their respective holders.
Roche Applied Science
© 2010 Roche Diagnostics GmbH. All rights reserved. 82372 Penzberg, Germany
Cell Polarity Goes with the Flow
PAGE 773
Planar polarization of epithelial cells allows the uniform alignment of hairs, cilia, and other cellular structures with tissue shape.
Now, Aigouy et al. combine experimental and theoretical approaches to show that polarity patterns in the Drosophila wing arise
during growth. Specifically, cell polarity is reoriented from a radial to a proximal-distal axis when mechanical stresses during
growth cause the cells to rotate or ‘‘flow’’ with respect to each other. Linking planar polarity to morphogenesis provides a simple
mechanism for coordinating the global polarity pattern with tissue shape.

Com-plexin’ with Semaphorins

PAGE 749
Semaphorins and their receptors, Plexins, are widely expressed protein families
that mediate repulsive signaling during cell guidance. Here, Liu et al. present two
X-ray crystal structures of PlexinC1, one in complex with the Semaphorin
Sema7A and another in complex with the Semaphorin mimetic A39R from the
smallpox virus. In both structures, the Semaphorin interacts with a PlexinC1
dimer in a novel edge-on, orthogonal geometry. These findings suggest that Plex-
ins are activated by ligand-induced dimerization during cell guidance.

Interspecies Organogenesis
PAGE 787
A goal of regenerative medicine is to derive organs from a patient’s pluripotent
stem cells (PSCs), but in vitro organogenesis is complex. Here, Kobayashi et al.
generate a functioning rat pancreas in mice without a pancreas by injecting rat
PSCs into mouse blastocysts. These interspecific chimeras provide proof of
principle for in vivo generation of organs derived from donor PSCs and for inter-
specific blastocyst complementation.

A Wormhole to the Origin of the Cortex

PAGE 800
The cerebral cortex or pallium controls the highest-order processing in mammals, but its evolution remains enigmatic. Now,
Tomer et al. develop an expression profiling technique to generate a gene expression map for the developing brain of the
segmented worm Platynereis dumerilii. Comparison of this map with that observed for the developing cerebral cortex
suggests a common evolutionary origin for the mammalian cortex and the worm’s mushroom body, a brain region in inver-
tebrates that processes sensory input.

Will the Real Chromosomal Proteins Please

Stand Up
PAGE 810
Proteomic analysis of large cellular structures is frequently hindered by the
presence of contaminants. In their analysis of mitotic chromosomes, Ohta
et al. overcome this problem by integrating proteomics with additional quanti-
tative and bioinformatic data—effectively adding a final purification step in silico.
This approach successfully pinpoints hitchhikers from amidst the 4,000 iden-
tified proteins and provides insight into the functional relationships among the
genuine constituents, including evidence that many more kinetochore-associ-
ated proteins exist than recognized previously.

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 651

Next-gen GWAS.
NOW.
It’s a content revolution.
Up to 50% more coverage of common and rare
variants than all other arrays. Maximum power for
any population.

The Omni family of microarrays can propel

your studies into true next-gen GWAS. With
a clearly defined product path to the future.
Immediate utility. Future flexibility.

Get on the path to next-gen GWAS.

Now is the time. Get started at

www.illumina.com/GWAS
Leading Edge

Molecular Biology Select

The central importance of the tumor suppressor Retinoblastoma protein (Rb) in cell-cycle progression makes
its regulation a focal point for diverse biological processes, as evidenced by recent work described in this
issue’s Molecular Biology Select. These findings reveal new insight into Rb’s involvement in tissue regener-
ation and differentiation, as well as previously unrecognized mechanisms of Rb regulation.

Short-Term Inactivation, Lasting Benefits

Tissue regeneration in mammals is greatly restricted, whereas
many other types of vertebrates display astonishing feats of tissue
replacement, including regrowth of major structures such as
limbs. According to new findings by Pajcini et al. (2010), mamma-
lian innovations in the realm of tumor suppression are one factor
that is likely to contribute to this pronounced deficiency. The
authors show that combined inactivation of the tumor suppressors
retinoblastoma (Rb) and ARF reverses the differentiation of mouse
muscle cells, turning postmitotic cells into proliferating myoblasts,
and when these induced myoblasts are transplanted into donor
mice they successfully fuse into existing myofibers. Prior work in
newts has shown that Rb phosphorylation, which inactivates the
Dedifferentiated myocytes redifferentiate and fuse protein, promotes the reentry of myotubes into the cell cycle, a crit-
into existing muscles in vivo (visualized with green
ical initiating event in limb regeneration. The inspiration to addition-
fluorescent protein).
ally inactivate ARF was motivated by available evidence suggesting
its potential exclusivity to mammals and birds, and thus the authors
reasoned that ARF might be particularly important in mediating differences in regeneration potential between
vertebrate species. Could a similar strategy of Rb and ARF inactivation be used to promote cell-cycle entry of
endogenous cells for regeneration therapy? Although this remains to be seen, the observation that only tran-
sient inactivation of these tumor suppressors is needed for successful creation of regenerative cells may go
some way toward allaying concerns that such an approach would invariably promote cancer. Future efforts
are also likely to address whether this intervention triggers cell-cycle reentry for a range of mammalian cell
types.
K.V. Pajcini et al. (2010). Cell Stem Cell 7, 198–213.

Fat Chance for Bone Formation

Previous in vitro studies suggested that Rb plays a critical role in the
decision of meschenchymal cells to become adipocyte or bone
cells, but in vivo evidence for this hypothesis has been lacking.
Calo et al. (2010) now show that Rb gives meschenchymal cells
the extra nudge they need to commit to becoming bone-forming
osteoblasts. Without Rb, these cells are more likely to differentiate
into brown fat cells, leading to reduced levels of calcified bones
and increased levels of brown adipose tissue in mice. To sort out Deletion of Rb in the embryo proper using Meox2-
how Rb regulates the fate of the meschenchymal cells during devel- Cre reduces the level of calcified bone as detected
opment, the authors engineered mice with the RB1 gene and/or by Alizarin Red staining. Image courtesy of J. Lees.
the p53 gene deleted only in uncommitted meschenchymal cells.
As expected, animals missing the tumor suppressor p53 develop multiple types of tumors, including osteosar-
comas. Combining the p53 mutation with a deletion of one RB1 allele increases the frequency of osteosarcomas,
whereas deletion of both RB1 alleles shifts the tumor distribution away from osteosarcomas towards brown fat
tumors. Thus, Rb regulates the fate of meschenchymal cells in a dose-dependent manner. These results are
surprising given that the majority of human osteosarcomas contain RB1 mutations. The authors speculate
that osteosarcoma tumors most likely arise from cells already committed to becoming osteoblasts, and muta-
tions in RB1 promote dedifferentiation of these cells and thus tumorigenesis.
E. Calo et al. (2010). Nature. Published online August 4, 2010. 10.1038/nature09264.

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 653

 .%7

Reach Your Ideal Candidate!

så0OSTåONEåORåMULTIPLEåå
JOBåOPENINGSå

så0ROMOTEåANåUPCOMINGåå
CONFERENCEåORåEVENTå

så!NNOUNCEåAåGRANTAWARDå

så%LEVATEåYOURåå
ORGANIZATIONSåPROlLE

careers.cell.com

X-ray crystal structure of Rb (magenta) in complex
with the PP1 catalytic domain (gray). Image cour-
tesy of S. Rubin.
Rb Gets Caught in a Custody Battle
Rb is caught in a tug-of-war between cyclin-dependent kinases
(Cdks) that inactivate it by phosphorylation to permit cell-cycle
progression and phosphatases that remove the modification
to promote cell-cycle arrest. New findings of Hirschi et al. (2010)
demonstrate that this conflict is waged over the same binding
interface of Rb. They provide structural evidence that the phospha-
tase PP1 interacts with Rb in a region previously shown critical for
the interaction of Rb with Cdks. They further show that PP1 can
suppress the activity of Cdk2-cyclin A toward Rb to block cell-
cycle progression in a human osteosarcoma cell line, and that
the complex of PP1 and Rb appears to be stable, or at least
more prevalent, at mitotic exit. Among the interesting questions
this work raises is, what factors determine the outcome of PP1
and Cdk competition? The authors propose that concentration
and subcellular localization of the competing proteins likely play
a role, but it remains unclear how this molecular dispute is settled
under specific biological circumstances, for example after DNA
damage.
A. Hirschi et al. (2010). Nat. Struct. Mol. Bio. Published online August
8, 2010. 10.1038/nsmb.1868.

Methylation Moves to the Front of the Line

Although first reported more than 30 years ago, methylation
of protein N termini (a-N-methylation) remains a poorly understood
protein modification. Thus, the recent identification by Schaner-Too-
ley et al. (2010) of an enzyme that catalyzes the reaction promises to
accelerate understanding of the modification’s functions. The
authors use methylation of a known target of a-N-methylation called
RCC1 (a Ran guanine nucleotide exchange factor) as an indicator of
the presence or absence of the enzymatic activity from fractionated
HeLa cell nuclear extracts. Fractions with methyltransferase activity
were subjected to mass spectrometry, leading to the identification of
the methyltransferase responsible, which the authors name NMRT
(N-terminal RCC1 methyltransferase). The NMRT crystal structure
facilitated the modeling of substrate recognition, and further analysis
defining the consensus sequence for target recognition suggested
Rb as a potential substrate. Subsequent assays provide evidence NMRT structure with RCC peptide modeled in the
active site. Image courtesy of I. Macara.
that Rb is modified by a-N-methylation, at least in some cell types.
a-N-methylation of RCC1 promotes stable association with chro-
matin, and loss of NMRT or the absence of the RCC1 methylation results in defects in mitosis. In contrast, the
purpose of Rb a-N-methylation remains a tantalizing mystery. Regardless of whether Rb a-N-methylation is rele-
vant to its roles in cell-cycle control, the discovery of NMRT opens a door through which others will likely follow.
Are there other a-N-methyltransferases? And if so, is their substrate specificity similar to NMRT? The answer may
be an indicator of whether this modification is an exotic posttranslational event or might instead be considerably
more common than previously appreciated. Another compelling question is whether there are a-N-demethylases
that reverse the modification.
C.E. Schaner-Tooley et al. (2010). Nature. Published online July 29, 2010. 10.1038/nature09343.

Robert P. Kruger

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 655

 ,OOKING¬FOR¬WHAT¬#ELL¬
HAS¬PUBLISHED¬IN¬¬
)MMUNOLOGY¬LATELY


PHOIDå#ELLS åå åPPån

ONåINå,YM Då-OTIF
Så!CTIVATI Lå  å*AN OSINE "ASE
CTERIZEå2A SE å* 0 å# EL IC å4 YR
Så#HARA OO LA SM
å(YSTERESI KRABORTYå!
+ å2 Nå#YTOP n
NALINGåAND EISSå!å#HA å#$EPSILO  åPPå
$IGITALå3IG OV ER Nå # å9ANGå-å7 N E å" IN D INGåOFåTHE Lå   å. OVåå
ANå*å' ICå-EMB RA å# EL
-å:IKHERM åBYå$YNAM RPFENNIGå+7 LOPMENT
$ASå*å(Oå !CTIVATION **å7UCHE Cå#ELLå$EVE HUANGå9å2
EIZISå
Rå å#6å#HOUå RI TI
Få4 C E LLå2 E C E P TO
IE TE RS å# $ å# AR M AN
A C YT O ID å$ E N D
OL TE Rå7 å2AUCHå!å:
2EGULATIO
NåO LLå*2å3CH W RåOFå0LASM 3å+ANTå3'
å(
NONå%å#AL
Lå-%å3CHNE Cå2EGULATO å(OLLANDERå.
8Uå#å'AG E N TI A LåA NDå3PECIl RG å$ å: WEIERå#åDEN
åISåANå%SS LEYå2å(OLM
BE ISEASE
RI P TI O N å& ACTORå%  å4å3 CH EUå3å,OCKS A M M A TO RYå"OWELå$
4RANSC AE DA UMANå)N m
EHNERå-å- å2ISKåFORå( BERGå23åå
ATONå-,å, 
RSå'ENETIC ERå,(å"LUM
#ISSEå"å# å å   åPPån N D å# O N FE å3 CH RE IB ER å3å'LIMCH
"å#ELLå
å/CT ATIONåA å(IGGINSå$
%
ALå)NmAMM ENHUISå%%
2 å3 TR E SS åTOå)NTESTIN SS IG å3 å4 ILGå(å.IEUW
Så% KMANå*.å:
EI
8"0å,INK NKEå!å'LIC
EEå!(å&RA 
+ASERå!å, åPPån
EPåå MUNITY n
#EL Lå   å3 åOFå!UTOIM  åPPå EOSTASIS
)N TR IN SI C å)NITIATION   å! UGåå MUNEå(OM
VE N TS å# E LL Vå 2 å# EL Lå
IN TA IN Så )M
4REXå0RE EIDMANNå4
å-EDZH ITO THATå-A å#HENå9(åå
å+Oå*3å( å)MMUNITYå (å9ANGå8
3TETSONå$" N A TE åA N Då!DAPTIVE å(ILLIARDå"å(Uå3å3HENå
Få)N Gå,å8Uå, RACTIONS
EGULATORåO å3UNå*å+ON OMICå)NTE
EGATIVEå2 ILLIARDå!å,Iå, ANGEå'EN
4)0%åAå. OD Yå 2 * å( Rå , O N G 2
Gå3å#ARM TIONSåFO REå#åå
3UNå(å'ON åPPån
 Så)MPLICA CHå4!å-UR
  å- AYåå å( E A VY # HAINå,OCU å' RO SV ELDå&'å+NO
Lå GLOBULIN **
#EL
Eå)MMUNO ANå$ONGENå
å2IBLETå2åV JURY
U CTUREåOFåTH å#UTCHINå3 TEå,UNGå)N
4H E å $ å3 TR
å: EL M å- # å0 EA Kå - -
E Yå 0 A TH W AYåOFå!CU Gå ( å, IUå(å3UNå9
åå
ALAå3åVAN n GåASåAå+ NGå9(å7AN NINGERå*-åå
*HUNJHUNW å PP å  å3 IG N A LIN ZE Nå 2 å, EU Rå# * å0 EN
å!PRåå
 ECEPTORå VAå-å6ELDH
UI
IANGå#å"IN
DE
#ELLå å4OLL ,IKEå2 å'å%RMOLAE ICHOLLSå*å*
D A TI VE å3 TRESSåAND ER KM AN Nå 4åVANå,OO ST å- å( OL MDAHLå2å.
TIONåOFå/XI Iå2å0 VI
AN -ALHAM KIRAå3å(ULTQ
)DENTIlCA 'å9AGHUBI EIRISå*3å3LU
TSKYå!3å!
å9 å+ UB Aå +å.EELYå' AV AR Iå3 å0
)MAI å-ECHå#å"
-å+OPFå-
0ASPARAKISå å   åPPå
n
 å! PR å 
#EL Lå 

,OOKå!GAIN
$ISCOVERå-ORE
WWWCELLCOM
“ TRUSTED PARTNERS
LIKE MILLIPORE ARE
RARE BUT CENTRAL
TO OUR SUCCESS.”

Dr. Michael West

CEO, BioTime, Inc.
Renowned thought leader in
stem cell therapeutics

Millipore is the source for sterile filtration you trust.

Stericup® and Steritop® sterile filters combine the superior flow rates and low non-
specific binding you need to protect everything you’ve poured into your cell culture work.
With over 50 years of filtration expertise, Millipore continues to set the industry standard
for high-performance membrane technology and its application in sterile filtration.

Find out
Go to themore atat
source
www.millipore.com/trustStericup1
www.millipore.com/amicon

Millipore, Advancing Life Science Together, Stericup and Steritop are registered trademarks of Millipore Corporation.
The M mark is a trademark of Millipore Corporation. ©2010 Millipore Corporation. All rights reserved. LS-SBU-10-03205
Stay Connected
Join ASHG Today!
Benefits of Membership:
• Subscription to AJHG, now published
by Cell Press
• Reduced registration fees for the
Annual Meeting
• Discounted subscriptions to Cell
Press journals
• Access to educational materials
• Influence on public policy process

www.ashg.org
MAGPIX®—THE NEW WAVE IN MULTIPLEXING
The magnetic-bead, low-cost platform that attracts multiplexing to
every lab. There’s never been a better time to jump in.
Accelerate your research at a price you can afford, using the xMAP® technology you trust.
The new MAGPIX instrument easily detects up to 50 analytes simultaneously in a small sample.
To complete this flexible solution, choose from Millipore’s wide selection of magnetic bead
kits encompassing metabolic disease, inflammation, toxicity, neuroscience and cell signaling.

ADVANCING LIFE SCIENCE TOGETHER® Catch the wave at

Research. Development. Production. www.millipore.com/MAGPIX2

Millipore and Advancing Life Science Together are registered trademarks of Millipore Corporation. The M Mark
is a trademark of Millipore Corporation. MAGPIX and xMAP are registered trademarks of Luminex Corporation.
©2010 Millipore Corporation. All rights reserved.
 ¬,OOK¬!GAIN

$ISCOVER¬-ORE
s¬!CCESS¬TO¬THE¬¬#ELL¬0RESS¬PRIMARY¬RESEARCH¬JOURNALS¬AND¬
¬4RENDS¬REVIEWS¬TITLES¬ALL¬ON¬THE¬SAME¬PLATFORM
s¬)MPROVED¬MORE¬ROBUST¬ARTICLE¬AND¬AUTHOR¬SEARCH
s¬6IDEO¬ANIMATIONS¬AND¬SOUND¬lLES
s¬%ASY¬TO¬NAVIGATE¬HOME¬PAGE¬ARTICLES¬PAGES¬AND¬ARCHIVE

WWWCELLCOM
Leading Edge
Essay

Phosphotyrosine Signaling: Evolving a New

Cellular Communication System
Wendell A. Lim1,2,* and Tony Pawson3,4,*
1Howard Hughes Medical Institute
2Department of Cellular and Molecular Pharmacology
University of California, San Francisco, San Francisco, CA 94158, USA
3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario M5G 1X5, Canada
4Department of Molecular Genetics, University of Toronto, Toronto, Ontario M5S 1A8, Canada

*Correspondence: lim@cmp.ucsf.edu (W.A.L.), pawson@lunenfeld.ca (T.P.)

DOI 10.1016/j.cell.2010.08.023

Tyrosine phosphorylation controls many cellular functions. Yet the three-part toolkit that regulates
phosphotyrosine signaling—tyrosine kinases, phosphotyrosine phosphatases, and Src Homology
2 (SH2) domains—is a relatively new innovation. Genomic analyses reveal how this revolutionary
signaling system may have originated and why it rapidly became critical to metazoans.

Throughout human history, new technolo-
gies and technological platforms have
constantly been invented. Only a small
fraction of these technologies go on to
be widely adopted, but these can
ultimately have transformational conse-
quences. In the evolutionary history of
living organisms, we know that innovative
molecular systems have appeared at key
points in time, and these are thought to
have played a transformative role in major
evolutionary transitions in the tree of life.
But how do such innovative molecular
systems emerge, and how and why do
some proliferate and become stably
adopted by subsequent lineages?
An example of such an innovative
molecular system is phosphotyrosine
(pTyr)-based signal transduction. This
molecular system for transmitting cellular
regulatory information is estimated to
have appeared relatively recently in the
history of life—600 million years ago,
just prior to the emergence of multicellular
animals (King et al., 2003; Pincus et al.,
2008; Manning et al., 2008). The pTyr sig-
naling system has become an essential
part of metazoan biology. For example,
pTyr signaling plays a central role in many
cell-to-cell communication pathways,
including those that regulate proliferation,
differentiation, adhesion, hormone res-
ponses, and immune defense (Hunter,
2009).
In modern metazoans, pTyr signaling is
mediated by a toolkit of three distinct
functional modules: tyrosine kinases
(TyrK) phosphorylate specific target
tyrosine residues, phosphotyrosine phos-
phatases (PTP) remove the phosphates,
and Src Homology 2 (SH2) domains
recognize the modifications (Pawson
1995). Together, these three modules
form the ‘‘writer,’’ ‘‘eraser,’’ and ‘‘reader’’
toolkit that is common to many diverse
cellular information processing platforms
(Figure 1A). A rich array of diverse and
complex regulatory schemes can be
achieved through the dynamic interplay
of these three modular functions (Pawson
et al., 1993; Pawson, 1995; Bhattachar-
yya et al., 2006; Kholodenko, 2006).
A combination of these modules can
lead to higher-order functions (Figure 1B).
For example, there are several proteins
containing a combination of SH2 and
kinase domains that can generate
positive feedback (phosphorylation of
tyrosine sites leads to SH2-mediated
recruitment of the kinase, and subse-
quently, more extensive phosphorylation)
(Pawson, 2004). Similarly SH2-phospha-
tase domain combinations can generate
negative feedback (Tonks and Neel,
2001).
The three-part pTyr signaling toolkit
thus raises a classic question in evolu-
tionary biology: how do complex, interde-
pendent systems arise? It is clear why
a new system encompassing a writer,
eraser, and reader might be extremely
useful. But, given their interdependence,
how could these individual components
arise in a stepwise fashion consistent
with an evolutionary process? Proteins
that bind or remove a posttranslational
modification would seem useless without
an enzyme to generate the modification
and, in principle, would not provide
a fitness advantage leading to its retention
and spread. The pTyr signaling platform
provides a case study to look for plausible
stepwise pathways of the evolution of
a multipart system.
Here, we reconstruct a possible history
for the evolution of pTyr signaling. This
reconstruction is based on the recent
sequencing of the genomes of a number
of organisms that originated both before
and after the emergence of metazoans
from single-celled eukaryotic ancestors
(King et al., 2008). The genome sequence
of the choanoflagellate, Monosiga brevi-
collis, has been particularly illuminating
as choanoflagellates are thought to be
one of the closest single-celled relatives
of metazoans. We present a model for
how this three-part signaling system
could have plausibly evolved in a stepwise
manner. We propose that once the
complete three-part system was in place,
it may have rapidly taken hold in subse-
quent lineages because it could generate
new regulatory behaviors without signifi-
cant cross-interference with existing
regulatory circuits. We also discuss the
possible role of this new communication
system in facilitating the transformative
evolutionary shift to multicellularity.
Given the incomplete record, however,
such an evolutionary reconstruction is

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 661


Figure 1. The Writer, Reader, Eraser pTyr Toolkit
(A) In pTyr signaling, the tyrosine kinase (TyrK), Src Homology 2 (SH2), and phosphotyrosine phosphatase
(PTP) domains form a highly interdependent signaling platform. This platform serves as the writer, reader,
and eraser modules, respectively, for processing pTyr marks.
(B) Components of pTyr signaling can be used to build complex circuits. For example, recruitment of an
SH2-TyrK protein to an initiating pTyr site can lead to amplification of tyrosine phosphorylation through
a positive feedback loop.
highly speculative. For example, we
cannot rule out more complex paths
involving cycles of evolutionary gain and
loss of components, nor the possibility
that similar components in distinct line-
ages have independent origins. Nonethe-
less, this model may provide a useful
framework for focusing studies of pTyr
signaling origins and the origins of analo-
gous multicomponent signaling plat-
forms.
We describe three possible stages in
the emergence of the modern pTyr
signaling toolkit, each represented by an
extant model organism (Figure 2). These
stages are representative; we do not
claim to define the exact path of evolution,
but rather focus on identifying the domi-
nant classes of stable intermediates that
can exist in the broader evolutionary land-
scape.
PTPs in a Pre-Tyrosine
Kinase World
What came first, TyrK, PTP, or SH2
domains? Sequence analysis suggests
that it was PTP domains. The genome of
a simple single-celled eukaryote like the
budding yeast Saccharomyces cerevisiae
shows no TyrK proteins and one proto-
SH2 domain, but a handful of PTP
proteins (Pincus et al., 2008) (Figure 2,
Stage 1). Most fungi have no more than
five PTP proteins, and several of these
662 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
have tyrosine phosphatase activity. We
refer to the single putative ‘‘SH2’’ domain
in yeast, found within the gene Spt6, as
a proto-SH2 domain because it does not
show pTyr binding (the domain has been
reported to show phospho-Ser/Thr
binding; Dengl et al., 2009). Thus, func-
tionally, it cannot be considered a
‘‘reader’’ domain that is part of a pTyr
regulatory system. These observations
suggest a simple model: the first step in
the evolution of the three-part pTyr
signaling machinery was likely to have
been the emergence of a functional tyro-
sine phosphatase. But why would PTPs
arise in the pre-tyrosine kinase world?
What functional use and fitness advan-
tage would this eraser domain provide in
organisms lacking a writer domain?
The answer may lie in the fact that some
Ser/Thr kinase domains, which are more
ancient than tyrosine kinases (dating
back close to the origins of eukaryotes),
can carry out sporadic but functionally
important phosphorylation of tyrosines.
Phosphoamino acid analysis of yeast
reveals a small but significant population
of pTyr (Schieven et al., 1986). Moreover,
certain events, such as the activation
of mitogen-activated protein kinases
(MAPKs) and inhibition of the cell-cycle
kinase Cdk1, are known to involve phos-
phorylation of tyrosine residues (for acti-
vation, a MAPK must be phosphorylated
by an upstream Ser/Thr kinase on both
a Thr and Tyr residue within its activation
loop; Cdk1 is phosphorylated on Tyr14
by the inhibitory kinase Wee1). These
tyrosine modifications are clearly not
recognized by SH2 domains, but they
exert direct allosteric effects within the
proteins in which they occur. Thus, PTP
domains may have provided a fitness
benefit by negatively modulating these
rare but functionally important phosphor-
ylation events. Consistent with this model,
the proteins PTP2 and PTP3 in yeast
clearly have a functionally important role
in downregulating MAPK-mediated sig-
naling in response to pheromones or
osmolarity changes, explaining their
fitness benefit (Pincus et al., 2008). In
addition, PTPs may have played a general
role buffering against the occasional
harmful stray phosphorylation of function-
ally important tyrosines.
Where did these PTPs come from?
PTPs are likely to have arisen from
a common ancestor of the related dual-
specificity phosphatases, which are also
found in most single-celled eukaryotes
(Kennelly, 2001; Alonso et al., 2004).
Dual-specificity phosphatases are cata-
lytic domains that can dephosphorylate
both pSer/Thr and pTyr substrates. The
PTP and dual-specificity phosphatase
catalytic domains are distinct but are
evolutionarily related. They share a
common fold and the core catalytic motif
HC(X)5R, in which a phosphocysteine
enzyme intermediate is generated during
catalysis. (Sometimes, both dual-speci-
ficity phosphatases and classical PTPs
are referred to as PTPs; here, we use
this nomenclature only for the classical
PTP domains that act only on pTyr). The
domains of dual-specificity phosphatases
have a shallower active site than classical
PTPs, which may explain why they can
dephosphorylate either Tyr or Thr/Ser
residues. In some lineages, dual-speci-
ficity phosphatases have functionally
diverged further, giving rise to members
that can act on lipid substrates, such as
the phosphoinositide phosphatases
PTEN and the myotubularins (Alonso
et al., 2004). Thus, the PTPs appear to
have arisen from a somewhat promis-
cuous class of multifunctional phospha-
tases.
Despite the presence of PTP proteins in
fungi, there are striking differences
Figure 2. Evolution of pTyr Signaling
Shown is a possible path for the emergence of phosphotyrosine (pTyr) signaling. We postulate three successive stages, each represented by what is observed in
a modern organism. The thickness of the tree reflects the approximate degree of usage of pTyr signaling (thicker lines mean more usage). Stage 1 (exemplified by
the budding yeast Saccharomyces cerevisiae) reflects the situation in early eukaryotes, in which PTPs emerged but were limited in number and complexity. They
were most likely used to reverse or process sporadic cross-phosphorylation of tyrosine residues by Ser/Thr kinases. S. cerevisiae has fewer than five PTP
proteins and no functional SH2 or TyrK domains. Stage 2 reflects systems in which functional SH2 domains emerged that were able to bind to pTyr motifs.
Together with Ser/Thr kinases with increased cross-reactivity for Tyr (such as tyrosine kinase-like or dual specificity Ser/Thr kinases), these systems may reflect
the most primitive of pTyr writer/reader/eraser systems. However, the lack of a dedicated Tyr kinase may have limited the utility and expansion of this toolkit. This
stage is potentially represented by the slime mold, Dictyostelium discoideum. Stage 3 reflects systems that evolved after the emergence of the modern TyrK
domain. We postulate that the full writer/reader/eraser system was of so much greater utility that its use expanded dramatically. This likely resulted in many
more proteins in these families, as well as much more complex, multidomain architectures than those seen in the earlier stages. This stage is represented by
both the multicellular metazoan and unicellular choanoflagellate lineages.

between these proteins and those
found in metazoans (Figure 2). For
example, there are far fewer PTPs in
fungi (5/genome versus 40/genome
in metazoans) and they are considerably
less complex in domain architecture
(Pincus et al., 2008). Metazoan PTP
proteins tend to be large multidomain
proteins in which the PTP module has
been functionally recombined with multi-
ple other signaling modules. In contrast,
in fungi, the PTP domains are all either
in simple single-domain proteins or in
combination with a single rhodanase-like
domain (a putative regulatory domain
that is homologous to a class of sulfur
transfer enzymes; Bordo and Bork,
2002). Thus, fungal PTP proteins are
very simple (one to two domains) and
lack the combinatorial complexity of
metazoan PTP proteins. The simplicity
and low number of PTPs in yeast sug-
gests that in early single-celled eukary-
otes, PTP domains had fairly limited

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 663

functional utility, especially when com-
pared to their broad and complex usage
in metazoans.
Unlike PTPs, there are no known pTyr-
binding SH2 domains in fungi, although
there is one clearly homologous domain
found in the yeast protein SPT6. This
protein, which has a domain with an
SH2-like sequence and fold, is involved
in the regulation of transcription elonga-
tion, and the SH2 domain binds to the
Ser/Thr phosphorylated C-terminal tail of
RNA polymerase II. The domain does
not bind to pTyr (Dengl et al., 2009). Inter-
estingly, a single SPT6 ortholog, with the
same overall domain architecture, is
found in all eukaryotes, including all fungi
and metazoans (but not prokaryotes). This
finding suggests that in early eukaryotes,
a proto-SH2 domain emerged to perform
a highly specialized function—one that
was unrelated to the flexible modular
pTyr recognition function of the modern
SH2 domain. This proto-SH2 domain
most likely did not ‘‘read’’ pTyr modifica-
tions, but instead recognized a special-
ized related modification. Thus, although
SPT6 is likely to represent an early
ancestor or relative that eventually gave
rise to modern SH2 domains, it cannot
be considered a functional part of a pTyr
regulatory toolkit. We therefore postulate
that early eukaryotes had only a pTyr
eraser function (mediated by PTPs) with
no specialized complementary reader or
writer functions.
In summary, the PTP domain and
a structural ancestor of the SH2 domain
appear to have arisen in early single-
celled eukaryotes, but are likely to have
functional origins that are not directly
related to their later function in modern
pTyr regulatory systems. These compo-
nents may have provided a limited but
incremental fitness advantage, even in
the absence of a specialized tyrosine
kinase domain.
Toward a Write/Read/Erase System
In the early days of a more sophisticated
pTyr-signaling system, we suggest that
a proto-SH2 domain (mostly likely a
homolog of the yeast Spt6 protein) in
a single-celled organism acquired the
new and functionally beneficial ability to
bind to pTyr-containing peptide motifs.
The slime mold Dictyostelium discoideum
has the simplest repertoire of bona fide
664 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
pTyr-binding SH2 domains and may
therefore provide a living representative
of this second evolutionary stage (Fig-
ure 2, Stage 2). Dictyostelium is a soil-
living amoeba that has a unicellular
lifestyle in the presence of bacterial
food. However, when food is depleted,
individual cells aggregate in response to
the chemoattractant cAMP to form
a multicellular structure, which then
develops into a fruiting body through the
differentiation of stalk and spore cells.
The rudimentary pTyr-SH2 system in
Dictyostelium is important for aspects of
this differentiation process, including
intracellular responses to both cAMP
and the morphogen differentiation in-
ducing factor or DIF (which induces the
differentiation of prestalk cells), as well
as for transcriptional regulation in res-
ponse to hyperosmotic stress. These
observations are consistent with early
pTyr-SH2 signaling playing a role in
cellular responses to changing environ-
mental conditions.
The Dictyostelium genome specifies 13
proteins with SH2 domains (as well as
a single Spt6 homolog). These 13 proteins
cluster into five basic domain architec-
tures, two of which are homologous to
metazoan SH2 proteins. Notably, Dic-
tyostelium has four STAT (signal trans-
ducers and activators of transcription)
proteins that are very similar to metazoan
STAT transcription factors (Kay, 1997;
Kawata et al., 1997). For example, they
all have an SH2 domain juxtaposed to
a DNA-binding region; they are inducibly
phosphorylated on tyrosine residues in
response to stress or the extracellular
signaling molecule DIF; they undergo
pTyr/SH2-mediated dimerization and
then translocate to the nucleus to regulate
the expression of specific genes. Dictyos-
telium also has an ortholog of the
mammalian E3 ubiquitin ligase Cbl, which
uses SH2 and Ring domains to couple
pTyr signals to the ubiquitination ma-
chinery (Langenick et al., 2008). The re-
maining three domain architectures of
Dictyostelium SH2 proteins are distinct
from those found in other sequenced
organisms. The LrrB protein has an SH2
domain linked to a leucine-rich repeat
domain (Sugden et al., 2010), whereas
the FbxB protein has an F-box followed
by an SH2 domain and ankyrin repeats.
In addition, the Shk proteins have a protein
kinase domain followed by an SH2
domain, a domain combination that is
somewhat similar to metazoan cyto-
plasmic tyrosine kinases like Src (Monia-
kis et al., 2001). The Shk catalytic domain,
however, lacks motifs characteristic of
bona fide tyrosine kinases and biochemi-
cally displays dual specificity toward
serine/threonine and tyrosine residues.
Indeed, Dictyostelium differs from
metazoans and choanoflagellates in that
its genome does not encode any modern
tyrosine-specific protein kinases. For
example, metazoan STAT proteins are
usually phosphorylated by Janus tyrosine
kinases (JAKs), but there are no JAKs in
Dictyostelium (Kay, 1997). This suggests
the possibility that signaling proteins con-
taining SH2 domains such as STATs
evolved before the modern tyrosine
kinases with which they are associated
in metazoans. The identity of the kinase
responsible for STAT tyrosine phosphory-
lation, and the consequent formation of
SH2-binding sites, remains mysterious.
How, then, is tyrosine phosphorylation
of Dictyostelium proteins such as the
STATs controlled? Thus far, genetic anal-
ysis has not identified a specific relevant
kinase, and it has been proposed that, in
contrast to mammalian STATs, there
may be basal constitutive phosphoryla-
tion of Dictyostelium STAT tyrosine sites,
which is regulated by changes in PTP
activity in response to extracellular
signals (Langenick et al., 2008). One of
the PTPs in Dictyostelium, PTP3, binds
and dephosphorylates STATc, thereby
blocking SH2-mediated dimerization and
STATc accumulation in the nucleus.
Signaling induced by the DIF morphogen
appears to transmit signals by inhibiting
PTP3 activity and consequently boosting
STATc tyrosine phosphorylation and
STATc-dependent gene expression.
Although Dictyostelium lacks true tyro-
sine kinases, it is noteworthy that its
genome has a significant expansion in
the number of putative dual-specificity
protein kinases (there are 70, also known
as tyrosine kinase-like or TKL kinases)
(Manning et al., 2008). This set includes
the Shk catalytic domain, described
above. It is unlikely that any of these
kinases are precursors of modern tyrosine
kinases. However, it is plausible that these
represent the first evolutionary form of
the ‘‘writer’’ function in a prototype pTyr
three-part regulatory system. The union of
an SH2 domain and a dual specificity
kinase domain, as found in the Shk
proteins, may be an early example of link-
ing ‘‘reader’’ and ‘‘writer’’ modules to
achieve more complex functions such as
positive feedback. Nonetheless, the
limited functionality of the dual specificity
kinases in carrying out tyrosine phosphor-
ylation may have limited the capabilities of
this early system. This may explain the
very modest expansion of pTyr signaling
in organisms such as Dictyostelium.
These observations paint the following
picture of Dictyostelium pTyr signaling
and, by extension, of an early phase in
the evolution of pTyr communication.
SH2 domains have acquired pTyr-binding
activity and are found in several distinct
combinations with other types of signaling
domains. Among these, the STAT and Cbl
proteins are shared with metazoans,
whereas the LrrB, FbxB, and Shk proteins
are unique to Dictyostelium. But no dedi-
cated modern tyrosine kinases have
been found, and the dynamic control of
tyrosine phosphorylation may be primarily
regulated by PTPs. Although functionally
important for aggregation and differentia-
tion, the pTyr signaling system has not
acquired the pervasive influence evident
in M. brevicollis and metazoans, perhaps
because of the lack of an efficient tyrosine
kinase. Put another way, Dicytostelium
has effective pTyr readers and erasers,
but the writer is poorly developed.
Invention of TyrK and Expansion
of the pTyr Toolkit
Current analysis suggests that the
modern tyrosine kinases arose just prior
to the evolution of the metazoans. Aside
from metazoans, canonical tyrosine
kinases have thus far only been observed
in the choanoflagellates, which appear to
be the closest known single-celled rela-
tives of metazoans (King et al., 2008).
The absence of significant numbers of
such tyrosine kinases in any other branch
of life suggests that this new catalytic
domain evolved in a recent common
ancestor of choanoflagellates and meta-
zoans, most likely as a branch of the older
Ser/Thr kinases. Some bacteria do have
specialized tyrosine kinases (BY kinases),
but these resemble P loop NTPases
(nucleotide triphosphatases) and are
structurally unrelated to eukaryotic tyro-
sine kinases (Lee and Jia, 2009). It is
therefore probable that BY kinases
evolved separately from metazoan tyro-
sine kinases and operate in a different
fashion.
The new eukaryotic tyrosine kinase
domain appears to have been a game
changing innovation (Figure 2, Stage 3).
The total number of tyrosine kinase
proteins in both choanoflagellate and
metazoan species is in the range of 30–
150 per genome (Pincus et al., 2008;
Manning et al., 2008). Among sequenced
genomes, there is a striking absence of
species with only a small number of TyrK
proteins. This all-or-none sudden jump in
the number of TyrK proteins suggests
their importance as they appear to have
undergone rapid expansion and subse-
quent retention.
What is perhaps more striking is the
observation that the emergence of the
TyrK domain and its rapid expansion
correlates with an equally rapid expansion
of PTP and SH2 domains within the same
genomes (Pincus et al., 2008). Although
fungi and Dictyostelium have 5 PTP
proteins, metazoans, and choanoflagel-
lates have 30–40 per genome. Similarly,
Dictyostelium has approximately ten
SH2 domain-containing proteins (fungi
have none), whereas metazoans and
choanoflagellates have 100 each.
Thus, both PTP and SH2 proteins
undergo a roughly 10-fold increase in
number per genome after the emergence
of the TyrK domain. Moreover, the
proteins containing SH2 and PTP
domains become far more complex and
varied (Jin et al., 2009). For example, in
yeast and Dictyostelium, SH2 and PTP
proteins normally are very simple one or
two domain proteins. However, in line-
ages that have modern TyrK proteins,
SH2 and PTP proteins almost always
comprise three to ten domains.
These observations are consistent with
the following model. When a far more
efficient TyrK domain—or ‘‘writer’’ func-
tion—emerged, this dramatically increased
the functional utility of the pre-existing PTP
(eraser) and SH2 (reader) domains. As
a three-part toolkit—a catalytic domain to
generate pTyr, an interaction domain to
bind to these pTyr sites, and an enzyme
to dephosphorylate them—this domain
set could be used to encode and execute
a far wider and diverse range of regulatory
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 665
functions, thus leading to the subsequent
expansion of the complete set. Although
the PTP and SH2 domains had utility in
simpler organisms, their much larger
functional potential was not unleashed
until the emergence of the TyrK domain.
The rapid expansion of the pTyr
signaling machinery in the ancestors of
choanoflagellates and animals is reminis-
cent of how technology expands in
quantum jumps, especially in situations
involving codependent technologies. For
example, the value of the laser expanded
dramatically after the later invention of
the complementary technology of fiber
optics. This codependent technology al-
lowed lasers to be repurposed to rapidly
displace electrical transmission via
copper wires as the backbone of global
communication (Alwayn, 2004). Thus,
although lasers had standalone utility,
their major application had to await the
introduction of complementary tech-
nology. The expansion of molecular com-
ponents in biology is likely to be similar.
A toolkit of writer, reader, and eraser
functions may be of full use only when all
components are present. Thus, it may
be common for any system of this type
to show a quantum ‘‘all-or-none’’ expan-
sion only when the final piece of the toolkit
emerges.
Applying the New pTyr Toolkit
to Different Functions
Although both choanoflagellate and
metazoan lineages show a large expan-
sion of the three-part pTyr regulatory
machinery, the way in which these
components are used appears to be quite
different. When one examines the domain
types that co-occur with TyrK, SH2, or
PTP domains, one finds many distinct
combinations that are unique to each
lineage (Pincus et al., 2008; Manning
et al., 2008). These differences in domain
combinations imply distinct functions for
proteins containing these domains in the
choanoflagellate and metazoan lineages
(Li et al., 2009). Assuming that the evolu-
tion of new TyrK, SH2, and PTP proteins
occurred by recombination with new
accessory domains (Jin et al., 2009; Pei-
sajovich et al., 2010), this observation
also implies that the complete signaling
toolkit emerged only shortly before the
divergence of metazoans and choanofla-
gellates (i.e., shortly before the evolution
of metazoan multicellularity) and that
much of the divergent expansion of these
domain families occurred after the lineage
split.
Thus, earlier assumptions that pTyr
signaling is only used in metazoan cell-
cell communication are clearly incorrect.
Choanoflagellates do not form the com-
plex and permanent cell-cell organization
that metazoans do, yet surprisingly they
have a comparable (if not greater) number
of pTyr signaling proteins (Manning et al.,
2008). Sequencing of other organisms
that arose near the origins of metazoans
is ongoing. Preliminary data also suggest
a large number of pTyr signaling proteins
in other single-celled relatives of meta-
zoans. Thus, it may be more reasonable
to view the pTyr signaling system as an
innovative but generic information pro-
cessing system that could potentially be
used for transmitting many different types
of information.
Orthogonal Signaling: A Platform
for Biological Innovation
When the three-part pTyr system first
emerged, it presented a new platform
with which to transmit information that
was orthogonal to pre-existing signaling
systems. Because it was based on a
distinct covalent modification, new regu-
latory circuits could be assembled with
these components without significant
cross-interference with pre-existing net-
works. Thus, this brand new signaling
apparatus probably had a high encoding
potential for evolving dramatically new
functions, such as those involved in multi-
cellularity. One possible problem that
could be caused by the expansion of the
new pTyr signaling enzymes might be
excessive general phosphorylation of
tyrosine residues throughout the pro-
teome. Interestingly, however, organisms
using pTyr signaling may have developed
a simple solution to deal with this
problem—a decrease in the tyrosine
content of proteins across the proteome
is observed to correlate with tyrosine
kinase expansion (Tan et al., 2009).
A new orthogonal signaling system like
the pTyr signaling platform can be viewed
as analogous to a newly opened region in
the telecommunications spectrum. New
frequencies provide the opportunity for
transmitting large amounts of information
as there is little interference from existing
666 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
communication. Because of this valuable
high encoding potential, there is extreme
pressure to quickly fill this region of the
spectrum. Moreover, the exact type of
information carried by each region of the
spectrum is flexible—for example, the
same region of the spectrum can be as-
signed to different functions in different
countries. We hypothesize that the new
pTyr signaling system that emerged prior
to metazoans presented similar new
opportunities to transmit more informa-
tion. This virgin system was rapidly
exploited, though the way it has been
used appears to be different in the two
branches (metazoans and choanoflagel-
lates) that emerged after the complete
toolkit was established.
It is tempting to speculate that the
emergence of a new signaling system
with high encoding potential may have
played a key role in the emergence of
a new, complex biological function such
as metazoan multicellularity. Such large-
scale phenotypic evolutionary innova-
tions may require and coincide with
innovations in basic molecular compo-
nents (King, 2004; Rokas, 2008).
Indeed, we speculate that pTyr
signaling may provide a more general
model for the generation of multicompo-
nent biological systems, involving first a
limited stepwise development of elements
that together have a rudimentary biolog-
ical utility, followed by an explosive
spread, once all of the components of
the mature system are in place. Explora-
tion of this concept, and further analysis
of the evolution of pTyr signaling, will be
assisted by the increasing sequence infor-
mation being gathered for both unicellular
and multicellular eukaryotes (Srivastava
et al., 2010), which will no doubt yield
surprises akin to the discovery of exten-
sive pTyr signaling in M. brevicollis. More-
over, as genomic information bracketing
other major evolutionary transitions
becomes available, it will be interesting
to see whether these innovations are
also associated with the explosive expan-
sion of new molecular toolkits.
A key point here is that the specific
emergence of the pTyr toolkit may not
have been essential for the evolution of
multicellularity, but rather, any number of
new orthogonal signaling toolkits with
the same high encoding potential could
have served a similar role. Other analo-
gous new molecular information curren-
cies could have, in principle, been able
to serve as the substrate for dramatic
phenotypic innovation. In this context,
plants make extensive use of protein
phosphorylation and have numerous
transmembrane receptor Ser/Thr kinases,
but they lack conventional tyrosine
kinases, indicating that pTyr-based
signaling is not the only mechanism of
information transfer through which organ-
isms can achieve multicellularity.
Is pTyr Signaling Saturated?
How close is the pTyr signaling system to
being saturated? Is there still available en-
coding potential that could be tapped for
the evolution of new pathways and
behaviors? It is difficult to answer these
questions. However, the fact that new
pTyr signaling proteins appear to be asso-
ciated with advanced processes like
adaptive immunity suggests that there
was still some remaining encoding poten-
tial in the system as late as the evolution of
mammals. The evolutionary history re-
constructed here begs many questions.
Are there new regulatory toolkits evolving
now or in the future? Will these new tool-
kits be the substrate required for the
next big evolutionary innovation?
The importance of new molecular tool-
kits is conversely also very relevant to
the emerging field of synthetic biology, in
which the goal is to engineer cellular
systems with new functions. A major
potential limitation is how to build such
new functions in a reliable fashion that
does not cross-interfere in unanticipated
ways with existing systems (Lim, 2010).
Can we develop new synthetic molecular
signaling currencies that are orthogonal
to existing natural ones, and would these
systems dramatically facilitate our ability
to reliably and predictably endow cells
with innovative new functions?
Conclusions
Current data suggest that PTP and SH2
domains evolved before modern TyrK
domains, most likely to process pTyr
modifications sporadically catalyzed by
Ser/Thr kinases. However, the PTP and
SH2 domain protein families did not
expand dramatically until the emergence
of an efficient TyrK. We postulate that
only with the complete toolkit of writer
(TyrK), reader (SH2), and eraser (PTP)
domains, was the full encoding potential
of this system unleashed, leading to rapid
expansion and elaboration of these
domain families. This type of explosive
increase in component usage may prove
to be common to all multipart molecular
systems. The emergence of the modern
TyrK maps just prior to the split between
metazoans and choanoflagellates. These
two lineages appear to have used this
new molecular communication system in
distinct ways—multicellular metazoans
used it for cell-cell coordination, whereas
unicellular choanoflagellates used it for
distinct but as yet uncharacterized
functions.
Thus, we are able to reconstruct a plau-
sible model by which the pTyr signaling
machinery could have evolved in a rela-
tively simple stepwise manner into what
today is a complex and highly interdepen-
dent system. In this model, evolution is
opportunistic and forward looking,
borrowing, and repurposing machinery
that pre-exists. The first simple PTP
proteins likely arose from the more
ancient Ser/Thr phosphatase family and
may have been maintained initially as
a way to reverse the unavoidable occa-
sional tyrosine phosphorylation event
catalyzed by a Ser/Thr kinase. In some
cases, like the MAPKs, which are present
in all eukaryotes, these tyrosine phos-
phorylation events appear to have
become exploited and fixed as actual
parts of signal transmission, alongside
Ser/Thr phosphorylation events. SH2
domains also likely arose from a pre-ex-
isting fold in the SPT6 protein, which is
found in all eukaryotes but has no pTyr
binding activity. But this fold, once co-
opted for this function, began to expand,
most likely because of its ability to
contribute to a wider range of modular
signaling events. But the full utility of these
components was only unleashed upon
the emergence of the modern TyrK
domain, which led to the highly expanded
three-part system. One cannot help but
wonder what other simple pieces of
molecular machinery may be lying around
in today’s biological systems, of limited
utility now, but awaiting the emergence
of some as yet unknown complementary
component that will generate a complete
toolkit that will help to drive future evolu-
tionary innovation.
ACKNOWLEDGMENTS
We thank D. Pincus, B. Mayer, P. Beltrao, O. Hoeller,
R.Linding, G. Superti-Furga, N.King, N.Helman, L.
Holt, A. Horwitz, T. Miller, G.Manning, T. Hunter,
D.Morgan, J. Williams, H. Bourne, and I. Ernberg
for helpful comments. This work was supported by
the Howard Hughes Medical Institute (W.L.), the
National Institutes of health (GM55040, GM62583,
GM081879, and EY016546—W.L.), the Packard
Foundation (W.L.), the National Science Foundation
Synthetic Biology Engineering Research Center
(W.L.), the Canadian Institutes for Health Research
(T.P.—MOP-6849), Genome Canada (T.P.), and
the Canadian Cancer Society Research Institute
(T.P.).
REFERENCES
Alonso, A., Sasin, J., Bottini, N., Friedberg, I.,
Friedberg, I., Osterman, A., Godzik, A., Hunter,
T., Dixon, J., and Mustelin, T. (2004). Cell 117,
699–711.
Alwayn, V. (2004). Optical Network Design and Im-
plementation (Indianapolis, IN: Cisco Press).
Bhattacharyya, R.P., Reményi, A., Yeh, B.J., and
Lim, W.A. (2006). Annu. Rev. Biochem. 75,
655–680.
Bordo, D., and Bork, P. (2002). EMBO Rep. 3, 741–
746.
Dengl, S., Mayer, A., Sun, M., and Cramer, P.
(2009). J. Mol. Biol. 389, 211–225.
Hunter, T. (2009). Curr. Opin. Cell Biol. 21,
140–146.
Jin, J., Xie, X., Chen, C., Park, J.G., Stark, C.,
James, D.A., Olhovsky, M., Linding, R., Mao, Y.,
and Pawson, T. (2009). Sci. Signal. 2, ra76.
Kawata, T., Shevchenko, A., Fukuzawa, M.,
Jermyn, K.A., Totty, N.F., Zhukovskaya, N.V., Ster-
ling, A.E., Mann, M., and Williams, J.G. (1997). Cell
89, 909–916.
Kay, R.R. (1997). Curr. Biol. 7, R723–R725.
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 667
Kennelly, P.J. (2001). Chem. Rev. 101, 2291–2312.
Kholodenko, B.N. (2006). Nat. Rev. Mol. Cell Biol.
7, 165–176.
King, N. (2004). Dev. Cell 7, 313–325.
King, N., Hittinger, C.T., and Carroll, S.B. (2003).
Science 301, 361–363.
King, N., Westbrook, M.J., Young, S.L., Kuo, A.,
Abedin, M., Chapman, J., Fairclough, S., Hellsten,
U., Isogai, Y., Letunic, I., et al. (2008). Nature 451,
783–788.
Langenick, J., Araki, T., Yamada, Y., and Williams,
J.G. (2008). J. Cell Sci. 121, 3524–3530.
Lee, D.C., and Jia, Z. (2009). Trends Biochem. Sci.
34, 351–357.
Li, W., Scarlata, S., and Miller, W.T. (2009).
Biochemistry 48, 5180–5186.
Lim, W.A. (2010). Nat. Rev. Mol. Cell Biol. 11,
393–403.
Manning, G., Young, S.L., Miller, W.T., and Zhai, Y.
(2008). Proc. Natl. Acad. Sci. USA 105, 9674–9679.
Moniakis, J., Funamoto, S., Fukuzawa, M.,
Meisenhelder, J., Araki, T., Abe, T., Meili, R.,
Hunter, T., Williams, J., and Firtel, R.A. (2001).
Genes Dev. 15, 687–698.
Pawson, T. (1995). Nature 373, 573–580.
Pawson, T. (2004). Cell 116, 191–203.
Pawson, T., Olivier, P., Rozakis-Adcock, M.,
McGlade, J., and Henkemeyer, M. (1993). Philos.
Trans. R. Soc. Lond. B Biol. Sci. 340, 279–285.
Peisajovich, S.G., Garbarino, J.E., Wei, P., and
Lim, W.A. (2010). Science 328, 368–372.
Pincus, D., Letunic, I., Bork, P., and Lim, W.A.
(2008). Proc. Natl. Acad. Sci. USA 105, 9680–9684.
Rokas, A. (2008). Annu. Rev. Genet. 42, 235–251.
Schieven, G., Thorner, J., and Martin, G.S. (1986).
Science 231, 390–393.
Srivastava, M., Simakov, O., Chapman, J., Fahey,
B., Gauthier, M.E., Mitros, T., Richards, G.S., Con-
aco, C., Dacre, M., Hellsten, U., et al. (2010).
Nature 466, 720–726.
Sugden, C., Ross, S., Bloomfield, G., Ivens, A.,
Skelton, J., Mueller-Taubenberger, A., and Wil-
liams, J.G. (2010). J. Biol. Chem. 285, 22927–
22935.
Tan, C.S., Pasculescu, A., Lim, W.A., Pawson, T.,
Bader, G.D., and Linding, R. (2009). Science 325,
1686–1688.
Tonks, N.K., and Neel, B.G. (2001). Curr. Opin. Cell
Biol. 13, 182–195.

What Controls T Cell Receptor
Phosphorylation?
In a study published in Cell, Xu et al. (2008)
raised the intriguing possibility that tyro-
sine residues present in the immunore-
ceptor tyrosine-based activation motifs
(ITAMs) of the cytoplasmic domain of the
CD33 subunit of the T cell receptor (TCR)
are sequestered in the membrane. Here,
they are inaccessible to tyrosine kinases
that would otherwise drive signaling.
As a consequence, the authors proposed
that ‘‘triggering’’ of the TCR—that is, the
events leading to phosphorylation of the
receptor—must be dependent upon
some, as yet undefined, process that
liberates the ITAMs from the membrane.
In an accompanying Preview, Kuhns and
Davis (2008) invoked the idea that mem-
brane association constitutes a ‘‘safety
catch’’ that prevents untoward triggering
of the TCR. Such effects add an additional
layer of complexity to the TCR triggering
problem.
Using equilibrium dialysis, Xu et al.
showed that the cytoplasmic domain of
CD33 (CD33CD) partitioned into acidic
rather than zwitterionic lipids. They then
went on to derive a structural model for
ITAMs bound by acidic lipids using
nuclear magnetic resonance (NMR)-
based methods. A new in vivo FRET-
based assay suggested that the associa-
tion of the ITAM with the membrane was
dependent on a cluster of positively
charged residues in the membrane-proxi-
mal amino-terminal region of CD33CD;
mutation of these residues reduced the
association of CD33CD with the mem-
brane as detected by the FRET assay.
The obvious question was, does mem-
brane association prevent kinases from
accessing ITAMs in vivo? Xu et al. showed
that the wild-type CD33 ITAM is not phos-
phorylated in resting T cells, implying that
it may indeed be completely inaccessible
to kinases. Somewhat surprisingly, how-
ever, they did not go on to clinch the argu-
ment by showing that there was increased
phosphorylation of the membrane nonas-
sociating CD33CD mutant. The results
of this experiment turn out to be inconsis-
tent with the predictions of Xu and
colleagues.
668 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
Correspondence
To show that CD33CD is not phosphor-
ylated in resting Jurkat T cells, Xu et al.
used a chimeric protein consisting of
murine CD33CD fused to the extracellular
and transmembrane domains of the
human natural killer cell inhibitory recep-
tor KIR2DL3. We generated FLAG-tagged
forms of this protein and a second
chimera consisting of the extracellular
and transmembrane domains of KIR2DL3
fused with CD33CD mutated at the set of
clustered positively charged residues
proposed to mediate membrane asso-
ciation, that is, the mutant Emut1+2
described by Xu et al. (see Experimental
Procedures). We expressed the two
chimeras at similar levels in Jurkat
T cells using a lentivirus expression
system (Figure S1A) and probed for
phosphorylation of CD33CD by western
blotting (Figure S1B). Neither KIR2DL3/
CD33CD nor KIR2DL3/CD33CDEmut1+2
immunoprecipitated with anti-FLAG anti-
body showed detectable phosphorylation
in resting T cells (Figure S1B, lanes 1
and 2). Even after very long exposures of
the blots, we were unable to detect phos-
phorylation over background levels of
cross-reactivity of the antibody and we
saw no differences in signal for the wild-
type versus the mutant proteins (Fig-
ure S1B). This was also the case for
wild-type and Emut1+2-mutated forms
of full-length CD33 incorporated into the
TCR complex (data not shown).
These findings suggest either that
the Emut1+2 mutations do not release
the ITAMs from the membrane or that,
in vivo, the association with the mem-
brane is weaker than expected and
some other process prevents CD33CD
phosphorylation in resting cells. We sus-
pected that tyrosine phosphatases might
be responsible for the lack of phosphory-
lation of the wild-type and mutant forms of
CD33CD. To test this, we incubated cells
expressing the chimeric proteins with
the tyrosine phosphatase inhibitor, perva-
nadate (Swarup et al., 1982). Both the
wild-type and mutated forms of CD33CD
were heavily phosphorylated following
incubation of the cells for 20 min with
Leading Edge
the inhibitor (Figure S1B, lanes 3 and 4).
Time course experiments indicated that
rather than enhancing it, the Emut1+2
mutations somewhat slowed the kinetics
of CD33CD phosphorylation (data not
shown). In vivo, therefore, the wild-type
form of CD33CD seemed to be at least
as accessible to tyrosine kinases as
the mutated, membrane nonassociating
form of the protein when kinase activity
could be observed by inhibiting phospha-
tase activity. Finally, CD33 was among the
more heavily tyrosine phosphorylated
proteins present in whole-cell lysates
from pervanadate-treated wild-type
Jurkat T cells (Figure S1C), emphasizing
its accessibility to kinases relative to other
kinase targets.
How can our observations be recon-
ciled with those of Xu et al.? Our experi-
ments suggest that membrane associa-
tion, if it occurs, has less impact on
CD33CD phosphorylation than might
have been expected on the basis of the
Xu et al. work. The strongest evidence for
lipid association is, of course, the NMR
structure of CD33CD stably bound to
acidic lipids. It should be borne in mind,
however, that we know little about the
actual distribution of inner leaflet
lipids around receptors embedded in the
membrane. Recent work by Zech et al.
(2009) shows that the acidic phospholipid
phosphatidylserine is enriched in mem-
brane sheets isolated from activated
T cells using beads coated with anti-CD3
antibodies, but the extent to which this
reflects the distribution of lipids in the
immediate vicinity of the TCR in resting
or activated cells is unclear. The density
or properties of the acidic lipids adjacent
to CD33 in resting cells might be insuffi-
cient to sustain an interaction of the type
demonstrated in vitro by Xu et al. Whatever
interaction ITAMs have with membranes
in vivo might be too weak and dynamic to
prevent access of the kinases.
The reasons for the low levels of CD33CD
phosphorylation in resting T cells remain to
be fully worked out. In any discussion of the
possible constraints on tyrosine phosphor-
ylation of the TCR, some consideration
must be given to the protein tyrosine phos-
phatases (PTPs) present at the cell surface,
such as CD45 (Hermiston et al., 2003).
Leukocytes express as many as 105 CD45
molecules, i.e., 2–3 for every TCR (Wil-
liams and Barclay, 1986), each with very
broad substrate specificity (Barr et al.,
2009); the catalytic activities of PTPs have
been estimated to be 10- to 1000-fold
higher than that of tyrosine kinases
(discussed by Fischer et al., 1991). Another
contributing factor is the activity of inhibi-
tory kinases such as Csk, which also
constrain T cell activation (Schoenborn
et al., 2009). Progress in understanding
TCR triggering will require the teasing
apart of these and other competing
factors and the identification of those
factors with the largest impact on net
phosphorylation of the TCR. The effects
of pervanadate on phosphorylation levels
in resting T cells observed by ourselves
(Figures S1B and S1C) and by others
many years ago (e.g., O’Shea et al., 1992)
suggest that PTPs could be key. For us,
the sheer weight of the numbers also
warrants their serious consideration.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Experimental
Procedures and one figure and can be found with
this article online at doi:10.1016/j.cell.2010.08.018.
Response
Multilayered Control of T Cell
Receptor Phosphorylation
We would like to respond to the Corre-
spondence by Fernandes et al., but first
we will summarize the data in our Cell
paper (Xu et al., 2008). Our study used
a biophysical approach to examine the
binding of the CD33 cytoplasmic domain
(CD33CD) to the plasma membrane. We
reported a new FRET assay to examine
CD33CD binding to the inner leaflet of
the plasma membrane in live Jurkat cells,
a transformed T cell line. Furthermore, we
developed an approach to determine
the NMR structure of CD33CD bound in
a lipid bilayer environment. The structure
showed that the two tyrosines of the im-
ACKNOWLEDGMENTS
This work was funded by the Wellcome Trust, the
United Kingdom Medical Research Council, and
the Portuguese Fundação para a Ciência e a Tec-
nologia.
Ricardo A. Fernandes,1,4 Chao Yu,1,4
Alexandre M. Carmo,2
Edward J. Evans,1
P. Anton van der Merwe,3
and Simon J. Davis1,*
1Nuffield Department of Clinical Medicine
and Medical Research Council Human
Immunology Unit, The Weatherall Institute of
Molecular Medicine, University of Oxford,
Oxford OX3 9DU, UK
2Group of Cell Activation and Gene
Expression, Institute for Molecular and
Cellular Biology, 4150-180 Porto, Portugal
and Abel Salazar Institute for Biomedical
Sciences, University of Porto,
Porto 4099-003, Portugal
3Sir William Dunn School of Pathology,
The University of Oxford, Oxford OX1 3RE,
UK
4These authors contributed equally to this
work
*Correspondence: simon.davis@ndm.ox.
ac.uk
DOI 10.1016/j.cell.2010.08.018
munoreceptor tyrosine-based activation
motif (ITAM) partitioned into the hydro-
phobic core of the bilayer in a dynamic
manner, with substantial movement of
the two tyrosines and other elements of
the cytoplasmic domain (Figure 6, Xu
et al., 2008). Microscopy data yielded
similar FRET values for CD33CD tagged
with C-terminal teal fluorescent protein
(TFP) and for a positive control in which
TFP was in close proximity to the plasma
membrane (there was a three amino acid
linker between the transmembrane and
TFP domains). These data indicated that
most of the cytoplasmic domain was
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 669
REFERENCES
Barr, A.J., Ugochukwu, E., Lee, W.H., King, O.N.,
Filippakopoulos, P., Alfano, I., Savitsky, P.,
Burgess-Brown, N.A., Müller, S., and Knapp, S.
(2009). Cell 136, 352–363.
Fischer, E.H., Charbonneau, H., and Tonks, N.K.
(1991). Science 253, 401–406.
Hermiston, M.L., Xu, Z., and Weiss, A. (2003).
Annu. Rev. Immunol. 21, 107–137.
Kuhns, M.S., and Davis, M.M. (2008). Cell 135,
594–596.
O’Shea, J.J., McVicar, D.W., Bailey, T.L., Burns,
C., and Smyth, M.J. (1992). Proc. Natl. Acad. Sci.
USA 89, 10306–10310.
Schoenborn, J.R., Zhang, C., Shokat, K.M., and
Weiss, A. (2009). J. Immunol. 182, Abstract 35.4.
Swarup, G., Cohen, S., and Garbers, D.L. (1982). Bi-
ochem. Biophys. Res. Commun. 107, 1104–1109.
Williams, A.F., and Barclay, A.N. (1986). Glycopro-
tein antigens of the lymphocyte surface and their
purification by antibody affinity chromatography.
In Handbook of Experimental Immunology (Oxford:
Blackwell Scientific Publications), pp. 22.1–22.24.
Xu, C., Gagnon, E., Call, M.E., Schnell, J.R., Schwi-
eters, C.D., Carman, C.V., Chou, J.J., and
Wucherpfennig, K.W. (2008). Cell 135, 702–713.
Zech, T., Ejsing, C.S., Gaus, K., de Wet, B., Shev-
chenko, A., Simons, K., and Harder, T. (2009).
EMBO J. 28, 466–476.
membrane bound in these transfected
nonstimulated Jurkat cells. However, we
did not claim that all CD33 cytoplasmic
domains in a given T cell are completely
bound to the plasma membrane. This
would be impossible in a biological sys-
tem because binding events always
follow equilibrium conditions, with the
size of the bound versus free fractions
being determined by the ratio of on and
off rates. It follows that changes in equilib-
rium can result in dissociation of CD33CD
from the membrane. Our data showed
a structure with substantial mobility in
which changes in equilibrium, such as
recruitment of the tyrosine kinase Lck,
could enable phosphorylation.
We wish to point out that we did not
claim that binding of CD33CD to the
membrane was the only mechanism that
prevents spontaneous T cell signaling.
It is well established that there is
multilayered control of T cell receptor
(TCR) signaling because complete
phosphorylation of even one or a few TCRs
is sufficient to result in T cell activation
(Bergman et al., 1992; Davis and van der
Merwe, 2006). Without highly effective
mechanisms to prevent spontaneous
signaling, rampant chronic inflammation
and autoimmunity would result. Sponta-
neous signaling is inhibited by the Csk
kinase, which phosphorylates Lck kinase
at inhibitory tyrosine 505, as well as
multiple phosphatases (Bergman et al.,
1992; Davis and van der Merwe, 2006). In
agreement with Fernandes et al., we did
not detect phosphorylation of the
CD33CD Emut1+2 mutant protein in Jurkat
cells (Figure S1B, CD33-Mut time 0), which
was a negative control for the FRET exper-
iments. However, if binding of CD33
to the membrane together with other inhib-
itory mechanisms guards against sponta-
neous phosphorylation, this result is not
surprising. Phosphorylation may only be
favored with appropriate TCR localization
into microclusters that exclude the CD45
phosphatase (Varma et al., 2006), as well
as simultaneous colocalization of active
Lck bound to CD4/CD8 coreceptors close
to the clustered TCR-CD3 complexes.
Fernandes et al. assume that the
CD33CD Emut1+2 mutant protein would
normally interact with Lck. However, the
loss of six basic residues substantially
changes its charge properties. We
therefore directly compared the phos-
phorylation of the mutant and wild-type
CD33CD proteins in an in vitro phosphor-
ylation assay with the purified cyto-
plasmic domains of the wild-type and
Emut1+2 proteins and purified Lck in the
absence of liposomes (Figure S1A). We
observed greatly reduced phosphoryla-
tion of the mutant protein, even though
equal amounts of wild-type and mutant
proteins were used. The mutant protein
was not detectably phosphorylated with
100 ng of Lck kinase and was only phos-
phorylated at low levels with 400 ng Lck;
the wild-type protein was robustly phos-
phorylated under both conditions. Fer-
nandes et al. also mention delayed phos-
phorylation kinetics for this mutant
protein in cells treated with pervanadate,
which inhibits a broad range of phospha-
tases (data not shown). We confirmed
this finding and observed substantially
delayed phosphorylation of the mutant
protein after pervanadate treatment: the
wild-type protein was phosphorylated
670 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
after 2 min, whereas phosphorylation of
the mutant was delayed, with a 3.5- to
5-fold reduction at 10 min (Figure S1B).
However, both proteins were phosphory-
lated to a similar degree at a later time
point (30 min), as Fernandes et al. also
showed. The greatly reduced interaction
of the mutant protein with Lck as well as
the multiple control mechanisms regu-
lating T cell activation could account for
the absence of observable phosphoryla-
tion of this particular CD33 mutant in non-
stimulated Jurkat cells. It is not known
why mutation of six basic residues in the
N-terminal part of CD33CD reduces the
interaction with Lck, but conformational
changes in the cytoplasmic domain or
loss of long-range charge interactions
with Lck could be involved.
Pervanadate oxidizes the active site
cysteines of many different phosphatases
and is generated by mixing hydro-
gen peroxide with sodium orthovanadate
(Huyer et al., 1997). Given the potential
for unanticipated effects, we used our
FRET assay to examine the conse-
quences of pervanadate treatment on
CD33CD membrane binding (Xu et al.,
2008). The assay measures the interac-
tion of a TFP domain attached to the
C terminus of CD33CD with a lipophilic
dye R18 (octadecyl rhodamine B) incor-
porated in the plasma membrane; label-
ing the cells with R18 caused a decrease
in the TFP signal (quenching) due to
energy transfer from TFP to R18 (Fig-
ure S1C). In the absence of pervanadate,
the FRET efficiency was high (average
50%; Figure S1D), which is similar to
the FRET signal obtained with the positive
control in which TFP is positioned close
to the plasma membrane (a three amino
acid linker between the transmembrane
domain and TFP) (Xu et al., 2008). After
20 min of pervanadate treatment, the
FRET signal was reduced on average
to 15% (Figure S1D), similar to the
negative control with a 50 amino acid flex-
ible linker between the transmembrane
domain and TFP (Xu et al., 2008). Perva-
nadate treatment thus results in dissocia-
tion of the CD33 cytoplasmic domain from
the membrane.
We next examined the potential mech-
anisms that could contribute to this effect.
We had previously shown that basic resi-
dues of CD33CD are critical for its binding
to the plasma membrane. Therefore, we
assessed whether the negative charge of
the inner leaflet of the plasma membrane
could be affected. Phosphatidylserine is
the most abundant negatively charged
lipid in the inner leaflet. We used a highly
specific calcium-independent phosphati-
dylserine probe, the Lactadherin C2
domain (Lact-C2), to study the dynamics
of phosphatidylserine distribution (Yeung
et al., 2008). Jurkat cells were transduced
with a lentivirus containing the Lact-C2
probe, and the probe’s localization in live
cells was analyzed by confocal micros-
copy (Figure S1E). Under resting condi-
tions, the Lact-C2 probe predominantly
associated with the plasma membrane in
75% of cells. In contrast, pervanadate
treatment reduced the Lact-C2 signal
at the plasma membrane and increased
the amount of this phosphatidylserine
probe in internal membranes in 92%
of cells, suggesting that the plasma
membrane inner leaflet had a reduced
negative charge (Figures S1E and S1F).
Phosphatidylserine at the cell surface
was also slightly increased, as measured
by annexin-V labeling (Figure S1G). We
conclude that pervanadate treatment
reduces the negative charge of the inner
leaflet and induces release of CD33CD
from the plasma membrane. These find-
ings complicate the interpretation of
experiments using pervanadate to assess
whether CD33CD is bound to the plasma
membrane.
A recent study confirmed that CD33CD
binds to the plasma membrane (Deford-
Watts et al., 2009), and prior work has
shown that CD3z is membrane bound
(Aivazian and Stern, 2000). Furthermore,
other cytoplasmic peptides have been
shown to interact with the inner leaflet
by similar mechanisms. McLaughlin and
colleagues showed that the MARCKS
peptide binds to the inner leaflet using
clusters of basic amino acids and five
phenylalanine residues (Zhang et al.,
2003). NMR measurements demon-
strated directly that these phenylalanine
residues are located in the hydrophobic
core of the lipid bilayer, similar to the tyro-
sines of CD33CD (Zhang et al., 2003).
Also, the cytoplasmic small GTPase Rit
localizes to the plasma membrane using
three clusters of basic amino acids
and interspersed hydrophobic residues;
mutation of a tryptophan in this segment
results in loss of membrane binding
(Heo et al., 2006). Therefore, usage of
both basic and hydrophobic residues
for plasma membrane binding is a more
general theme that extends beyond the
CD33 and z cytoplasmic domains of the
TCR-CD3 complex.
In their Correspondence, Fernandes
et al. raise questions about the functional
relevance of the binding of CD33CD to the
plasma membrane during TCR activation.
It will be important to further study the
functional significance of the binding of
CD33CD to the plasma membrane in
a physiological setting using primary
T cells because the early events in TCR
triggering are very complex, and subtle
changes may result in receptor activation
after ligand binding. Analysis of multiple
CD33CD mutant proteins will be useful,
including mutants with reduced rather
than complete loss of membrane binding.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Experimental
Procedures and one figure and can be found with
this article online at doi:10.1016/j.cell.2010.08.019.
Etienne Gagnon,1 Chenqi Xu,2
Wei Yang,2 H. Hamlet Chu,1
Matthew E. Call,3 James J. Chou,3
and Kai W. Wucherpfennig1,4,*
1Department of Cancer Immunology & AIDS,
Dana-Farber Cancer Institute, Boston,
MA 02115, USA
2Institute of Biochemistry and Cell Biology,
Shanghai Institutes for Biological Sciences,
Chinese Academy of Sciences,
Shanghai 200031, China
3Department of Biological Chemistry and
Molecular Pharmacology
4Department of Neurology and Program in
Immunology
Harvard Medical School, Boston, MA 02115,
USA
*Correspondence: kai_wucherpfennig@dfci.
harvard.edu
DOI 10.1016/j.cell.2010.08.019
REFERENCES
Aivazian, D., and Stern, L.J. (2000). Nat. Struct.
Biol. 7, 1023–1026.
Bergman, M., Mustelin, T., Oetken, C., Partanen,
J., Flint, N.A., Amrein, K.E., Autero, M., Burn, P.,
and Alitalo, K. (1992). EMBO J. 11, 2919–2924.
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 671
Davis, S.J., and van der Merwe, P.A. (2006). Nat.
Immunol. 7, 803–809.
Deford-Watts, L.M., Tassin, T.C., Becker, A.M.,
Medeiros, J.J., Albanesi, J.P., Love, P.E., Wulfing,
C., and van Oers, N.S. (2009). J. Immunol. 183,
1055–1064.
Heo, W.D., Inoue, T., Park, W.S., Kim, M.L., Park,
B.O., Wandless, T.J., and Meyer, T. (2006).
Science 314, 1458–1461.
Huyer, G., Liu, S., Kelly, J., Moffat, J., Payette, P.,
Kennedy, B., Tsaprailis, G., Gresser, M.J., and
Ramachandran, C. (1997). J. Biol. Chem. 272,
843–851.
Varma, R., Campi, G., Yokosuka, T., Saito, T., and
Dustin, M.L. (2006). Immunity 25, 117–127.
Xu, C., Gagnon, E., Call, M.E., Schnell, J.R., Schwi-
eters, C.D., Carman, C.V., Chou, J.J., and
Wucherpfennig, K.W. (2008). Cell 135, 702–713.
Yeung, T., Gilbert, G.E., Shi, J., Silvius, J., Kapus,
A., and Grinstein, S. (2008). Science 319, 210–213.
Zhang, W., Crocker, E., McLaughlin, S., and Smith,
S.O. (2003). J. Biol. Chem. 278, 21459–21466.
 Leading Edge
Previews

Fishing Out a Sensor for Anti-inflammatory Oils

Alan R. Saltiel1,*
1Life Sciences Institute, Departments of Internal Medicine and Molecular and Integrative Physiology, University of Michigan Medical School,

Ann Arbor, MI 48109, USA

*Correspondence: saltiel@lsi.umich.edu
DOI 10.1016/j.cell.2010.08.022

The u-3 fatty acids have anti-inflammatory and antidiabetic effects in humans. Now, Oh et al. (2010)
demonstrate that the G protein-coupled receptor GPR120 is a receptor for u-3 fatty acids on
macrophages and fat cells. Activation of GPR120 by u-3 fatty acids inhibits multiple inflammation
cascades in macrophages and reverses insulin resistance in obese mice.

Evidence providing an inflammatory link
between obesity and type 2 diabetes is
accumulating. In numerous animal and
clinical studies, obesity is associated
with a state of low-grade, chronic inflam-
mation in liver and adipose tissue,
which includes activation of the innate
immune system and the appearance of
proinflammatory immune cells (Hotamisli-
gil, 2006; Shoelson and Goldfine, 2009).
Most notably, macrophages conspire
with increased levels of inflammatory
cytokines to attenuate insulin action and
increase lipid accumulation.
Recent studies indicate that the NF-kB
and JNK (JUN N-terminal kinase) path-
ways play important roles in the com-
munication among macrophages, adipo-
cytes, and liver cells (Arkan et al., 2005;
Chiang et al., 2009; Solinas et al., 2007).
However, key questions still remain about
the initial establishment of this inflamma-
tory state. Do fat and liver cells first sense
an overload of energy and respond by
secreting chemokines, which then recruit
macrophages to the liver and fat? Or do
fatty acids in the diet directly initiate the
inflammatory cascade, and, if so, which
ones? In this issue of Cell, Oh et al.
(2010) address the latter question by iden-
tifying a sensor for u-3 fatty acids that is
upregulated in obese mice. Furthermore,
activation of this receptor exerts potent
anti-inflammatory effects that improve
insulin resistance and other symptoms of
metabolic syndrome in mice.
Previous studies suggest that saturated
fatty acids promote inflammation by acti-
vating Toll-like receptor 4 (TLR4) on fat
cells and macrophages (Shi et al., 2006).
In contrast, most unsaturated fats are
metabolically neutral. However, recent
evidence indicates that u-3 fatty acids
derived from fish oils, such as docosahex-
anoic acid (DHA) and eicosapentanoic
acid (EPA), have an anti-inflammatory
effect (Serhan et al., 2008).
To sort out the metabolic impact of
various types of fatty acids, Oh et al.
characterized the tissue expression pat-
terns of five G protein-coupled receptors
(GPCRs) known to bind and respond
to fatty acids. Among these receptors,
GPR120 was the only one with an expres-
sion profile that correlated well with a
potential role in regulating metabolism.
They found that GPR120 is highly ex-
pressed in adipose tissue macrophages,
fat cells, and specialized macrophages
in the liver called Kupffer cells. Moreover,
a high-fat diet increases the expression of
this receptor on macrophages, suggest-
ing that GPR120 might be controlled by
inflammatory signals.
GPR120 is an orphan receptor for
which no endogenous ligands are known.
Using a heterologous reporter system, Oh
et al. now find that the u-3 fatty acids
DHA, EPA, and palmitoleate are agonists
of GPR120. Furthermore, activation of
GPR120 by DHA antagonizes the proin-
flammatory effects of TNFa and lipopoly-
saccharide in a macrophage cell line.
DHA not only blocks the NFkB and JNK
pathways but also prevents expression
of cytokines (Figure 1, bottom right).
GPR120 is known to couple with the
Gq/11 family of G proteins. After ligand binds
and Gq/11 is released, G protein receptor
kinases phosphorylate the receptor. This
generates binding sites for b-arrestins,
which mediate internalization and down-
regulation of the receptors. However,
b-arrestins can also interact with down-
stream signaling molecules (Rajagopal
et al., 2010). In a series of gene silencing
experiments, Oh and colleagues demon-
strate that b-arrestin2 is essential for
the anti-inflammatory effects of u-3 fatty
acids in macrophage cells, but Gq is
surprisingly dispensable for this process.
Moreover, b-arrestin2 inhibits both the
JNK and NF-kB pathways by seques-
tering the TAK1 binding protein TAB1.
The inhibition of TAB1 prevents phos-
phorylation and thus activation of IkB
kinase upstream of NFkB and MKK4
(mitogen-activated protein kinase kinase 4)
upstream of JNK.
These new insights into the sensing and
signaling of GPR120 offered Oh and
colleagues a unique opportunity to study
the molecular mechanisms underlying
the metabolic benefits of u-3 fatty acids
in vivo. First, Oh and colleagues establish
that the high-fat diet given to mice in the
laboratory is low in u-3 fatty acids. They
then show that supplementing this diet
with DHA and EPA reverses the delete-
rious effects that the high-fat diet has on
glucose homeostasis and lipid storage
(i.e., steatosis). Although the authors do
not address whether u-3 fatty acids can
prevent insulin resistance or glucose intol-
erance in mice, they do demonstrate that
DHA and EPA reverse insulin resistance
caused by the high-fat diet. Moreover,
disruption of the GPR120 gene abolishes
the benefits of u-3 fatty acids on glucose
homeostasis and insulin sensitivity in
mice. These results demonstrate the
crucial role GPR120 plays in the meta-
bolic benefits of DHA and EPA. Interest-
ingly, mice receiving bone marrow
transplants from the GPR120-deficient
mice are also resistant to the beneficial

672 Cell 142, September 3, 2010 ª2010 Elsevier Inc.

 GPR120 expression, and what is its phys-
iological role in adipose and liver tissue?
Also, does GPR120 respond to endoge-
nous ligands to control macrophage
activity? Although the new study by Oh
and colleagues explains how activation
of GPR120 inhibits inflammatory path-
ways, it is still unknown how this receptor
increases the presence of anti-inflamma-
tory M2 macrophages in adipose tissue
and from where these cells arise. Do M2
macrophages (or their precursors)
express GPR120 in order to develop or
maintain the anti-inflammatory pheno-
type, or does activation of GPR120
induce the transdifferentiation of inflam-
matory M1 macrophages to anti-inflam-
matory M2 macrophages in situ?
Finally, previous studies demonstrate
that fish oils have diverse benefits on
multiple tissues. For example, u-3 fatty
acids can inhibit the production of proin-
flammatory eicosanoids and serve as
precursors for resolvins, (i.e., protective
Figure 1. A Model for How u-3 Fatty Acids Preserve Insulin Sensitivity through GPR120 lipids that help reduce inflammation)
A high-fat diet with a disproportionate ratio of saturated fatty acids to u-3 fatty acids triggers activation of
Toll-like receptor 4 (TLR4) in adipocytes and circulating immune cells. This launches an inflammatory
(Serhan et al., 2008). In addition, fish oils
cascade that results in the recruitment of proinflammatory M1 macrophages, increased secretion of help prevent cardiovascular disease and
TNFa, and insulin resistance in adipocytes. The addition of u-3 fatty acids to the diet activates the have positive effects on many inflamma-
G protein-coupled receptor GPR120 on proinflammatory M1 macrophages (Oh et al., 2010), which in
tory disorders, such as arthritis, asthma,
turn attenuates the inflammatory response and recruits anti-inflammatory M2 macrophages to adipose
tissue. Eventually, these M2 macrophages restore secretion of interleukin-10 and improve insulin and ulcerative colitis. Is GPR120 the only
sensitivity. receptor responsible for these various
benefits? Future studies are also needed
to determine whether dietary supple-
properties of DHA and EPA. Thus, the u-3 These cells produce cytokines, such as ments and ingestion of fatty fish can
fatty acids appear to act primarily through TNFa, which further activate the macro- provide high enough concentrations of
macrophages. phages and attenuate insulin action in circulating u-3 fatty acids to promote
Taken together with previous data, adipocytes. Eventually, this leads to GPR120 activation. Nevertheless, the
these findings from Oh and colleagues local and then systemic insulin resistance. new insights presented by Oh and
support a model in which dietary fatty However, these activated M1 macro- colleagues into the anti-inflammatory
acids control the inflammatory properties phages also express elevated levels of mechanisms of u-3 fatty acids provide
of macrophages in adipose tissue by GPR120. Thus, addition of u-3 fatty acids a platform for investigating these impor-
regulating the activity and expression to the diet stimulates GPR120 and tant questions. Plus, the identification of
of opposing receptors (Figure 1). With initiates a signaling pathway through GPR120 pinpoints a new therapeutic
a normal diet containing a balanced b-arrestin2, which blocks the effects of target for treating the inflammatory state
ratio of saturated and u-3 unsaturated TLR4 and inflammatory cytokine recep- associated with obesity and type 2 dia-
fatty acids, anti-inflammatory M2 macro- tors. This reduces the inflammatory state betes. This alone is cause for a bit of
phages protect adipose cells by damp- of these cells and simultaneously pro- excitement.
ening excess inflammation and maintain- motes the return of anti-inflammatory
ing insulin sensitivity of fat cells (Lumeng M2 macrophages to adipose tissue, REFERENCES
et al., 2007). When mice are given which leads to the restoration of insulin
a high-fat diet with excess calories sensitivity. Arkan, M.C., Hevener, A.L., Greten, F.R., Maeda,
S., Li, Z.W., Long, J.M., Wynshaw-Boris, A., Poli,
and little u-3 fatty acids, TLR4 is left This model raises many new questions
G., Olefsky, J., and Karin, M. (2005). Nat. Med.
unchecked in fat cells. The activated about the crosstalk between adipose 11, 191–198.
receptor induces expression and release tissue and immune cells in obesity. First, Chiang, S.H., Bazuine, M., Lumeng, C.N., Geletka,
of chemokines, such as MCP-1 (mono- could GRP120 activation serve as L.M., Mowers, J., White, N.M., Ma, J.T., Zhou, J.,
cyte chemotactic protein-1), which then a homeostatic mechanism in macro- Qi, N., Westcott, D., et al. (2009). Cell 138,
recruit proinflammatory M1 macrophages phages to resolve inflammation in obese 961–975.
into adipose tissue (Lumeng et al., 2007). animals? If so, then what controls Hotamisligil, G.S. (2006). Nature 444, 860–867.

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 673

Lumeng, C.N., Bodzin, J.L., and Saltiel, A.R.
(2007). J. Clin. Invest. 117, 175–184.
Oh, D.Y., Talukdar, S., Bae, E.J., Imamura, T.,
Morinaga, H., Fan, W., Li, P., Lu, W.J., Watkins,
S.M., and Olefsky, J.W. (2010). Cell 142, this issue,
687–698.
Rajagopal, S., Kim, J., Ahn, S., Craig, S., Lam,
C.M., Gerard, N.P., Gerard, C., and Lefkowitz,
R.J. (2010). Proc. Natl. Acad. Sci. USA 107,
628–632.
Serhan, C.N., Chiang, N., and Van Dyke, T.E.
(2008). Nat. Rev. Immunol. 8, 349–361.
Shi, H., Kokoeva, M.V., Inouye, K., Tzameli, I., Yin,
H., and Flier, J.S. (2006). J. Clin. Invest. 116, 3015–
3025.
Shoelson, S.E., and Goldfine, A.B. (2009). Nat.
Med. 15, 373–374.
Solinas, G., Vilcu, C., Neels, J.G., Bandyopad-
hyay, G.K., Luo, J.L., Naugler, W., Grivennikov,
S., Wynshaw-Boris, A., Scadeng, M., Olefsky,
J.M., and Karin, M. (2007). Cell Metab. 6,
386–397.

A New Spin on Planar Cell Polarity

Patricio Olguin1 and Marek Mlodzik1,*
1Department of Developmental and Regenerative Biology, Mount Sinai School of Medicine, New York, NY 10029, USA

*Correspondence: marek.mlodzik@mssm.edu
DOI 10.1016/j.cell.2010.08.025

The generation of planar cell polarity (PCP) and tissue shape during morphogenesis is tightly linked,
but it is not clear how. Aigouy et al. (2010) now show in the developing Drosophila wing that PCP
initially has a radial orientation that becomes realigned to the proximal-distal axis of organ shape
by mechanical forces and cell rearrangements mediated by Dachsous.

Most tissues and organs composed of and
organized as epithelial cell layers display,
in addition to the common apical-basal
epithelial polarity, a polarity within the
epithelial plane. This is commonly referred
to as planar cell polarity (PCP). Genetic
studies in the fruit fly Drosophila mela-
nogaster have established that there are
two molecular systems coordinating the
cellular asymmetries in the plane of
tissues. These include the Frizzled/PCP
signaling pathway containing the Van-
Gogh (Vang, also known as Strabismus/
Stbm) protein and other factors (Strutt,
2003; Seifert and Mlodzik, 2007) and
a pathway mediated by the protocadher-
ins Fat and Dachsous (Lawrence et al.,
2007). Although the molecular relation-
ships between these two systems are
unclear, there is strong evidence that
they act in parallel, probably affect
different effectors, and may compensate
for each other in some tissues (Lawrence
et al., 2007; Wu and Mlodzik, 2009).
In the developing Drosophila wing,
cellular asymmetries in the plane of the
epithelium are first detected at later pupal
stages along the proximal-distal axis
(at around 24–30 hr after puparium forma-
tion). The core Frizzled/PCP factors form
two distinct complexes that become
localized asymmetrically to either the
proximal (Vang/Stbm and associated
proteins) or the distal side of pupal wing
cells (Frizzled and associated proteins).
These complexes are stabilized by feed-
back loop interactions among themselves
(Seifert and Mlodzik, 2007; Strutt, 2003). In
addition, Frizzled and Vang/Stbm protein
complexes may be required earlier to
coordinate global tissue polarity/PCP
within the wing epithelium (Classen et al.,
2005; Wu and Mlodzik, 2009).
Supporting this idea, the new study by
Aigouy et al. (2010) in this issue of Cell
helps to establish that subcellular asym-
metries among the Frizzled/PCP core
group proteins are already present at
early pupal stages during wing develop-
ment (14–15 hr after puparium formation
or earlier). Strikingly, the Frizzled/PCP
complexes display radial polarity that is
perpendicular to the wing margin
(Figure 1A), confirming that coordination
of global Frizzled/PCP signaling is estab-
lished early in pupal fly wings. The authors
further demonstrate that these early
asymmetries indeed depend on Frizzled-
Vang/Stbm signaling as the nonautono-
mous behavior of frizzled mutant cell
patches (clones) affecting the polarity of
wild-type cells flanking the frizzled mutant
cells (Vinson and Adler, 1987) is already
observed at this stage. Strikingly, in
contrast to the nonautonomous effects
observed at the distal side of frizzled
mutant cell patches in late pupal and adult
wings (Vinson and Adler, 1987), early friz-
zled clones influence the polarity of wild-
type cells residing between the wing
margin and the clone within the radial
polarity axis. This confirms a ‘‘signaling
axis’’ toward the wing margin at early
stages. Taken together, the observations
of Aigouy et al. (2010) indicate that (1)
PCP, mediated by Frizzled-Vang/Stbm
signaling, is established during late-
larval/early-pupal stages in a radial axis
perpendicular to the margin and (2) the
polarity/PCP seen in the adult wing is
a result of cellular rearrangements during
wing morphogenesis within the prox-
imal-distal axis that are dependent on
Dachsous.
How is polarity realigned along the
proximal-distal axis as morphogenesis
proceeds? As PCP is already established

674 Cell 142, September 3, 2010 ª2010 Elsevier Inc.

at early stages, its final align- cues’’ that orient the initial
ment from the radial orienta- polarity of Frizzled/PCP
tion to the proximal-distal complexes, but this hypoth-
axis must be redirected esis has been challenged by
through active relocalization strong genetic evidence
of Frizzled and Vang/Stbm showing that the Fat/Dachs-
complexes and/or through ous and Frizzled/PCP
the shifting or moving of the systems act in parallel (Law-
cells as a whole. For rence et al., 2007). During fly
example, polarity could be larval stages, the Fat/Dachs-
achieved by rotation of the ous system is required to
cells toward the distal axis regulate the growth and
as happens with the rotation shape of the wing, the latter
of photoreceptor cell clusters (at least in part) by orienting
(ommatidia) in the Drosophila the axis of cell division
eye toward the anterior- perpendicular to the margin
posterior axis (Seifert and (Baena-López et al., 2005).
Mlodzik, 2007). But how Aigouy et al. (2010) show
would such a rotation be that high expression of
regulated? At early stages of Dachsous in the hinge region
pupal development, the is not required for its contrac-
proximal half of the wing tion, but correct Dachsous
epithelium (hinge) and the levels in the wing blade are
wing blade are similar in size Figure 1. Planar Cell Polarity in the Fly Wing required for the wing blade
(A) During fly pupal development, the initial axis of planar cell polarity (PCP) is
(Figure 1A). Subsequently, radial, that is, oriented toward the wing margin (black arrows).
to respond to the anisotropic
preceding and coinciding (B) As development proceeds, the hinge region (blue) contracts creating an aniso- mechanical stress that
with PCP realignment, the tropic mechanical stress on the wing blade, resulting in movements of wing cells orients cell elongation along
and realignment of PCP to the proximal-distal axis. Green arrows indicate the
hinge contracts and gener- the proximal-distal axis.
direction of cell movement, and red arrows show the direction of cellular rotations.
ates an anisotropic mechan- These processes take place between 14 and 24 hr after puparium formation. Moreover, cell polarity
ical stress on the blade, (C) The final orientation of PCP is in the proximal-distal axis in late pupal/adult fly defects correlate with the
which leads to its elongation wings (black arrows). inversion of local tissue rota-
in the proximal-distal axis tion in wild-type wings, with
(Figure 1B). Strikingly, epithelial cell elon- a proximal-distal axis within the wing (Fig- altered levels of Dachsous in the posterior
gation could drive the realignment of ure 1C). Interestingly, during this remodel- compartment, suggesting that Dachsous
cortical microtubules with the proximal- ing process, the global coordination/long- imbues cells with the ability to respond
distal axis, which appears essential for range coherence of PCP is diminished, coordinately to mechanical stress. Inter-
the delivery of Frizzled to the distal side which may be the reason why previous estingly, Harumoto et al. (2010) show
of cells (Shimada et al., 2006). New work studies of PCP during development have that Dachsous and Fat are required to
appearing in Developmental Cell by missed the early asymmetry/polarity of align cortical microtubules along the prox-
Uemura and colleagues (Harumoto et al., PCP core proteins. As pupal wing cells re- imal-distal axis and that Dachsous biases
2010) shows that, prior to hinge contrac- pack as they acquire a hexagonal shape, the direction of microtubule growth from
tion, cortical microtubules align perpen- the global coordination/long-range coher- high to low Dachsous levels, similar to its
dicular to the margin in the proximal ence of PCP increases again. This role in cell orientation. Whether the Fat/
region of the wing blade. This supports phenomenon is a consequence of the Dachsous system regulates cell remodel-
a general role for the orientation of cortical persistence of the Vang/Stbm and Frizzled ing by controlling the polarity of cortical
microtubules in PCP and in the realign- complexes at boundaries formed in the microtubules or vice versa remains to be
ment of PCP later in development. early stages of development and of the resolved.
Quantitative analyses of time-lapse proximal-distal alignment of new bound- In their elegant new study, Aigouy et al.
imaging of the pupal wing by Aigouy et al. aries. On the other hand, the Frizzled/ (2010) analyzed the timeline of events for
(2010) show that, in response to the aniso- PCP core factors are required for hexag- the establishment of PCP in the devel-
tropic stress, cells move with respect to onal cell packing, probably by polarizing oping Drosophila wing. They have
each other in a proximal direction and membrane trafficking along the proximal- provided evidence that the early Friz-
inwards with different velocities (Figure 1B). distal axis (Classen et al., 2005). Thus, zled/PCP core polarization toward the
This behavior causes shear and the local both early polarity and cellular packing wing margin (in a radial orientation) is real-
rotation of cells, mainly clockwise in the would feed in to one another to shape igned along the proximal-distal axis by
anterior and anticlockwise in the posterior and repolarize the epithelia. anisotropic mechanical stress and
half of the blade (Figure 1B). As a conse- It has been proposed that the Fat/ Dachsous-mediated tissue remodeling.
quence, PCP is reoriented from a radial to Dachsous system would provide ‘‘global These conclusions are consistent with,

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 675

and supported by, the phenotypic PCP
features of Frizzled/PCP core group
genes on one side and that of the Fat/
Dachsous system on the other. Flies
carrying mutations in Frizzled/PCP core
proteins exhibit defects in PCP
throughout the wing. In contrast, the
Fat/Dachsous system mainly affects
polarity in the proximal half of the wing,
as this area strongly depends on cellular
realignment and rotation during the
switch to the proximal-distal PCP axis.
Together, these observations provide an
exciting new framework for under-
standing the generation of PCP and its
relation to new mechanisms that sculpt
the shape of organs in general.
REFERENCES
Aigouy, B., Farhadifar, R., Staple, D.B., Sagner, A.,
Röper, J., Jülicher, F., and Eaton, S. (2010). Cell
142, this issue, 773–786.
Baena-López, L.A., Baonza, A., and Garcı́a-Bel-
lido, A. (2005). Curr. Biol. 15, 1640–1644.
Classen, A.K., Anderson, K.I., Marois, E., and Ea-
ton, S. (2005). Dev. Cell 9, 805–817.
Harumoto, T., Ito, M., Shimada, Y., Kobayashi,
T.J., Ueda, H.R., Lu, B., and Uemura, T. (2010).
Dev. Cell 19. Published online September 2,
2010. 10.1016/j.devcel.2010.08.004.
Lawrence, P.A., Struhl, G., and Casal, J. (2007).
Nat. Rev. Genet. 8, 555–563.
Seifert, J.R., and Mlodzik, M. (2007). Nat. Rev.
Genet. 8, 126–138.
Shimada, Y., Yonemura, S., Ohkura, H., Strutt, D.,
and Uemura, T. (2006). Dev. Cell 10, 209–222.
Strutt, D. (2003). Development 130, 4501–4513.
Vinson, C.R., and Adler, P.N. (1987). Nature 329,
549–551.
Wu, J., and Mlodzik, M. (2009). Trends Cell Biol. 19,
295–305.

Viable Rat-Mouse Chimeras:

Where Do We Go from Here?
Davor Solter1,*
1Institute
of Medical Biology, A*STAR, 138648 Singapore, Republic of Singapore
*Correspondence: davor.solter@imb.a-star.edu.sg
DOI 10.1016/j.cell.2010.08.021

In a tour-de-force study, Kobayashi et al. (2010) describe the first viable rat-mouse chimeras and
demonstrate that rat induced pluripotent stem (iPS) cells can rescue organ deficiency in mice.
Rat iPS cells formed a fully functional pancreas when injected into mouse blastocysts lacking the
Pdx1 gene required for pancreas formation.

Experimentally produced chimeras
between different mouse strains (Tarkow-
ski, 1961) have been an exceedingly
useful tool for developmental biologists,
contributing to our understanding of the
establishment of cell lineages, cell deter-
mination, and the development of the
immune system and other organs. In this
issue of Cell, Kobayashi et al. (2010)
dramatically extend the potential of mam-
malian chimeras with their report of viable
rat-mouse chimeras that can develop to
term and become fully functional adults.
In their study, Kobayashi and col-
leagues relied on previous knowledge
but also added a few new wrinkles. They
first derived mouse and rat embryonic
stem (ES) cells and induced pluripotent
stem (iPS) cells using standard methods
for the mouse and capitalizing on the
recent isolation of rat ES and iPS cells
(Buehr et al., 2008; Li et al., 2008). Both
mouse and rat cells were tagged with
different fluorescent dyes, allowing the
authors to follow their distribution in the
developing chimeras. The authors wanted
to prove, at least in principle, that xenoge-
neic organ complementation could be
achieved, that is, that donor cells of one
species could rescue a defect in organ
development in a recipient of a different
species. So, as a first step, they set out
to produce viable chimeras between rats
and mice, even though many previous
efforts to make such chimeras had failed.
The only viable intergeneric chimera—
that is, a hybrid between animals from
different genera—reported so far is
the geep between a sheep (Ovis aries)
and a goat (Capra hircus) (Fehilly et al.,
1984).
To test the possibility that viable rat-
mouse chimeras could be formed,
Kobayashi et al. (2010) injected fluores-
cently labeled mouse or rat iPS cells into
rat or mouse blastocysts, respectively,
and returned them to blastocyst-compat-
ible pseudopregnant females (that is,
foster mothers of the same species as
the blastocysts). The authors then exam-
ined the resulting fetuses, newborns, and
adults and found evidence of a substantial
contribution of donor stem cells to tis-
sues and organs of the host (Figure 1A).
Despite a big contribution of donor cells,
the size of newborn and adult chimeras
(with one exception) was determined by

676 Cell 142, September 3, 2010 ª2010 Elsevier Inc.

Figure 1. Generating Rat-Mouse Chimeras
(A) Induced pluripotent stem (iPS) cells were derived from adult mouse and rat cells and were labeled with different fluorescent proteins. Rat (blue) and mouse
(red) iPS cells were injected into reciprocal blastocysts (mouse into rat and vice versa) to produce intergeneric chimeras. From these blastocysts, several
chimeras were born and some survived to adulthood. The contribution of injected donor stem cells was observed throughout the body of the host. The size
and morphology of the newborn and adult chimeras was determined by the host blastocyst.
(B) Fluorescently labeled rat iPS cells (blue) were injected into normal mouse blastocysts (left) or blastocysts lacking the Pdx1 gene (right), which encodes the
transcription factor pancreatic and duodenal homeobox 1 that is required for pancreas development. Chimeras derived from normal or Pdx1-deficient mouse
blastocysts showed an extensive contribution of rat cells to all tissues. However, in the Pdx1-deficient chimeras, the entire pancreas was derived from donor
rat cells (inset, blue) and was fully functional, including production of insulin by b islet cells.

the species of the host blastocyst. It is not
clear whether it is the embryo itself or the
uterine environment that determines the
extent of chimera growth. To distinguish
between these possibilities, one would
have to transfer chimeric embryos into
the uterus of pseudopregnant females of
the same species as the donor stem cells
(not the blastocysts). Previous studies
suggest that such experiments would fail
because of the need for compatibility
between the fetal part of the placenta and
the uterus (Rossant et al., 1982).
Besides controlling the size and growth
of the chimera, the host blastocyst seems
to impose additional morphogenetic reg-
ulation. The postimplantation develop-
ment of normal rat and mouse embryos
is very similar, but there are differences
in organ morphogenesis. One of the
most noticeable differences is the pres-
ence of a gall bladder in mice and its
absence in rats. In all adult chimeras
produced by combining rat stem cells
and mouse blastocysts, the resulting
‘‘mouse-like’’ chimeras had a gall bladder
despite the significant contribution of rat
cells to abdominal organs. Reciprocal
chimeras were ‘‘rat-like’’ and, again,
despite a significant contribution from
mouse cells to abdominal organs, the
gall bladder was absent. These results
suggest that cells of the blastocyst inner
cell mass possess a ‘‘morphogenetic’’
capacity that controls the behavior of
injected stem cells at all developmental
stages.
This may explain why Kobayashi et al.
were able to successfully inject rat stem
cells into mouse blastocysts, whereas
insertion of the rat inner cell mass into
the mouse blastocyst cavity did not result
in viable rat-mouse chimeras (Gardner
and Johnson, 1973). This notion could
be tested further using tetraploid comple-
mentation (that is, donor ES or iPS cells
are injected into tetraploid blastocysts)
to produce rat-mouse chimeras (Nagy
et al., 1993). Tetraploid blastocyst cells
cannot participate in formation of the
embryo proper; thus, the resulting fetus
(and adult) is derived entirely from the
injected cells, whereas the placenta and
extraembryonic membranes are derived
from the tetraploid blastocyst. It remains
to be seen whether this approach could
produce a fetus derived entirely from
mouse ES cells after their injection into
a rat tetraploid blastocyst that then
develops in the uterus of a pseudopreg-
nant rat female.
A major goal of the Kobayashi et al.
study was to determine whether stem cells
from a xenogeneic donor mammal could
correct a genetic defect in a recipient
mammal of a different species. So, in their
next set of experiments, the authors in-
jected rat iPS cells into recipient mouse
blastocysts that lacked the Pdx1 gene,

Cell 142, September 3, 2010 ª2010 Elsevier Inc. 677

which encodes a transcription factor
(pancreatic and duodenal homeobox 1)
that is essential for development of the
pancreas and formation of insulin-
producing b islet cells. Although they
observed a substantial contribution of rat
cells to different organs and tissues, most
importantly, the pancreas of the rat-mouse
chimeras was composed exclusively of rat
cells (Figure 1B). Thus, cells derived from
rat iPS cells were able to completely
rescue the genetic deficiency of the host
mouse blastocyst. These rat-mouse
chimeras developed into adult animals
with a normal functional pancreas, demon-
strating that xenogeneic organ comple-
mentation is achievable. This is a remark-
able accomplishment.
So where do we go from here? Although
human ES and iPS cells offer hope for
tissue and cell replacement therapies in
the not too distant future, the replacement
of complex organs—lung, kidney, liver,
gut, and, of course, pancreas—is likely to
be much more difficult. Several strategies
for organ replacement are being tested.
These include the growth of organs
in vitro with mixtures of different stem cells
and biocompatible scaffolds or the gener-
ation of ‘‘humanized’’ pigs engineered to
lack certain antigens so that their organs
can be used for transplantation in human
patients with a reduced chance of immune
rejection. Could production of human
organs in, for example, human-pig
chimeras be an alternative approach?
Although production of viable rat-mouse
chimeras could be viewed as a first step
in this direction, as Kobayashi et al.
propose, there are huge biological and
678 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
technical challenges. For example, the
mouse and rat are developmentally very
similar (apart from size), but it is not clear
that chimeras between animals belonging
to different phylogenetic families or orders
would be viable. Indeed, the only attempts
to make such chimeras (between a mouse
and a bank vole) have failed (Mystkowska,
1975). In this experiment, the mouse-bank
vole chimeras were made by aggregation
of embryos; it is possible that injection of
bank vole stem cells into mouse blasto-
cysts, followed by their development
in the uterus of mouse foster mothers
might yield positive results. Successful
chimerism between members of different
orders (the pig and human, for example)
seems very unlikely, and attempts to
produce early postimplantation human-
mouse chimeras have not been encour-
aging (James et al., 2006). Even if we
succeed in developing organ-deficient
pigs by genetic manipulation and pro-
ducing pig-human chimeras in which the
parenchymal cells of the specific organ
are entirely derived from human cells,
immune rejection will still be a problem
because the human organ carried by the
pig will contain pig-derived stromal cells
and blood vessels.
Finally, there are huge legal and ethi-
cal barriers to creating human-animal
chimeras and, indeed, their production is
forbidden in most countries. However, it
is possible that injecting human ES or
iPS cells into a mouse blastocyst and al-
lowing limited (early postimplantation)
development of human-mouse chimeras
would be approved for the specific
purpose of testing the differentiation
potential of human stem cells. Yet such
experiments will be complicated, time
consuming, difficult to interpret, and, I
suspect, will never become part of the
standard protocols regulating the medical
use of human stem cells. Although xeno-
geneic organ complementation is unlikely
to be a viable strategy for regenerative
medicine, the elegant work of Kobayashi
et al. is a boon for researchers seeking
to better understand the biology of stem
cells and mammalian development.
REFERENCES
Buehr, M., Meek, S., Blair, K., Yang, J., Ure, J.,
Silva, J., McLay, R., Hall, J., Ying, Q.-L., and Smith,
A. (2008). Cell 135, 1287–1298.
Fehilly, C.B., Willadsen, S.M., and Tucker, E.M.
(1984). Nature 307, 634–636.
Gardner, R.L., and Johnson, M.H. (1973). Nat. New
Biol. 246, 86–89.
James, D., Noggle, S.A., Swigut, T., and Brivanlou,
A.H. (2006). Dev. Biol. 295, 90–102.
Kobayashi, T., Yamaguchi, T., Hamanaka, S.,
Kato-Itoh, M., Yamazaki, Y., Ibata, M., Sato, H.,
Lee, Y.-S.,, Usui, J.-i., Knisely, A.S., et al. (2010).
Cell 142, this issue, 787–799.
Li, P., Tong, C., Mehrian-Shai, R., Jia, L., Wu, N.,
Yan, Y., Maxson, R.E., Schulze, E.N., Song, H.,
Hsieh, C.-L., et al. (2008). Cell 135, 1299–1310.
Mystkowska, E.T. (1975). J. Embryol. Exp. Mor-
phol. 33, 731–744.
Nagy, A., Rossant, J., Nagy, R., Abramow-
Newerly, W., and Roder, J.C. (1993). Proc. Natl.
Acad. Sci. USA 90, 8424–8428.
Rossant, J., Mauro, V.M., and Croy, B.A. (1982).
J. Embryol. Exp. Morphol. 69, 141–149.
Tarkowski, A.K. (1961). Nature 190, 857–860.
Leading Edge
Previews
‘Fore Brain:
A Hint of the Ancestral Cortex
Lora B. Sweeney1,2,3 and Liqun Luo1,2,3,*
1Neurosciences Program
2Department of Biology
3Howard Hughes Medical Institute
Stanford University, Stanford, CA 94305, USA
*Correspondence: lluo@stanford.edu
DOI 10.1016/j.cell.2010.08.024
By combining gene expression profiling with image registration, Tomer et al. (2010) find that the
mushroom body of the segmented worm Platynereis dumerilii shares many features with the
mammalian cerebral cortex. The authors propose that the mushroom body and cortex evolved
from the same structure in the common ancestor of vertebrates and invertebrates.
The mammalian cerebral cortex underlies
many higher-order processes, such as
perception, memory, language, and ad-
vanced motor skills. With its intricate
furrows and ridges (i.e., the sulci and
gyri), the complexity of the cerebral cortex
is evident even at its surface. Beneath the
surface, the cerebral cortex is separated
into layers of densely packed neurons
with axons reaching deep into the white
matter. These layers are divided further
into functional regions that correspond to
the body plan. How does such a complex
structure develop such precise organiza-
tion? Many researchers have approached
this question from a developmental
perspective by identifying and perturbing
molecular components that contribute
to the structure of the cerebral cortex
(Hébert and Fishell, 2008). Alternatively,
one can take an evolutionary approach
and search for an ancestral precursor to
the cortex. In this issue of Cell, Tomer
et al. (2010) follow the latter approach
and identify a brain region of the seg-
mented worm Platynereis dumerilii (an
annelid) called the mushroom body that
shares the same ‘‘molecular fingerprint’’
as the developing mammalian cortex.
The cerebral cortex derives from the
pallium. From the Latin for ‘‘cloak,’’
pallium refers to the outer layer of the
brain. The mammalian palium consists of
the cerebral cortex, olfactory cortex, and
the hippocampus. The evolutionary origin
of the vertebrate pallium has fascinated
biologists for centuries because our high-
est mental functions originate from it. To
identify structures in the developing brain
of Platynereis that are possibly related
to the vertebrate pallium, Tomer et al.
characterized the expression patterns of
many genes at different stages of neural
development in the worm. The standard
methods for characterizing gene expres-
sion are in situ hybridizations and immuno-
staining with antibodies. However, these
techniques are generally restricted to
one or two genes at a time and thus are
unable to provide direct comparisons of
expression patterns for a large number of
genes.
To overcome this technical hurdle,
Tomer et al. use advanced image registra-
tion methods (including linear transforma-
tion and nonlinear warping), in which
multiple microscopy images are aligned
to one coordinate system, standardized
to one size, and smoothed to correct arti-
facts due to stretches during sample
preparation (Rueckert et al., 1999; Rohlf-
ing et al., 2001; Kurylas et al., 2008).
Specifically, Tomer et al. use the axon
scaffold (i.e., bundles of axons in the
developing brain) as a landmark to align
individual gene expression patterns from
multiple worms onto a standard brain
template. This allows the simultaneous
mapping of expression profiles for an
unlimited number of genes during neural
development. In this manner, the authors
define the spatial relationship for the
expression of more than fifteen genes in
Platynereis that were previously shown
to regulate the patterning of the cortex in
mammals. These include many transcrip-
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 679
tion factors such as Bf1 (brain factor 1,
also known as Foxg1) and Pax 6 (paired
box gene 6) (Hébert and Fishell, 2008).
With this technique, Tomer et al. iden-
tify a structure in Platynereis with gene
patterns that mirror those observed in
the vertebrate pallium. The authors then
follow this structure during the develop-
ment of Platynereis and find that its
neurons develop into the mushroom
body, a brain region in insects and worms
involved in sensory processing and
memory (Figure 1). Moreover, this embry-
onic structure in the worm gives rise
to similar types of neurons as found
in the pallium (e.g., glutamatergic and
GABAergic neurons).
Similarities between the mammalian
cortex and the invertebrate mushroom
body have been noted previously, but
such clear parallels in their gene pattern-
ing have not been observed (Strausfeld
et al., 1998; Farris, 2008). Why not? Mush-
room bodies are well characterized in in-
sects, especially in the fruit fly Drosophila,
and choosing to study an annelid was
critical to making this new connection.
Based on the arrangement of the axon
fibers and the function of its neurons, the
mushroom body of insects has been
loosely compared to the mammalian
cerebral cortex, the hippocampus, or the
cerebellum (Strausfeld et al., 1998).
However, insects are fast-evolving inver-
tebrates and thus may deviate quickly
from their ancestors. By contrast, anne-
lids evolve slowly and therefore make
excellent organisms for comparing the
 the vertebrate pallium and invertebrate
mushroom body evolved from the same
structure in the last common ancestor
of these organisms (Figure 1). Moreover,
this structural precursor may also have
processed sensory information, as the
cortex and mushroom body do.
That said, however, the vertebrate
pallium and invertebrate mushroom body
are considerably different. First, they have
different shapes (Figure 1). This may result
in part from considerable differences
in neural developmental programs in
vertebrates and invertebrates. Second,
it is unclear whether the mushroom body
exhibits the same type of layering and
functional specialization as the cortex.
Third, gene expression patterns appear
to remain constant over time during
development of the annelid mushroom
body, but the patterns are quite dynamic
during the development of the mamma-
lian cortex. This temporal variation com-
plicates the interpretation of the homolo-
Figure 1. A Common Origin for the Mushroom Body and Cerebral Cortex? gous gene expression patterns detected
In mammals, the pallium develops into the cerebral cortex, olfactory cortex, and hippocampus. Gene for the two structures. Finally, the expres-
expression profiling combined with image registration in the developing brain of the segmented worm
Platynereis dumerilii identified a brain region with gene expression patterns similar to that of the mouse sion profiles are not identical for the
cortex (Tomer et al., 2010). This region in the worm develops into the mushroom body, which is involved two structures. For example, expression
in sensory processing and memory formation in insects and likely worms. These results suggest that the of Wnt3a was present in the vertebrate
mushroom body and the cerebral cortex evolved from the same structure in a common ancestor of verte-
brates and annelids 600 million years ago (mya). The estimates of divergence follow Peterson et al.
pallium but not in the developing
(2004). mushroom body. Such differences are
probably not surprising, given that verte-
evolution of invertebrates with that of ing olfactory inputs and for memory brates and annelids diverged from a
vertebrates. For example, an analysis formation (Heisenberg, 2003). Indeed, common precursor 600 million years
of intron abundance across the entire Tomer and colleagues found that a subset ago (Figure 1).
genomes of vertebrates, annelids, and of genes that share similar expression In summary, the data presented by
insects found that two-thirds of the patterns in the annelid mushroom body Tomer and colleagues provide a tanta-
human introns were also present in and the mammalian cortex are also lizing hint that one of the highest-order
annelids but were largely absent from expressed in the Drosophila mushroom processing centers of the human brain
insects, confirming that insects have body. shares an evolutionary origin with the
evolved faster than annelids (Raible Two distinct mechanisms can explain mushroom body of worms and insects.
et al., 2005). Moreover, annelids are ex- the molecular similarities of the inverte- In principle, the techniques presented
tremely amenable to experimental anal- brate mushroom body and the vertebrate by the authors for profiling expression
ysis because they are easy to breed and cerebral cortex. They could arise inde- patterns of numerous genes simulta-
observe in vivo. Further, they develop pendently through convergent evolution, neously can be used in other organisms
with highly stereotyped structures that or they could evolve from the same to garner further support for this hypoth-
vary little between individuals. structure in a common ancestor. Tomer esis. Furthermore, comparative struc-
Given the similarity in the gene expres- and colleagues present a statistical anal- ture-function analyses may also provide
sion patterns of the developing annelid ysis indicating that the similarities in additional insights into the function of
mushroom body and the mammalian gene expression that they observed are this ancestral structure, including how its
cerebral cortex, do these structures unlikely to occur by convergent evolution. form has adapted over the last 600 million
share similar functions? The cerebral Although these arguments suffer from years.
cortex transforms sensory information the caveat that expression patterns of
into action. The mushroom body of anne- different genes may not have evolved REFERENCES
lids also processes sensory information, independently, the number of genes
receiving input from chemosensory cells studied by Tomer and colleagues do Farris, S.M. (2008). Brain Behav. Evol. 72, 106–122.
on their mouthparts. In insects, the analo- indeed favor the possibility of a common Hébert, J.M., and Fishell, G. (2008). Nat. Rev.
gous structure is best known for process- origin. Together, their data suggest that Neurosci. 9, 678–685.

680 Cell 142, September 3, 2010 ª2010 Elsevier Inc.

Heisenberg, M. (2003). Nat. Rev. Neurosci. 4,
266–275.
Kurylas, A.E., Rohlfing, T., Krofczik, S., Jenett, A.,
and Homberg, U. (2008). Cell Tissue Res. 333,
125–145.
Peterson, K.J., Lyons, J.B., Nowak, K.S., Takacs,
C.M., Wargo, M.J., and McPeek, M.A. (2004).
Proc. Natl. Acad. Sci. USA 101, 6536–6541.
Raible, F., Tessmar-Raible, K., Osoegawa, K.,
Wincker, P., Jubin, C., Balavoine, G., Ferrier, D.,
Benes, V., de Jong, P., Weissenbach, J., et al.
(2005). Science 310, 1325–1326.
Rohlfing, T., Brandt, R., Maurer, C.R., Jr., and
Menzel, R. (2001). IEEE Workshop on Mathemat-
ical Methods in Biomedical Image Analysis,
L. Staib, ed. Kauai, HI, IEEE Computer Society,
Los Alamitos, CA, pp. 187–194.
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 681
Rueckert, D., Sonoda, L.I., Hayes, C., Hill, D.L.,
Leach, M.O., and Hawkes, D.J. (1999). IEEE Trans.
Med. Imaging 18, 712–721.
Strausfeld, N.J., Hansen, L., Li, Y., Gomez, R.S.,
and Ito, K. (1998). Learn. Mem. 5, 11–37.
Tomer, R., Denes, A.S., Tessmar-Raible, K., and
Arendt, D. (2010). Cell 142, this issue, 800–809.

The Language of Histone Crosstalk
Jung-Shin Lee,1 Edwin Smith,1 and Ali Shilatifard1,*
1Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110, USA
*Correspondence: ash@stowers.org
DOI 10.1016/j.cell.2010.08.011
It has been suggested that a specific pattern of histone posttranslational modifications and their
crosstalk may constitute a code that determines transcriptional outcomes. However, recent studies
indicate that histone modifications have context-dependent effects, making their interplay more
like a language within the chromatin signaling pathway than a code.
For almost two decades, a primary focus in the field of transcrip-
tional regulation was to identify DNA elements that control the
expression of genes. These efforts were in part motivated by the
expectation that it would one day be possible to look at a gene
and its regulatory sequences and predict when and where
a gene was going to be transcribed. We then learned that along
with the sequence-specific binding of activators and repressors,
there is an additional world of factors that modify, interact, and
remodel chromatin to regulate gene expression. The identification
of a multitude of histone modifications—some correlated with
activation, some with repression—led to the proposal that the
modifications constitute a code that could be recognized by tran-
scription factors to determine the transcriptional state of a gene.
However, additional research has since added layers of
complexity, revealing a nuanced and intriguing language, not
a strict code, as the basis for transcriptional regulation through
the chromatin signaling pathway. Here, we review the complex
crosstalk among histone modifications, including recent studies
that illustrate how the context and timing of these modifications
are critical for a particular transcriptional readout.
Histone Crosstalk and Gene Activity
In eukaryotic cells, gene expression can be regulated at the level
of chromatin structure. Numerous residues within the histone
tails and several residues within the histone globular domains
can be modified in a variety of ways, including acetylation,
phosphorylation, ubiquitination, and methylation. A well-charac-
terized posttranslational modification regulating chromatin
structure is the acetylation of histone N-terminal tails, which is
thought to facilitate transcriptional activation either by charge
neutralization of the tails’ interaction with DNA or by forming
a binding site for bromodomain-containing transcription factors,
some of which can remodel nucleosomes. Another well-studied
histone modification is the methylation of lysine 4 of histone H3
(H3K4), a modification generally associated with transcriptionally
active genes and a binding site for a variety of factors that include
histone-modifying and -remodeling activities (Shilatifard, 2006).
More complex scenarios arise when histone modifications act
combinatorially in a context-dependent manner to facilitate or
repress chromatin-mediated transcription. In some cases the
modification of one residue can alter the ability of a second
residue to be implemented by its modifying enzyme(s) (Figure 1).
The first example of histone crosstalk falls into this category: the
682 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
Leading Edge
Minireview
phosphorylation of serine 10 on histone H3 stimulates the ability
of Gcn5 to acetylate histone H3 at lysine 14 (H3K14) (Cheung
et al., 2000; Lo et al., 2000) (Figure 1A). Another well-character-
ized example is the requirement of histone H2B monoubiquitina-
tion for proper H3K4 methylation by the H3K4 methylase
complex COMPASS (Figure 1B) (Shilatifard, 2006). This process,
initially discovered in yeast (Shilatifard, 2006), is now known to
be highly conserved among eukaryotes (Kim et al., 2009). Addi-
tionally, a recent comprehensive mutation analysis of all histone
residues reveals that a single point mutation in histone H3K14,
a site of acetylation, results in the specific loss of H3K4 trimethy-
lation, but not mono- or dimethylation (Nakanishi et al., 2008).
This screen also demonstrated that H3K4 trimethylation is regu-
lated by both monoubiquitination-dependent and monoubiquiti-
nation-independent processes (Nakanishi et al., 2008).
Given that histone-modifying enzymes are often found in mul-
tisubunit complexes, modification of nearby residues can create
binding sites for the components of the complex helping to
anchor an enzyme to a nucleosome. For example, the PHD finger
of Yng1, a subunit of the NuA3 histone acetyltransferase
complex, recognizes methylated H3K4 and helps recruit this
histone acetyltransferase complex for acetylation of H3K14
(Martin et al., 2006; Taverna et al., 2006). The Yng1-related
ING2 can also bind methylated H3K4; however, it is present in
a histone deacetylase complex (Shi et al., 2006). Therefore,
H3K4 methylation can serve as a landing platform for a variety
of histone-modifying enzymes with opposing activities.
Modifications of nearby residues can also prevent the recog-
nition of a substrate by an enzyme, as recently reported to occur
when methylation of histone H3 arginine 2 (H3R2) interferes with
H3K4 methylation by Set1/COMPASS in yeast and COMPASS-
like complexes in mammalian cells (Guccione et al., 2007;
Kirmizis et al., 2007) (Figure 1B). Histone modifications can
also prevent the recruitment of factors other than enzymes.
For example, heterochromatin protein 1 (HP1), which binds
methylated H3K9, cannot do so when the adjacent serine 10
(H3S10) is phosphorylated during mitosis or during gene activa-
tion (Fischle et al., 2005; Mateescu et al., 2008).
Multiple types of histone crosstalk, involving numerous histone-
modifying complexes, can occur at any one gene. A major chal-
lenge is to understand the events that regulate changes in gene
expression through these modifications/crosstalk. One strategy
has been to profile histone modifications genomewide, with the

Figure 1. Examples of Histone Crosstalk
(A) The first characterized example of histone crosstalk is the stimulation of
acetyltransferase activity of GCN5 toward the histone H3 tail by prior phos-
phorylation (P) of serine 10. Acetylation, Ac.
(B) Crosstalk among histone modifications can span more than one histone.
Monoubiquitination of histone H2B on lysine 120 of the C-terminal helix can
lead to the trimethylation of lysine 4 in the histone 3 tail (H3K4) by Set1/
COMPASS. However, H3K4 methylation by COMPASS and COMPASS-like
complexes can be blocked if the nearby arginine of H3 is already methylated.
expectation that a given pattern will indicate a transcriptional
outcome due to the recruitment of specific proteins by these modi-
fications. However, some recent examples of trans-histone cross-
talk illustrate that transcriptional readout depends on context and
timing by which these modifications are introduced. Simply put,
just looking at the pattern of chromatin modifications at a locus
is not sufficient to determine its gene expression status. These
studies provide new insight into the language of histone crosstalk.
From Histone Phosphorylation to Transcription
Elongation
A novel form of crosstalk was recently discovered by Zippo and
colleagues, who studied the transcriptional control of FOSL1,
location, and perhaps identity of the H3 kinase, and not the H3S10 modification alone, determines downstream events.
(B) Another example of how the order of implementation of histone modifications can affect transcription comes from work from Wang et al. (2009). They report
that despite correlations between histone acetylation and H3K4 methylation, artificially increasing acetylation through treatment of cells with deacetylase inhib-
itors (HDACs) does not lead to productive transcription, despite the presence of H3K4 methylation and Pol II recruitment. Therefore, patterns of histone modi-
fications cannot simply be ‘‘read’’ but instead have distinct effects depending on the cellular context and upstream signaling events.
a gene activated in response to serum (Zippo et al., 2009) (Fig-
ure 2A). They present evidence for a transcription activation
pathway in which the phosphorylation of H3 tails leads to the
acetylation of H4 tails. In turn, acetylation of H4 tails is required
for the recruitment of the RNA Pol II positive transcription elonga-
tion factor, P-TEFb (Figure 2A). Previously, the authors found
that activation of the FOSL1 gene requires PIM1, a proto-onco-
gene whose kinase activity is activated through MAP kinase
signaling. Numerous cellular substrates of PIM1 have been iden-
tified, including H3S10. Other H3S10 kinases, such as MSK1 and
MSK2 (MSK1/2), are also implicated in the phosphorylation of
histones at serum-responsive genes, including FOSL1. Zippo
and colleagues find that the spatiotemporal pattern of H3S10
phosphorylation differs for PIM1 and MSK1/2. MSK1/2 mediates
the phosphorylation of H3S10 at the promoter of FOSL1 at early
time points of gene expression, whereas PIM1 is required for
H3S10 phosphorylation at a FOSL1 enhancer after the MSK1/
2-mediated phosphorylation of H3S10 (Figure 2A).
Screening for other histone modifications specifically associ-
ated with the FOSL1 enhancer shows that the acetylation of
H4K16 coincides with H3S10 phosphorylation. RNA interfer-
ence-mediated knockdown of PIM1 results in loss of H4K16
acetylation, suggesting a trans-tail crosstalk from H3S10 phos-
phorylation to H4K16 acetylation. Zippo and colleagues asked
whether 14-3-3 g, 3, and z proteins, previously shown to bind
phosphorylated H3S10, are recruited to the promoter and the
enhancer of FOSL1 in response to serum. They find that 14-3-
33 and 14-3-3z are recruited to both the promoter and enhancer
of FOSL1 after serum stimulation. However, 14-3-3 is required
for recruiting the H4K16 histone acetyltransferase MOF only to
the enhancer and not to the promoter of FOSL1. Recruitment
of MOF to the enhancer results in H4K16 acetylation, which
can be bound by the bromodomain-containing protein, Brd4.
Brd4 associates with P-TEFb, a kinase that phosphorylates Pol
Figure 2. Context-Dependent Outcomes of
Histone Crosstalk
(A) Zippo and colleagues (Zippo et al., 2009)
uncover a new form of histone crosstalk by
studying the transcriptional control of FOSL1,
a gene activated in response to serum. Activation
requires the binding of PIM1 to the FOSL1
enhancer. PIM1 is a kinase responsible for phos-
phorylation (P) of serine 10 on the histone H3 tail
(H3S10). Phosphorylated H3S10 creates a binding
site for 14-3-3, a phosphoserine binding protein.
Acetylation (Ac) of lysine 16 on the H4 (H4K16)
tail occurs through interaction of 14-3-3 with the
histone acetyltransferase MOF. Acetylated
H4K16 can in turn form a binding site for the bro-
modomain-containing protein Brd4, a component
of P-TEFb, a kinase that phosphorylates the
C-terminal domain of RNA Pol II to facilitate tran-
scription elongation. However, at an earlier stage
of serum stimulation, an MSK1/2 kinase is re-
cruited to the promoter where it phosphorylates
H3S10. 14-3-3 is then recruited to the promoter,
but unlike the situation at the enhancer, MOF is
not recruited to the promoter. Thus, the timing,
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 683
II to facilitate transcription elongation (Figure 2A). Thus, Zippo
and colleagues propose that crosstalk between modifications
on two different histone tails regulate productive transcription
elongation through the sequential recruitment of proteins that
bind these modifications.
One question raised by this study is why H3S10 phosphoryla-
tion produces different results at the enhancer than at the
promoter of FOSL1 even though 14-3-3 is recruited to both sites.
At the enhancer, 14-3-3 recruits the histone acetyltransferase
MOF. At the promoter, it does not. What is the difference
between 14-3-3 at the promoter and at the enhancer? Interest-
ingly, 14-3-33 and 14-3-3z are thought to be regulated via lysine
acetylation (Choudhary et al., 2009) and an acetyltransferase,
Tip60, is preferentially recruited to the promoter of FOSL1. One
possibility is that Tip60 acetylates 14-3-3 and prevents its inter-
action with MOF.
Another intriguing aspect of the study by Zippo and colleagues
is the link between H3S10 phosphorylation and H4K16 acetyla-
tion. These two modifications were previously linked in studies of
dosage compensation in the fruit fly Drosophila. In Drosophila
dosage compensation, MOF is recruited to the coding region
of X-linked genes in males where it mediates H4K16 acetylation
in a process thought to facilitate transcription elongation. Coloc-
alizing with MOF on the male X chromosome is the JIL-1 kinase,
an MSK1/2-related kinase that mediates the phosphorylation of
H3S10 on this chromosome. In the case of Drosophila dosage
compensation, recruitment of the JIL-1 kinase to the male X
chromosome occurs later than H4K16 acetylation (Wang et al.,
2001), a reversal of the order of the addition of these marks at
FOSL1 in response to serum. Concordantly, the MOF complex
that mediates acetylation in coding regions is likely to be distinct
from the MOF complex that mediates promoter/enhancer acet-
ylation of genes (Li et al., 2009). Thus, by all appearances, these
two examples of the coexistence of both H3S10 phosphorylation
and H4K16 acetylation are unrelated in their order of implemen-
tation and in their biological meaning. This suggests that
descriptions of histone modification patterns, without under-
standing the mechanisms leading to the implementation of these
marks, should be interpreted with caution. Importantly, the study
by Zippo and colleagues begins to determine the role of histone
modifications in the activation of FOSL1, with a spatial and
temporal dissection of how a cascade of histone modifications
can lead to a particular transcriptional outcome.
Trimethylation Converses with Acetylation
Another example of trans-tail crosstalk was proposed by Wang
et al. (2009). In this case the communication takes place between
the H3 and H4 tails and, like the example provided by Zippo et al.
(2009), involves H4K16 acetylation and FOSL1 transcription. By
analyzing genomewide profiles of several histone acetyltrans-
ferases, deacetylases, and modifications, these investigators
find a link between H3K4 methylation and H4K16 acetylation at
some inducible genes, including FOSL1. The authors show
that a subset of transcriptionally quiescent genes, marked by
the presence of H3K4 methylation, display a marked increase
in histone acetylation at H3K9 and H4K16 after addition of de-
acetylase inhibitors. In contrast, quiescent genes not marked
684 Cell 142, September 3, 2010 ª2010 Elsevier Inc.
with H3K4 methylation rarely show this increase in acetylation
in response to deacetylase inhibitors.
In order to determine whether H3K4 methylation is functionally
linked to H4K16 acetylation, Wang and colleagues use RNA
interference-mediated knockdown of WDR5, a common compo-
nent of the Set1 and MLL (mixed-lineage leukemia) COMPASS-
like H3K4 methyltransferase complexes. Upon knockdown of
WDR5, they observe reduced levels of histone acetylation at
the subset of transcriptionally quiescent genes marked by
H3K4 methylation. Based on this information, Wang and
colleagues suggest that H3K4 methylation primes certain genes
for an increase in H3K9 and H4K16 acetylation. Interestingly,
what was not considered by Wang and colleagues is the fact
that WDR5 is also a component of complexes that can acetylate
H4K16 or H3K9. WDR5 associates with the H4K16 acetyltrans-
ferase MOF in the NSL/MSL1v complex (Cai et al., 2010; Li
et al., 2009) as well as with the H3K9/K14 acetyltransferase
Gcn5 in the ATAC complex (Suganuma et al., 2008; Wang
et al., 2008). As such, the effect of WDR5 knockdown could be
a consequence of WDR5’s role as a subunit of the H3K4 methyl-
ases, or WDR5’s role as a subunit of the H3 and H4 acetyltrans-
ferase complexes, or a combination of the two. Given that WDR5
is part of the H3 and H4 acetyltransferase complexes, the exis-
tence of crosstalk between H3K4 and H4K16 needs to be further
characterized.
One surprising finding of the study by Wang and colleagues is
that transcription is rarely induced at the genes tested, although
Pol II is recruited following treatment with a deacetylase inhibitor
(Figure 2B). Thus, Pol II recruitment does not lead to the antici-
pated increase in transcription, despite the fact that H3K9 and
H4K16 acetylation are increased. These modifications coincide
with transcriptional activation at FOSL1 upon serum treatment
(Zippo et al., 2009). Together with the studies of H3S10 phos-
phorylation and H4K16 acetylation at the FOSL1 enhancer, it is
clear that knowing the mechanism and timing of these modifica-
tions is necessary for determining the transcriptional outcome.
The Emerging Grammar of Histone Crosstalk
The existence of a histone modification code was proposed
10 years ago as a way to approach the study of the quickly
growing number of histone modifications involved in the regula-
tion of gene expression and other DNA-templated processes,
such as replication, repair, and recombination. New ‘‘words’’ of
histone modifications are being discovered, and they continue
to appear in interesting combinations. However, discovering
the exact roles these modifications play in gene expression
has been complicated by finding a counterexample for almost
every example of crosstalk, such as the case of H3K4 and
H3K36 methylation recruiting both histone acetyltransferases
and deacetylases.
A common theme of recent research on histone crosstalk is
that the order and mechanism of the addition and removal of
histone modifications are important for the transcriptional
readout of a gene. The recent examples of histone crosstalk
that we have addressed here illustrate this point. In one study,
the implementation of H3S10 phosphorylation at two different
locations, by two different enzymes, and at two different times
after serum stimulation had disparate effects on subsequent
histone acetylation at the respective locations (Zippo et al., 2009)
(Figure 2A). Zippo and colleagues were able to propose a mech-
anism of gene activation by identifying the histone-modifying
enzymes, the histone modifications, and a set of proteins that
recognized these modifications on the FOSL1 gene after serum
stimulation. In another study, Wang and colleagues found
that artificially recreating histone modifications that correlate
with gene expression could result in the recruitment of RNA
Pol II, but this was not sufficient for transcription (Figure 2B).
Thus, simply mapping histone modification patterns without
understanding the recruitment, regulation, and interactions of
the complexes implementing these marks is not sufficient
to understand the mechanisms regulating gene expression.
Genomewide profiling techniques have now become widely
adopted, providing the ability to map histone modifications,
the enzymes implementing these marks, and the factors that
recruit them under different experimental conditions.
The study of the regulation of gene expression has grown from
identifying transcription factors and their binding sites to include
a wide variety of other binding events associated with the modi-
fications of histones that package the DNA. Future progress will
require us to learn much more about how the words comprising
the dictionary of histone crosstalk are used in a particular order
to provide the grammar of this complex language.
ACKNOWLEDGMENTS
We thank Dr. E. Park for the critical reading of the manuscript. We also thank
L. Shilatifard for editorial assistance. Studies in the Shilatifard laboratory are
supported in part by grants from the NIH: R01CA89455, R01GM069905, and
R01CA150265.
REFERENCES
Cai, Y., Jin, J., Swanson, S.K., Cole, M.D., Choi, S.H., Florens, L., Washburn,
M.P., Conaway, J.W., and Conaway, R.C. (2010). J. Biol. Chem. 285, 4268–
4272.
Cheung, P., Tanner, K.G., Cheung, W.L., Sassone-Corsi, P., Denu, J.M., and
Allis, C.D. (2000). Mol. Cell 5, 905–915.
Choudhary, C., Kumar, C., Gnad, F., Nielsen, M.L., Rehman, M., Walther, T.C.,
Olsen, J.V., and Mann, M. (2009). Science 325, 834–840.
Fischle, W., Tseng, B.S., Dormann, H.L., Ueberheide, B.M., Garcia, B.A., Sha-
banowitz, J., Hunt, D.F., Funabiki, H., and Allis, C.D. (2005). Nature 438, 1116–
1122.
Guccione, E., Bassi, C., Casadio, F., Martinato, F., Cesaroni, M., Schuchlautz,
H., Luscher, B., and Amati, B. (2007). Nature 449, 933–937.
Kim, J., Guermah, M., McGinty, R.K., Lee, J.S., Tang, Z., Milne, T.A., Shilati-
fard, A., Muir, T.W., and Roeder, R.G. (2009). Cell 137, 459–471.
Kirmizis, A., Santos-Rosa, H., Penkett, C.J., Singer, M.A., Vermeulen, M.,
Mann, M., Bahler, J., Green, R.D., and Kouzarides, T. (2007). Nature 449,
928–932.
Li, X., Wu, L., Corsa, C.A., Kunkel, S., and Dou, Y. (2009). Mol. Cell 36,
290–301.
Lo, W.S., Trievel, R.C., Rojas, J.R., Duggan, L., Hsu, J.Y., Allis, C.D., Marmor-
stein, R., and Berger, S.L. (2000). Mol. Cell 5, 917–926.
Martin, D.G., Baetz, K., Shi, X., Walter, K.L., MacDonald, V.E., Wlodarski, M.J.,
Gozani, O., Hieter, P., and Howe, L. (2006). Mol. Cell. Biol. 26, 7871–7879.
Mateescu, B., Bourachot, B., Rachez, C., Ogryzko, V., and Muchardt, C.
(2008). EMBO Rep. 9, 267–272.
Nakanishi, S., Sanderson, B.W., Delventhal, K.M., Bradford, W.D., Staehling-
Hampton, K., and Shilatifard, A. (2008). Nat. Struct. Mol. Biol. 15, 881–888.
Shi, X., Hong, T., Walter, K.L., Ewalt, M., Michishita, E., Hung, T., Carney, D.,
Pena, P., Lan, F., Kaadige, M.R., et al. (2006). Nature 442, 96–99.
Shilatifard, A. (2006). Annu. Rev. Biochem. 75, 243–269.
Suganuma, T., Gutierrez, J.L., Li, B., Florens, L., Swanson, S.K., Washburn,
M.P., Abmayr, S.M., and Workman, J.L. (2008). Nat. Struct. Mol. Biol. 15,
364–372.
Taverna, S.D., Ilin, S., Rogers, R.S., Tanny, J.C., Lavender, H., Li, H., Baker, L.,
Boyle, J., Blair, L.P., Chait, B.T., et al. (2006). Mol. Cell 24, 785–796.
Wang, Y., Zhang, W., Jin, Y., Johansen, J., and Johansen, K.M. (2001). Cell
105, 433–443.
Wang, Y.L., Faiola, F., Xu, M., Pan, S., and Martinez, E. (2008). J. Biol. Chem.
283, 33808–33815.
Wang, Z., Zang, C., Cui, K., Schones, D.E., Barski, A., Peng, W., and Zhao, K.
(2009). Cell 138, 1019–1031.
Zippo, A., Serafini, R., Rocchigiani, M., Pennacchini, S., Krepelova, A., and
Oliviero, S. (2009). Cell 138, 1122–1136.
Cell 142, September 3, 2010 ª2010 Elsevier Inc. 685
GPR120 Is an Omega-3 Fatty Acid Receptor
Mediating Potent Anti-inflammatory
and Insulin-Sensitizing Effects
Da Young Oh,1,4 Saswata Talukdar,1,4 Eun Ju Bae,1 Takeshi Imamura,2 Hidetaka Morinaga,1 WuQiang Fan,1 Pingping Li,1
Wendell J. Lu,1 Steven M. Watkins,3 and Jerrold M. Olefsky1,*
1Department of Medicine, Division of Endocrinology and Metabolism, University of California, San Diego, La Jolla, CA 92093, USA
2Divisionof Pharmacology, Shiga University of Medical Science, Tsukinowa, Seta, Otsu-city, Shiga, 520-2192 Japan
3Tethys Bioscience, 3410 Industrial Boulevard, Suite 103, West Sacramento, CA 95691, USA
4These authors contributed equally to this work
*Correspondence: jolefsky@ucsd.edu
DOI 10.1016/j.cell.2010.07.041
SUMMARY
Omega-3 fatty acids (u-3 FAs), DHA and EPA, exert
anti-inflammatory effects, but the mechanisms are
poorly understood. Here, we show that the G
protein-coupled receptor 120 (GPR120) functions as
an u-3 FA receptor/sensor. Stimulation of GPR120
with u-3 FAs or a chemical agonist causes broad
anti-inflammatory effects in monocytic RAW 264.7
cells and in primary intraperitoneal macrophages.
All of these effects are abrogated by GPR120 knock-
down. Since chronic macrophage-mediated tissue
inflammation is a key mechanism for insulin resis-
tance in obesity, we fed obese WT and GPR120
knockout mice a high-fat diet with or without u-3 FA
supplementation. The u-3 FA treatment inhibited
inflammation and enhanced systemic insulin sensi-
tivity in WT mice, but was without effect in GPR120
knockout mice. In conclusion, GPR120 is a functional
u-3 FA receptor/sensor and mediates potent insulin
sensitizing and antidiabetic effects in vivo by repres-
sing macrophage-induced tissue inflammation.
INTRODUCTION
Chronic activation of inflammatory pathways plays an important
role in the pathogenesis of insulin resistance and the macro-
phage/adipocyte nexus provides a key mechanism underlying
the common disease states of decreased insulin sensitivity
(Schenk et al., 2008). This involves migration of monocytes/
macrophages to adipose tissue (including intramuscular fat
depots) and liver with subsequent activation of macrophage
proinflammatory pathways and cytokine secretion. Through
paracrine effects, these events promote inflammation and
decreased insulin sensitivity in nearby insulin target cells
(Shoelson et al., 2007; Schenk et al., 2008). In these studies,
we explore the interlocking biology between proinflammatory
and anti-inflammatory molecules within the specialized popula-
tion of proinflammatory tissue macrophages.
G protein-coupled receptors (GPCRs) are important signaling
molecules for many aspects of cellular function. They are
members of a large family that share common structural motifs
such as seven transmembrane helices and the ability to acti-
vate heterotrimeric G proteins, such as Gas, Gai, and Gaq.
Ligands bind specifically to GPCRs to stimulate and induce
a variety of cellular responses via several second messenger
pathways; e.g., modulation of cAMP production, the phospho-
lipase C pathway, ion channels, and mitogen-activated protein
kinases (Ulloa-Aguirre et al., 1999; Gether, 2000; Schulte and
Fredholm, 2003. Recently, several groups reported that five
orphan receptors, GPR40, GPR41, GPR43, GPR84, and
GPR120, can be activated by free fatty acids (FFAs). Short-
chain fatty acids (FAs) are specific agonists for GPR41 and
GPR43 (Tazoe et al., 2008) and medium-chain FAs for GPR84
(Wang et al., 2006a). Long-chain FAs can activate GPR40
(Itoh et al., 2003) and GPR120 (Hirasawa et al., 2005). Hirasawa
et al. showed that the stimulation of GPR120 by FFAs resulted
in elevation of [Ca2+]i and activation of the ERK cascade which
suggests interactions with the Gaq family of G proteins.
However, other physiological functions of GPR120 remain to
be explored. In the current study, we found that GPR120 was
highly expressed in adipose tissue, and proinflammatory
macrophages. The high expression level of GPR120 in mature
adipocytes and macrophages indicates that GPR120 might
play an important role in these cell types. We stimulated
GPR120 with a synthetic agonist (GW9508) and omega-3 fatty
acids (u-3 FAs) and examined whether activation of GPR120
affected LPS- and TNF-a-induced inflammatory signaling
responses. While SFAs are proinflammatory and unsaturated
FAs are generally neutral, we found that u-3 FAs (docosahex-
aenoic acid (C22:6n3, DHA) and eicosapentaenoic acid
(C20:5n3, EPA)), the major ingredients in fish oil, exert potent
anti-inflammatory effects through GPR120.
b-arrestins can serve as scaffold or adaptor proteins for a wide
range of GPCRs, as well as a selected group of other receptor
subtypes (Miller and Lefkowitz, 2001). After ligand binding,
b-arrestins can associate with the cytoplasmic domains of
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 687

GPCRs and couple the receptor to specific downstream
signaling pathways, as well as mediate receptor endocytosis
(Luttrell and Lefkowitz, 2002). Here we find that b-arrestin2 asso-
ciates with ligand-stimulated GPR120 and participates in down-
stream signaling mechanisms.
Since chronic inflammation is a mechanistic feature of
obesity-related insulin resistance, we postulated that the anti-
inflammatory effect of GPR120 stimulation could promote insulin
sensitization. In the present study, we elucidate the role of
GPR120 activation in integrating anti-inflammatory and insulin
sensitizing effects in vitro and in vivo.
RESULTS
GPR120 Expression
Fatty acids (FAs) can function as endogenous ligands modu-
lating inflammatory responses, but not all FAs work in the
same way. In general, saturated FAs (SFAs) are proinflammatory,
unsaturated FAs are weakly proinflammatory or neutral, and u3-
FAs can be anti-inflammatory (Lee et al., 2003; Calder, 2005;
Solinas et al., 2007). Because of the importance of inflammation
in a number of chronic human diseases including insulin resis-
tance, obesity, and type 2 diabetes mellitus, we surveyed the
family of FA sensing GPCRs (GPR40, 41, 43, 84, and 120). Based
on its tissue expression pattern, GPR120 emerged as a receptor
of particular interest. As seen in Figure 1, GPR120 is the only lipid
688 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
Figure 1. Expression Level of GPR120 and
GPR120-Mediated Anti-inflammatory Res-
ponse in RAW 264.7 Cells
(A and B) (A) The mRNA expression pattern of
various lipid sensing GPCRs is shown in adipose
tissue, (B) CD11c+ bone marrow-derived dendritic
cells (BMDCs), bone marrow-derived macro-
phages (BMDMs), IPMacs, 3T3-L1 preadipocytes,
differentiated 3T3-L1 adipocytes, RAW 264.7
cells, and L6 myocytes. Ribosomal protein S3
(RPS3) was used as internal control.
(C) Expression of GPR120 in SVF, adipocytes and
hepatic Kupffer cells from chow (NC)- or HFD-fed
mice was examined by q-PCR. Data are ex-
pressed as the mean ± SEM of at least three
independent experiments in triplicate. *p < 0.05
versus NC.
(D) RAW 264.7 cells, transfected with scrambled
(Scr) or GPR120 #2 siRNA (GPR120 KD), were
treated with 100 mM of GW9508 for 1 hr prior to
LPS (100 ng/ml) treatment for 10 min and then
subjected to western blotting. Left panel is a repre-
sentative image from three independent experi-
ments, and the scanned bar graph (right panel)
shows fold induction over basal conditions.
Knockdown efficiency of GPR120 siRNA is shown
in Figure S1.
(E) Cytokine secretion level was measured in RAW
264.7 cells by ELISA. Data are expressed as the
mean ± SEM of three independent experiments.
*p < 0.05 versus LPS treatment in scrambled
siRNA transfected cells. See also Figures S1
and S2.
sensing GPCR which is highly expressed in adipose tissue,
proinflammatory CD11c+ macrophages (BMDCs), mature adipo-
cytes, and monocytic RAW 264.7 cells (Figure 1A and 1B).
GPR120 is induced in the stromal vascular fraction (SVF) of
adipose tissue (which contains the macrophages), as well as in
hepatic Kupffer cells, during high-fat diet (HFD) feeding in mice
(Figure 1C). GPR120 is also expressed in enteroendocrine L cells
with negligible expression in muscle (Figure S4C available
online), hepatocytes or other cell types (Hirasawa et al., 2005;
Gotoh et al., 2007).
Ligand-Stimulated GPR120 Exerts Anti-inflammatory
Effects
It has been previously reported that GPR120 signals via
a Gaq/11-coupled pathway and can respond to long chain FAs
(Hirasawa et al., 2005). To pursue the biology of GPR120,
a tool compound was needed, and, some years ago, Glaxo pub-
lished GW9508 as a GPR40 selective agonist. However, this
compound was not specific and also stimulated GPR120 (Bris-
coe et al., 2006). Since macrophages and adipocytes do not
express GPR40 (this was confirmed by repeated q-PCR and
RT-PCR measures, Figure S1A), GW9508 is a functional
GPR120 specific compound in these cell types. Using this
approach, we found that GW9508 treatment broadly and mark-
edly repressed the ability of the TLR4 ligand LPS to stimulate
inflammatory responses in RAW 264.7 cells (Figure 1D and E).

Thus, GW9508 inhibited LPS stimulated phosphorylation of IKKb
and JNK, prevented IkB degradation, and inhibited TNF-a and
IL-6 secretion. All of these effects of GW9508 were completely
abrogated by siRNA mediated knockdown of GPR120 (Figures
1D and 1E and Figure S1F).
Based on these remarkable anti-inflammatory effects of
GPR120 stimulation, we established a cell based reporter
system by transfecting HEK293 cells with constructs for
GPR120 along with a serum response element-luciferase
promoter/reporter (SRE-luc). Since GPR120 is a Gaq/11-
coupled receptor, it stimulates both PKC and MAP kinase, and
both of these biologic effects are detected by the SRE-driven
reporter system (Oh et al., 2005). The reporter cells were treated
with various FAs and the synthetic GW9508 ligand. We found
that GW9508, the u-3 FAs (DHA and EPA) and palmitoleate
(C16:1n7), all activated the SRE-luc reporter with an EC50 of
1-10 mM (Figure 2A), while SFAs were without effect. GW9508
and DHA were used at 100 mM in all subsequent studies to
achieve maximal action. The u-3 FAs (DHA and a-linolenic
acid), and SFA (palmitic acid (C16:0)) activated ERK phosphory-
Figure 2. Omega-3 FA Stimulates GPR120
and Mediates Anti-inflammatory Effects
(A–D) GPR120-mediated SRE-luc activity after
treatment with various FAs. Results are fold
activities over basal. Each data point represents
mean ± SEM of three independent experiments
performed in triplicate. Black lines indicate SRE-
luc activities without GPR120 transfection or with
non-stimulating FAs. DHA inhibits LPS-induced
inflammatory signaling (B), cytokine secretion (C),
and inflammatory gene mRNA expression level
(D) in RAW 264.7 cells, but not in GPR120 knock-
down cells.
(E and F) GPR120 stimulation inhibits LPS-induced
inflammatory response in WT primary macro-
phage. Data are expressed as the mean ± SEM
of three independent experiments. *p < 0.05
versus LPS treatment in scrambled siRNA trans-
fected cells or WT IPMacs. See also Figure S2.
lation in RAW 264.7 cells, but only DHA-
and a-linolenic acid-mediated ERK phos-
phorylation were abolished by GPR120
knockdown (Figure S2A). These results
indicate that u-3 FAs, but not SFAs,
specifically activated ERK via GPR120.
The activation of GPR120 by u-3 FAs,
as well as its expression in adipocytes
and macrophages, led us to study
whether DHA, a representative u-3 FA,
can modulate inflammation through
GPR120 in these cells. To examine this,
we pretreated RAW 264.7 cells and
3T3-L1 adipocytes with GW9508 or DHA
for 1 hr, followed by LPS (TLR4), TNF-a,
TLR2, or TLR3 stimulation, respectively.
We found that GW9508 and, more impor-
tantly, DHA, strongly inhibited LPS-
induced phosphorylation of JNK and IKKb, IkB degradation,
cytokine secretion and inflammatory gene expression level in
RAW 264.7 cells (Figures 2B–2D) as well as TNF-a, TLR2 and
TLR3-induced JNK and IKKb phosphorylation in 3T3-L1 adipo-
cytes (Figure S2B) or RAW 264.7 cells (Figure S2C). All of the
effects of GW9508 and DHA were completely prevented by
GPR120 knockdown, demonstrating that these anti-inflamma-
tory effects were specifically exerted through GPR120 (Figure 1,
Figure 2, Figure S1, and Figure S2). Similar results were seen in
primary wild-type (WT) intraperitoneal macrophages (IPMacs)
and GPR120 knockout (KO) IPMacs (Figures 2E and 2F). These
data argue that GPR120 is an u-3 FA receptor or sensor, and
provide a molecular mechanism for the anti-inflammatory effects
of this class of FAs.
Role of b-arrestin2 in GPR120 Signaling
Given these potent cell selective anti-inflammatory effects,
it was of interest to understand the specific mechanisms
whereby signals from GPR120 inhibit inflammatory pathways.
To further assess this, we used RNA interference to examine
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 689

molecules involved in generation of GPR120 signals. As seen in
Figures 3A and 3B, LPS signaling was not affected by b-ar-
restin1, -2, or Gaq/11 knockdown. However, with b-arrestin2
knockdown, DHA-mediated anti-inflammatory signaling was ex-
tinguished, while b-arrestin1 and Gaq/11 knockdown were
without effect (Figure 3A).
Figure 3A and Figure S2 show that GPR120 stimulation inhibits
both TLR4- and TNF-a mediated inflammatory responses. Since
the TNF-a and TLR signaling cascades converge downstream of
GPR120 activation, these results indicate that the site of
GPR120-induced inhibition is either at, or upstream, of JNK/
IKKb. LPS activates inflammation through the TLR4 pathway
by engaging the serine kinase IRAK, leading to phosphorylation
of transforming growth factor-b activated kinase 1 (TAK1) which
is upstream of MKK4/JNK and IKKb (Kawai and Akira, 2006,
Figure 3I). TNF-a and TLR2/3 also leads to stimulation of
TAK1, resulting in activation of IKKb and JNK (Takaesu et al.,
2003). Consequently, we determined whether DHA stimulation
of GPR120 inhibited TAK1 and MKK4. As seen in Figure 3C,
DHA treatment abrogated LPS-induced TAK1 and MKK4 phos-
690 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
Figure 3. GPR120 Internalization with
b-arrestin2 Mediates Anti-inflammatory
Effects
(A) RAW 264.7 cells were transfected with siRNA as
indicated and stimulated with or without 100 mM of
DHA 1 hr prior to LPS (100 ng/ml) treatment for
10 min and then subjected to western blotting.
(B) TNF-a secretion was measured in RAW 264.7
cell cultured media with or without RNA interfer-
ence as indicated.
(C) Phosphorylation of TAK1 and MKK4 in RAW
264.7 cells with or without siRNA transfection as
indicated.
(D) HEK293 cells were cotransfected with HA-
GPR120 and b-arrestin2$GFP to analyze
GPR120 internalization after DHA stimulation for
the indicated times. GPR120 (red) and b-arrestin2
(green) were localized by confocal microscopy.
(E–H) (E) Coimmunoprecipitation between GPR120
and b-arrestin2 with DHA stimulation for 30 min
in RAW 264.7 cells and, (F) HEK293 cells (HA-
GPR120 and b-arrestin2$GFP), respectively. Lysate
indicates 1/10 input in each experiment. Interaction
between TAB1 and b-arrestin2 (G) and interaction
between TAB1 and TAK1 (H) were detected by
coimmunoprecipitation and the scanned bar graph
quantitates the association in RAW 264.7 cells.
(I) Schematic diagram of the b-arrestin2 and
GPR120-mediated anti-inflammatory mechanism.
Red colored letters and arrows indicate the DHA-
mediated anti-inflammatory effect, and black
colored letters and arrows indicate the LPS- and
TNF-a-induced inflammatory pathway. See also
Figure S2.
phorylation in a GPR120 and b-arrestin2-
dependent manner. Since TLR2/3/4 and
TNF-a signaling were inhibited by
GPR120 activation, these results indicate
that DHA signaling intersects at TAK1 and
inhibits all upstream input activating signals via a GPR120/b-
arrestin2 interaction (Figure 3I).
After ligand stimulation, b-arrestin2 can translocate to
a number of GPCRs where it mediates receptor internalization
and signaling (Barak et al., 1997). We transfected HEK293 cells
with b-arrestin2$GFP to visualize intracellular trafficking of b-ar-
restin2 following activation of GPR120 (Figure 3D). In the basal
state, GPR120 was localized to the plasma membrane as
assessed by immunostaining (red fluorescence, Figure 3D),
while b-arrestin2 exhibited a diffuse, largely cytoplasmic stain-
ing pattern (green, Figure 3D). Following DHA treatment for
5 min, b-arrestin2$GFP translocated from the cytosol to the
plasma membrane and can be seen colocalized with GPR120
(merged, right fields). After 30 min of DHA treatment, much of
the GPR120 is internalized, as visualized by punctate intracel-
lular staining (lower left panel), and b-arrestin2$GFP is now
colocalized with the intracellular GPR120 (lower right panel,
Figure 3D). DHA-stimulated binding of b-arrestin2 to activated
GPR120 was also detected by coimmunoprecipitation
(Figure 3E and F).

LPS or TNF-a signaling activate TAK1 by causing the associ-
ation of TAK1 binding protein 1 (TAB1) with TAK1. Figure 3H
shows that LPS stimulation of RAW 264.7 cells causes TAB1/
TAK1 association. DHA treatment leads to the association of
b-arrestin2 with TAB1 (Figure 3G) and largely blocks TAK1/
TAB1 association (Figure 3H). To further examine the interaction
site of b-arrestin2 and GPR120 or TAB1, we pursued coimmuno-
precipitation with a series of b-arrestin2 truncation/deletion
mutants (Figure S2D). Full-length b-arrestin2 was able to bind
to GPR120 and TAB1 only in the presence of DHA, clearly
showing the DHA dependency of this interaction. Interestingly,
only the full-length b-arrestin2 coprecipitated with GPR120 and
TAB1, while a series of deletion/truncation b-arrestin2 mutants
did not, indicating that the interactions are dependent on the
complete tertiary structure of b-arrestin2 (Figure S2D; Luttrell
et al., 1999). Taken together, these results suggest that
GPR120 activation leads to association of b-arrestin2 with the
receptor and that this complex subsequently internalizes, where-
upon b-arrestin2 can bind to TAB1. The data further suggest that
association of b-arrestin2 with TAB1 blocks TAB1/TAK1 binding,
resulting in inhibition of TAK1 phosphorylation and activation
(Figure 3I).
Figure 4. GPR120 Activation Enhances
GLUT4 Translocation and Glucose Uptake
(A) 3T3-L1 adipocytes were transfected with
a dually tagged HA-GLUT4-GFP construct. Total
GLUT4 expression was determined by GFP fluo-
rescence, and GLUT4 translocation to the cell
surface after 100 ng/ml insulin or 100 mM DHA
stimulation for 30 min was determined by indirect
immunofluorescence of the HA-conjugated with
Alexa 594 in fixed cells. Translocation following
insulin stimulation was expressed as a percentage
of the maximum response. The bar graph
represents the mean ± SEM data from four inde-
pendent experiments. *p < 0.05 versus vehicle
treatment. (B) Glucose uptake was measured in
WT and GPR120 KO mouse primary adipose
tissue and in (C–H) 3T3-L1 adipocytes ± siRNA
with the indicated treatment. Data are expressed
as mean ± SEM of three independent experiments
in triplicate. *p < 0.05 versus basal activity. The
indicated siRNA knockdown efficiency was vali-
dated by western blotting. See also Figure S3.
GPR120 Activation Enhances
Glucose Uptake in 3T3-L1
Adipocytes
Since our data show that GPR120 is
expressed in mature adipocytes and sig-
nals through Gaq/11 in these cells, we
assessed the effects of GPR120 stimula-
tion on insulin sensitivity in primary adi-
pose tissue cultures and in 3T3-L1 adipo-
cytes. Primary adipose tissue explants
and 3T3-L1 adipocytes were pretreated
for 30 min with GW9508 or DHA, followed
by measurement of basal and insulin
stimulated GLUT4 translocation (Fig-
ure 4A) and 2-deoxyglucose (2-DOG) transport (Figures 4B–
4H). Ligand-stimulation of GPR120 led to an increase in glucose
transport and translocation of GLUT4 to the plasma membrane
in adipocytes, but was without effect in muscle cells (Figure S4D)
which don’t express GPR120 (Figure 1B and Figure S4C). This
stimulatory effect of DHA and GW9508 was blocked when
GPR120 or Gaq/11 was depleted by siRNA knockdown
(Figure 4D and Figure 4F). GLUT4 knockdown also blocked the
effects of DHA and GW9508, while the effects of insulin were
decreased by 90% (Figure 4E). This, along with the GLUT4
translocation data provided in Figure 4A, indicates that the
stimulatory effects of GPR120 are indeed working through
GLUT4. Further assessment of this pathway showed that DHA
had a modest effect to stimulate phosphorylation of Akt, but
that this was abrogated with GPR120 knockdown (Figure S3A).
The effects of DHA to stimulate Akt were blocked by inhibiting
PI3 kinase with LY294002 (Figure S3B). Finally, DHA did not
stimulate IRS-1 phosphorylation (Figure S3C), indicating that
its glucose transport stimulatory effects were downstream of
IRS-1. Knockdown of Gaq/11 also completely blocked the
effects of DHA to stimulate glucose transport (Figure 4F), while
b-arrestin1 or -2 knockdown was without effect (Figures 4G
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 691

and 4H). Interestingly, Gaq/11 knockdown not only inhibited
DHA and GW9508 stimulated glucose transport, but it also atten-
uated insulin stimulatory effects, and the latter is fully consistent
with previous publication (Imamura et al., 1999) showing the role
of Gaq/11 in insulin signaling to glucose transport in adipocytes.
This scheme is shown in Figure S3D.
In Vivo Metabolic Studies in GPR120 KO Mice
Since chronic tissue inflammation can cause insulin resistance,
we hypothesized that deletion of GPR120 would enhance the
proinflammatory tone, promoting glucose intolerance and
decreased insulin sensitivity. To test this idea, GPR120
KO mice and WT littermates were evaluated on normal chow
diet (NC). Body weights were similar in both groups, and as
summarized in Figure 5, glucose tolerance tests (GTT) showed
a mild degree of impairment in GPR120 KO animals compared
to WTs (Figure 5A). More impressively, insulin secretion was
more than 2-fold greater in the KO animals, and the combination
of hyperinsulinemia and mild glucose intolerance indicates the
presence of insulin resistance (Figures 5B and 5C). This was
confirmed by performing hyperinsulinemic/euglycemic clamp
studies in the chow fed WT and KO mice (Figure 5D). These
studies revealed a 31% decrease in the glucose infusion rate
692 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
Figure 5. In Vivo Metabolic Studies in
GPR120 KO Mice
(A) GTT in NC-fed WT and GPR120 KO mice. n = 7
per group.
(B and C) Insulin concentration was measured at
the indicated time points and (C) area-under-curve
analysis of the insulin data shows a significant differ-
ence between WT and GPR120 KO mice on NC.
(D) Hyperinsulinemic/euglycemic clamp studies in
chow-fed WT and GPR120 KO mice.
(E) Clamp studies in HFD, u-3 FA supplemented
(+u3), and Rosiglitazone treated HFD mice
(+Rosi). n = 8 per group, *p < 0.05 compared to
HFD-fed WT group.
(F) Mean ± SEM plasma concentration (mole (%))
of DHA and EPA for each diet in WT and GPR120
KO mice. n = 7 per each group. *p < 0.05,
compared to NC, and **p < 0.05 compared to
HFD. Data are represented as mean ± SEM. See
also Figure S4, Figure S5, Figure S6, and Table S2.
(GIR) required to maintain euglycemia in
the KO mice. Since 70%–80% of total
body insulin stimulated glucose disposal
is accounted for by skeletal muscle
glucose uptake (Baron et al., 1988), the
decreased insulin stimulated (IS)-glucose
disposal rate (GDR) provides direct
evidence for skeletal muscle insulin resis-
tance in the KO mice. Likewise, the
GPR120 KO mice exhibited a marked
decrease in the ability of insulin to sup-
press hepatic glucose production (HGP),
demonstrating the presence of hepatic
insulin resistance. Thus, the decreased
GIR was 50% related to muscle and 50% due to liver insulin
resistance, respectively. Since the chow diet contains exoge-
nous u-3 FAs, we conclude that blunted u-3 FA signaling in the
KO mice, accounts for the decreased insulin sensitivity.
Since u-3 FA administration can improve insulin sensitivity in
rats (Buettner et al., 2006), we reasoned that u-3 FA supplemen-
tation could alleviate HFD/obesity-induced insulin resistance in
WT mice, but would be ineffective in GPR120 KOs. Accordingly,
WT and GPR120 KO mice were placed on 60% HFD for 15 weeks.
At this point, separate groups of 15 mice each, were treated for
five additional weeks with 60% HFD or an isocaloric HFD diet con-
taining 27% fish oil supplementation enriched in u-3 FAs. This diet
provided 50 and 100 mg of DHA and EPA, respectively, per
mouse, per day. Figure 5E shows that administration of the u-3
FA diet led to improved insulin sensitivity with increased glucose
infusion rates, enhanced muscle insulin sensitivity (increased IS-
GDR), greater hepatic insulin sensitivity (increased HGP suppres-
sion), and decreased hepatic steatosis (Figures S6A and S6B).
Importantly, the u-3 FA diet was completely without effect in the
GPR120 KO mice. A separate group of WT mice were treated
with the insulin sensitizing thiazolidinedione Rosiglitazone, and
the effects of u-3 FAs were equal to or greater (HGP suppression)
than the effects of this clinically used insulin sensitizing drug.
 In addition to improving hepatic insulin sensitivity, u-3 FA
treatment had a beneficial effect on hepatic lipid metabolism,
causing decreased liver triglycerides, DAGs, along with reduced
SFA and u-6 FA content in the various lipid classes (Figures
S6A–S6C and Table S2). The u-3 FA supplementation was
entirely without effect, or much less effective, at reducing hepatic
lipid levels in the GPR120 KOs.
Interestingly, in the absence of u-3 FA supplementation,
GPR120 KO mice were just as susceptible to HFD-induced
insulin resistance as were the WT mice. We hypothesize that
this was because the 60% HFD is relatively deficient of exoge-
nous u-3 FAs, so that ligands for GPR120 were relatively absent
in these animals. To assess this, we performed a lipomics anal-
ysis of the various fatty acid classes in the chow and HFD-fed WT
and KO mice. As predicted, circulating concentrations of u-3
FAs were much lower on HFD compared to chow diets, and
the administration of the u-3 FA supplement to the HFD led to
a large increase in plasma u-3 FA content in both genotypes
(Figure 5F). This would account for the relative lack of effect of
GPR120 KO on HFD alone, since u-3 FA ligand stimulation is
negligible, while the KO animals displayed an insulin resistant
phenotype on chow diets when a moderate level of u-3 FAs
was provided. Importantly, the GPR120 KO mice are completely
refractory to the insulin sensitizing effects of u-3 FA administra-
tion on HFD.
To address the contribution of macrophages to the overall
in vivo phenotype, we performed bone marrow transplantation
(BMT) from GPR120 KOs into irradiated WT mice (adoptive
transfer) to generate hematopoietic cell deletion of GPR120.
The studies in the BMT WT and BMT GPR120 KO mice on
chow diet revealed a highly significant 20%–30% decrease in
GIR in the KOs, with a more dramatic impairment in the ability
of insulin to suppress hepatic glucose production (Figure S4A).
Thus, the studies in the BMT animals on the chow diet are
comparable to the results (Figure 5D) observed in WT versus
whole body GPR120 KOs on chow diet. When studied on the
HFD ± u-3 FA supplementation (Figure S4B), the u-3 FA supple-
mented BMT GPR120 KO animals exhibited a 30% decrease in
GIR compared to the u-3 FA supplemented BMT WTs. This was
explained by skeletal muscle insulin resistance (decreased IS-
GDR) and hepatic insulin resistance (decreased HGP suppres-
sion) in the GPR120 KOs compared to the WT BMT mice on
the u-3 FA supplemented HFD. These data are fully consistent
with the results in the global KOs (Figure 5E) and reinforce the
concept that the in vivo phenotype we observed can be largely
traced to hematopoietic cells/macrophages.
Omega-3 FAs Reduce Inflammatory Macrophages
in Adipose Tissue
We conducted histologic examination of adipose tissue
macrophages (ATMs) from WT and GPR120 KO mice on HFD
or the u-3 FA enriched HFD by immunostaining for the M1
macrophage marker F4/80 and the M2 macrophage marker
MGL1 (Lumeng et al., 2008) (Figure 6A). Consistent with previous
studies (Weisberg et al., 2003; Xu et al., 2003; Nguyen et al.,
2007), HFD induced a large increase in F4/80 positive ATMs,
which form crown-like structures (CLS) around adipocytes in
both WT and GPR120 KO mice. In contrast, MGL1 staining
was minimal in both groups on HFD (Figure 6A). On the u-3 FA
diet, we observed a decrease in F4/80 staining, along with
a marked increase in MGL1 positive cells in WT mice. Impor-
tantly, no change in F4/80 or MGL1 staining was noted in the
GPR120 KO mice on the u-3 FA diet. SVFs were prepared
from adipose tissue and analyzed by flow cytometry to quanti-
tate the total number of ATMs, as well as the content of
CD11b+ and CD11c+ and negative macrophage subpopulations
(Figure 6B). HFD led to a large but comparable increase in
CD11b+ and CD11c+ ATM content in WT and GPR120 KO
mice (Figure 6B, middle panel). Treatment with the u-3 FA-en-
riched HFD caused a striking decrease in CD11b+ and CD11c+
ATMs in WT mice, but was without effect in the GPR120 KO
group (Figure 6B, right panel). Thus, the FACS analysis was fully
consistent with the histological results. Interestingly, CD11c+
ATM content was also greater in the GPR120 KOs on the chow
diet relative to WT consistent with the insulin resistance in the
KO animals.
It seemed possible that the reduction in ATM content in WT
animals on the u-3 FA diet reflected decreased chemotaxis of
macrophages. To test this hypothesis, we measured the migra-
tory capacity of IPMacs from WT and GPR120 KO mice using
an in vitro transwell chemotaxis assay. As seen in Figure 6C,
macrophages from both groups readily migrated toward condi-
tioned media (CM) harvested from 3T3-L1 adipocytes. Pretreat-
ment of macrophages with DHA for 3 hr before exposure to CM
led to an 80% inhibition of chemotactic capacity in WT macro-
phages, but had no significant effect on IPMacs obtained from
the GPR120 KO mice. Similar experiments were performed
using the specific chemokine, monocyte chemotactic protein-1
(MCP-1) as a chemoattractant, rather than CM, and these exper-
iments yielded identical results (Figure 6D). These data indicate
that u-3 FAs cause decreased macrophage chemotaxis by
acting through the GPR120 receptor, contributing to the differ-
ences in ATM content seen in Figures 6A and 6B.
Omega-3 FAs Decrease M1 Proinflammatory Gene
and Increase M2 Anti-inflammatory Gene Expression
in Adipose Tissue
As shown in Figure 7A, expression of M1 inflammatory genes
such as IL-6, TNF-a, MCP-1, IL-1b, iNOS, and CD11c was
increased by HFD compared to chow diet in both genotypes,
and was reduced in the u-3 FA treated WT mice, but not in
the GPR120 KO mice. Even on chow diet, expression of several
inflammatory genes was higher in GPR120 KOs compared to
WT, consistent with the insulin resistance observed in the
chow-fed KO mice. Expression of the M2 anti-inflammatory
genes, arginase 1, IL-10, MGL1, Ym-1, Clec7a, and MMR
was increased by u-3 FAs in WT, but not in the GPR120 KO
adipose tissue (Figure 7B). These results are consistent with
Figure 6 and demonstrate that the dietary change from HFD
to u-3 FA supplemented HFD resulted in an overall decreased
proinflammatory profile in adipose tissue from WT, but not in
GPR120 KO mice. These changes in gene expression were
predominantly manifested in the SVF, except for MCP-1 and
IL-6, which are known to be readily expressed in adipocytes
(Figure S7). Qualitatively similar results were seen in the liver
(Figures S6D and S6E).
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 693

DISCUSSION
In this report we show that GPR120 functions as an u-3 FA
receptor/sensor in proinflammatory macrophages and mature
adipocytes. By signaling through GPR120, DHA and EPA (the
major natural u-3 FA constituents of fish oil), mediate potent
anti-inflammatory effects to inhibit both TLR and TNF-a inflam-
matory signaling pathways. The mechanism of GPR120-medi-
ated anti-inflammation involves inhibition of TAK1 through a
b-arrestin2/TAB1 dependent effect. Since chronic tissue inflam-
mation is an important mechanism causing insulin resistance
(Xu et al., 2003; Shoelson et al., 2007; Schenk et al., 2008), the
anti-inflammatory actions of u-3 FAs exert potent insulin sensi-
tizing effects. The in vivo anti-inflammatory and insulin sensi-
tizing effects of u-3 FAs are dependent on expression of
GPR120, as demonstrated in studies of obese GPR120 KO
animals and WT littermates. Thus, GPR120 is highly expressed
in proinflammatory macrophages and functions as an u-3 FA
receptor, mediating the anti-inflammatory effects of this class
694 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
Figure 6. Omega-3 FA Enriched Diet Decreases
Inflammatory Macrophage Infiltration in Adipose
Tissue
(A) Confocal merged images from epididymal fat pads
from HFD and u-3 FA enriched HFD (HFD+u3)-fed WT
and GPR120 KO mice, costained with anti-F4/80 (green)
and anti-Caveolin1 (blue) antibodies, left 4 panels, or
anti-MGL1 (green) and anti-Caveolin1 (red) antibodies,
right 4 panels. The image is representative of similar
results from three to four independent experiments. Scale
bar represents 100 mm.
(B) Dot plot representation of CD11b versus CD11c
expression for FACS data obtained from adipose tissue
SVF of NC, HFD or HFD+u3-fed WT and GPR120 KO.
Scattergram is representative from three independent
mice from each group.
(C and D) Migratory capacity of IPMacs from WT and
GPR120 KO mice as measured using an in vitro transwell
chemotaxis assay as described under supplemental
experimental procedures. Data are expressed as mean ±
SEM of three independent experiments in triplicate. *,
p<0.05 versus CM treatment.
of FAs to inhibit both the TLR2/3/4 and the
TNF-a response pathways and cause systemic
insulin sensitization.
GPR120 is a Gaq/11-coupled receptor, and
since it is expressed in enteroendocrine L cells,
past interest in this receptor has focused on its
potential ability to stimulate L cell GLP-1 secre-
tion. In the current study, we show that, in addi-
tion to L cells, GPR120 is highly expressed in
proinflammatory, M1-like macrophages and
mature adipocytes, with negligible expression
in muscle, pancreatic b-cells, and hepatocytes
(Gotoh et al., 2007). In the HFD/obese mouse
model, GPR120 expression is highly induced in
ATMs as well as resident liver macrophages
(Kupffer cells). To explore the biology around
GPR120, we established an artificial reporter
cell assay and found that the u-3 FAs, DHA, and EPA, are ligands
for GPR120, and comparable to the effects of a non-selective
GPR120 tool compound (GW9508), the u-3 FAs exert potent
anti-inflammatory effects in macrophages. Our results also re-
vealed the molecular mechanisms underlying these anti-inflam-
matory effects. Thus, DHA stimulation of GPR120 inhibits both
the TLR2/3/4 and TNF-a proinflammatory cascade. Since acti-
vation of IKKb and JNK are common to TLR and TNF-a signaling,
this indicates that the locus of GPR120 inhibition is at or proximal
to these kinases. TAK1 activation stimulates both the IKKb/NFkB
and JNK/AP1 pathways, and the TLR and TNF-a signaling path-
ways converge at this step. Our data show that stimulation of
GPR120 specifically inhibits TAK1 phosphorylation and activa-
tion providing a common mechanism for the inhibition of both
TLR and TNF-a signaling.
Beta-arrestins can serve as important adaptor and scaffold
molecules mediating the functions of a number of different
GPCRs, as well as other receptor subtypes (Miller and Lefkowitz,
2001). The C-terminal region of GPR120 contains several

putative b-arrestin2 binding motifs [(S/T)X4-5(S/T); Cen et al.,
2001], but whether b-arrestins play any role in GPR120 function
was unknown. Here we find that activation of GPR120 by DHA
stimulation leads to association of the receptor with b-arrestin2,
but not b-arrestin1, and that the anti-inflammatory effects of
GPR120 are completely b-arrestin2 dependent. Functional
immunocytochemical studies showed that DHA stimulation
leads to recruitment of b-arrestin2 to the plasma membrane
where it colocalizes with GPR120. This is followed by receptor
and b-arrestin2 internalization, where the two are now colocal-
ized in the cytoplasmic compartment. TAB1 is the activating
protein for TAK1 and our results show that following DHA-stim-
ulated internalization of the GPR120/b-arrestin2 complex,
b-arrestin2 can now associate with TAB1, as measured in coim-
munoprecipitation experiments; only full-length b-arrestin2 was
capable of interacting with GPR120 and TAB1. This apparently
blocks the association of TAB1 with TAK1, inhibiting TAK1
activation and downstream signaling to the IKKb/NFkB and
JNK/AP1 system. These results provide a mechanism for the
b-arrestin2-mediated inhibition of TLR4, TNF-a, and TLR2/3
action. Other studies in the literature are consistent with these
findings, since it has been shown that b-arrestin2 can inhibit
NFkB signaling in other systems (Gao et al., 2004; Wang et al.,
2006b). Furthermore, Lefkowitz’ group has recently published
an extensive proteomics analysis of b-arrestin2 interacting
partners, and among the 266 proteins they identified, TAB1
was represented on the list (Xiao et al., 2007).
Interestingly, the anti-inflammatory effects mediated by
GPR120 were entirely dependent on b-arrestin2, but indepen-
dent of Gaq/11, despite the fact that GPR120 can be a Gaq/
Figure 7. M1 and M2 Inflammatory Gene Expres-
sion Levels in Adipose Tissue from WT versus
GPR120 KO Mice
Relative mRNA levels for M1 proinflammatory genes (A)
and M2 anti-inflammatory genes (B) in NC, HFD, or
HFD+u3 (+u3)-fed WT and GPR120 KO mice, as measured
by q-PCR. Data are expressed as mean ± SEM of three
independent experiments in triplicate. n=7 per group, *,
p<0.05 compared to the HFD-fed WT group. **, p<0.05
compared to the WT versus GPR120 KO on NC. See also
Figure S7. Primer sequences are shown in Table S1.
11-coupled receptor in other contexts. This
provides further evidence demonstrating the
concept that a single GPCR can independently
signal through multiple pathways. In previous
studies, we have demonstrated that Gaq/11
activation can lead to stimulation of GLUT4
translocation in adipocytes. Since GPR120
was expressed in mature adipocytes, but not
preadipocytes, we explored the potential role
of GPR120 in glucose transport control. Inter-
estingly, we found that DHA stimulation of
GPR120 in 3T3-L1 adipocytes increased
GLUT4 translocation to the cell surface with
a subsequent increase in glucose transport
into the cells. RNA interference studies showed
that the DHA effect on glucose uptake was GPR120, GLUT4, and
Gaq/11 dependent, but independent of b-arrestin2. This effect
was about 30%–50% as great as the effect of insulin and the
actions of DHA on glucose uptake were additive to those of
a submaximally stimulating concentration of insulin. From this,
it is possible to propose that these insulinometic effects con-
tribute to the overall insulin sensitizing actions of u-3 FAs.
However, muscle glucose uptake accounts for the great majority
(70%–80%) of insulin stimulated glucose disposal. Furthermore,
GPR120 is not expressed in muscle, and DHA did not stimulated
glucose uptake in L6 myocytes (Figures S4C and S4D). In addi-
tion, acute administration of DHA had no stimulatory effects on
IS-GDR (Figure S4E). This reports the conclusion that the
in vivo stimulatory effects of DHA on GDR are related to
anti-inflammation, and that the glucose transport stimulatory
effects in adipocytes contribute little to the overall phenotype.
Since chronic, low grade tissue inflammation is an important
cause of obesity-related insulin resistance, we reasoned that
the anti-inflammatory effects of GPR120 stimulation should be
coupled to insulin sensitizing actions in vivo. This idea was
confirmed in studies of WT and GPR120 KO mice. On a chow
diet, the lean GPR120 KO mice were glucose intolerant, hyperin-
sulinemic and displayed decreased skeletal muscle and hepatic
insulin sensitivity, as measured during glucose clamp studies.
They also displayed increased ATM content, relative to WT
mice, and a 2- to 5-fold higher expression level of the M1 proin-
flammatory markers, MCP-1, iNOS, and IL-6 (Figure 7A). On
HFD, GPR120 KO and WT mice became equally obese and
insulin resistant. Importantly, u-3 FA supplementation markedly
increased insulin sensitivity in WT mice but was without effect in
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 695
the GPR120 KOs. Consistent with these results, u-3 FA treat-
ment led to a decrease in ATM accumulation with reduced
adipose tissue markers of inflammation in WT, but not in KO
mice. In addition to direct anti-inflammatory effects in macro-
phages, DHA treatment inhibited the ability of primary WT
macrophages to migrate toward adipocyte CM. This could be
due to DHA-induced decreased chemokine secretion or down-
regulation of chemokine receptors, or both. In addition, it is
possible that DHA, by signaling through GPR120, can mediate
heterologous desensitization of other GPCR chemokine recep-
tors. We also observed a concomitant increase in M2 markers,
such as IL-10, arginase 1, MGL1, Ym-1, Clec7a, and MMR.
This latter finding raises the possibility that u-3 FAs can redirect
ATMs from an M1 to an M2 polarization state. Taken together,
these mechanisms account for the decreased inflammatory
state. The in vivo anti-inflammatory actions of u-3 FAs are
consistent with the insulin sensitizing effects of these agents
and are fully dependent on the presence of GPR120, indicating
a causal relationship. Finally, the adoptive transfer studies
showed that hematopoietic cell GPR120 deletion results in
a comparable insulin resistant, u-3 FA non-responsive pheno-
type as seen in the global GPR120 KOs, indicating that this
phenotype can be traced back to inflammatory events in
macrophages.
We also performed a detailed in vivo lipidomic analysis of FAs
in the different lipid classes in the liver (Table S2). The results
showed that HFD leads to an increase in total TAGs, DAGs, total
SFAs, monounsaturated FAs and u-6 FAs in WT mice, while all of
these lipid changes are ameliorated with u-3 FA treatment. In the
GPR120 KO mice, all of these lipids are elevated on HFD to the
same extent as in WT mice, but, u-3 FA supplementation was
either ineffective or much less effective. These results are
consistent with the view that the reversal of steatosis/non-
alcoholic fatty liver disease (NAFLD) by u-3 FA treatment is
mediated, in large part, by GPR120 and that the GPR120 KO
mice are predisposed toward NAFLD even in the context of
u-3 FA supplementation.
Dietary DHA is rapidly esterified into chylomicrons during the
process of gastrointestinal absorption, and is also packaged
into VLDL triglycerides by the liver. DHA can also be esterified
into phospholipids and cholesterol esters associated with
circulating lipopoproteins and only a small proportion (5%) of
total plasma DHA is found in the FFA pool. Through the action
of lipoprotein lipase bound to the luminal surface of endothelial
cells, u-3 FAs are cleaved from circulating triglycerides where
they can act as ligands or be taken up by peripheral tissues
(Polozova and Salem Jr., 2007). Recent studies have also indi-
cated that metabolic products derived from u-3 FAs, such as
17S-hydroxy-DHA, resolvins, and protectins may play a role in
the long term resolution of inflammation and this might attenuate
insulin resistance in the context of obesity (González-Périz et al.,
2009). If this proves to be correct, then this could provide an
additional mechanism for long term u-3 FA-induced anti-inflam-
matory, insulin sensitizing effects. However, in the current
studies, we found that these u-3 FA derivatives were unable to
stimulate GPR120 activation in our reporter cell assay (data not
shown), indicating that GPR120 functions as a receptor for u-3
FAs and not their biochemical products. Resolution of inflamma-
696 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
tion versus anti-inflammatory actions are distinct processes, and
it is certainly possible that the products derived from u-3 FA
metabolism work on the former but not the latter.
In summary, we have found that GPR120 functions as an u-3
FA receptor/sensor and mediates robust and broad anti-inflam-
matory effects, particularly in macrophages. After ligand stimula-
tion, GPR120 couples to b-arrestin2 which is followed by
receptor endocytosis and inhibition of TAB1-mediated activation
of TAK1, providing a mechanism for inhibition of both the TLR
and TNF-a proinflammatory signaling pathways. Since chronic
tissue inflammation is linked to insulin resistance in obesity, we
used GPR120 KO mice to demonstrate that u-3 FAs cause
GPR120-mediated anti-inflammatory and insulin sensitizing
effects in vivo. Overall, these results strongly argue that anti-
inflammatory effects can ameliorate insulin resistance in obesity.
Taken together, GPR120 emerges as an important control point
in the integration of anti-inflammatory and insulin sensitizing
responses, which may prove useful in the future development
of new therapeutic approaches for the treatment of insulin resis-
tant diseases.
EXPERIMENTAL PROCEDURES
Chemicals and Reagents
GW9508 was purchased from Tocris bioscience (Ellisville, MO) and DHA was
from Cayman chemical (Ann Arbor, MI). All other chemicals were purchased
from Sigma unless mentioned otherwise.
Animal Care and Use
Male C57Bl/6 or GPR120 KO littermates were fed a normal chow (13.5% fat;
LabDiet) or high-fat diet (60% fat; Research Diet) ad libitum for 15–20 weeks
from 8 weeks of age. GPR120 KO mice and WT littermates were provided
by Taconic Inc. (Hudson, NY). After 15 weeks on HFD, WT and GPR120 KO
mice were switched to an isocaloric HFD-containing 27% menhaden fish oil
replacement (wt/wt; menhaden fish oil: 16% EPA (C20:5n3), 9%, DHA
(C22:6n3), Research Diet) (Jucker et al., 1999; Neschen et al., 2007) and fed
for 5 weeks. Mice received fresh diet every 3rd day, and food consumption
and body weight were monitored. Animals were housed in a specific path-
ogen-free facility and given free access to food and water. All procedures
were approved by the University of California, San Diego animal care and
use committee. In vivo metabolic studies were performed as described under
supplemental experimental procedures.
Data Analysis
Densitometric quantification and normalization were performed using the
ImageJ 1.42q software. The values presented are expressed as the means ±
SEM. The statistical significance of the differences between various treat-
ments was determined by one-way ANOVA with the Bonferroni correction
using GraphPad Prism 4.0 (San Diego, CA). The p < 0.05 was considered
significant.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, seven
figures, and two tables and can be found with this article online at doi:10.1016/
j.cell.2010.07.041.
ACKNOWLEDGMENTS
We thank Jachelle M. Ofrecio and Sarah Nalbandian for their help with animal
maintenance and Elizabeth J. Hansen for editorial assistance. We are grateful
to Dr. Robert Lefkowitz (Howard Hughes Medical Institute, Duke University) for
the gift of FLAG-tagged serial mutant b-arrestin2 constructs and to Dr. Maziyar
Saberi at NGM Bio Inc. (San Francisco, CA) for GLP-1 measurements. We
thank the Flow Cytometry Resource and Neal Sekiya for assistance with
FACS analysis at the VA San Diego hospital, the UCSD Histology Core lab
for technical help with processing liver specimens, and UCSD Microscope
Resource for microscopy analysis. This study was funded in part by the
National Institutes of Health grants NIDDK DK033651 (J.M.O.), DK063491
(J.M.O.), DK 074868 (J.M.O.), and the Eunice Kennedy Shriver NICHD/NIH
through a cooperative agreement U54 HD 012303-25 as part of the specialized
Cooperative Centers Program in Reproduction and Infertility Research.
Received: January 20, 2010
Revised: May 24, 2010
Accepted: July 19, 2010
Published: September 2, 2010
REFERENCES
Barak, L.S., Ferguson, S.S., Zhang, J., and Caron, M.G. (1997). A beta-
arrestin/green fluorescent protein biosensor for detecting G protein-coupled
receptor activation. J. Biol. Chem. 272, 27497–27500.
Baron, A.D., Brechtel, G., Wallace, P., and Edelman, S.V. (1988). Rates and
tissue sites of non-insulin- and insulin-mediated glucose uptake in humans.
Am. J. Physiol. 255, E769–E774.
Briscoe, C.P., Peat, A.J., McKeown, S.C., Corbett, D.F., Goetz, A.S., Littleton,
T.R., McCoy, D.C., Kenakin, T.P., Andrews, J.L., Ammala, C., et al. (2006).
Pharmacological regulation of insulin secretion in MIN6 cells through the fatty
acid receptor GPR40: identification of agonist and antagonist small molecules.
Br. J. Pharmacol. 148, 619–628.
Buettner, R., Parhofer, K.G., Woenckhaus, M., Wrede, C.E., Kunz-Schughart,
L.A., Scholmerich, J., and Bollheimer, L.C. (2006). Defining high-fat-diet rat
models: metabolic and molecular effects of different fat types. J. Mol. Endocri-
nol. 36, 485–501.
Calder, P.C. (2005). Polyunsaturated fatty acids and inflammation. Biochem.
Soc. Trans. 33, 423–427.
Cen, B., Xiong, Y., Ma, L., and Pei, G. (2001). Direct and differential interaction
of beta-arrestins with the intracellular domains of different opioid receptors.
Mol. Pharmacol. 59, 758–764.
Gao, H., Sun, Y., Wu, Y., Luan, B., Wang, Y., Qu, B., and Pei, G. (2004).
Identification of beta-arrestin2 as a G protein-coupled receptor-stimulated
regulator of NF-kappaB pathways. Mol. Cell 14, 303–317.
Gether, U. (2000). Uncovering molecular mechanisms involved in activation of
G protein-coupled receptors. Endocr. Rev. 21, 90–113.
González-Périz, A., Horrillo, R., Ferre, N., Gronert, K., Dong, B., Moran-
Salvador, E., Titos, E., Martinez-Clemente, M., Lopez-Parra, M., Arroyo, V.,
and Claria, J. (2009). Obesity-induced insulin resistance and hepatic steatosis
are alleviated by omega-3 fatty acids: a role for resolvins and protectins.
FASEB J. 23, 1946–1957.
Gotoh, C., Hong, Y.H., Iga, T., Hishikawa, D., Suzuki, Y., Song, S.H., Choi,
K.C., Adachi, T., Hirasawa, A., Tsujimoto, G., et al. (2007). The regulation of
adipogenesis through GPR120. Biochem. Biophys. Res. Commun. 354,
591–597.
Hirasawa, A., Tsumaya, K., Awaji, T., Katsuma, S., Adachi, T., Yamada, M.,
Sugimoto, Y., Miyazaki, S., and Tsujimoto, G. (2005). Free fatty acids regulate
gut incretin glucagon-like peptide-1 secretion through GPR120. Nat. Med. 11,
90–94.
Imamura, T., Vollenweider, P., Egawa, K., Clodi, M., Ishibashi, K., Nakashima,
N., Ugi, S., Adams, J.W., Brown, J.H., and Olefsky, J.M. (1999). G alpha-q/11
protein plays a key role in insulin-induced glucose transport in 3T3-L1 adipo-
cytes. Mol. Cell. Biol. 19, 6765–6774.
Itoh, Y., Kawamata, Y., Harada, M., Kobayashi, M., Fujii, R., Fukusumi, S., Ogi,
K., Hosoya, M., Tanaka, Y., Uejima, H., et al. (2003). Free fatty acids regulate
insulin secretion from pancreatic beta cells through GPR40. Nature 422,
173–176.
Jucker, B.M., Cline, G.W., Barucci, N., and Shulman, G.I. (1999). Differential
effects of safflower oil versus fish oil feeding on insulin-stimulated glycogen
synthesis, glycolysis, and pyruvate dehydrogenase flux in skeletal muscle:
a 13C nuclear magnetic resonance study. Diabetes 48, 134–140.
Kawai, T., and Akira, S. (2006). TLR signaling. Cell Death Differ. 13, 816–825.
Lee, J.Y., Plakidas, A., Lee, W.H., Heikkinen, A., Chanmugam, P., Bray, G., and
Hwang, D.H. (2003). Differential modulation of Toll-like receptors by fatty
acids: preferential inhibition by n-3 polyunsaturated fatty acids. J. Lipid Res.
44, 479–486.
Lumeng, C.N., DelProposto, J.B., Westcott, D.J., and Saltiel, A.R. (2008).
Phenotypic switching of adipose tissue macrophages with obesity is gener-
ated by spatiotemporal differences in macrophage subtypes. Diabetes 57,
3239–3246.
Luttrell, L.M., Ferguson, S.S., Daaka, Y., Miller, W.E., Maudsley, S., Della
Rocca, G.J., Lin, F., Kawakatsu, H., Owada, K., Luttrell, D.K., et al. (1999).
Beta-arrestin-dependent formation of beta2 adrenergic receptor-Src protein
kinase complexes. Science 283, 655–661.
Luttrell, L.M., and Lefkowitz, R.J. (2002). The role of beta-arrestins in the
termination and transduction of G-protein-coupled receptor signals. J. Cell
Sci. 115, 455–465.
Miller, W.E., and Lefkowitz, R.J. (2001). Expanding roles for beta-arrestins as
scaffolds and adapters in GPCR signaling and trafficking. Curr. Opin. Cell Biol.
13, 139–145.
Neschen, S., Morino, K., Dong, J., Wang-Fischer, Y., Cline, G.W., Romanelli,
A.J., Rossbacher, J.C., Moore, I.K., Regittnig, W., Munoz, D.S., et al. (2007).
n-3 Fatty acids preserve insulin sensitivity in vivo in a peroxisome
proliferator-activated receptor-alpha-dependent manner. Diabetes 56,
1034–1041.
Nguyen, M.T., Favelyukis, S., Nguyen, A.K., Reichart, D., Scott, P.A., Jenn, A.,
Liu-Bryan, R., Glass, C.K., Neels, J.G., and Olefsky, J.M. (2007). A subpopula-
tion of macrophages infiltrates hypertrophic adipose tissue and is activated by
free fatty acids via Toll-like receptors 2 and 4 and JNK-dependent pathways. J.
Biol. Chem. 282, 35279–35292.
Oh, D.Y., Song, J.A., Moon, J.S., Moon, M.J., Kim, J.I., Kim, K., Kwon, H.B.,
and Seong, J.Y. (2005). Membrane-proximal region of the carboxyl terminus
of the gonadotropin-releasing hormone receptor (GnRHR) confers differential
signal transduction between mammalian and nonmammalian GnRHRs. Mol.
Endocrinol. 19, 722–731.
Polozova, A., and Salem, N., Jr. (2007). Role of liver and plasma lipoproteins
in selective transport of n-3 fatty acids to tissues: a comparative study of
14C-DHA and 3H-oleic acid tracers. J. Mol. Neurosci. 33, 56–66.
Schenk, S., Saberi, M., and Olefsky, J.M. (2008). Insulin sensitivity: modulation
by nutrients and inflammation. J. Clin. Invest. 118, 2992–3002.
Schulte, G., and Fredholm, B.B. (2003). Signaling from adenosine receptors to
mitogen-activated protein kinases. Cell Signal 15, 813–827.
Shoelson, S.E., Herrero, L., and Naaz, A. (2007). Obesity, inflammation, and
insulin resistance. Gastroenterology 132, 2169–2180.
Solinas, G., Vilcu, C., Neels, J.G., Bandyopadhyay, G.K., Luo, J.L., Naugler,
W., Grivennikov, S., Wynshaw-Boris, A., Scadeng, M., Olefsky, J.M., and
Karin, M. (2007). JNK in hematopoietically derived cells contributes to diet-
induced inflammation and insulin resistance without affecting obesity. Cell
Metab. 6, 386–397.
Takaesu, G., Surabhi, R.M., Park, K.J., Ninomiya-Tsuji, J., Matsumoto, K., and
Gaynor, R.B. (2003). TAK1 is critical for IkappaB kinase-mediated activation of
the NF-kappaB pathway. J. Mol. Biol. 326, 105–115.
Tazoe, H., Otomo, Y., Kaji, I., Tanaka, R., Karaki, S.I., and Kuwahara, A. (2008).
Roles of short-chain fatty acids receptors, GPR41 and GPR43 on colonic func-
tions. J. Physiol. Pharmacol. 59 (Suppl 2), 251–262.
Ulloa-Aguirre, A., Stanislaus, D., Janovick, J.A., and Conn, P.M. (1999).
Structure-activity relationships of G protein-coupled receptors. Arch. Med.
Res. 30, 420–435.
Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc. 697
Wang, J., Wu, X., Simonavicius, N., Tian, H., and Ling, L. (2006a). Medium-
chain fatty acids as ligands for orphan G protein-coupled receptor GPR84.
J. Biol. Chem. 281, 34457–34464.
Wang, Y., Tang, Y., Teng, L., Wu, Y., Zhao, X., and Pei, G. (2006b). Association
of beta-arrestin and TRAF6 negatively regulates Toll-like receptor-interleukin 1
receptor signaling. Nat. Immunol. 7, 139–147.
Weisberg, S.P., McCann, D., Desai, M., Rosenbaum, M., Leibel, R.L., and
Ferrante, A.W., Jr. (2003). Obesity is associated with macrophage accumula-
tion in adipose tissue. J. Clin. Invest. 112, 1796–1808.
698 Cell 142, 687–698, September 3, 2010 ª2010 Elsevier Inc.
Xiao, K., McClatchy, D.B., Shukla, A.K., Zhao, Y., Chen, M., Shenoy, S.K.,
Yates, J.R., 3rd, and Lefkowitz, R.J. (2007). Functional specialization of
beta-arrestin interactions revealed by proteomic analysis. Proc. Natl. Acad.
Sci. USA 104, 12011–12016.
Xu, H., Barnes, G.T., Yang, Q., Tan, G., Yang, D., Chou, C.J., Sole, J., Nichols,
A., Ross, J.S., Tartaglia, L.A., and Chen, H. (2003). Chronic inflammation in fat
plays a crucial role in the development of obesity-related insulin resistance. J.
Clin. Invest. 112, 1821–1830.
Anti-CD47 Antibody Synergizes with
Rituximab to Promote Phagocytosis and
Eradicate Non-Hodgkin Lymphoma
Mark P. Chao,1,10,* Ash A. Alizadeh,1,2,3,10 Chad Tang,1 June H. Myklebust,3,9 Bindu Varghese,3 Saar Gill,5 Max Jan,1
Adriel C. Cha,1 Charles K. Chan,1 Brent T. Tan,4 Christopher Y. Park,1,4 Feifei Zhao,1 Holbrook E. Kohrt,2,3
Raquel Malumbres,6 Javier Briones,7 Randy D. Gascoyne,8 Izidore S. Lossos,6 Ronald Levy,3 Irving L. Weissman,1,4,10
and Ravindra Majeti1,2,10
1Institutefor Stem Cell Biology and Regenerative Medicine, Stanford Cancer Center, and Ludwig Center at Stanford
2Department of Internal Medicine, Division of Hematology
3Division of Oncology
4Department of Pathology
5Division of Blood and Bone Marrow Transplantation

Stanford University, Palo Alto, CA 94304, USA

6Department of Medicine, Division of Hematology-Oncology, University of Miami Miller School of Medicine, Miami, FL 33136, USA
7Department of Hematology, Hospital Santa Creu i Sant Pau, Autonomous University of Barcelona, Barcelona 08193, Spain
8Department of Pathology, British Columbia Cancer Agency, Vancouver V5Z 1L3, Canada
9Department of Immunology, Institute for Cancer Research, Oslo University Hospital/Centre for Cancer Biomedicine, University of Oslo,

Oslo 0310, Norway

10These authors contributed equally to this work

*Correspondence: mpchao@stanford.edu
DOI 10.1016/j.cell.2010.07.044

SUMMARY modality for cancer treatment (Adams and Weiner, 2005).

Monoclonal antibodies are standard therapeutics for
several cancers including the anti-CD20 antibody
rituximab for B cell non-Hodgkin lymphoma (NHL).
Rituximab and other antibodies are not curative
and must be combined with cytotoxic chemotherapy
for clinical benefit. Here we report the eradication
of human NHL solely with a monoclonal antibody
therapy combining rituximab with a blocking anti-
CD47 antibody. We identified increased expression
of CD47 on human NHL cells and determined that
higher CD47 expression independently predicted
adverse clinical outcomes in multiple NHL subtypes.
Blocking anti-CD47 antibodies preferentially enabled
phagocytosis of NHL cells and synergized with ritux-
imab. Treatment of human NHL-engrafted mice with
anti-CD47 antibody reduced lymphoma burden and
improved survival, while combination treatment
with rituximab led to elimination of lymphoma and
cure. These antibodies synergized through a mecha-
nism combining Fc receptor (FcR)-dependent and
FcR-independent stimulation of phagocytosis that
might be applicable to many other cancers.
INTRODUCTION
Emerging evidence has demonstrated that monoclonal anti-
bodies (mAbs) either alone or in combination are an effective
Although therapies combining a mAb with chemotherapeutic
agents are effective in several human cancers, antibodies alone
are not curative. Antibodies effective against cancer are believed
to function by several mechanisms including antibody-depen-
dent cellular cytotoxicity (ADCC), stimulation of complement-
dependent cytotoxicity (CDC), inhibition of signal transduction,
or direct induction of apoptosis (Cheson and Leonard, 2008).
Non-Hodgkin lymphoma (NHL) is the fifth most common
cancer in the United States consisting of indolent and aggressive
subtypes with a 5 year overall survival ranging from 25%–75%
(The International Non-Hodgkin’s Lymphoma Prognostic
Factors Project, 1993). The anti-CD20 antibody, rituximab (Rit-
uxan), is a standard therapy for many CD20-positive B cell
lymphomas and significantly improves long-term survival in
combination with conventional chemotherapy (Cheson and Leo-
nard, 2008). As a single agent or in combination with chemo-
therapy, rituximab is not curative in the majority of B cell NHL
patients and rituximab resistance has been observed (reviewed
in Cheson and Leonard, 2008). Multiple lines of evidence indicate
that rituximab acts at least in part by engaging Fc receptors
(FcRs) on immune effector cells, such as NK cells and macro-
phages, and stimulating effector functions such as ADCC (Glen-
nie et al., 2007; Nimmerjahn and Ravetch, 2007). Although resis-
tance has been reported to occur through several mechanisms
(Cartron et al., 2004), there has been limited development of
agents that can overcome this resistance.
Immune effector cells, including NK cells and phagocytes, are
critical to the efficacy of many anticancer antibodies. Phagocytic
cells, including macrophages and dendritic cells, express signal
regulatory protein alpha (SIRPa), which binds CD47, a widely

Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 699

expressed transmembrane protein (Brown and Frazier, 2001).
CD47-mediated activation of SIRPa initiates a signal transduc-
tion cascade resulting in inhibition of phagocytosis (reviewed in
Jaiswal et al., 2010). In identifying a role for CD47 in cancer path-
ogenesis, we previously demonstrated that forced expression of
mouse CD47 on a human leukemia cell line facilitated tumor
engraftment in immunodeficient mice through the evasion of
phagocytosis (Jaiswal et al., 2009). We further demonstrated
that this mechanism could be targeted therapeutically in human
acute myeloid leukemia (AML) with a blocking anti-CD47 anti-
body that enabled phagocytosis and eliminated AML stem cells
(Majeti et al., 2009). Based on this antibody mechanism, we
hypothesized that the combination of a blocking anti-CD47
antibody with a second FcR-activating antibody would both
prevent an inhibitory signal and deliver a positive stimulus,
resulting in the synergistic phagocytosis and elimination of tar-
get cells. Here, we tested this antibody synergy hypothesis by
investigating the combination of a blocking anti-CD47 mAb
with rituximab against human NHL.
RESULTS
CD47 Expression Is Increased on NHL Cells Compared
to Normal B Cells
We examined CD47 protein expression on primary human NHL
samples and normal B cells by flow cytometry. Compared to
both normal peripheral blood and germinal center B cells,
CD47 was more highly expressed on a large subset of primary
patient samples from multiple B cell NHL subtypes (Figure 1A
and Figure S1A available online), including diffuse large B cell
lymphoma (DLBCL), B cell chronic lymphocytic leukemia
(B-CLL), mantle cell lymphoma (MCL), follicular lymphoma (FL),
marginal zone lymphoma (MZL), and pre-B acute lymphoblastic
leukemia (pre-B ALL). Across NHL subtypes, we found differing
levels of CD47 expression that also varied within each NHL
subtype (Figure 1B).
Increased CD47 Expression Correlates with a Worse
Clinical Prognosis and Adverse Molecular Features
in Multiple NHL Subtypes
Having previously shown a correlation between CD47 mRNA
and protein expression (Majeti et al., 2009), we assessed CD47
mRNA expression across NHL subtypes for associations with
morphologic and molecular subgroups using gene expression
data from previously described patient cohorts (Table S1) and
investigated the prognostic implications of increased CD47
expression in disease outcome. Higher CD47 expression was
associated with adverse prognosis in DLBCL, B-CLL, and MCL
(Figures 1C–1E). In patients with DLBCL, whether treated with
or without rituximab-based combination chemotherapy (Fig-
ure 1C and Figure S1B), higher CD47 expression was signifi-
cantly associated with risk of death. This increased risk was
largely due to disease progression (Figure S1C), a finding vali-
dated in an independent cohort of patients using quantitative
RT-PCR on fixed archival specimens (Figure S1D).
We next investigated whether increased CD47 expression was
associated with known adverse molecular features in NHL.
In DLBCL, prior studies have identified two distinct subgroups
700 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.
based on the presumed cell of origin of tumors: normal germinal
center B cells (GCB-like), which are associated with a favorable
clinical outcome, or activated blood memory B cells (ABC-like),
which are associated with a poor clinical outcome (Alizadeh
et al., 2000; Rosenwald et al., 2002). CD47 expression was sig-
nificantly higher in ABC-like DLBCL (Figure 1F and Figure S1E).
CD47 expression was not found to have independent prog-
nostic value within GCB and ABC subtypes, suggesting a strong
association with the cell-of-origin classification of DLBCL.
Higher CD47 expression was also associated with unmutated
immunoglobulin heavy chain variable regions (IgVH) in CLL
(Figure 1G and Figure S1E) and significantly correlated with
the proliferative index in MCL (Figure 1H), both known adverse
prognostic factors (Katzenberger et al., 2006; Oscier et al.,
2002; Rosenwald et al., 2003). In multivariate analyses, CD47
expression remained prognostic of disease progression inde-
pendent from the international prognostic index in DLBCL (The
International Non-Hodgkin’s Lymphoma Prognostic Factors
Project, 1993) and two major prognostic factors in CLL: IgVH
mutation status and ZAP-70 status (Figure S1G). Within the
small MCL cohort, a multivariate model did not find independent
prognostic value for CD47 when considering the proliferation
index (data not shown).
Blocking Anti-CD47 Antibodies Enable Phagocytosis
of NHL Cells by Macrophages and Synergize
with Rituximab In Vitro
We first tested the ability of anti-human CD47 antibodies to
enable phagocytosis of human NHL cell lines, primary NHL cells,
and normal peripheral blood (NPB) cells by human macrophages
in vitro. Incubation of NHL cells in the presence of IgG1 isotype
control or anti-CD45 IgG1 antibody did not result in significant
phagocytosis; however, two different blocking anti-CD47 anti-
bodies (B6H12.2 and BRIC126) enabled phagocytosis of NHL
cells but not NPB cells (Figures 2A and 2B).
Next, we repeated the in vitro phagocytosis assays with
mouse macrophages. Incubation of NHL cells in the presence
of IgG1 isotype control or anti-CD45 IgG1 antibody did not result
in significant phagocytosis; however, phagocytosis of NHL cells
was observed with blocking antibodies to CD47 (B6H12.2 and
BRIC126), whereas no phagocytosis was observed with a non-
blocking antibody (2D3) (Figure 2B). Disruption of the CD47-
SIRPa interaction with an anti-mouse SIRPa antibody also
resulted in significant phagocytosis (Figure 2B).
Given variable expression of CD47 on primary NHL, we inves-
tigated by two independent methods whether CD47 expression
levels correlated with the degree of anti-CD47 antibody-medi-
ated phagocytosis. First, lentiviral shRNA vectors were used to
knock down expression of CD47 in Raji cells. Several clones
were generated with a range of CD47 knockdown (Figures S2A
and S2B). Those clones with a greater than 50% reduction in
CD47 expression (shCD47-1 and shCD47-2) demonstrated
a significant reduction in anti-CD47 antibody-mediated phago-
cytosis (Figure S3C). In the second approach, a statistical anal-
ysis demonstrated a positive correlation between CD47 expres-
sion and degree of anti-CD47 antibody-mediated phagocytosis
with both mouse and human macrophage effector cells
(Figure S2D).
Figure 1. CD47 Expression Is Increased on NHL Cells Compared to Normal B Cells, Confers a Worse Clinical Prognosis, and Correlates with
Adverse Molecular Features in Multiple NHL Subtypes
(A) CD47 expression on normal peripheral blood (PB) B cells (CD19+), normal germinal center (GC) B cells (CD3
CD5
CD20+CD10+CD38+), and NHL B cells
(CD19+) was determined by flow cytometry, and mean fluorescence intensity was normalized for cell size. Each point represents a different patient sample:
DLBCL = 2, CLL = 15, MCL = 4, FL = 6, MZL = 2, and pre-B ALL = 1 (****p < 0.0001). Normalized mean expression (and range) for each population were: normal
PB B cells 420.9 (267.3–654.0), normal GC B cells 482.5 (441.1–519.9), and NHL 888.5 (270.1–1553).
(B) CD47 expression across NHL subtypes including DLBCL (DL, n = 15), MCL (n = 34), FL (n = 28), and B-CLL (n = 14) was determined as in (A). Normalized mean
expression (and range) for each population were: DL 725.7 (261.2–1344), MCL 1055 (444.2–2196), FL 825.1 (283.6–1546), CLL 713.6 (432.8–1086) (*p < 0.05).
(C–E) Prognostic influence of CD47 mRNA expression is shown on overall (C and E) and event-free (D) survival of patients with DLBCL, B-CLL, and MCL.
For DLBCL and CLL, stratification into low and high CD47 expression groups was based on an optimal threshold determined by microarray analysis; this cutpoint
was internally validated for both DLBCL and CLL and also externally validated in an independent DLBCL cohort. For MCL, stratification relative to the median was
employed as an optimal cutpoint could not be defined. Significance measures are based on log-likelihood estimates of the p value, when treating CD47 expres-
sion as a continuous variable, with corresponding dichotomous indices also provided in Table S1.
(F–H) CD47 mRNA expression is shown in relation to cell-of-origin classification for DLBCL (F), immunoglobulin heavy chain mutation status (IgVH) for CLL (G),
and proliferation index for MCL (H). Error bars represent upper and lower quartiles (F and G). Analyses for (C)–(H) employed publicly available datasets for NHL
patients (Table S1). NGC = normal germinal center, ABC = activated B cell-like, GCB = germinal center B cell-like.
See also Figure S1 and Table S1.

Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 701

Figure 2. Blocking Antibodies against CD47 Enable Phagocytosis of NHL Cells by Macrophages and Synergize with Rituximab In Vitro
(A) CFSE-labeled NHL cells were incubated with human macrophages and the indicated antibodies and examined by immunofluorescence microscopy to detect
phagocytosis (arrows). Photomicrographs from a representative NHL sample are shown.
(B) Phagocytic indices of primary human NHL cells (blue), normal peripheral blood (NPB) cells (red), and NHL cell lines (purple, orange, and green) were deter-
mined using human (left) and mouse (middle) macrophages. Antibody-induced apoptosis (right panel) was tested by incubating NHL cells with the indicated anti-
bodies or staurosporine without macrophages and assessing the percentage of apoptotic and dead cells (% annexin V and/or PI positive).

702 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.

 It has been reported that immobilized or crosslinked anti-
bodies against CD47 induce apoptosis of primary human
B-CLL cells, as well as several malignant lymphoid cell lines
(Kikuchi et al., 2004, 2005; Mateo et al., 1999; Uno et al.,
2007). Therefore, anti-CD47 antibodies might be predicted to
directly induce apoptosis of NHL cells that are then recognized
by macrophages and phagocytosed. Contrary to this prediction,
when NHL cells were incubated with anti-CD47 antibody in
the absence of macrophages, no induction of apoptosis was
observed when cells were incubated in suspension for either
2 hr (Figure 2B, right) or 8 hr (Figures S2E and S2F). Incubation
of NHL cells with immobilized anti-CD47 antibody resulted in
increased apoptosis compared to controls (Figures S2G and
S2H), consistent with prior reports (Mateo et al., 1999). Given
that phagocytosis of NHL cells occurs in the presence of soluble
anti-CD47 mAbs, it is unlikely that these mAbs induce apoptosis
of NHL cells that are then secondarily phagocytosed.
Next, we tested the ability of a blocking anti-CD47 mAb to syn-
ergize with rituximab in the phagocytosis of NHL cells. We exam-
ined CD20 expression on NHL cells and found no difference
between normal B cells and NHL cells (Figures S2I and S2J).
Incubation of NHL cells with rituximab in the presence of mouse
or human macrophages resulted in significant phagocytosis
(Figures 2D and S2E). We then tested the synergy hypothesis
by isobologram analysis (Chou, 2008; Tallarida, 2006). Using
Raji, SUDHL4, and NHL17 cells, which express varying levels
of both CD47 and CD20 (Figure S2K), anti-CD47 antibody
synergized with rituximab or anti-human CD20 (mouse IgG2a)
antibody, as indicated by combination indices less than 1 (Fig-
ure 2C). In a second approach, in vitro phagocytosis assays
were conducted with primary NHL cells incubated with either
anti-CD47 antibody or rituximab alone, or both in combination
at half of the single agent dose. NHL cells exhibited a significant
increase in phagocytosis when incubated with the combination
compared to either antibody alone when using mouse (Fig-
ure 2D) or human (Figure 2E) macrophage effectors. No phago-
cytosis of NPB cells was observed with either antibody alone or
in combination with human macrophages (Figure 2E).
Ex Vivo Coating of NHL Cells with an Anti-CD47 Antibody
Inhibits Tumor Engraftment
Next, the ability of blocking anti-CD47 antibody to eliminate NHL
in vivo either alone or in combination with rituximab was explored
by two separate treatment strategies. First, the effect of ex vivo
anti-CD47 antibody coating on the engraftment of human NHL
cells was tested. Luciferase-expressing Raji and SUDHL4 cell
lines were precoated ex vivo with anti-CD47, IgG1 isotype
control, or anti-CD45 antibody and transplanted intravenously
into SCID mice. Coating with anti-CD47 antibody prevented
engraftment of both cell lines (Figures 3A–3F). Coating of Raji
cells with rituximab also inhibited engraftment when trans-
(C) Synergistic phagocytosis by anti-CD47 antibody (B6H12.2) and anti-CD20 mAbs was examined by isobologram analysis and determination of combination
indices (CI). CIobs indicates observed results, and the dashed line indicates the expected results if antibodies were additive.
(D and E) Synergy between anti-CD47 antibody and rituximab in the phagocytosis of NHL and NPB cells was assessed by determining the phagocytic index when
incubated with a combination of both antibodies compared to either antibody alone at twice the dose, with either mouse (D) or human (E) macrophages. NHL17*:
cell line derived from primary sample NHL17.
***p < 0.001, ****p < 0.0001, *****p < 0.00001. Figure 2B p values represent comparison against IgG1 isotype control antibody. See also Figure S2.
planted into SCID mice (Figure S3). In addition to these cell lines,
we identified a primary NHL patient specimen that engrafted in
NSG mice in the bone marrow upon intravenous transplantation
(Figures S5A and S5B). As with the cell lines, ex vivo coating of
these primary cells with anti-CD47 antibody, but not controls
(Figure 3G), resulted in complete inhibition of bone marrow
engraftment (Figure 3H).
Combination Therapy with Anti-CD47 Antibody and
Rituximab Eliminates Lymphoma in Both Disseminated
and Localized Human NHL Xenotransplant Models
In the second treatment strategy, mice were first engrafted with
NHL and then administered single or combination antibody
therapy. To best model NHL, we established disseminated and
localized xenotransplantation models in NSG mice that are defi-
cient in T, B, and NK cells (Shultz et al., 2005) but retain phago-
cyte effector cells. In the disseminated model, luciferase-
expressing Raji cells were transplanted intravenously into adult
NSG mice. Two weeks later, these mice were administered daily
injections of either control mouse IgG, anti-CD47 antibody, ritux-
imab, or anti-CD47 antibody and rituximab. Anti-CD47 antibody
treatment decreased the lymphoma burden in these mice (Fig-
ures 4A and 4B) and significantly prolonged survival compared
to control IgG, although all mice eventually died (Figure 4C and
Table S2). Similar results were seen with rituximab and were
not statistically different compared to anti-CD47 antibody (Fig-
ures 4A–4C and Table S2). In contrast, combination therapy
with anti-CD47 antibody and rituximab eliminated lymphoma in
60% of mice as indicated by long-term survival (Figure 4C) and
the absence of luciferase-positive lymphoma (data not shown)
more than 4 months after the end of treatment. In humans, ritux-
imab efficacy is thought to be primarily mediated by both macro-
phages and NK cells (Nimmerjahn and Ravetch, 2007; Taylor
and Lindorfer, 2008). Given that NSG mice lack NK cells, we con-
ducted a similar experiment in NK cell-containing SCID mice and
observed similar therapeutic responses as in NSG mice (Figures
S4A and S4B).
In the localized NHL model, luciferase-expressing Raji cells
were transplanted subcutaneously in the right flank of NSG
mice. Once tumors were palpable (approximately 2 weeks),
mice were treated with antibody therapy. Mice treated with
anti-CD47 antibody or rituximab demonstrated a decreased
rate of lymphoma growth compared to control IgG-treated
mice as measured by both luciferase signal and tumor volume
(Figures 4D–4F) but, like controls, eventually had to be sacrificed
due to enlarged tumor growth. In contrast, mice treated with the
combination of anti-CD47 antibody and rituximab demonstrated
complete elimination of lymphoma, with 86% of treated mice
having no measurable mass or luciferase-positive lymphoma
4 weeks after the end of therapy (Figures 4D–4F and Figures
S4C–S4E). Moreover, all showed no evidence of tumor growth,
Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 703
Figure 3. Ex Vivo Coating of NHL Cells with an Anti-CD47 Antibody Inhibits Tumor Engraftment
(A–F) Luciferase-expressing Raji (A) and SUDHL4 (C) cells were assessed for ex vivo antibody coating by flow cytometry. SCID mice transplanted with Raji (B) and
SUDHL4 (D) were subject to bioluminescent imaging (1-IgG1 isotype control, 2-anti-CD45, 3 and 4-anti-CD47, 5-luciferase negative control). Bioluminescence
for Raji (E) and SUDHL4 (F) engrafted mice was quantified (n = 3 per antibody condition). No tumor engraftment was observed in mice transplanted with anti-
CD47-coated cells compared to IgG-coated cells (p < 0.05) for both Raji and SUDHL4, as assessed by bioluminescent imaging. Data are represented as
mean ± standard deviation (SD).

704 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.

remained relapse free, and were alive at over 197 days after
tumor engraftment. Out of six mice achieving a complete remis-
sion, five remained relapse free whereas one mouse died of non-
tumor-related causes (Figure 4E). For both disseminated and
localized xenograft models, expression of CD47 and CD20 in
transplanted Raji cells did not differ from Raji cells in culture
(Figure S4F).
Combination Therapy with Anti-CD47 Antibody
and Rituximab Eliminates Lymphoma in Primary Human
NHL Xenotransplant Mouse Models
NHL cell lines have been valuable for the evaluation of therapeu-
tics, but they may not accurately recapitulate the heterogeneity
of the primary disease. We report here two new mouse xenograft
models for NHL in which intravenous transplantation of cells
from a DLBCL patient (NHL7/SUNHL7) and a FL patient
(NHL31/SUNHL31) give rise to robust lymphoma engraftment
in the bone marrow and/or peripheral blood (Figure S1A and
Figure S5A). Primary DLBCL cells were transplanted into mice,
which 2 weeks later were treated with daily injections of anti-
bodies for 14 days. Treatment with anti-CD47 antibody either
alone or in combination with rituximab eliminated human
lymphoma in the bone marrow, whereas treatment with rituxi-
mab resulted in a reduction of disease in 60% of mice (Figures
5A and 5B). Mouse hematopoietic cells were unaffected by anti-
body therapy (Figure S5E). Treatment was then discontinued,
and all mice were monitored for survival. Mice treated with either
anti-CD47 antibody or rituximab alone had a significantly longer
survival compared to mice treated with control IgG, but all even-
tually died secondary to disease due to widespread organ
dissemination on autopsy (Figure 5C, data not shown). Most
significantly, 8 out of 9 mice (89%) administered combination
antibody treatment were cured of lymphoma, as indicated by
long-term disease-free survival more than 4 months after the
end of treatment (Figure 5C and Table S3) with no detectable
lymphoma in the bone marrow (data not shown). In a second
primary model, FL cells were transplanted intravenously in a
similar manner. CD20+CD10+ lymphoma engraftment in the
peripheral blood and bone marrow was detected after 8 weeks.
At this time, mice were treated with a single intraperitoneal injec-
tion of either control IgG, anti-CD47 antibody, rituximab, or the
combination and then followed for disease progression. A single
dose of anti-CD47 antibody alone or in combination with rituxi-
mab eliminated lymphoma both in the peripheral blood (Fig-
ure 5D) and in bone marrow (Figure 5E). In contrast, a single
dose of rituximab enabled a partial reduction in tumor burden
that rebounded back to baseline levels in the peripheral blood
with no tumor reduction observed in the bone marrow. The differ-
ence in anti-CD47 antibody clearance of lymphoma as a single
dose in FL-engrafted mice (Figures 5D and 5E) compared to
multiple dose therapy in mice engrafted with DLBCL (Fig-
ure 5B) or Raji (Figure 4) may be due to cell-intrinsic differences
in antibody sensitivity between different NHL subtypes or due to
(G) Bulk lymphoma cells from a human NHL patient were assessed for ex vivo antibody coating by flow cytometry.
(H) Compared to IgG1 isotype control, anti-CD47 antibody pretreatment inhibited engraftment of NHL cells (p < 0.0001) whereas anti-CD45-coated cells
engrafted similarly to controls (p = 0.54). All p values were determined using Fisher’s exact test.
Error bars represent SD (E and F). See also Figure S3.
different anti-CD47 antibody potencies in distinct tissue com-
partments (peripheral blood versus bone marrow (BM) versus
soft tissue compartments).
To assess the ability of these two primary NHL xenotransplant
models to model the disease, we compared histological sections
of the primary NHL specimen and the transplanted tumor. Similar
DLBCL and FL morphology was observed for NHL7 and NHL31,
respectively (Figure S5B). We next determined whether the per-
centages of macrophages infiltrating the tumor differed between
the primary patient and the xenografted tumor. For NHL31, the
percentage of infiltrating human macrophages (CD68+) in the
primary lymph node was similar to the percentage of infiltrating
mouse macrophages (F4/80+) in bone marrow of transplanted
mice (Figure S5C). Analyzing infiltrating macrophage frequency
by flow cytometry, no difference was observed between the
primary specimen and xenograft for either NHL7 or NHL31
(Figure S5D).
Synergy between Anti-CD47 Antibody and Rituximab
Does Not Occur through NK Cells or Complement
Rituximab can eliminate malignant cells via apoptosis, NK cell-
mediated ADCC, and CDC (reviewed in Smith, 2003). However,
it is not known whether anti-CD47 antibody also enables ADCC
or CDC in addition to phagocytosis. Therefore, we investigated
whether anti-CD47 antibody alone could induce antitumor
effects by macrophage-independent mechanisms, and whether
synergy with rituximab could occur through these modalities.
First, to investigate possible synergy in direct apoptosis, NHL
cells were incubated with either anti-CD47 antibody/rituximab
alone or in combination without macrophages, and cell death
was measured. No synergistic apoptosis was observed when
NHL cells were incubated with soluble (Figure 6A and Fig-
ure S6A) or immobilized (Figures S6B and S6C) antibodies.
Furthermore, crosslinking of soluble anti-CD47 antibody alone
or in combination with rituximab by macrophages did not induce
increased apoptosis of nonphagocytosed NHL cells compared
to IgG1 isotype control, whereas rituximab induced a slight
increase in apoptosis (Figure S6D). No synergistic apoptosis
was observed in this context (Figure S6D). The small increase
in apoptosis upon antibody treatment was not FcR dependent
given that results were similar with macrophages lacking the
Fcg subunit (Takai et al., 1994) (Figure S6D). Second, we inves-
tigated whether NK cells could mediate tumor elimination by
anti-CD47 antibody alone or in synergy with rituximab. A prior
report observed increased NK cell cytotoxicity of cancer cell
lines with an anti-CD47 antibody in vitro, though the mechanism
of targeting was not elucidated (Kim et al., 2008). We first deter-
mined whether human or mouse NK cells expressed SIRPa and
found that both human NK cells, CD3
CD56+CD7+ (Milush et al.,
2009), as well as mouse NK cells, CD3
DX5+, expressed minimal
to no SIRPa (Figure 6B). Next, the ability of anti-CD47 antibody
to induce NK cell-mediated ADCC through FcRs or by CD47-
SIRPa blockade was investigated. Utilizing human NK cells as
Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 705
Figure 4. Combination Therapy with Anti-CD47 Antibody and Rituximab Eliminates Lymphoma in Both Disseminated and Localized Human
NHL Xenotransplant Mouse Models
(A) NSG mice transplanted intravenously with luciferase-expressing Raji cells were treated with the indicated antibodies (n = 8 per treatment group). Luciferase
imaging of representative mice from pre- and 10 days post-treatment are shown (A) and averaged for all mice in each treatment group (B).
(C) Kaplan-Meier survival analysis was performed (Table S2). p values compare IgG control to combination antibody treatment or anti-CD47 antibody/rituximab
single antibody to combination. Arrows indicate start (day 14) and stop (day 35) of treatment.
(D) Luciferase-expressing Raji cells were transplanted subcutaneously in the flank of NSG mice. When palpable tumors (0.1 cm3) formed, treatment
began with the indicated antibodies. Luciferase imaging of representative mice from each treatment group is shown before (day 0) and during (day 14)
treatment.
(E) Quantified bioluminescence was determined and averaged for all mice in each treatment group (n = 7).
(F) Tumor volume was measured with average volume shown. p values were derived from a two-way ANOVA and compared to IgG treatment.
*p < 0.05, ***p < 0.001, ****p < 0.0001. Complete remission was defined as the number of mice with no evidence of tumor at the indicated date. Relapse was
defined as the number of mice achieving a complete remission that later developed recurrence of tumor growth. For (E), one mouse that achieved a complete
remission died of non-tumor-related causes. Data presented in (B), (E), and (F) are mean values ± SD. See also Figure S4 and Table S2.

706 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.

Figure 5. Combination Therapy with Anti-CD47 Antibody and Rituximab Eliminates Lymphoma in Primary Human NHL Xenotransplant
Mouse Models
(A and B) Sublethally irradiated NSG mice were transplanted intravenously with cells from a primary DLBCL patient (NHL7) and treated with the indicated anti-
bodies. Flow cytometry of human lymphoma engraftment in the bone marrow of two representative mice is shown pre- and 14 days post-antibody treatment in
(A). Data from all mice are included in (B). **p < 0.01, comparing pre- and post-treatment values for each respective antibody treatment.
(C) Kaplan-Meier survival analysis (Table S3) of DLBCL-engrafted mice from each antibody treatment cohort is shown (n R 10 per antibody group), with p values
calculated comparing control IgG to combination antibody treatment or anti-CD47 antibody/rituximab single antibody to combination treatment. Arrows indicate
start (day14) and stop (day 28) of treatment.
(D and E) Mice engrafted intravenously with a primary FL patient sample (NHL31) were treated with a single dose of the indicated antibodies. Compared to IgG
control and rituximab, anti-CD47 antibody alone or in combination with rituximab eliminated tumor burden in the peripheral blood (p = 0.04, two-way ANOVA) and
bone marrow (p < 0.001, t test).
(E) Lyphoma engraftment in the bone marrow was determined 14 days post-treatment. Each antibody treatment group consisted of three mice.
For mice reported in (D) and (E), human lymphoma chimerism was between 5% and 25% in the peripheral blood and bone marrow. Error bars represent SD
(D and E). See also Figure S5 and Table S3.

Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 707

Figure 6. Synergy between Anti-CD47 Antibody and Rituximab Does Not Occur through NK Cells or Complement
(A) NHL cells were incubated with the indicated soluble antibodies for 2 hr and the percentage of dead cells was calculated (% annexin V+ and/or 7-AAD+). No
statistically significant difference in % dead cells was observed with the combination of anti-CD47 antibody and rituximab compared to either anti-CD47 antibody
alone (p = 0.24) or rituximab alone (p = 0.95).
(B) SIRPa expression is shown for both human and mouse NK cells as determined by flow cytometry.
(C and D) Chromium release assays measuring ADCC were performed in triplicate with human (C) and mouse (D) at an effector:target ratio of 17.5:1, and percent-
specific lysis is reported. Antibodies were incubated at 10 mg/ml except anti-CD47 full-length or F(ab0 )2 antibody+rituximab (5 mg/ml).
(E) CDC assay with human complement was performed in duplicate. Compared to IgG1 isotype control, anti-CD47 antibody did not enable CDC (p > 0.2),
whereas rituximab did (p < 0.001) by two-way ANOVA for both SUDHL4 and NHL17*. Combination treatment with anti-CD47 antibody and rituximab did not
enable greater levels of CDC compared to rituximab (p = 0.78).
(F) CDC assay with mouse complement was performed in duplicate. Compared to IgG1 isotype control, anti-CD47 antibody did not enable CDC (p > 0.25)
whereas rituximab did (p = 0.03, p = 0.08, respectively) for both SUDHL4 and NHL17*. No difference in CDC between CD47 antibody+rituximab and rituximab
alone was observed (p > 0.13) for both SUDHL4 and NHL17*.
*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars represent SD (C–F). NHL17* = primary NHL17 cells expanded in culture. See also Figure S6.

708 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.

effectors, anti-CD47 antibody did not induce increased ADCC of
Raji or SUDHL4 cells compared to IgG1 isotype control (Fig-
ure 6C). Although rituximab did enable ADCC of these two
NHL cell lines, no synergistic effect between anti-CD47 antibody
and rituximab was observed (p = 0.77, Figure 6C). As anti-CD47
antibody (B6H12.2) is a mouse IgG1 isotype, we repeated these
assays with mouse NK cells. Anti-CD47 antibody caused
increased ADCC of these two NHL cell lines compared to isotype
control, whereas rituximab induced ADCC to a lesser degree
(Figure 6D). To test whether anti-CD47 antibody-mediated
ADCC was Fc dependent, we generated a F(ab0 )2 fragment of
the anti-CD47 antibody (Figures S7B–S7I). The F(ab0 )2 fragment
did not enable ADCC, indicating that the increased ADCC was
likely mediated through FcRs (Figure 6D). The combination of
anti-CD47 antibody or F(ab0 )2 fragment with rituximab did not
induce a statistically significant increase in ADCC compared to
single agent therapy, indicating no synergistic effect. Third, we
investigated the role of complement in anti-CD47 antibody-
mediated cytotoxicity. Using either human (Figure 6E) or mouse
(Figure 6F) complement, anti-CD47 antibody did not induce CDC
of either an NHL cell line or a primary NHL sample, whereas
rituximab did induce significant CDC against both of these
samples. Moreover, the combination of anti-CD47 antibody
and rituximab did not induce increased CDC compared to ritux-
imab alone. Fourth, we investigated the relative contribution of
major components of macrophages, NK cells, and complement
in mediating the therapeutic effects of anti-CD47 antibody and
rituximab in vivo. Luciferase-labeled Raji cells were engrafted
intravenously into SCID mice, which have functional macro-
phages, NK cells, and complement. Mice were then separated
into cohorts receiving selective depletion of either macrophages
by clodronate, NK cells by anti-asialoGM1 antibody, comple-
ment by cobra venom factor, or a vehicle control. These cohorts
were then treated with combination anti-CD47 antibody and rit-
uximab therapy for 3 days, and tumor burden was measured
by bioluminescent imaging pre- and post-treatment. Compared
to vehicle control, NK cell and complement depletion had no
effect on tumor elimination by combination antibody therapy
(Figure S6E). In contrast, macrophage depletion significantly
abrogated the therapeutic effect, indicating that macrophages,
and not NK cells or complement, are required for combination
anti-CD47 antibody and rituximab-mediated elimination of NHL
in vivo.
Anti-CD47 Antibody Synergizes with Rituximab through
FcR-Independent and FcR-Dependent Mechanisms
We hypothesize that the observed synergy between an anti-
CD47 antibody and rituximab occurs through the combination
of two separate mechanisms for stimulating phagocytosis: (1)
FcR-independent through blockade of an inhibitory phagocytic
signal by anti-CD47 antibody, and (2) FcR-dependent through
delivery of a positive phagocytic signal by rituximab. We utilized
four independent methods to investigate this hypothesis. First,
synergistic phagocytosis was observed with the combination
of anti-SIRPa antibody and rituximab by isobologram analysis
(Figure 7A), and with a large panel of primary NHL samples (Fig-
ure 7B). Second, mouse macrophages lacking the Fcg receptor,
thereby incapable of enabling FcR-dependent phagocytosis
(Takai et al., 1994), but still expressing SIRPa (Figure S7A),
were utilized as effector cells for phagocytosis of NHL cells incu-
bated with either anti-CD47 antibody, anti-SIRPa antibody, ritux-
imab, or anti-CD47/anti-SIRPa antibody in combination with
rituximab. As predicted, anti-CD47 antibody and anti-SIRPa
antibody, but not rituximab, enabled phagocytosis of NHL cells,
without evidence of synergistic phagocytosis (Figure 7C).
Third, F(ab0 )2 fragments of both anti-CD47 antibody and
rituximab were generated (Figures S7B–S7I) and utilized in
phagocytosis assays with NHL cells and wild-type macro-
phages. Anti-CD47 F(ab0 )2 antibody synergized with rituximab
as demonstrated by isobologram analysis (Figure 7D). Addition-
ally, anti-CD47 F(ab0 )2, but not rituximab F(ab0 )2, enabled phago-
cytosis of NHL cells (Figure 7E). Consistent with the proposed
mechanism, synergistic phagocytosis was observed with the
combination of either full-length anti-CD47 or anti-CD47 F(ab0 )2
with full-length rituximab, but not with any combination involving
rituximab F(ab0 )2 (Figure 7E).
Fourth, synergistic phagocytosis was investigated in vivo
using GFP+ Raji cells engrafted in NSG mice. As single agents,
anti-CD47 antibody and rituximab enabled phagocytosis of
Raji cells engrafted in the liver as evidenced by an increased
percentage of mouse macrophages containing phagocytosed
GFP+ Raji cells (Figure 7F). Most significantly, combination
anti-CD47 antibody and rituximab treatment enabled signifi-
cantly increased phagocytosis compared to either single agent
demonstrating that synergistic phagocytosis occurred in vivo
(Figure 7F).
DISCUSSION
In this report, we identify a distinct mechanism of synergy
between mAbs in cancer therapy leading to cures in the absence
of chemotherapy. Specifically, we utilized a blocking anti-CD47
antibody in combination with the anti-CD20 antibody rituximab
to eradicate human NHL through a mechanism of synergy
involving FcR-independent enabling of phagocytosis by anti-
CD47 antibody and FcR-dependent stimulation of phagocytosis
by rituximab. In addition, the identification of CD47 expression
as a prognostic factor could be incorporated into standard clin-
ical prognostic considerations across multiple subtypes of NHL
and may be useful in risk-adapted therapy decision-making.
Although it is thought that many therapeutic mAbs for human
malignancies, including rituximab, function primarily through
NK cell-mediated ADCC, several lines of evidence indicate that
the therapeutic effect of anti-CD47 antibody alone or in combi-
nation with rituximab is mediated primarily through macrophage
phagocytosis. First, synergistic macrophage phagocytosis was
observed with combination anti-CD47 antibody and rituximab
in vitro, whereas no synergy was observed for direct apoptosis,
ADCC, or CDC (Figures 2C–2E, Figure S6, Figure 6A, and Figures
6C–6F). Second, phagocytosis of NHL cells in vivo was observed
with either anti-CD47 antibody or rituximab alone and, most
importantly, significantly increased with combination therapy
(Figure 7F). Third, the therapeutic effect of combination antibody
treatment was similar in an NHL xenotransplant model using
complement and NK cell-deficient NSG mice (Figure 4C) as in
complement and NK cell-competent SCID mice (Figures S4A
Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 709
Figure 7. Anti-CD47 Antibody Synergizes with Rituximab through FcR-Independent and FcR-Dependent Mechanisms
(A) Isobologram analysis of phagocytosis induced by anti-SIRPa antibody and rituximab is shown for Raji cells and mouse macrophages.
(B and C) NHL cells were incubated in vitro with the indicated antibodies in the presence of wild-type (B) or Fcgr
/
(C) mouse macrophages, and the phagocytic
index was determined.
(D) Isobologram analysis of phagocytosis induced by anti-CD47 F(ab0 )2 antibody and rituximab is shown for Raji cells and mouse macrophages.
(E) NHL cells were incubated with wild-type mouse macrophages in the presence of the indicated full-length or F(ab0 )2 antibodies (single antibodies at 10 mg/ml,
combination antibodies at 5 mg/ml each) and the phagocytic index was determined.
(F) The level of in vivo phagocytosis measured as the percentage of mouse macrophages containing phagocytosed GFP+ Raji cells (hCD45
GFP+F4/80+) was
determined by flow cytometry of livers from mice engrafted with GFP+ Raji cells and then treated with the indicated antibodies (see Experimental Procedures),
with each treatment group performed in duplicate. Compared to IgG control, anti-CD47 antibody and rituximab enabled increased levels of phagocytosis.
Compared to anti-CD47 antibody alone, combination anti-CD47 antibody and rituximab enabled higher levels of phagocytosis.
*p < 0.05, **p < 0.01, ****p < 0.0001. Error bars represent SD (E and F). See also Figure S7.

710 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.

and S4B), suggesting that macrophages alone are sufficient to
mediate the therapeutic effect. Fourth, depletion of macro-
phages, but not complement or NK cells, abrogated the syner-
gistic effect of anti-CD47 antibody in combination with rituximab
(Figure S6E). These studies highlight the importance of macro-
phages as effectors of anti-CD47 antibody therapy in human
NHL.
This study describes a mechanism of antibody synergy in the
elimination of NHL in the absence of chemotherapy. Combina-
tion antibody therapies for NHL have previously been investi-
gated, mostly in combination with rituximab, with some pro-
gressing to clinical trials. These include a humanized antibody
targeting the B cell antigen CD22 (epratuzumab) and galiximab,
a chimeric antibody targeting the costimulatory ligand CD80.
Phase I/II studies with either epratuzumab or galiximab in combi-
nation with rituximab demonstrate relative safety and clinical
responses equal to or greater than single agent therapy alone
(Leonard et al., 2005, 2007, 2008). Based on these results, phase
III trials are underway. Antibody combinations involving anti-
CD20 antibodies and antibodies to proapoptotic receptors are
also being explored preclinically (Daniel et al., 2007; Maddipatla
et al., 2007). These studies highlight the clinical potential of
combination antibody approaches in NHL patients.
Combination therapy with two or more mAbs possesses
several advantages compared to monotherapies in NHL or other
malignancies. First, therapy solely with monoclonal antibodies
targeting cancer-specific antigens could result in decreased
off-target toxicity compared to current therapeutic regimens
that utilize chemotherapy. Second, synergy between two distinct
antibody effector mechanisms, FcR-independent and FcR-
dependent as shown here, would result in increased therapeutic
efficacy. Third, antibody targeting of two distinct cell-surface
antigens would be more likely to eliminate cancer cells with
pre-existing epitope variants or epitope loss, such as those
reported in rituximab-refractory/resistant NHL patients (Foran
et al., 2001; Hiraga et al., 2009; Kennedy et al., 2004). Fourth,
a bispecific FcR-engaging antibody with one arm binding and
blocking CD47 and the other binding to a validated cancer anti-
body target (CD20) could reduce potential antibody toxicity,
while retaining the synergy effect, especially as CD47 is
expressed in multiple normal tissue types. Although we demon-
strated that an anti-mouse CD47 antibody is relatively nontoxic
to wild-type mice (Majeti et al., 2009), a clinical anti-human
CD47 antibody may have a different human toxicity profile that
could be overcome by a bispecific antibody.
In addition to its application in NHL, the reported mechanism
of antibody synergy provides proof-of-principle that a blocking
mAb directed against CD47 can synergize with an FcR-acti-
vating antibody to provide superior therapeutic efficacy. This
finding raises the possibility of potential synergy between an
anti-CD47 antibody and other clinically approved therapeutic
antibodies that may activate FcRs on immune effector cells for
the treatment of diverse human malignancies, including trastu-
zumab (Herceptin) for HER2-positive breast carcinomas, cetux-
imab (Erbitux) for colorectal carcinomas and head and neck
squamous cell carcinomas, alemtuzumab (Campath) for CLL
and T cell lymphoma, and others in development (Finn, 2008).
To date, we have demonstrated effective anti-CD47 antibody
targeting of several human cancers including AML (Majeti
et al., 2009), bladder cancer (Chan et al., 2009), and now NHL,
leading us to speculate that CD47 targeting will be effective
against a wide range of human cancers.
EXPERIMENTAL PROCEDURES
Cell Lines
A Burkitt’s lymphoma cell line (Raji) and a DLBCL cell line (SUDHL4) were
obtained from the American Type Culture Collection or generated in the lab.
The NHL17* cell line was generated from a patient with DLBCL by culturing
bulk cells in vitro with IMDM supplemented with 10% human AB serum for
1.5 months.
Human Samples
Normal human peripheral blood and human NHL samples were obtained from
the Stanford University Medical Center (Stanford, CA, USA) with informed
consent, according to an IRB-approved protocol (Stanford IRB# 13500) or
with informed consent from the Norwegian Radium Hospital (Oslo, Norway)
according to a Regional Ethic Committee (REK)-approved protocol (REK#
2.2007.2949). Normal tonsils for germinal center B cell analysis were obtained
from discarded tonsillectomy specimens from consented pediatric patients at
Stanford University Medical Center according to an IRB-approved protocol
(Stanford IRB# 13500).
Flow Cytometry Analysis
For analysis of normal peripheral blood cells, germinal center B cells, and
primary NHL cells, the following antibodies were used: CD19, CD20, CD3,
CD10, CD45, CD5, CD38 (Invitrogen, Carlsbad, CA, USA and BD Biosciences,
San Jose, CA, USA). Analysis of CD47 expression was performed with an anti-
human CD47 FITC antibody (clone B6H12.2, BD Biosciences). Cell staining
and flow cytometry analysis was performed as previously described (Majeti
et al., 2009).
Evaluation of Prognostic Value of CD47 in NHL
Gene expression and clinical data were analyzed for eight previously
described cohorts of adult NHL patients, including four studies of patients
with DLBCL, three with B-CLL, and one with MCL (detailed in Table S1). The
clinical end points analyzed included overall (OS), progression free (PFS),
and event-free survival (EFS), with events defined as the interval between
study enrollment and need for therapy or death from any cause, with data
censored for patients who did not have an event at the last follow-up visit.
See Extended Experimental Procedures for a detailed description of the
analyses.
Therapeutic Antibodies
Rituximab (anti-CD20, human IgG1) was obtained from the Stanford University
Medical Center, mouse anti-human CD20, IgG2a from Beckman Coulter
(Miami, FL, USA), and anti-CD47 antibody BRIC126, IgG2b from AbD Serotec
(Raleigh, NC, USA). Other anti-CD47 antibodies were used as in Majeti et al.
(2009). All in vivo antibody experiments were performed using the anti-CD47
B6H12.2 antibody.
In Vitro Isobologram Studies
In vitro phagocytosis assays were conducted with NHL cells incubated with
anti-CD47 antibody (B6H12.2), anti-CD20 IgG2a, or rituximab either alone or
in combination at concentrations from 1 mg/ml to 10 mg/ml. The concentration
of each antibody required to produce a defined single-agent effect (phagocytic
index) was determined for each cell type. Concentrations of the two antibodies
combined to achieve this same phagocytic index were then plotted on an
isobologram and the combination index (CI) determined. The CI was calcu-
lated from the formula CI = (d1/D1) + (d2/D2), whereby d1 = dose of drug
1 in combination to achieve the phagocytic index, d2 = dose of drug 2 in
combination to achieve the phagocytic index, D1 = dose of drug 1 alone to
achieve the phagocytic index, D2 = dose of drug 2 alone to achieve the
Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 711
phagocytic index. A CI of less than, equal to, and greater than 1 indicates
synergy, additivity, and antagonism, respectively.
Annexin V Apoptosis Assays
Assays were performed as previously described (Majeti et al., 2009).
Preparation of Human and Mouse Immune Effector Cells, Immune
Effector Cytotoxicity Assays, and In Vivo Depletion of Immune
Effector Cells
See Extended Experimental Procedures.
Preparation of F(ab0 )2 Fragments
See Extended Experimental Procedures.
Generation of Luciferase-Positive Cell Lines and Luciferase Imaging
Analysis
See Extended Experimental Procedures.
In Vivo Precoating Engraftment Assay
Assay were performed as previously described (Majeti et al., 2009). Precoated
cells were transplanted intravenously into SCID mice or sublethally irradiated
(200 rads) NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG). All experiments involving
mice were performed according to Stanford University institutional animal
guidelines.
In Vivo Antibody Treatment in a Disseminated Lymphoma
Xenograft Model
1.5 3 106 luciferase-labeled Raji cells were injected intravenously into the
retro-orbital sinus of 6- to 10-week-old SCID or NSG mice. Those mice with
luciferase-positive lymphoma were given daily intraperitoneal injections of
200 mg mouse IgG control, anti-CD47 antibody, rituximab, or 200 mg anti-
CD47 antibody + 200 mg rituximab for 3 weeks. Antibody treatment was
then stopped and mice were followed for survival analysis. A complete remis-
sion (CR) was defined as no evidence of lymphoma by bioluminescence at the
end of treatment. A relapse was defined as evidence of lymphoma by biolumi-
nescence after the end of treatment in a mouse with a prior CR.
In Vivo Antibody Treatment in a Localized Lymphoma
Xenograft Model
3 3 106 luciferase-labeled Raji cells were injected subcutaneously into the
right flank of 6- to 10-week-old NSG mice. Those mice with luciferase-positive
lymphoma were given daily intraperitoneal injections of 400 mg mouse IgG
control, 400 mg anti-CD47, 200 mg rituximab, or 400 mg anti-CD47 + 200 mg rit-
uximab for 4 weeks. Tumor volume was measured every 3–4 days using the
formula (length*width)/2. Antibody treatment was then stopped and mice
were followed for survival analysis.
In Vivo Antibody Treatment of Primary NHL-Engrafted Mice
2 3 106 NHL cells were transplanted intravenously via retro-orbital plexus into
sublethally irradiated NSG mice. Two to ten weeks later, bone marrow was
aspirated from these mice and those mice with evidence of human lymphoma
engraftment (hCD45+CD19/CD10+ bone marrow cells) were then treated with
the same antibody regimen as in the disseminated lymphoma model. After
14 days, bone marrow cells from these mice were aspirated and antibody
treatment was stopped and mice followed for survival analysis. A CR was
defined as no evidence of lymphoma in the BM at end of treatment. A relapse
was defined as evidence of lymphoma in the BM after end of treatment in
a mouse with a prior CR.
In Vivo Phagocytosis
In vivo phagocytosis was performed as previously described (Majeti et al.,
2009) analyzing mice transplanted with GFP+ Raji cells into adult NSG mice.
Mice were given a single dose of antibody and analyzed 4 hr later for in vivo
phagocytosis.
712 Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, seven
figures, and three tables and can be found with this article online at doi:10.
1016/j.cell.2010.07.044.
ACKNOWLEDGMENTS
The authors acknowledge Dr. Christopher Contag for providing luciferase
constructs, Dr. Robert Negrin for assistance, Dr. Yaso Natkunam for immuno-
histochemistry, Libuse Jerabek and Theresa Storm for lab management, and
Adriane Mosely for animal husbandry. We also acknowledge the patients and
surgeons including Drs. Wapnir, Chang, Cheng, Janisiewicz, Koltai, Liebowitz,
and Messner for providing research specimens. M.P.C. is supported by the
HHMI and the Stanford Cancer Biology Program, A.A.A. by an NIH T32 Ruth
L. Kirschstein National Research Service Award (HL007970), R. Malumbres
by a fellowship from Fundación Caja Madrid, I.S.L. by NIH grant CA122105,
and R. Majeti by a grant from the AACR. R. Majeti holds a Career Award for
Medical Scientists from the Burroughs Wellcome Fund. I.L.W., M.P.C.,
A.A.A., and R. Majeti have filed U.S. Patent Application Serial No. 12/321,215
entitled ‘‘Methods For Manipulating Phagocytosis Mediated by CD47.’’ This
research is supported by NIH grant P01CA139490 to I.L.W. and funding from
the Smith Family Fund.
M.P.C., A.A.A., I.L.W., and R. Majeti designed the experiments, and M.P.C.,
A.A.A., I.L.W., and R. Majeti wrote the manuscript. M.P.C., A.A.A., C.T.,
J.H.M., S.G., M.J., A.C.C., C.K.C., B.T.T., C.Y.P, F.Z., R. Malumbres, and
I.S.L. performed the experiments and analyzed data. J.B., R.D.G., R.L., and
I.S.L provided patient samples and clinical data. H.E.K. provided reagents.
B.V. generated F(ab0 )2 fragments of anti-CD47 antibody and rituximab. All
authors endorse the full content of this work.
Received: January 7, 2010
Revised: April 23, 2010
Accepted: July 6, 2010
Published: September 2, 2010
REFERENCES
Adams, G.P., and Weiner, L.M. (2005). Monoclonal antibody therapy of cancer.
Nat. Biotechnol. 23, 1147–1157.
Alizadeh, A.A., Eisen, M.B., Davis, R.E., Ma, C., Lossos, I.S., Rosenwald, A.,
Boldrick, J.C., Sabet, H., Tran, T., Yu, X., et al. (2000). Distinct types of diffuse
large B-cell lymphoma identified by gene expression profiling. Nature 403,
503–511.
Brown, E.J., and Frazier, W.A. (2001). Integrin-associated protein (CD47) and
its ligands. Trends Cell Biol. 11, 130–135.
Cartron, G., Watier, H., Golay, J., and Solal-Celigny, P. (2004). From the bench
to the bedside: ways to improve rituximab efficacy. Blood 104, 2635–2642.
Chan, K.S., Espinosa, I., Chao, M., Wong, D., Ailles, L., Diehn, M., Gill, H.,
Presti, J., Jr., Chang, H.Y., van de Rijn, M., et al. (2009). Identification, molec-
ular characterization, clinical prognosis, and therapeutic targeting of human
bladder tumor-initiating cells. Proc. Natl. Acad. Sci. USA 106, 14016–14021.
Cheson, B.D., and Leonard, J.P. (2008). Monoclonal antibody therapy for
B-cell non-Hodgkin’s lymphoma. N. Engl. J. Med. 359, 613–626.
Chou, T.C. (2008). Preclinical versus clinical drug combination studies. Leuk.
Lymphoma 49, 2059–2080.
Daniel, D., Yang, B., Lawrence, D.A., Totpal, K., Balter, I., Lee, W.P., Gogineni,
A., Cole, M.J., Yee, S.F., Ross, S., and Ashkenazi, A. (2007). Cooperation of the
proapoptotic receptor agonist rhApo2L/TRAIL with the CD20 antibody rituxi-
mab against non-Hodgkin lymphoma xenografts. Blood 110, 4037–4046.
Finn, O.J. (2008). Cancer immunology. N. Engl. J. Med. 358, 2704–2715.
Foran, J.M., Norton, A.J., Micallef, I.N., Taussig, D.C., Amess, J.A., Rohatiner,
A.Z., and Lister, T.A. (2001). Loss of CD20 expression following treatment with
rituximab (chimaeric monoclonal anti-CD20): a retrospective cohort analysis.
Br. J. Haematol. 114, 881–883.
Glennie, M.J., French, R.R., Cragg, M.S., and Taylor, R.P. (2007). Mechanisms
of killing by anti-CD20 monoclonal antibodies. Mol. Immunol. 44, 3823–3837.
Hiraga, J., Tomita, A., Sugimoto, T., Shimada, K., Ito, M., Nakamura, S., Kiyoi,
H., Kinoshita, T., and Naoe, T. (2009). Down-regulation of CD20 expression in
B-cell lymphoma cells after treatment with rituximab-containing combination
chemotherapies: its prevalence and clinical significance. Blood 113, 4885–
4893.
Jaiswal, S., Jamieson, C.H., Pang, W.W., Park, C.Y., Chao, M.P., Majeti, R.,
Traver, D., van Rooijen, N., and Weissman, I.L. (2009). CD47 is upregulated
on circulating hematopoietic stem cells and leukemia cells to avoid phagocy-
tosis. Cell 138, 271–285.
Jaiswal, S., Chao, M.P., Majeti, R., and Weissman, I.L. (2010). Macrophages
as mediators of tumor immunosurveillance. Trends Immunol. 31, 212–219.
Katzenberger, T., Petzoldt, C., Holler, S., Mader, U., Kalla, J., Adam, P., Ott,
M.M., Muller-Hermelink, H.K., Rosenwald, A., and Ott, G. (2006). The Ki67
proliferation index is a quantitative indicator of clinical risk in mantle cell
lymphoma. Blood 107, 3407.
Kennedy, A.D., Beum, P.V., Solga, M.D., DiLillo, D.J., Lindorfer, M.A., Hess,
C.E., Densmore, J.J., Williams, M.E., and Taylor, R.P. (2004). Rituximab infu-
sion promotes rapid complement depletion and acute CD20 loss in chronic
lymphocytic leukemia. J. Immunol. 172, 3280–3288.
Kikuchi, Y., Uno, S., Yoshimura, Y., Otabe, K., Iida, S., Oheda, M., Fukushima,
N., and Tsuchiya, M. (2004). A bivalent single-chain Fv fragment against CD47
induces apoptosis for leukemic cells. Biochem. Biophys. Res. Commun. 315,
912–918.
Kikuchi, Y., Uno, S., Kinoshita, Y., Yoshimura, Y., Iida, S., Wakahara, Y.,
Tsuchiya, M., Yamada-Okabe, H., and Fukushima, N. (2005). Apoptosis
inducing bivalent single-chain antibody fragments against CD47 showed anti-
tumor potency for multiple myeloma. Leuk. Res. 29, 445–450.
Kim, M.J., Lee, J.C., Lee, J.J., Kim, S., Lee, S.G., Park, S.W., Sung, M.W., and
Heo, D.S. (2008). Association of CD47 with natural killer cell-mediated cytotox-
icity of head-and-neck squamous cell carcinoma lines. Tumour Biol. 29,
28–34.
Leonard, J.P., Coleman, M., Ketas, J., Ashe, M., Fiore, J.M., Furman, R.R.,
Niesvizky, R., Shore, T., Chadburn, A., Horne, H., et al. (2005). Combination
antibody therapy with epratuzumab and rituximab in relapsed or refractory
non-Hodgkin’s lymphoma. J. Clin. Oncol. 23, 5044–5051.
Leonard, J.P., Friedberg, J.W., Younes, A., Fisher, D., Gordon, L.I., Moore, J.,
Czuczman, M., Miller, T., Stiff, P., Cheson, B.D., et al. (2007). A phase I/II study
of galiximab (an anti-CD80 monoclonal antibody) in combination with rituxi-
mab for relapsed or refractory, follicular lymphoma. Ann. Oncol. 18, 1216–
1223.
Leonard, J.P., Schuster, S.J., Emmanouilides, C., Couture, F., Teoh, N.,
Wegener, W.A., Coleman, M., and Goldenberg, D.M. (2008). Durable complete
responses from therapy with combined epratuzumab and rituximab: final
results from an international multicenter, phase 2 study in recurrent, indolent,
non-Hodgkin lymphoma. Cancer 113, 2714–2723.
Maddipatla, S., Hernandez-Ilizaliturri, F.J., Knight, J., and Czuczman, M.S.
(2007). Augmented antitumor activity against B-cell lymphoma by a combina-
tion of monoclonal antibodies targeting TRAIL-R1 and CD20. Clin. Cancer Res.
13, 4556–4564.
Majeti, R., Chao, M.P., Alizadeh, A.A., Pang, W.W., Jaiswal, S., Gibbs, K.D.,
Jr., van Rooijen, N., and Weissman, I.L. (2009). CD47 is an adverse prognostic
factor and therapeutic antibody target on human acute myeloid leukemia stem
cells. Cell 138, 286–299.
Mateo, V., Lagneaux, L., Bron, D., Biron, G., Armant, M., Delespesse, G., and
Sarfati, M. (1999). CD47 ligation induces caspase-independent cell death in
chronic lymphocytic leukemia. Nat. Med. 5, 1277–1284.
Milush, J.M., Long, B.R., Snyder-Cappione, J.E., Cappione, A.J., 3rd, York,
V.A., Ndhlovu, L.C., Lanier, L.L., Michaelsson, J., and Nixon, D.F. (2009). Func-
tionally distinct subsets of human NK cells and monocyte/DC-like cells identi-
fied by coexpression of CD56, CD7, and CD4. Blood 114, 4823–4831.
Nimmerjahn, F., and Ravetch, J.V. (2007). Antibodies, Fc receptors and
cancer. Curr. Opin. Immunol. 19, 239–245.
Oscier, D.G., Gardiner, A.C., Mould, S.J., Glide, S., Davis, Z.A., Ibbotson, R.E.,
Corcoran, M.M., Chapman, R.M., Thomas, P.W., Copplestone, J.A., et al.
(2002). Multivariate analysis of prognostic factors in CLL: clinical stage,
IGVH gene mutational status, and loss or mutation of the p53 gene are inde-
pendent prognostic factors. Blood 100, 1177–1184.
Rosenwald, A., Wright, G., Chan, W.C., Connors, J.M., Campo, E., Fisher, R.I.,
Gascoyne, R.D., Muller-Hermelink, H.K., Smeland, E.B., Giltnane, J.M., et al.
(2002). The use of molecular profiling to predict survival after chemotherapy
for diffuse large-B-cell lymphoma. N. Engl. J. Med. 346, 1937–1947.
Rosenwald, A., Wright, G., Wiestner, A., Chan, W.C., Connors, J.M., Campo,
E., Gascoyne, R.D., Grogan, T.M., Muller-Hermelink, H.K., Smeland, E.B.,
et al. (2003). The proliferation gene expression signature is a quantitative inte-
grator of oncogenic events that predicts survival in mantle cell lymphoma.
Cancer Cell 3, 185–197.
Shultz, L.D., Lyons, B.L., Burzenski, L.M., Gott, B., Chen, X., Chaleff, S., Kotb,
M., Gillies, S.D., King, M., Mangada, J., et al. (2005). Human lymphoid and
myeloid cell development in NOD/LtSz-scid IL2R gamma null mice engrafted
with mobilized human hemopoietic stem cells. J. Immunol. 174, 6477–6489.
Smith, M.R. (2003). Rituximab (monoclonal anti-CD20 antibody): mechanisms
of action and resistance. Oncogene 22, 7359–7368.
Takai, T., Li, M., Sylvestre, D., Clynes, R., and Ravetch, J.V. (1994). FcR
gamma chain deletion results in pleiotrophic effector cell defects. Cell 76,
519–529.
Tallarida, R.J. (2006). An overview of drug combination analysis with isobolo-
grams. J. Pharmacol. Exp. Ther. 319, 1–7.
Taylor, R.P., and Lindorfer, M.A. (2008). Immunotherapeutic mechanisms of
anti-CD20 monoclonal antibodies. Curr. Opin. Immunol. 20, 444–449.
The International Non-Hodgkin’s Lymphoma Prognostic Factors Project.
(1993). A predictive model for aggressive non-Hodgkin’s lymphoma. N. Engl.
J. Med. 329, 987–994.
Uno, S., Kinoshita, Y., Azuma, Y., Tsunenari, T., Yoshimura, Y., Iida, S., Kiku-
chi, Y., Yamada-Okabe, H., and Fukushima, N. (2007). Antitumor activity of
a monoclonal antibody against CD47 in xenograft models of human leukemia.
Oncol. Rep. 17, 1189–1194.
Cell 142, 699–713, September 3, 2010 ª2010 Elsevier Inc. 713
A C-Type Lectin Collaborates with a
CD45 Phosphatase Homolog to Facilitate
West Nile Virus Infection of Mosquitoes
Gong Cheng,1 Jonathan Cox,1 Penghua Wang,1 Manoj N. Krishnan,1 Jianfeng Dai,1 Feng Qian,2 John F. Anderson,3
and Erol Fikrig1,4,*
1Sectionof Infectious Diseases
2Sectionof Rheumatology
Department of Internal Medicine, Yale University School of Medicine, New Haven, CT 06520, USA
3Department of Entomology, Connecticut Agricultural Experiment Station, New Haven, CT 06504, USA
4Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA
*Correspondence: erol.fikrig@yale.edu
DOI 10.1016/j.cell.2010.07.038
SUMMARY
West Nile virus (WNV) is the most common
arthropod-borne flavivirus in the United States;
however, the vector ligand(s) that participate in infec-
tion are not known. We now show that an Aedes
aegypti C-type lectin, mosGCTL-1, is induced by
WNV, interacts with WNV in a calcium-dependent
manner, and facilitates infection in vivo and in vitro.
A mosquito homolog of human CD45 in A. aegypti,
designated mosPTP-1, recruits mosGCTL-1 to
enable viral attachment to cells and to enhance viral
entry. In vivo experiments show that mosGCTL-1 and
mosPTP-1 function as part of the same pathway
and are critical for WNV infection of mosquitoes.
A similar phenomenon was also observed in Culex
quinquefasciatus, a natural vector of WNV, further
demonstrating that these genes participate in WNV
infection. During the mosquito blood-feeding pro-
cess, WNV infection was blocked in vivo with
mosGCTL-1 antibodies. A molecular understanding
of flaviviral-arthropod interactions may lead to strat-
egies to control viral dissemination in nature.
INTRODUCTION
West Nile virus (WNV) is maintained in a bird-mosquito transmis-
sion cycle and has become the most common arthropod-borne
flavivirus in the United States. Humans, horses, and other nona-
vian vertebrates are incidental hosts (Monath and Heinz, 1996).
Infection in man can result in fever, meningitis, or encephalitis,
among other symptoms (Hubálek and Halouzka, 1999; Leis
et al., 2002). Approved human vaccines or therapeutics are not
available and preventive measures largely focus upon mosquito
control (Reisen and Brault, 2007).
The ability of different mosquito species to transmit WNV
varies widely. Culex spp. are the major vectors for WNV world-
714 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.
wide (Nir et al., 1968; Turell et al., 2000). In the United States,
the most important vectors are Culex pipiens in the east, Culex
tarsalis in the midwest and west, and Culex quinquefasciatus in
the southeast (Hayes et al., 2005). WNV has also been isolated
from Aedes, Ochlerotatus, and Culisetam mosquitoes (http://
www.cdc.gov/ncidod/dvbid/westnile/mosquitoSpecies.htm).
Aedes aegypti, a member of the Culicinae subfamily, is a major
vector for numerous flaviviruses (Gould and Solomon, 2008).
A. aegypti is ideal for viral pathogenesis studies because these
mosquitoes are easy to cultivate and the genome has been char-
acterized (Gubler, 1998; Nene et al., 2007; Halstead, 2008).
A. aegypti is readily susceptibility to infection with WNV in
the laboratory, and the virus rapidly disseminates throughout
most of the mosquito after the blood meal. As with Culex spp.,
A. aegypti is a threat for the transmission of WNV to humans
(Vanlandingham et al., 2007).
C-type lectins are a group of carbohydrate-binding proteins
(Zelensky and Gready, 2005). Several members of this family
are highly expressed by immune cells, including monocytes,
macrophages, and dendritic cells (DCs), and play a central
role in activating host defenses (Robinson et al., 2006). Human
mannose-binding lectin (MBL) is a pattern recognition molecule
of the innate immune system that binds to sugars on the surface
of invading pathogens, leading to opsonization, phagocytosis,
and activation of the complement pathway (Neth et al., 2002).
In contrast, some C-type lectins are recruited to facilitate flavivi-
ral infection. In mammals, two membrane C-type lectins,
DC-SIGN (CD209) and L-SIGN (CD209L), interact with flavivi-
ruses via high mannose glycans on viral glycoproteins (Klimstra
et al., 2003) and are essential host cell factors exploited by
dengue virus (DENV) and WNV to invade immature DCs and
macrophages (Geijtenbeek et al., 2000; Soilleux et al., 2002;
Tassaneetrithep et al., 2003; Davis et al., 2006). Another
C-type lectin, the mannose receptor (MR), also interacts with
the DENV envelope protein and may enhance viral attachment
to phagocytes (Miller et al., 2008). A recent study identified
C-type lectin domain family 5, member A (CLEC5A), as
a DENV receptor. The association between CLEC5A and
DENV does not result in viral entry, but rather stimulates the
Figure 1. Effect of WNV Infection on mosGCTL-1 Expression
(A) Viral distribution in A. aegypti. WNV (1 3 103 M.I.D50) was inoculated into the mosquito thorax.
(B–E) mosGCTL-1 messenger RNA (mRNA) was induced by viral infection in whole A. aegypti (B), salivary glands (C), hemolymph (D), and midgut (E). Total RNA
was isolated from various tissues or whole mosquitoes at 5 time points after viral infection.
(A–E) The viral load and mosGCTL-1 mRNA levels were determined by Taqman RT-QPCR and normalized with A. aegypti actin (AAEL011197). WNV (1 3 103
M.I.D50) was microinjected into each mosquito. Data are shown as the mean ± standard error (SEM). The experiments were repeated three times.
(F) Immunoblot to detect mosGCTL-1 in WNV-infected mosquitoes. Three WNV infected or control mosquitoes were pooled and homogenized. The supernatant
was then isolated, separated by SDS-PAGE, and probed with rabbit mosGCTL-1 antisera. UI, uninfected mosquitoes; I, WNV-infected mosquitoes; dpi, days
postinfection. Fifty micrograms of protein from mosquito lysates was loaded into each lane.
See also Figure S1 and Table S1.

release of proinflammatory cytokines, potentially contributing to
the pathogenesis of dengue hemorrhagic fever (Chen et al.,
2008).
Protein tyrosine phosphatases (PTPs) remove phosphate
groups from phosphorylated tyrosine residues and play critical
roles in cell communication, shape, motility, proliferation, and
differentiation (Alonso et al., 2004; Mustelin et al., 2005). One
well-known PTP, protein tyrosine phosphatase receptor type C
(PTPRC, CD45), is important for thymocyte development and
T cell activation (Byth et al., 1996; Trowbridge and Thomas,
1994) and is expressed on all nucleated cells of hemopoietic
origin (Thomas, 1989). The association between MBL and
CD45 in immature T cells influences thymocyte development
(Baldwin and Ostergaard, 2001).

Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 715

Figure 2. The Function of mosGCTL-1 in WNV Infection
(A and B) mosGCTL-1 silencing. The mock group was treated with the same amount of GFP dsRNA. mosGCTL-1- or GFP-dsRNA-treated mosquitoes were used
to isolate total RNA at several time points post dsRNA injection. mRNA levels were determined by SYBR Green RT-QPCR (A). mosGCTL-1-dsRNA-treated or
mock-treated mosquitoes were collected at 6 days after gene silencing. The supernatant was separated by SDS-PAGE and probed with rabbit mosGCTL-1
antisera (B).
(C and D) Silencing mosGCTL-1 impairs WNV (C), but not dengue virus (D), infection. The viral burden was examined at day 6 after infection. WNV or dengue virus
(10 M.I.D50) was used to challenge mosquitoes. The viral load was determined by Taqman RT-QPCR and normalized with A. aegypti actin. The result shown was
representative of four independent experiments.
(E) The role of the mosGCTL paralogues in WNV infection. The sample number was no less than 12 mosquitoes in each group. The viral burden was determined by
Taqman RT-QPCR and normalized with A. aegypti actin. *p < 0.05. The results were pooled from two independent experiments.
(F) WNV induces mosGCTL-1 homolog expression in C. quinquefasciatus. WNV-infected and mock mosquitoes were collected at 6 days after infection. Culex
mosGCTL-1 mRNA was determined by SYBR Green RT-QPCR and normalized with C. quinquefasciatus actin (CPIJ012570). The experiment was repeated three
times with similar results.
(G) Efficiency of Culex mosGCTL-1 RNA interference. The mock group was treated with the same amount of GFP dsRNA. mRNA levels were determined by SYBR
Green RT-QPCR.
(H) Silencing of Culex mosGCTL-1 decreases the WNV burden in C. quinquefasciatus. The viral burden was determined by Taqman RT-QPCR. The results were
combined from three independent experiments.

716 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.

 Genome-wide RNA interference (RNAi) screening studies
have revealed several hundred host factors that influence WNV
or DENV infection in human or Drosophila cell lines and have
identified cellular pathways that have a role in viral internaliza-
tion, replication, assembly, or secretion (Krishnan et al., 2008;
Sessions et al., 2009). However, the relationship between flavivi-
ruses and mosquitoes is not well understood. We now examine
WNV-mosquito interactions by using A. aegypti and Culex quin-
quefasciatus, characterize the expression profile of mosquito
homologs of human susceptibility proteins (Krishnan et al.,
2008) in response to viral infection, and identify a lectin-based
pathway that is critical for viral infection of mosquitoes.
RESULTS
A recent RNAi screening study characterized 283 human
proteins that facilitate WNV infection (Krishnan et al., 2008).
We have now identified 215 homologs of these genes in
A. aegypti. The expression of 32 genes was altered by the pres-
ence of WNV in mosquitoes, potentially suggesting a role in viral
infection of the arthropod (Table S1 available online). To examine
their function, we therefore silenced each of these 32 genes in
mosquitoes with RNAi and assessed the effect on viral load.
Knockdown of 13 genes significantly altered the viral burden in
the vector (Table S1 and Figure S1). One of these genes,
AAEL000563, exhibited the most dramatic reduction in viral
load after silencing and was selected as the target for further
investigation. AAEL000563 belongs to the A. aegypti galac-
tose-specific binding C-type lectin family, shares 26% amino
acid identity with human mannose-binding lectin, and was
designated as mosGCTL-1 (mosquito galactose-specific binding
C-type lectin).
mosGCTL-1 Expression Increases during WNV Infection
of A. aegypti
To further investigate the relationship between WNV infection
and mosGCTL-1 expression, we determined the viral load and
mosGCTL-1 level in selected A. aegypti tissues after the inocula-
tion of WNV into female mosquitoes. WNV was readily detect-
able 4 days after infection, and the viral level subsequently
increased. The salivary glands and hemolymph had the highest
levels of WNV, while the viral load in the midgut was substantially
lower (Figure 1A). mosGCTL-1 expression was induced in
diverse tissues over time, including the salivary glands, hemo-
lymph, and midgut (Figures 1B–1E). Immunoblots also demon-
strated an increased amount of mosGCTL-1 in WNV-infected
mosquitoes (Figure 1F).
mosGCTL-1 Facilitates WNV Infection in A. aegypti
Double-stranded RNA (dsRNA)-mediated gene silencing studies
then assessed the role of mosGCTL-1 in WNV infection of
(A, C, D and H) Statistical analysis was done with the Mann-Whitney test in all experiments. Each dot represents the mRNA levels in an individual mosquito.
The horizontal line depicts the medians.
(E–G) The Mann-Whitney test was used for statistical analysis. Data are shown as the mean ± standard error (SEM).
See also Figure S2 and Table S2.
mosquitoes. Total RNA from mock-injected and mosGCTL-1
dsRNA-injected mosquitoes was analyzed by RT-QPCR. Com-
pared to the mock group, the target gene was silenced 6- to
10-fold from day 3 through 9 (Figures 2A and 2B). WNV was
therefore inoculated into mosquitoes on day 3 after dsRNA treat-
ment, and the viral load was quantified on day 9. There was
a 3-fold reduction in the WNV burden in mosGCTL-1 silenced
A. aegypti (Figure 2C, p < 0.0001) compared to controls.
In contrast, the burden of a related flavivirus, dengue virus,
was unaffected by mosGCTL-1 silencing (Figure 2D), suggesting
that mosGCTL-1 has a specific role in WNV infection of
mosquitoes.
As mosGCTL-1 is part of a multigene family with 25 members
and shares 22%–48% identity with many of these genes, we
examined the role of the mosGCTL paralogues in WNV and
DENV infection. Twenty-one of these 25 genes were expressed
in adult female A. aegypti (Tables S1 and S2). Since AAEL014382
and AAEL014390 share greater than 95% identity with
AAEL011607 and AAEL011619, respectively, AAEL011607 and
AAEL011619 were selected for functional assays. We silenced
each of the expressed mosGCTL genes and then infected the
mosquitoes with virus. In the WNV challenge, two additional
mosGCTL subtypes (AAEL000556 and AAEL011610) showed
the similar phenotype to mosGCTL-1 (Figure 2E), and in DENV
infection, silencing of 5 mosGCTL-related genes reduced the
DENV burden (Figure S2).
Culex mosquitoes are a common vector for WNV in nature.
We therefore extended the studies to Culex quinquefasciatus,
a example of this species that is abundant in the southeast
U.S. We identified a mosGCTL-1 homolog (CPIJ010995, Culex
mosGCTL-1) in C. quinquefasciatus that has a higher degree
of similarity (64%) with mosGCTL-1 than other homologs.
We then determined whether Culex mosGCTL-1 had a similar
role in WNV infection. Culex mosGCTL-1 was upregulated by
WNV infection (Figure 2F). Silencing of Culex mosGCTL-1 in
C. quinquefasciatus reduced the WNV burden compared to the
control group (Figures 2G and 2H, p < 0.0001), suggesting that
Culex mosGCTL-1 plays a role in facilitating WNV infection in
Culex mosquitoes.
mosGCTL-1 Interacts with WNV
Our results show that silencing mosGCTL-1 in A. aegypti and
C. quinquefasciatus reduces WNV infection. The mechanism
by which mosGCTL-1 facilitates viral infection in mosquitoes
was, therefore, investigated. We first generated A. aegypti
mosGCTL-1 protein in a Drosophila expression system (Fig-
ure 3A) and investigated whether mosGCTL-1 associates with
WNV envelope (E) protein. Immunoprecipitation experiments
revealed that these two proteins strongly interacted in a
calcium-dependent manner (Figure 3B). We then determined,
by ELISA, that mosGCTL-1 bound to WNV virions. mosGCTL-1
Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 717
Figure 3. mosGCTL-1 Interacts with WNV
(A) Recombinant mosGCTL-1, produced in Drosophila cells, and purified with a Ni-His column (left). Glycosylated mosGCTL-1 was detected with a V5-HRP mAb
(right). The control was the supernatant of mock-infected S2 cells.
(B) mosGCTL-1 interacted with WNV E protein in a coimmunoprecipitation assay. The protein complex was pulled down with a flavivirus E mAb and probed with
anti-V5-HRP mAb. The experiments were repeated three times.
(C) mosGCTL-1 captured West Nile virions by ELISA. The interaction was determined with a flavivirus E mAb. Data are expressed as the mean ± standard error
from three independent experiments.
(D) Comicroinjection of the purified mosGCTL-1 with WNV enhanced virus infection in A. aegypti. The various amount of purified mosGCTL-1 were premixed with
WNV (10 M.I.D50 per mosquito) for 1 hr at 4 C. The mixture was microinjected into mosquito and compared to the control group inoculated with the same amount
WNV. The mosquitoes were collected at 3 days (i), 6 days (ii), and 9 days (iii) after WNV inoculation. Total RNA was isolated to determine the viral burden by
Taqman RT-QPCR and normalized with A. aegypti actin. Each dot represents the mRNA level in one mosquito. The horizontal line represents the medians.
n.s., nonsignificance (p > 0.05). The Mann-Whitney test was used for analysis. Three independent experiments yielded similar data.
(E) The association between mosGCTL-1 and WNV in A .aegypti hemolymph. Hemolymph was collected from WNV-infected or mock-infected mosquitoes
for immunofluorescence staining. WNV E protein was stained with anti-mouse IgG Alexa-488 (green), and mosGCTL-1 was identified with anti-rabbit
IgG Alexa-546 (red). Nuclei were stained blue with To-Pro-3 iodide. The white arrow represents the induced expression of mosGCTL-1 in WNV-infected

718 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.

efficiently captured virus in the presence of calcium, and
this interaction was inhibited by EDTA (Figure 3C). Since this
association suggests a role during infection, we premixed
mosGCTL-1 and WNV and then coinjected the combination
into A. aegypti and determined the viral burden. mosGCTL-1
protein significantly enhanced the viral load at days 3 (Fig-
ure 3D, i, p = 0.0001) and 6 (Figure 3D, ii, p < 0.0001) after viral
challenge.
To further characterize the association between mosGCTL-1
and WNV infection in vivo, we used immunofluorescence to
examine the hemolymph and salivary glands of WNV-infected
A. aegypti at different time points. Colocalization of mosGCTL-1
and WNV was clearly observed in both tissues at various inter-
vals after WNV infection (Figure 3E and Figure S3). In the hemo-
lymph, some hemocytes were highly infected by WNV, and
mosGCTL-1 was induced in these infected hemocytes
(Figure 3E). We counted the number of mosGCTL-1-positive
and WNV-infected hemocytes by microscopy. All these infected
cells stained positive for mosGCTL-1. The virus spread rapidly to
the salivary glands after inoculation (Figure S3), consistent with
the Q-PCR data (Figure 1A). mosGCTL-1 was identified on the
basement membrane of salivary glands in uninfected mosqui-
toes, and throughout this tissue after infection, suggesting that
mosGCTL-1 was induced by WNV infection of the salivary
glands (Figure S3).
mosPTP-1 Captures mosGCTL-1 onto the Cell Surface
Human mannose-binding lectin (MBL) (Zelensky and Gready,
2005) and mosGCTL-1 are both secreted. Human MBL is
thought to interact with several surface receptors to exert pleo-
trophic effects (Baldwin and Ostergaard, 2001; Arnold et al.,
2006). We therefore postulated that secreted mosGCTL-1
captures WNV and presents it to a ligand on the cell surface,
thereby facilitating viral entry. To assess this, we identified 11
A. aegypti homologs of human proteins that putatively interact
with human MBL (Table S3) from the Human Protein Reference
Database (http://www.hprd.org/) and examined their roles in
WNV infection of A. aegypti (Figure S4A). One mosquito CD45
homolog (AAEL013105, mosPTP-1) exhibited a phenotype
similar to that of mosGCTL-1. The viral burden was significantly
decreased in the mosPTP-1 dsRNA-treated mosquitoes (p <
0.002) (Figures 4A and 4B). To further determine whether
mosPTP-1 interacts with mosGCTL-1, we cloned and expressed
the extracellular region of mosPTP-1 (mosPTP-1-Ex) in a
Drosophila cell line (Figure 4C). mosGCTL-1 strongly interacted
with mosPTP-1-Ex in a coimmunoprecipitation assay (Figure 4D).
We then expressed the mosPTP-1 gene, including the trans-
membrane region (91 bp–2202 bp), in S2 cells (Figure 4C) and
examined the ability of mosPTP-1 to bind mosGCTL-1 on the
cell surface. A mock DNA vector transfected stable S2 cell line
was used as control. More than 40% of the mosPTP-1-express-
ing cells bound mosGCTL-1 and showed double-positive stain-
ing in FACS (Figure 4E, i) compared with controls (Figure 4E, ii,
iii). Confocal microscopy demonstrated that cells expressing
hemocytes. The yellow arrows show the infected hemocytes. Images were examined with a Zeiss LSM 510 Meta Confocal Microscope 633 objective
lens.
See also Figure S3.
high levels of mosPTP-1 in the cytoplasm also had substantial
mosGCTL-1 on the cell surface (Figure 4F). These results
suggest that mosGCTL-1 binds mosPTP-1.
To test whether mosPTP-1 has a conserved role in Culex spp.,
we identified the mosPTP-1 homolog from the Culex quinquefas-
ciatus genome (CPIJ014098, Culex mosPTP-1). Silencing of this
gene in Culex also influenced the viral burden (Figures S4B and
S4C), similar to mosPTP-1 in A. aegypti (Figure 4B), suggesting
that mosPTP-1 in both these mosquito species have a similar
role in WNV infection.
The mosGCTL-1/mosPTP-1 Pathway Has a Dominant
Role in WNV Infection
mosGCTL-1 and mosPTP-1 each facilitate WNV infection of
A. aegypti in vivo and in vitro. We therefore assessed whether
mosGCTL-1 and mosPTP-1 cooperate to enable viral infection.
WNV E protein, mosGCTL-1, and mosPTP-1-Ex form a complex
in which mosGCTL-1 is the key factor linking the other two
proteins (Figure 5A). WNV E protein did not interact directly
with mosPTP-1-Ex (Figure 5A). We then examined whether
mosPTP-1-expressing cells could recruit more WNV in the pres-
ence of mosGCTL-1. mosGCTL-1 and WNV were added to cells
and incubated at 4 C for membrane attachment. After washing,
the cells were transferred to room temperature and collected at
different time points to determine the viral burden (Krishnan
et al., 2007). mosPTP-1-expressing cells incorporated up to
5- to 10-fold more WNV in the presence of mosGCTL-1, com-
pared to control groups (Figure 5B). We then examined the role
of the mosGCTL-1/mosPTP-1 pathway for WNV infection in
A. aegypti. We knocked down the mosPTP-1 gene with dsRNA
and then inoculated the mosGCTL-1/WNV mixture at 3 days
after RNAi silencing. Silencing of mosPTP-1 interfered with the
ability of mosGCTL-1 to facilitate WNV infection (shown in Fig-
ure 3D) at 3 and 6 days after challenge (Figure 5C). The viral
burden of the mosPTP-1 RNAi/mosGCTL-1/WNV group was
2- to 3-fold lower than that of the mock/mosGCTL-1/WNV group
(p < 0.001) and decreased to a level similar to that of the group
that did not receive mosGCTL-1, suggesting that mosGCTL-1
and mosPTP-1 cooperate in the same pathway to enhance
WNV infection. To further determine the relationship between
mosGCTL-1 and mosPTP-1 in infection, we silenced both of
these genes with dsRNA. These genes were successfully
knocked down in mosquitoes in the cosilenced group (Figures
S5A and S5B). The decrease in the viral burden was similar in
the cosilenced and individually silenced groups (Figure S5C),
suggesting that mosPTP-1 is the dominant downstream
receptor for mosGCTL-1 in the process of WNV infection of
A. aegypti.
To better understand the association between mosGCTL-1,
mosPTP-1, and WNV in vivo, we examined the distribution of
mosPTP-1 in A. aegypti. mosPTP-1 was highly expressed in
various mosquito tissues but not induced by WNV infection
(Figure S4D). The salivary glands and hemolymph were sites of
abundant mosPTP-1 expression, while expression in midgut
Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 719
Figure 4. mosPTP-1 Captures mosGCTL-1 onto the Cell Surface
(A) RNAi efficiency for the mosPTP-1 gene at 6 days after dsRNA treatment. The amount of mosPTP-1 mRNA was determined by SYBR Green RT-QPCR and
normalized with A. aegypti actin. Data are represented as the mean ± standard error.
(B) Silencing mosPTP-1 decreased WNV infection. The viral burden was measured at day 6. WNV (10 M.I.D50) was injected into each mosquito. The viral load was
determined by Taqman RT-QPCR and normalized with A. aegypti actin. Statistical analysis was done with the Mann-Whitney test. The horizontal line depicts the
medians. The result shown is a combination of three independent experiments.
(C) Expression of recombinant mosPTP-1 and mosPTP-1-Ex in S2 cells. mosPTP-1 and mosPTP-1-Ex genes were isolated from complementary DNA (cDNA)
library of A. aegypti and expressed as recombinant proteins with an HA tag at the N terminus. The left panel is mosPTP-1, and the right panel is mosPTP-1-EX,
probed by anti-HA tag mAb in western blot. The control was the products from S2 cells transfected with mock DNA vector.
(D) mosGCTL-1 interacts with mosPTP-1-Ex peptide by coimmunoprecipitation. The protein complex was pulled down with a rabbit HA antibody and probed with
a V5-HRP mAb. The experiments were repeated three times.

720 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.

was comparatively lower (Figures S4E–S4G). We next generated
mosPTP-1 antibody in mice, which recognized native mosPTP-1
protein (Figure S5D) and the mosPTP-1-Ex expressed by S2
cells (Figure S5E). We then determined the relationship between
mosGCTL-1, mosPTP-1, and WNV in salivary glands by immu-
nofluorescence. mosPTP-1 was copiously expressed on the
cell surface of salivary glands (Figure 5D), thereby providing
additional data to complement and extend the initial QPCR
expression data (Figure S4E). Several regions in the salivary
glands with substantial mosPTP-1 also demonstrated staining
for mosGCTL-1 and WNV (Figure 5D).
mosGCTL-1 Antisera Interferes with WNV Infection
of Mosquitoes
Disruption of the transfer of WNV from the vertebrate to
arthropod host could theoretically diminish viral dissemination
in nature. We therefore investigated whether mosGCTL-1 anti-
sera reduces WNV infection in mosquitoes during the blood
meal. We generated mosGCTL-1 antisera in rabbits and showed
that mosGCTL-1 antisera interfered with mosGCTL-1 binding to
WNV E protein in vitro (Figure 6A). Then, we examined whether
mosGCTL-1 antisera influenced the ability of WNV to infect
mosquitoes during a blood meal. We mixed mosGCTL-1 antisera
and WNV with fresh whole blood and performed membrane
blood feeding with a Hemotek. Seven days later, mosquitoes
were sacrificed to determine the infectivity rate. mosGCTL-1
antisera efficiently blocked WNV infection of A. aegypti. The
number of infected mosquitoes was reduced in mosGCTL-1
antisera treated groups, compared to mock group, by QPCR
(Figure 6B) or a 50% tissue culture infective doses (TCID50) assay
(Figure 6C). Hence, a humoral response against mosGCTL-1 in
a vertebrate host may alter WNV infection of mosquitoes during
the feeding process. This hypothetically affords a strategy to
develop a transmission blocking vaccine to control WNV
dissemination in nature.
DISCUSSION
As mosquitoes are prominent vectors for flaviviruses, specific
interactions between the virus and arthropod likely enhance
pathogen survival. In the mammalian host, C-type lectins such
as DC-SIGN and the mannose receptor augment viral entry
into specific DCs and macrophages (Tassaneetrithep et al.,
2003; Davis et al., 2006; Miller et al., 2008). Our results show
that a secreted mosquito C-type lectin, mosGCTL-1, binds to
WNV in a calcium-dependent manner and enhances viral infec-
tion. A mosquito homolog of human CD45 (mosPTP-1) recruits
mosGCTL-1 to facilitate viral attachment to cells. Based on our
findings, we envision a model whereby WNV that is inoculated
into mosquitoes binds to secreted mosGCTL-1 in the hemo-
lymph, thereby forming a complex in the extracellular milieu
that has the ability to interact with the membrane protein,
mosPTP-1, to facilitate cellular entry. The virus rapidly replicates
in the mosquito thorax. This induces additional mosGCTL-1
expression, which accelerates formation of the mosGCTL-1/
WNV complex—enabling WNV to invade different mosquito
tissues and enhancing viral spread throughout the mosquito
body. This mechanism, which involves WNV associating with
mosGCTL-1 and then being captured by mosPTP-1 onto the
cell surface in mosquitoes, suggests that an extracellular soluble
protein is an important receptor for flavivirus in arthropods.
mosGCTL-1 shares homology with human MBL. In mammals,
MBL is a pattern recognition molecule that recognizes carbohy-
drate moieties on invading microbes (Neth et al., 2002). As exam-
ples, MBL interacts with HIV envelope protein (gp120) (Saifuddin
et al., 2000) and HBV surface antigen (HBsAg) (Chong et al., 2005)
and has a role in the opsonization of HIV (Ezekowitz et al., 1989).
In these processes, MBL associates with serine proteases,
MASPs, and activates the complement system (Neth et al.,
2002). Homologs of the proteins that associate with mammalian
MBL have not been found in A. aegypti (Table S3), suggesting that
the A. aegypti mosGCTL-1 may have different physiological func-
tions than mammalian MBL. Invertebrates lack antibody- and
interferon-based immune responses (Cheng et al., 2009). Since
lectin expression is significantly upregulated by microbial infec-
tion, these molecules are presumed to participate in nonself
recognition and pathogen resistance (Wilson et al., 1999; Tanji
et al., 2006). Indeed, recent studies have shown that a comple-
ment-like system exists in the hemolymph of Anopheles gambiae
and mediates parasite killing (Blandin et al., 2004; Povelones
et al., 2009). It is possible that mosGCTL-1 and other subtypes
in this family, similar to their mammalian homologs, may normally
recognize most pathogens and be involved in the arthropod
complement-like system. Nevertheless, our studies showed
that the expression of mosGCTL-1 is induced by WNV infection.
The induced mosGCTL-1 that then binds to virus amplifies WNV
infection. Overall, these suggest a critical role for mosGCTL-1 in
WNV infection of mosquitoes.
In the mammalian host, the association between MBL and the
CD45 external domain primarily occurs in immature T cells and
affect the development of thymocytes (Baldwin and Ostergaard,
2001). Mammalian CD45 is expressed on the hemopoietic-orig-
inated nucleated cells (Thomas, 1989); however, the mosquito
CD45 homolog, mosPTP-1, does not appear to be restricted to
particular cells. As a transmembrane protein, mosPTP-1 was
abundantly detected in the salivary glands and hemolymph of
mosquitoes. The pattern of mosPTP-1 expression correlated
with the distribution of WNV in A. aegypti. In our model, after

(E) mosPTP-1 captured mosGCTL-1 to the cell surface by flow cytometry. A stable cell line was generated to express mosPTP-1 in S2 cells. The purified
mosGCTL-1 was inoculated with mosPTP-1-expressing cells at 4 C. An empty DNA vector transfected stable S2 cell line was used as the control. The interaction
between mosPTP-1 and mosGCTL-1 was investigated by FACS. mosPTP-1 was stained by Alexa-488; mosGCTL-1 was stained by Phycoerythrin (PE). Three
independent experiments yielded similar results, and one representative study is shown in this figure.
(F) Confocal microscopy to examine for mosPTP-1 and mosGCTL-1. mosGCTL-1 was stained with Alexa-488 (green) and mosPTP-1 was identified with Alexa-
546 (red). Nuclei were stained by To-Pro-3 iodide (blue). The images were collected using a Zeiss LSM 510 Meta Confocal Microscope 633 objective lens.
The arrows represent the overlap between mosPTP-1 and mosGCTL-1.
See also Figure S4 and Table S3.

Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 721

Figure 5. The mosGCTL-1/mosPTP-1 Pathway in WNV Infection
(A) mosGCTL-1 associated with WNV E protein bound to mosPTP-1-Ex. The protein complex was pulled down with an HA antibody and probed with V5-HRP and
flavivirus E protein antibody. The experiment was repeated three times.
(B) mosPTP-1-expressing cells recruit WN virions in the presence of mosGCTL-1. mosGCTL-1 and WNV were added to cells and incubated at 4 C for 1 hr, for
membrane attachment. Cells were gently washed three times with cold PBS and were then moved to room temperature. At different time points—0, 15, and
60 min—cells were collected for total RNA isolation. Control cells were the empty DNA vector transfected stable S2 cell line. Viral attachment was determined
by SYBR Green RT-QPCR and normalized with Drosophila actin 5C (CG4027). Statistical analysis was done with ANOVA. Data are represented as the mean ±
standard error. The results are representative of three independent experiments.
(C) Silencing of mosPTP-1 impairs the function of mosGCTL-1. The mosPTP-1 gene was knocked down with dsRNA treatment. Then the mosGCTL-1/WNV
mixture was inoculated at 3 days after RNAi silencing. The virus burden was measured at 3 (i) and 6 (ii) days after the introduction of virus by Taqman RT-QPCR.
WNV (10 M.I.D50) and mosGCTL-1 (100 pg) were inoculated into each mosquito. Each dot represents the mRNA level in an individual mosquito. Statistical analysis
was done with the Mann-Whitney test. The horizontal line depicts the medians. The result shown is the combination of five independent experiments.
(D) Immunostaining of mosGCTL-1, mosPTP-1, and WNV in A. aegypti salivary glands. mosPTP-1 was stained with anti-mouse IgG Alexa-488 (green),
mosGCTL-1 was identified by anti-rabbit IgG Alexa-546 (red), and WNV E protein was probed with horse anti-E protein IgG and detected by anti-horse IgG
Alexa-633 (blue). The arrows show the sites of overlap between mosGCTL-1, mosPTP-1, and WNV at 6 days after infection. LL, lateral lobe; ML, median lobe
in female A.aegypti salivary glands. Images were examined with a Zeiss LSM 510 Meta Confocal Microscope 253 objective lens.
See also Figure S5.

722 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.


binding to mosGCTL-1, WNV binds to membrane bound
mosPTP-1. This implies that mosGCTL-1 and mosPTP-1 are
recruited as receptors to facilitate cellular invasion by WNV.
Mosquito control is a common strategy to influence
WNV numbers in nature (van der Meulen et al., 2005; Dauphin
and Zientara, 2007). The increase in viral spread and fatalities
over the last decade (Reisen and Brault, 2007; Lindsey et al.,
2009) (http://www.cdc.gov/ncidod/dvbid/westnile/index.htm)
suggests that additional strategies could assist in combating
WNV. For arthropod-borne microbes, vector ligands that interact
with pathogens are potential targets for interfering with
the successful acquisition of the microbe from the vertebrate
host. As an example, blocking the tick gut receptor for the
Lyme disease agent limits the colonization of ticks by Borrelia
burgdorferi (Pal et al., 2004). Our studies show that blocking
mosGCTL-1 within A. aegypti reduced the vector competence
for WNV and interrupted the infective cycle of WNV. These
results indicate that it is theoretically possible to develop a trans-
mission-blocking vaccine to interfere with the migration of WNV
from vertebrates to mosquito, thereby restricting viral dissemi-
nation in the environment.
In summary, we identified a lectin-based pathway to facilitate
flaviviral entry, in which mosGCTL-1 and mosPTP-1 are cascade
receptors in WNV infection of A. aegypti and C. quinquefascia-
tus. Characterization of mosquito ligands for WNV enhances
Figure 6. mosGCTL-1 Antiserum Interferes
with WNV Infection of Mosquitoes
(A) mosGCTL-1 antisera blocked the interaction
between mosGCTL-1 and WNV E protein.
mosGCTL-1 antisera or mock sera was diluted
1000-fold. The protein complex was pulled down
with a V5 mAb, and WNV E protein was detected
with an E protein antibody.
(B and C) mosGCTL-1 antisera interrupted WNV
infection during the blood meal. The antisera or
control sera were diluted 100- or 2000-fold with
fresh whole blood containing 5 3 106 pfu/ml
WNV. Membrane blood-feeding was then per-
formed with a Hemotek. Seven days later, mosqui-
toes were sacrificed to determine the infectivity
rate by Taqman RT-QPCR (B) and TCID50 (C).
Each group included 50 mosquitoes in the QPCR
assay and 32 mosquitoes in TCID50 assay. One
dot corresponds to a mosquito. n, the number of
mosquitoes in each group. The result is represen-
tative of three independent experiments.
our understanding of flavivirus-arthropod
interactions and may aid in the develop-
ment of strategies to target selected
points in the flaviviral life cycle and inter-
fere with these pathogens in nature.
EXPERIMENTAL PROCEDURES
Mosquitoes and Viruses
A. aegypti and C. quinquefasciatus mosquitoes
were maintained in a sugar solution at 27 C and
80% humidity according to standard rearing
procedures (Keene et al., 2004; Xi et al., 2008).
WNV strain 2471 and DENV-2 (DENV New Guinea C strain) were passaged
in mosquito C6/36 cells (Hanna et al., 2005; Lin et al., 2007; Krishnan et al.,
2008). The titer of WNV for cell culture was determined by a plaque formation
assay as described previously (Bai et al., 2007). The viruses for in vivo exper-
iments were titrated in mosquitoes through thoracic microinjection. The dose
of viruses was determined by 10-fold serial dilutions (i.e., 10 4, 10 5, 10 6, and
10 7) in PBS. The mosquitoes (12 in each group) were inoculated in the thorax
by microinjection with 300 nl of diluted virus. On day 6, the mosquitoes were
sacrificed, total RNA was isolated, and the viral load determined by RT-
QPCR (Figure S1). The 50% mosquito infective dose (M.I.D50) was estimated
by the Reed-Muench method (Pizzi, 1950).
Gene Silencing and Viral Challenge in Mosquitos
dsRNA synthesis was performed as described previously (Brackney et al.,
2008). The primers are shown in Table S4. For silencing the target genes, adult
female mosquitoes were kept on ice for 15 min and then transferred to a cold
tray to receive a systemic injection of dsRNA into the hemocoele. Two micro-
grams of dsRNA/300 nl in PBS was microinjected into the thorax of each
mosquito. After a 3 day recovery period, the mosquitoes were microinjected
with either WNV 10 M.I.D50/ 300 nl for functional studies or 1000 M.I.D50/
300 nl for expression profile assays.
Purification of mosGCTL-1 with the Drosophila Expression System
mosGCTL-1 was amplified by RT-PCR from adult female A. aegypti. The
primers are shown in Table S4. The PCR product was subcloned into the
pMT/BiP/V5-His A vector (Invitrogen, Carlsbad, CA) and transfected into
Drosophila S2 cells in combination with the hygromycin selection vector
pCo-Hygro for stable transfection. The cells were selected through the
use of 300 mg/ml Hygromycin-B (Invitrogen) for 4 weeks. The resistant cells
Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 723
were grown in spinner flasks, switched to Express Five serum-free medium
(GIBCO, Invitrogen) for 3 days, and induced with copper sulfate at a final
concentration of 500 mM for 4 days. The culture medium was cleared by
centrifugation at 1000 3 g for 5 min and collected for protein purification
with the Talon metal affinity resin (Clontech, Mountain View, CA). The
protein was eluted with 150 mM imidazole, extensively dialyzed against
PBS (pH 7.8), and concentrated by centrifugal filtration through a 5 kDa
filter (Millipore, Bedford, MA). The protein purity was checked by SDS-
PAGE, and immunoblots were performed with an anti-V5-HRP mouse
mAb (Invitrogen).
Isolation and Image of Mosquito Tissues
Salivary glands were dissected as the previously described (Coleman et al.,
2007). Tissues were isolated, placed on sialylated slides (PGC Scientific,
Gaithersburg, MD), washed in PBS, and fixed in 4% PFA at 37 C for 1 hr.
For hemolymph isolation, the mosquitoes were anesthetized on ice tray, and
then the proboscis was removed with forceps. Hemolymph was expelled in
a droplet using the tip of the proboscis upon pressure to the thorax. Only clear
droplets were collected, to avoid contamination by the fat body (Han et al.,
1999). The liquid was allowed to be dried on sialylated slides and fixed in
4% PFA. Samples were blocked in PBS with 1% BSA and 0.1% Triton
X-100 at room temperature for 2 hr before antibody incubations. After staining
by primary and secondary antibodies, slides were imaged with the Multi-Track
mode of a Zeiss LSM 510 Meta Confocal Microscope.
Membrane Blood Feeding
Fresh whole blood was obtained from C57BL/6J mice, placed in tubes with
anticoagulant, and centrifuged at 3000 rpm for 10 min to separate the serum
and cells. The serum was collected and heat inactivated at 57 C for 1 hr.
The blood cells were washed in PBS three times to remove the anticoagulant.
The cells were resuspended with serum. Five million plaque forming units per
milliliter WNV and antisera (or preimmune sera) were added to the treated
blood. About 1 ml of the virus/sera/blood mixture was used on each Hemotek
feeder. The feeder was placed on the mosquito container and the mosquitoes
were allowed to feed for 1 hr. The container was transferred to 4 C for 20 min to
anesthetize the mosquitoes. Carefully selected fed mosquitoes were placed
into new containers and reared for 7 days. The mosquitoes were sacrificed;
the whole mosquitoes were homogenized in RLT buffer (RNeasy Mini kit,
QIAGEN) to measure the viral load by RT-QPCR, or grinded in PBS for virologic
assays.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, five
figures, and four tables and can be found with this article online at doi:10.
1016/j.cell.2010.07.038.
ACKNOWLEDGMENTS
This work was supported by National Institutes of Health grants AI 50031 and
AI 070343. We thank Yue Zhang, Lei Liu, Lili Zhang, and Sukanya Narasimhan
for help with the biochemical studies, Debbie Beck for technical assistance,
and Michel Ledizet for providing the purified WNV E protein and the flaviviral
E protein antibodies. G.C. is a James Hudson Brown-Alexander Brown
Coxe Postdoctoral Fellow at the Yale School of Medicine. P.W. is supported
by a Career Development Award from Northeast Biodefense Center (U54-
AI057158-Lipkin). E.F. is an investigator with the Howard Hughes Medical
Institute. G.C. and E.F. designed the experiments and wrote the manuscript;
G.C. performed the majority of the experiments and analyzed data; J.C. assis-
ted the mosquito techniques; J.F.A. provided the A. aegypti and C. quinque-
fasciatus mosquitoes and helped with the mosquito feeding; J.D. and F.Q.
assisted in FACS and confocal microscopy; and P.W., M.N.K. and J.D.
provided suggestions for the project. All authors reviewed, critiqued, and
provided comments on the text.
724 Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc.
Received: September 12, 2009
Revised: April 5, 2010
Accepted: July 27, 2010
Published online: August 26, 2010
REFERENCES
Alonso, A., Sasin, J., Bottini, N., Friedberg, I., Friedberg, I., Osterman, A., God-
zik, A., Hunter, T., Dixon, J., and Mustelin, T. (2004). Protein tyrosine phospha-
tases in the human genome. Cell 117, 699–711.
Arnold, J.N., Dwek, R.A., Rudd, P.M., and Sim, R.B. (2006). Mannan binding
lectin and its interaction with immunoglobulins in health and in disease. Immu-
nol. Lett. 106, 103–110.
Bai, F.W., Town, T., Pradhan, D., Cox, J., Ashish, Ledizet, M., Anderson, J.F.,
Flavell, R.A., Krueger, J.K., Koski, R.A., and Fikrig, E. (2007). Antiviral peptides
targeting the west nile virus envelope protein. J. Virol. 81, 2047–2055.
Baldwin, T.A., and Ostergaard, H.L. (2001). Developmentally regulated
changes in glucosidase II association with, and carbohydrate content of, the
protein tyrosine phosphatase CD45. J. Immunol. 167, 3829–3835.
Blandin, S., Shiao, S.H., Moita, L.F., Janse, C.J., Waters, A.P., Kafatos, F.C.,
and Levashina, E.A. (2004). Complement-like protein TEP1 is a determinant
of vectorial capacity in the malaria vector Anopheles gambiae. Cell 116,
661–670.
Brackney, D.E., Foy, B.D., and Olson, K.E. (2008). The effects of midgut serine
proteases on dengue virus type 2 infectivity of Aedes aegypti. Am. J. Trop.
Med. Hyg. 79, 267–274.
Byth, K.F., Conroy, L.A., Howlett, S., Smith, A.J., May, J., Alexander, D.R., and
Holmes, N. (1996). CD45-null transgenic mice reveal a positive regulatory role
for CD45 in early thymocyte development, in the selection of CD4+CD8+
thymocytes, and B cell maturation. J. Exp. Med. 183, 1707–1718.
Chen, S.T., Lin, Y.L., Huang, M.T., Wu, M.F., Cheng, S.C., Lei, H.Y., Lee, C.K.,
Chiou, T.W., Wong, C.H., and Hsieh, S.L. (2008). CLEC5A is critical for
dengue-virus-induced lethal disease. Nature 453, 672–676.
Cheng, G., Zhao, X., Li, Z., Liu, X., Yan, W., Zhang, X., Zhong, Y., and Zheng, Z.
(2009). Identification of a putative invertebrate helical cytokine similar to the
ciliary neurotrophic factor/leukemia inhibitory factor family by PSI-BLAST-
based approach. J. Interferon Cytokine Res. 29, 461–468.
Chong, W.P., To, Y.F., Ip, W.K., Yuen, M.F., Poon, T.P., Wong, W.H., Lai, C.L.,
and Lau, Y.L. (2005). Mannose-binding lectin in chronic hepatitis B virus infec-
tion. Hepatology 42, 1037–1045.
Coleman, J., Juhn, J., and James, A.A.A. (2007). Dissection of midgut and
salivary glands from Ae. aegypti mosquitoes. J. Vis. Exp. 5, 228.
Dauphin, G., and Zientara, S. (2007). West Nile virus: recent trends in diagnosis
and vaccine development. Vaccine 25, 5563–5576.
Davis, C.W., Nguyen, H.Y., Hanna, S.L., Sánchez, M.D., Doms, R.W., and Pier-
son, T.C. (2006). West Nile virus discriminates between DC-SIGN and
DC-SIGNR for cellular attachment and infection. J. Virol. 80, 1290–1301.
Ezekowitz, R.A., Kuhlman, M., Groopman, J.E., and Byrn, R.A. (1989).
A human serum mannose-binding protein inhibits in vitro infection by the
human immunodeficiency virus. J. Exp. Med. 169, 185–196.
Geijtenbeek, T.B., Torensma, R., van Vliet, S.J., van Duijnhoven, G.C., Adema,
G.J., van Kooyk, Y., and Figdor, C.G. (2000). Identification of DC-SIGN, a novel
dendritic cell-specific ICAM-3 receptor that supports primary immune
responses. Cell 100, 575–585.
Gould, E.A., and Solomon, T. (2008). Pathogenic flaviviruses. Lancet 371,
500–509.
Gubler, D.J. (1998). Dengue and dengue hemorrhagic fever. Clin. Microbiol.
Rev. 11, 480–496.
Halstead, S.B. (2008). Dengue virus-mosquito interactions. Annu. Rev. Ento-
mol. 53, 273–291.
Han, Y.S., Chun, J., Schwartz, A., Nelson, S., and Paskewitz, S.M. (1999).
Induction of mosquito hemolymph proteins in response to immune challenge
and wounding. Dev. Comp. Immunol. 23, 553–562.
Hanna, S.L., Pierson, T.C., Sanchez, M.D., Ahmed, A.A., Murtadha, M.M., and
Doms, R.W. (2005). N-linked glycosylation of west nile virus envelope proteins
influences particle assembly and infectivity. J. Virol. 79, 13262–13274.
Hayes, E.B., Komar, N., Nasci, R.S., Montgomery, S.P., O’Leary, D.R., and
Campbell, G.L. (2005). Epidemiology and transmission dynamics of West
Nile virus disease. Emerg. Infect. Dis. 11, 1167–1173.
Hubálek, Z., and Halouzka, J. (1999). West Nile fever—a reemerging mosquito-
borne viral disease in Europe. Emerg. Infect. Dis. 5, 643–650.
Keene, K.M., Foy, B.D., Sanchez-Vargas, I., Beaty, B.J., Blair, C.D., and Olson,
K.E. (2004). RNA interference acts as a natural antiviral response to O’nyong-
nyong virus (Alphavirus; Togaviridae) infection of Anopheles gambiae. Proc.
Natl. Acad. Sci. USA 101, 17240–17245.
Klimstra, W.B., Nangle, E.M., Smith, M.S., Yurochko, A.D., and Ryman, K.D.
(2003). DC-SIGN and L-SIGN can act as attachment receptors for alphaviruses
and distinguish between mosquito cell- and mammalian cell-derived viruses.
J. Virol. 77, 12022–12032.
Krishnan, M.N., Sukumaran, B., Pal, U., Agaisse, H., Murray, J.L., Hodge,
T.W., and Fikrig, E. (2007). Rab 5 is required for the cellular entry of dengue
and West Nile viruses. J. Virol. 81, 4881–4885.
Krishnan, M.N., Ng, A., Sukumaran, B., Gilfoy, F.D., Uchil, P.D., Sultana, H.,
Brass, A.L., Adametz, R., Tsui, M., Qian, F., et al. (2008). RNA interference
screen for human genes associated with West Nile virus infection. Nature
455, 242–245.
Leis, A.A., Stokic, D.S., Polk, J.L., Dostrow, V., and Winkelmann, M. (2002).
A poliomyelitis-like syndrome from West Nile virus infection. N. Engl. J. Med.
347, 1279–1280.
Lin, C.C., Yang, C.F., Tu, C.H., Huang, C.G., Shih, Y.T., Chuang, C.K., and
Chen, W.J. (2007). A novel tetraspanin C189 upregulated in C6/36 mosquito
cells following dengue 2 virus infection. Virus Res. 124, 176–183.
Lindsey, N.P., Hayes, E.B., Staples, J.E., and Fischer, M. (2009). West Nile
virus disease in children, United States, 1999-2007. Pediatrics 123, e1084–
e1089.
Miller, J.L., de Wet, B.J., deWet, B.J., Martinez-Pomares, L., Radcliffe, C.M.,
Dwek, R.A., Rudd, P.M., and Gordon, S. (2008). The mannose receptor medi-
ates dengue virus infection of macrophages. PLoS Pathog. 4, e17.
Monath, T.P., and Heinz, F.X. (1996). Flaviviruses (Philadelphia: Lippincott-
Raven Publishers).
Mustelin, T., Vang, T., and Bottini, N. (2005). Protein tyrosine phosphatases
and the immune response. Nat. Rev. Immunol. 5, 43–57.
Nene, V., Wortman, J.R., Lawson, D., Haas, B., Kodira, C., Tu, Z.J., Loftus, B.,
Xi, Z., Megy, K., Grabherr, M., et al. (2007). Genome sequence of Aedes
aegypti, a major arbovirus vector. Science 316, 1718–1723.
Neth, O., Jack, D.L., Johnson, M., Klein, N.J., and Turner, M.W. (2002).
Enhancement of complement activation and opsonophagocytosis by
complexes of mannose-binding lectin with mannose-binding lectin-associ-
ated serine protease after binding to Staphylococcus aureus. J. Immunol.
169, 4430–4436.
Nir, Y., Goldwasser, R., Lasowski, Y., and Margalit, J. (1968). Isolation of West
Nile virus strains from mosquitoes in Israel. Am. J. Epidemiol. 87, 496–501.
Pal, U., Li, X., Wang, T., Montgomery, R.R., Ramamoorthi, N., Desilva, A.M.,
Bao, F., Yang, X., Pypaert, M., Pradhan, D., et al. (2004). TROSPA, an Ixodes
scapularis receptor for Borrelia burgdorferi. Cell 119, 457–468.
Pizzi, M. (1950). Sampling variation of the fifty percent end-point, determined
by the Reed-Muench (Behrens) method. Hum. Biol. 22, 151–190.
Povelones, M., Waterhouse, R.M., Kafatos, F.C., and Christophides, G.K.
(2009). Leucine-rich repeat protein complex activates mosquito complement
in defense against Plasmodium parasites. Science 324, 258–261.
Reisen, W., and Brault, A.C. (2007). West Nile virus in North America: perspec-
tives on epidemiology and intervention. Pest Manag. Sci. 63, 641–646.
Robinson, M.J., Sancho, D., Slack, E.C., LeibundGut-Landmann, S., and Reis
e Sousa, C. (2006). Myeloid C-type lectins in innate immunity. Nat. Immunol. 7,
1258–1265.
Saifuddin, M., Hart, M.L., Gewurz, H., Zhang, Y., and Spear, G.T. (2000). Inter-
action of mannose-binding lectin with primary isolates of human immunodefi-
ciency virus type 1. J. Gen. Virol. 81, 949–955.
Sessions, O.M., Barrows, N.J., Souza-Neto, J.A., Robinson, T.J., Hershey,
C.L., Rodgers, M.A., Ramirez, J.L., Dimopoulos, G., Yang, P.L., Pearson,
J.L., and Garcia-Blanco, M.A. (2009). Discovery of insect and human dengue
virus host factors. Nature 458, 1047–1050.
Soilleux, E.J., Morris, L.S., Leslie, G., Chehimi, J., Luo, Q., Levroney, E., Trow-
sdale, J., Montaner, L.J., Doms, R.W., Weissman, D., et al. (2002). Constitutive
and induced expression of DC-SIGN on dendritic cell and macrophage
subpopulations in situ and in vitro. J. Leukoc. Biol. 71, 445–457.
Tanji, T., Ohashi-Kobayashi, A., and Natori, S. (2006). Participation of a galac-
tose-specific C-type lectin in Drosophila immunity. Biochem. J. 396, 127–138.
Tassaneetrithep, B., Burgess, T.H., Granelli-Piperno, A., Trumpfheller, C.,
Finke, J., Sun, W., Eller, M.A., Pattanapanyasat, K., Sarasombath, S., Birx,
D.L., et al. (2003). DC-SIGN (CD209) mediates dengue virus infection of human
dendritic cells. J. Exp. Med. 197, 823–829.
Thomas, M.L. (1989). The leukocyte common antigen family. Annu. Rev.
Immunol. 7, 339–369.
Trowbridge, I.S., and Thomas, M.L. (1994). CD45: an emerging role as a protein
tyrosine phosphatase required for lymphocyte activation and development.
Annu. Rev. Immunol. 12, 85–116.
Turell, M.J., O’Guinn, M., and Oliver, J. (2000). Potential for New York mosqui-
toes to transmit West Nile virus. Am. J. Trop. Med. Hyg. 62, 413–414.
van der Meulen, K.M., Pensaert, M.B., and Nauwynck, H.J. (2005). West Nile
virus in the vertebrate world. Arch. Virol. 150, 637–657.
Vanlandingham, D.L., McGee, C.E., Klinger, K.A., Vessey, N., Fredregillo, C.,
and Higgs, S. (2007). Relative susceptibilties of South Texas mosquitoes to
infection with West Nile virus. Am. J. Trop. Med. Hyg. 77, 925–928.
Wilson, R., Chen, C.W., and Ratcliffe, N.A. (1999). Innate immunity in insects:
the role of multiple, endogenous serum lectins in the recognition of foreign
invaders in the cockroach, Blaberus discoidalis. J. Immunol. 162, 1590–1596.
Xi, Z., Ramirez, J.L., and Dimopoulos, G. (2008). The Aedes aegypti toll
pathway controls dengue virus infection. PLoS Pathog. 4, e1000098.
Zelensky, A.N., and Gready, J.E. (2005). The C-type lectin-like domain super-
family. FEBS J. 272, 6179–6217.
Cell 142, 714–725, September 3, 2010 ª2010 Elsevier Inc. 725
The ATAC Acetyltransferase Complex
Coordinates MAP Kinases
to Regulate JNK Target Genes
Tamaki Suganuma,1 Arcady Mushegian,1,2 Selene K. Swanson,1,3 Susan M. Abmayr,1,3 Laurence Florens,1
Michael P. Washburn,1,4 and Jerry L. Workman1,*
1Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110, USA
2Department of Microbiology, Molecular Genetics, and Immunology
3Department of Anatomy and Cell Biology
4Department of Pathology and Laboratory Medicine
University of Kansas Medical Center, Kansas City, KS 66160, USA
*Correspondence: jlw@stowers.org
DOI 10.1016/j.cell.2010.07.045
SUMMARY
In response to extracellular cues, signal transduction
activates downstream transcription factors like
c-Jun to induce expression of target genes. We
demonstrate that the ATAC (Ada two A containing)
histone acetyltransferase (HAT) complex serves as
a transcriptional cofactor for c-Jun at the Jun
N-terminal kinase (JNK) target genes Jra and chick-
adee. ATAC subunits are required for c-Jun occu-
pancy of these genes and for H4K16 acetylation at
the Jra enhancer, promoter, and transcribed se-
quences. Under conditions of osmotic stress, ATAC
colocalizes with c-Jun, recruits the upstream kinases
Misshapen, MKK4, and JNK, and suppresses further
activation of JNK. Relocalization of these MAPKs
and suppression of JNK activation by ATAC are
dependent on the CG10238 subunit of ATAC. Thus,
ATAC governs the transcriptional response to MAP
kinase signaling by serving as both a coactivator
of transcription and as a suppressor of upstream
signaling.
INTRODUCTION
Histone acetyltransferase complexes have been isolated from
multiple organisms and shown to be involved in nuclear events
that relate to chromatin biology (Kimura et al., 2005; Lee and
Workman, 2007). The Drosophila ATAC (Ada two A containing)
complex consists of 13 proteins and includes two distinct
histone acetyltransferases, Gcn5/KAT2 and Atac2/KAT14 (Fig-
ure 1A). Whereas Gcn5/KAT2 preferentially acetylates histones
H3K9 and H3K14, Atac2/KAT14 acetylates H4K16 (Ciurciu
et al., 2006; Guelman et al., 2006; Suganuma et al., 2008). The
Gcn5/KAT2, Ada3, and CG30390 (Sgf29 in yeast) subunits of
ATAC are shared with SAGA (Spt-Ada-Gcn5 acetyltransferase)
(Suganuma et al., 2008), an important transcriptional coactivator
726 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.
complex (Lee and Workman, 2007). However, it was unknown
whether ATAC functions as a transcription cofactor like SAGA.
Mitogen-activated protein kinases (MAPKs) and their target
kinases in the pathway lead to phosphorylation of target
transcription factors (Edmunds and Mahadevan, 2004; Thomson
et al., 1999). Exposure of cells to increases in extracellular osmo-
lality results in rapid activation of stress-activated MAPKs
(SAPKs), including c-Jun-NH2-terminal kinase (JNK/basket in
Drosophila) and p38 MAPKs (de Nadal et al., 2002; Edmunds
and Mahadevan, 2004; Kyriakis and Avruch, 2001). Osmotic
stress causes activation of JNK by phosphorylation which, in
turn, phosphorylates c-Jun and enhances its transcriptional
activity (Kayali et al., 2000; Thomson et al., 1999). MAPKs are
more intimately involved in the regulation of downstream target
genes than just phosphorylation of target transcription factors
(Dioum et al., 2009). For example, the yeast SAPK Hog1 is
recruited to target genes in chromatin and interacts with tran-
scription factors, cofactors, and RNA polymerase II (Alepuz
et al., 2001, 2003; de Nadal et al., 2004; Mas et al., 2009; Pokho-
lok et al., 2006; Proft et al., 2006; Zapater et al., 2007). Recently,
the ERK2 MAPK was found to bind with sequence specificity
to DNA and act as a transcriptional repressor of interferon-g-
induced genes (Hu et al., 2009). Moreover, multiple MAPKs,
including ERK1/2, p38, and JNK, bind to and regulate transcrip-
tion of the insulin gene (Lawrence et al., 2009). Thus, MAPKs play
important chromatin-associated functions in the regulation of
gene expression.
MBIP (MAPK upstream kinase [MUK] binding inhibitory
protein), a component of the human ATAC complex, was
identified as a MUK binding partner in a yeast two-hybrid screen
(Fukuyama et al., 2000). The Drosophila ATAC component
CG10238 encodes the Drosophila homolog of MoaE, a subunit
of molybdopterin (MPT) synthase, an essential enzyme involved
in the synthesis of the molybdenum cofactor (Moco). Moco binds
molybdenum in the active site of molybdenum enzymes, which
catalyze redox reactions as part of ancient and conserved
biosynthetic pathways (Iyer et al., 2006; Leimkuhler et al.,
2003; Rudolph et al., 2001; Schwarz and Mendel, 2006).
However, CG10238 also contains C-terminal sequences that
A Figure 1. Peptides from Proteins Related to
the JNK Pathway Were Detected in Purifica-
Purifications
tions of ATAC Subunits, and CG10238
CG10238 Atac2 CHRAC14 Ada2a WDS D12 Control
Proteins PEP (SC%) PEP (SC%) PEP (SC%) PEP (SC%) PEP (SC%) PEP (SC%) PEP (SC%) Inhibits JNK Activation by Osmotic Stress
ATAC CG10238 29 (72.21) 5 (21.25) 2 (9.81) 7 (25.61) 6 (30.25) 16 (53.13) 0 (0)
Gcn5/KAT 35 (46.69) 4 (6.15) 4 (6.4) 5 (8.24) 4 (7.38) 33 (33.95) 0 (0) in Drosophila Schneider’s S2 Cells
Atac1 14 (53.37) 2 (5.34) 2 (7.58) 6 (16.3) 0 (0) 9 (20.71) 0 (0) (A) Peptides of MAPK pathway proteins in addition
Ada3 21 (58.89) 1 (1.98) 1 (1.98) 2 (7.19) 1 (3.24) 10 (16.37) 0 (0)
Ada2a 20 (48.2) 2 (4.93) 2 (4.93) 9 (18.6) 0 (0) 18 (36.43) 0 (0) to ATAC subunits detected by MudPIT analysis for
HCF 66 (68.87) 11 (11.93) 16 (17.6) 17 (19.87) 19 (18.07) 54 (49.4) 1 (2.6)
D12 37 (45.86) 7 (9.08) 11 (19.81) 10 (13.62) 5 (9.7) 31 (38.29) 0 (0) affinity purifications of ATAC subunits with FLAG-
CG30390 14 (56.60) 1 (5.88) 0 (0) 6 (26.64) 0 (0) 13 (55.36) 0 (0)
Atac2/KAT14 26 (47.16) 10 (22.09) 7 (16.28) 5 (9.95) 5 (13.18) 19 (39.41) 0 (0)
HA-tagged CG10238, Atac2/KAT14, Ada2a,
CHARC14 5 (62.5) 1 (10.16) 1 (10.16) 3 (27.34) 1 (14.84) 7 (71.88) 0 (0) CHRAC14, WDS, and D12. Parental S2 cells
Atac3 14 (40.88) 1 (3.89) 0 (0) 3 (7.61) 0 (0) 9 (20.71) 0 (0)
NC2 beta 6 (60.66) 2 (14.21) 1 (7.1) 0 (0) 2 (14.21) 5 (27.32) 0 (0) were used in mock purification as a control. PEP,
WDS 18 (66.48) 7 (25.76) 5 (18.28) 8 (26.59) 20 (63.43) 13 (53.19) 1 (3.6)
Proteins Jra 2 (10.03) 0 (0) 3 (19.03) 0 (0) 0 (0) 2 (13.84) 0 (0)
peptide count; SC%, sequence coverage (%).
related to Misshapen 16 (21.34) 0 (0) 5 (6.12) 0 (0) 0 (0) 5 (5.85) 0 (0) (B) Homology domains of CG10238 and experi-
JNK pathway MKK4 4 (15.8) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
slik 5 (4.82) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) mental design of truncation mutants. The diagram
MEK3/MKK3 2 (8.08) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
Chickadee 2 (20.63) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
shows the domains of CG10238 that are homolo-
gous to human MOCS2B (hMoaE) and to human
MBIP, which are separate proteins. The full-length
B (FL), N-terminal that contains only MoaE domain
Drosophila melanogaster
(N), and C-terminal that contains only MBIP protein
1 137 171 367 a.a. sequences (C) were stably expressed in S2 cells
MoaE MBIP
(Figure 2; see Figure S1B).
Homo sapiens
1 188 a.a.
MoaE
(C) CG10238 inhibited the activation of JNK stimu-
Mocs2B
1 344 a.a. lated by osmotic stress in S2 cells. Stably ex-
MBIP pressed FLAG-HA-tagged CG10238 was induced
Design of deletion mutants of CG10238 by 0, 50, or 150 mM CuSO4 in S2 cells. Cells were
FL
1 367a.a.
stimulated with osmotic stress by addition of
1 137a.a.
-HF
500 mM sorbitol for 12 min before harvesting cells
N
137a.a. 367a.a. (C, lanes 6–10; Figure 2C, lanes 5–14; Figure 2D,
C -HF
lanes 1–12; see Figures S2A–S2C). The nuclear
extracts were examined by western blot probing
with anti-HA, anti-Active JNK, and anti-JNK
C (C; Figure 2C). Parental S2 cells of stable cell lines
CuSO4 (x10µM) - 15 - 5 15 - 15 0 5 15
cultured with 0 or 150 mM CuSO4 were used as
HA-CG10238 - - - - controls (C, lanes 1, 2, 6, and 7; see Figure S3).
Parental cells + + - - - + + - - - See also Figures S1, S2, and S3.
Sorbitol - - - - - + + + + +
1 2 3 4 5 6 7 8 9 10

anti-HA 53.9

53.9
anti-ActiveJNK

53.9
anti-JNK

are homologous to human MBIP. This observation, coupled with genetically with the JNK pathway (Jasper et al., 2001; Morrison
the presence of MAPK signaling proteins in ATAC purifications, et al., 2000).
led us to ask whether CG10238 and ATAC play a role in MAPK
signaling. The ATAC Subunit CG10238 Functions as an Inhibitor
of JNK Activation in Response to Osmotic Stress
RESULTS We tested whether CG10238 plays a role in MAPK signaling like
MBIP (Fukuyama et al., 2000). MAPK cascades can be activated
ATAC Interacts with Proteins Related to the MAPK by osmotic stress, resulting in activation of JNK by phosphoryla-
Pathway tion (Kayali et al., 2000; Yang et al., 2003). We therefore exam-
Affinity purifications of the ATAC complex revealed proteins that ined whether expression of CG10238 affected the activation of
are part of the MAPK signaling pathway (Figure 1A; see Fig- JNK under conditions of osmotic stress (Kayali et al., 2000).
ure S1A available online). Peptides from these proteins were We first titrated the cellular response to osmotic stress
found in purifications via the CG10238, CHRAC14, and D12 stimulated by sorbitol in S2 cells by western blot (Figures S2A
ATAC subunits. Peptides were identified from the transcription and S2B). Maximum activation of JNK was observed between
factor Jra (Jun-related antigen), the Drosophila homolog of 7 and 30 min after treatment with a minimum concentration of
c-Jun, and Misshapen (MSN), the Drosophila homolog of 500 mM sorbitol (Figure S2A, lane 5; Figure S2B, lanes 2–4).
Ste-20 kinase (Figure 1A) (Morrison et al., 2000; Su et al., 1998; We then examined the level of JNK activation in cells expressing
Treisman et al., 1997). We also found peptides from other STE CG10238 in the presence or absence of osmotic stress induced
kinases, such as MKK4, slik, and MEK3/MKK3, as well as by 500 mM sorbitol for 12 min. Expression of CG10238 was
peptides from Chickadee, which has been shown to interact inducible by CuSO4, and parental S2 cells were treated with

Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc. 727

A B Figure 2. MBIP-Related Sequences Incor-
Input nuclear extracts Elution
porate CG10238 into ATAC; However, the
- F N C F N C - (KDa)
CG10238-Domain N C MoaE Domain Inhibits JNK Activation
ATAC subunits PEP( SC%) PEP( SC%) 54.8
(A) FLAG-HA-tagged N terminus that contains only
CG10238-PA 7 (25.6) 17 (50.95)
Gcn5/KAT 0 (0) 8 (12.67)
the MoaE domain (N) and C terminus that contains
37.7
anti-HA only MBIP protein sequences (C) were stably ex-
Atac1 0 (0) 3 (9.55)
Ada3 0 (0) 3 (8.99) 29.4 pressed in S2 cells purified by FLAG affinity beads.
Ada2a 0 (0) 5 (13.09) The copurified proteins were examined by MudPIT
HCF 5 (1.47) 24 (26.8) 17.5
D12 0 (0) 15 (21.26) 97.3 analysis (see Figures S1A and S1B). The observed
anti-Atac1
CG30390 0 (0) 3 (13.15) PEP and SC (%) of each sample are indicated as in
Atac2/KAT14 0 (0) 5 (10.98) 118.2
anti-Atac2 Figure 1.
CHRAC14 0 (0) 0 (0)
anti-Atac3
97.3 (B) The copurified proteins of CG10238-full-length
Atac3 0 (0) 2 (6.55)
NC2beta 0 (0) 3 (14.21) 206.3 (F), -N, or -C (Figure 1B; Figure S1) were analyzed
anti-D12
118.2
WDS 1 (3.88) 6 (27.42) 118.2 by western blotting with antibodies against HA tag
anti-Gcn5
and ATAC subunits. As a control, untransfected
anti-Ada2a 54.8
parental S2 cells were mock purified and analyzed
anti-Hcf
97.3
by western blotting (-).
29.4 (C and D) The MoaE domain inhibits JNK activa-
anti-NC2 beta tion. The ability to inhibit JNK activation was
54.8 compared between FLAG-HA-tagged CG10238
anti-WDS
(FL), CG10238-N (MoaE domain), and CG10238-
37.6
anti-CG30390 C (MBIP domain) after the expression levels were
normalized (see also Figure S1B). Parental S2 cells
cultured with 150 mM CuSO4 were used as a posi-
tive control (C, lanes 1 and 5; D, lanes 1, 5, and 9).
C D Osmotic stress was induced as in Figure 1C
HA-C - - - + - - - - - - - (see also Figures S2A and S2B). For sorbitol-stim-
HA-N - - + - - - - - - - -
Active JNK ulated samples, the intensity of each band for
HA-FL - + - - - - - - - - -
Parental cells + - - - + - - - - - - - - - Active JNK in western blotting was quantified,
160
Sorbitol - - - - + + + + + + + + + +
1 2 3 4 5 6 7 8 9 10 11 12 13 14 140
and each sample from CG10238 (FL), CG10238-N,
120
and CG10238-C was individually compared with
100
the intensity of parental cells (P), and the ratios
Ratio (%)

80
(%) are shown in (D). The average of four indepen-
anti-HA
60
dent experiments is graphed. Error bars represent
40 standard deviation.
20 See also Figures S1 and S2.
0
anti-ActiveJNK 1 2 3 4 5 6 7 8 9 10 11 12

P P P
anti-JNK HA-FL HA-N HA-C

the same amount of CuSO4 and sorbitol as controls. We first examined the two parts of CG10238 to determine which might
confirmed by western blot that the expression level of HA- incorporate it into the ATAC complex. We generated S2 cell lines
tagged CG10238 in the stable cell line without induction was that stably expressed tagged truncated forms of CG10238 that
similar to that of endogenous CG10238 in the parental cells included only the N-terminal MoaE domain or the C-terminal
(Figure S2C). Active JNK was not detected in the absence of MBIP domain (Figure 1B). These tagged proteins were affinity
sorbitol treatment, or upon induction of CG10238 by addition purified from the stable cell lines after their expression levels
of CuSO4 (Figure 1C, lanes 1 and 2). In the presence of sorbitol, were normalized (Figure S1B). Proteins that copurified with
JNK was activated in the parental cells (Figure 1C, lanes 6 and 7). each domain of CG10238 were identified by multidimensional
However, induction of CG10238 expression by CuSO4 inhibited protein identification technology (MudPIT) analysis and
the activation of JNK in a dose-dependent manner (Figure 1C, confirmed by western blots (Figures 2A and 2B; Figure S1A).
lanes 8–10). Thus, CG10238 inhibits JNK activation by osmotic All ATAC subunits except CHRAC14 associated with the MBIP
stress in vivo. Because JNK is also activated by ultraviolet light domain. A few peptides from two of the ATAC subunits were de-
(UV), we examined whether the inhibitory activity of CG10238 tected by purification with the MoaE domain and three subunits
extended to this JNK activation mechanism (Angel et al., 1988; were detected weakly by westerns (Figure 2B). Thus, the MBIP
Devary et al., 1991; Rozek and Pfeifer, 1995). Expression of domain of CG10238 is responsible for incorporating the protein
CG10238 also inhibited activation of JNK by UV (Figure S3). into the ATAC complex.

The MBIP Domain of CG10238 Is Required The MoaE Domain of CG10238 Inhibits JNK Activation
for Incorporation into the ATAC Complex Analysis of the domain structure of MoaE and MBIP orthologs in
The finding that CG10238 prevented JNK activation when a variety of organisms indicates that many insects encode these
expressed in vivo led us to ask whether CG10238 served this two domains as a translational fusion like that of Drosophila
function in isolation or as part of the ATAC complex. We first CG10238. By contrast, orthologs in nematodes, plants, fungi,

728 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.

 B S2 Cells Figure 3. ATAC Inhibits JNK Signaling
A S2 Cells ta ta
nt nt c2 c2 t t be be S2 cells were transfected with dsRNA-Atac2, dsRNA-NC2
on Con NC2 NC2
Co A-Co -Ata A-Ata -C
A- A -
NA RNA RNA RNA
- -
b, or dsRNA-Control (dsRNA-Cont) (A and B) (see
RN sRN sRN sRN ds
R
ds ds ds
Transfected with ds d d d Transfected with
- + - + (KDa) Sorbitol - + - + Figure S4). 293T cells were transfected with dsRNA-
Sorbitol (KDa)
118.1
anti-NC2beta 29.4 hAtac2/CSRP2BP, dsRNA-MBIP, or dsRNA-Control
anti-Atac2
(dsRNA-Cont) (C and D). The knockdown cells were then
54.8 54.8
anti-ActiveJNK anti-ActiveJNK stimulated with or without 500 mM (for S2 cells) or
54.8 54.8
400 mM (for 293T cells) sorbitol as in Figure 1C (see
anti-JNK anti-JNK
Figure S2D). The nuclear extracts were examined by
anti-tubulin anti-tubulin western blot probing with anti-Active JNK, anti-JNK,
54.8 54.8
anti-Atac2 (A and C), anti-NC2 b (B), anti-MBIP (D), or
anti-tubulin antibodies as a loading control (A–D, top).
dsRNA-NC2beta
dsRNA-Atac2 The intensities of each band of JNK, Active JNK, Atac2,
800 120 3000 120
NC2 b, hAtac2, or MBIP in western blotting were quantified
Ratio-JNK or Ratio-ActJNK

Ratio-JNK or Ratio-ActJNK
700
100 2500 100
600
from four independent experiments and plotted as ratios to

Ratio-NC2 beta
80 2000 80 the untreated control cells (A–D, bottom). Dark bars show
Ratio-Atac2

500
JNK JNK
400 60 ActJNK 1500 60 ActJNK
the ratio of JNK intensities, light bars show the ratio of
300
Atac2 NC2beta
Active JNK intensities, and the dotted line shows the ratio
40 1000 40
200 of Atac2 (A), NC2 b (B), Atac2/CSRP2BP (C), or MBIP (D)
100
20 500 20
intensities. Error bars represent standard deviation.
0 0 0 0 See also Figures S2 and S4.
1 2 3 4 1 2 3 4

Sorbitol - + - + Sorbitol - + - +
dsRNA-Cont. dsRNA-Atac2 dsRNA-Cont. dsRNA-NC2beta

the inhibitory activities of the full-length protein

C 293T Cells BP 2BP D 293T Cells to that of the MoaE and MBIP domains (Fig-
t t P2 P nt nt IP BIP
on on SR SR Co Co MB ures 2C and 2D). Full-length CG10238 and the
A -C A-C A-C A-C A- NA- NA- NA-M
R N RN R N RN N
Transfected with ds ds ds ds Transfected with d sR ds
R
ds
R
ds
R MoaE domain alone inhibited activation of JNK
Sorbitol - + - + (KDa) Sorbitol - + - + (KDa)
46
(Figure 2C, lanes 6–11; Figure 2D, bars 1–8).
anti-Atac2/CSRP2BP anti-MBIP
80
However, the MBIP domain of CG10238 failed
anti-Active JNK
46
anti-Active JNK
46 to prevent the activation of JNK (Figure 2C,
46
lanes 12–14; Figure 2D, bars 9–12). Thus, the
46
anti-JNK anti-JNK active portion of CG10238 in inhibition of JNK
anti-tubulin anti-tubulin
activation is the MoaE domain, whereas the
46
46
C-terminal sequences connect this activity to
the ATAC complex.
dsRNA-CSRP2BP dsRNA-MBIP

600 140 350 140

ATAC Inhibits JNK Activation

Ratio-JNK or Ratio-ActJNK

Ratio-JNK or Ratio-ActJNK

120 300 120

500
The MoaE domain of CG10238 was sufficient to
Ratio-CSRP2BP

100 250 100

400
prevent JNK activation when expressed in vivo,
Ratio-MBIP

80 JNK 200 80 JNK

300
60
whereas the MBIP domain was required to
ActJNK
CSRP2BP 150 60
ActJNK
MBIP

200
40 incorporate CG10238 into ATAC (Figures 2A– 100 40

100
20 2D). These data raised the question of whether 50 20

0 0 ATAC itself plays a role in inhibition of the JNK 0 0

1 2 3 4 1 2 3 4

Sorbitol - + - +
pathway. To address this question, we exam-
Sorbitol - + - +
dsRNA-Cont dsRNA-CSRP2BP dsRNA-Cont dsRNA-MBIP ined JNK activation in S2 cells where endoge-
nous subunits of ATAC were knocked down by
and all prokaryotes except parasitic bacteria have a MoaE dsRNA interference. The expression level of Atac2 was reduced
homology domain but are missing the MBIP domain. Sequence 60% in cells expressing dsRNA-Atac2 (Figure 3A). Interestingly,
database searches using the PSI-BLAST program (Altschul JNK was partially activated upon reduction of Atac2 even in the
et al., 1997) with human MBIP (GenBank accession number absence of osmotic stress (Figure 3A). Moreover, activation of
119586267) as query revealed sequence matches between the JNK by osmotic stress was enhanced in the Atac2 knockdown
N-terminal portion of MBIP orthologs from Metazoa and MoaE cells (Figure 3A). We observed similar results upon knockdown
proteins. Thus, the N-terminal sequences of mammalian MBIPs of NC2 b or CG10238 subunits of ATAC by dsRNA (Figure 3B;
are related to the MoaE sequences. Hence, our bioinformatic Figure S4A). Although JNK activation was not observed in D12
analysis suggests the N-terminal sequences of MBIP evolved knockdown cells without osmotic stress, its activation was
from MoaE and raises the possibility that the MoaE domain of also increased in these cells under conditions of osmotic stress
CG10238 contributes to the JNK inhibition activity. (Figure S4B). Because human MBIP was recently shown to be a
We generated S2 cell lines that stably expressed tagged component of the human ATAC complex (Fukuyama et al., 2000;
truncated forms containing the MoaE domain (N-terminal) or Wang et al., 2008), it was of interest to determine whether human
MBIP domain (C-terminal) (Figure S1B), and then compared ATAC inhibited the activation of JNK by osmotic stress in human

Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc. 729

 A B Figure 4. ATAC Is Required for the Transcription
Input IP Input IP Regulation of JNK Target Genes
N
IP with - IgG Jra IP with - IgG M
S (A and B) Extracts from S2 cells endogenously expressing
1 2 3 1 2 3 ATAC subunits and JNK (B) were immunoprecipitated with
anti-Atac2 anti-Atac2 antibodies to endogenous Jra (A) or Misshapen (MSN) (B).
The presence of ATAC subunits, MSN, and JNK in the
anti-CG10238 anti-CG10238 input and immunoprecipitates (IP) was detected by
anti-NC2beta western blots.
anti-NC2 beta (C) Quantitative real-time RT-PCR (qRT-PCR) of Jra
anti-D12
and Chickadee mRNA in S2 cells transfected with
anti-Jra anti-MSN dsRNA-LacZ (Control), dsRNA-Atac2, dsRNA-CG10238,
dsRNA-NC2 b (NC2b), dsRNA-Jra, or dsRNA-MSN (see
anti-JNK
Figure S5A). Gene-specific mRNA levels from the cells
without sorbitol stimulation (C, top) and 30 min after
500 mM sorbitol stimulation (C, bottom) were measured
C
Jra Sorbitol (-) Chickadee Sorbitol (-)
and normalized to RpL-32 expression (Figure S5B), and
are represented as ratios by qRT-PCR measurements.
2 2
1.8 1.8 The average of three independent experiments is graphed.
Chickadee/Control

1.6 1.6
1.4 1.4
Error bars represent standard deviation.
Jra/Control

1.2 1.2 (D) ChIP assays were performed with an antibody against
1 1
0.8 0.8 Jra from S2 cells without sorbitol stimulation (D, left) or with
0.6 0.6
0.4 0.4
sorbitol stimulation (D, right) expressing dsRNA-LacZ
0.2 0.2 (Cont), dsRNA-Atac2, dsRNA-CG10238, or dsRNA-MSN
0 0
1 2 3 4 5 6 1 2 3 4 5 6 (D) (Figures 5B–5M). The association of Jra on the probed
RNAi b a RNAi
cZ c2 38 2 Jr MSN cZ c2 38 2b Jra MSN region of the Jra gene as indicated in Figure 5A, and input
La Ata 102 NC La Ata 102 NC
CG CG
chromatin was measured by qRT-PCR and normalized to
Jra Sorbitol (+) Chickadee Sorbitol (+)
input, and the ratios of quantities are represented (see
4 1.6 Figure S6). The average of three independent experiments
1.4
is graphed. Error bars represent standard deviation.
Chickadee/Control

3 1.2
See also Figures S2, S5, and S6.
Jra/Control

1
2 0.8
0.6
1 0.4
0.2
0 0

RNAi
1 2 3 4 5 6
RNAi
in vivo. Endogenous Atac2, CG10238, NC2 b,
1 2 3 4 5 6
2 8 2b Jra SN a
cZ tac 23 cZ c2 38 2b Jr MSN
La A G10 N C M La Ata 102 NC
and D12 coimmunoprecipitated with endoge-
C CG
nous MSN (Figure 4B).
If ATAC serves as a cofactor for Jra, then
D
ChIP-Jra on Jra Sorbitol (-) ChIP-Jra on Jra Sorbitol (+)
Jra-dependent transcription would require
dsRNA-Cont dsRNA-Atac2 dsRNA-CG10238 dsRNA-MSN dsRNA-Cont dsRNA-Atac2 dsRNA-CG10238 dsRNA-MSN ATAC and ATAC should be localized to Jra target
100.00%
90.00%
genes. Because the gene encoding c-Jun, the
80.00%

70.00%
80.00%
70.00%
mammalian homolog of Jra, is positively
60.00%
ChIP/Input

ChIP/Input

50.00%
60.00%
50.00%
regulated by c-Jun protein (Angel et al., 1988),
40.00%
40.00%
30.00% we examined whether ATAC functions in Jra
30.00%

20.00% 20.00%

10.00% transcription. We measured the levels of Jra

10.00%
0.00%

Probe A B C D E Probe A B C D E
transcripts by quantitative real-time RT-PCR
0.00%

(qRT-PCR) of control and ATAC knockdown S2

cells. As a control, we confirmed that Jra tran-
293T cells (Figure S2D). Indeed, JNK was activated to higher script levels were reduced in cells expressing dsRNA-Jra
levels upon knockdown of MBIP or CSRP2BP (human homolog compared with cells expressing control-dsRNA-LacZ (Figure 4C,
of Atac2) in 293T cells (Figures 3C and 3D). These data indicate upper left). Importantly, Jra transcript levels were significantly
that the ATAC complex is required for complete inhibition of JNK reduced in Atac2 and CG10238 knockdown cells and slightly
activation in both Drosophila and human cells. reduced in NC2 b knockdown cells (Figure 4C, upper left). We
did not detect effects on Jra transcripts in MSN knockdown cells.
ATAC Functions as Transcriptional Cofactor We confirmed that the transcripts of atac2, CG10238, nc2 b, and
for JNK Target Genes msn were reduced by expression of each dsRNA-Atac2, dsRNA-
The presence of peptides from the Jra transcription factor in CG10238, dsRNA-NC2 b, and dsRNA-MSN by qRT-PCR (Fig-
MudPIT analyses of the affinity-purified ATAC complex (Fig- ure S5A). We also observed that the expression of Atac2 was
ure 1A) suggests that ATAC interacts with Jra and may serve increased in Jra knockdown (see below).
as a transcriptional cofactor. Indeed, endogenous Jra was able We next examined chickadee, another Jra target gene
to coimmunoprecipitate with Atac2, CG10238, and NC2 b, sub- (Jasper et al., 2001). Transcript levels of chickadee decreased
units of ATAC (Figure 4A). We also sought to confirm the interac- significantly upon knockdown of Atac2 and somewhat more
tion of MSN (Figure 1A), the Drosophila Ste-20 kinase, with ATAC modestly upon knockdown of NC2 b or Jra (Figure 4C).

730 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.

Transcript levels of chickadee were not reduced significantly in
CG10238 knockdown cells. Of note, we did not observe changes
in the transcript levels of target genes of a different H4K16
acetyltransferase complex (MSL) upon knockdown of ATAC,
indicating that requirement for ATAC is specific for Jra and
chickadee (Figure S5B).
The JNK cascade is activated by osmotic stress and increases
c-Jun expression in mammals. Thus, we examined Jra and
chickadee transcript levels in ATAC knockdown cells upon
treatment with sorbitol. Interestingly, Jra transcripts were not
reduced in Atac2 knockdown cells upon osmotic stress but
were increased in CG10238 knockdown cells. The expression
of Chickadee was also increased in ATAC and MSN knockdown
cells under osmotic stress. Thus, ATAC positively impacts the
expression of JNK target genes in the absence of osmotic stress
but plays a predominantly negative role at JNK target genes
under conditions of osmotic stress. Moreover, the positive
effects of ATAC on expression of JNK target genes are
mediated primarily by Atac2, whereas its negative effects are
most strongly dependent on CG10238.
ATAC Functions as a Cofactor for the JNK Target Gene
Jra, and Coordinates Upstream MAPKs onto Jra
Previous studies have shown the direct association of MAPKs
with gene loci that they regulate (de Nadal and Posas, 2010;
Dioum et al., 2009). Thus, we measured the occupancy of
ATAC and upstream kinases directly on JNK target genes. We
examined the occupancy of ATAC on the Jra gene in ATAC
and MSN knockdown cells by chromatin immunoprecipitation
(ChIP) assay. Previous studies have shown that Jun can form a
homodimer or heterodimerize with ATF or Fos to form the AP-1
transcription factor. The c-Jun promoter contains AP-1-like
sequences (TTACCTCA and TGACATCA) in addition to an
NF-jun binding sequence (GGAGTCTCC) (Rozek and Pfeifer,
1993). c-Jun was found to occupy the c-Jun promoter without
external stimulus (Angel and Karin, 1991). We confirmed that
Jra actually bound to the Jra gene by ChIP assay (Figure 4D,
left). Jra was found to significantly occupy a site in the middle
of the coding region that contains an NF-jun recognition
sequence (Figure 4D, left). We analyzed regions of the Jra
gene that are comparable to the regulatory regions of the human
c-Jun gene (Figure 5A). We probed ATAC occupancy, with and
without osmotic stress, at an upstream enhancer (primer A),
the promoter containing the AP-1 binding sequence (primer B),
the 50 region of the coding sequence (primer C), the middle
of the coding sequence that contained an NF-jun binding
sequence (GGAGGCACC) (primer D), and a region 30 of the
coding sequence (primer E) (FlyBase and UCSC Genome
Browser) (Figure 5A) (Rozek and Pfeifer, 1993).
In the absence of osmotic stress, Atac2 was found to
occupy the enhancer and promoter regions of Jra (Figure 5B,
dsRNA-Cont). As expected, Atac2 occupancy of these regions
was reduced in the Atac2 knockdown cells (Figure 5B). Atac2
occupancy of the promoter was reduced in CG10238 but not
MSN knockdown cells. As controls for the specificity of ATAC
occupancy of Jra, we performed an ATAC-ChIP assay on two
target genes of the MSL complex (Figures S5 and S6). Back-
ground signals on these genes were not affected by the
knockdown of Atac2 or CG10238. We next investigated
whether ATAC occupancy affected acetylation of the nucleo-
some on the Jra gene by a ChIP assay for H4K16 acetylation.
In control cells without osmotic stress, both the enhancer and
promoter regions were enriched in H4K16 acetylation, which
was significantly reduced in the Atac2 and CG10238 knock-
down cells (Figure 5D). Thus, in the absence of osmotic
stress, ATAC occupied the enhancer and promoter of Jra,
and H4K16 acetylation at these sites was dependent on
ATAC.
In the presence of osmotic stress, ATAC occupancy and
H4K16 acetylation of the promoter were similar but were
reduced at the upstream enhancer (Figures 5C and 5E). Interest-
ingly, ATAC occupancy and H4K16 acetylation of the coding
region at primer D were stimulated by osmotic stress (Figures
5C and 5E). Jra was clearly localized on this region with or
without osmotic stress (Figure 4D). Jra occupancy at primer D
was strongly reduced in Atac2, CG10238, and MSN knockdown
cells in the absence of sorbitol. However, under osmotic stress,
Jra occupancy required primarily CG10238 (Figure 4D). Thus,
under conditions of osmotic stress, ATAC occupancy at the
upstream enhancer was reduced and instead increased in the
coding region where Jra was bound.
We next tested whether stress-activated upstream kinases
were recruited to the Jra gene. We probed Active JNK (phos-
pho-JNK) and phospho-MKK4/SEK1 (Ser257/Thr261) occu-
pancy on the Jra gene. We also tested for the occupancy of
the MSN kinase. In the absence of osmotic stress in control cells,
each of these kinases could be found on the Jra gene although
their locations varied. Active JNK was found at the enhancer
(primer A), the promoter (primer B), and downstream of the
coding region (primer E) (Figure 5F). Phospho-MKK4 was local-
ized to the enhancer, promoter, and downstream of the coding
region (primer E) and MSN was localized to the promoter (primer
B) (Figures 5H and 5J). Strikingly, like ATAC upon osmotic stress,
occupancy of each of these kinases was increased in the coding
region (primer D), the site of Jra binding (Figures 5G, 5I, and 5K).
This was most significant for Active JNK, which bound strongly
to the primer D region upon osmotic stress (note the difference
in scale between Figures 5F and 5G), and for MSN, which relo-
calized from the promoter to the coding region (primer D)
(Figures 5J and 5K).
The ChIP results from the knockdown cell lines indicate a
dynamic interplay between ATAC and the upstream kinases.
For example, in the absence of osmotic stress, H4K16 acetyla-
tion at the promoter (primer B) was independent of MSN (Fig-
ure 5D); however, upon osmotic stress, MSN suppressed
H4K16 acetylation at the promoter but was required for
maximum acetylation in the coding region (primer D) (Figure 5E).
In the absence of osmotic stress, the binding of Active JNK to the
enhancer, promoter, and downstream of the coding region
required ATAC but was suppressed by MSN (Figure 5F). By
contrast, under osmotic stress, the strong binding of Active
JNK to the coding region (primer D) required both ATAC and
MSN (Figure 5G). In the absence of osmotic stress, binding of
phospho-MKK4 to the enhancer was suppressed by ATAC and
MSN (Figure 5H) but upon osmotic stress, binding to the
promoter required MSN and CG10238 specifically (Figure 5I).
Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc. 731
A Figure 5. Occupancy of ATAC, H4K16
Jra gene span Jra CDS Acetylation, MAPKs, and H3S10P on the
Enhancer Promoter Coding Coding E
Jra Gene with and without Osmotic Stress
Probes 100bp
A B C D (A) Schematic representation of the Jra gene
Enhancer
B ChIP-Atac2 on Jra Sorbitol (-) C ChIP-Atac2 on Jra Sorbitol (+) AP-1
NF-jun
indicating the position of different probes tested
dsRNA-Contt dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN dsRNA-Cont dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
(A–E on x axis for B–M). The gray box indicates
1.00% 1.00%

0.80% 0.80%
the coding sequence (CDS), the gray circles indi-

ChIP/Input
cate the enhancer, the white circle indicates pre-
ChIP/Input

0.60% 0.60%

0.40% 0.40%
dicted AP-1 sites, and the black square indicates
0.20%
0.20%

0.00%
the NF-jun-like sequence.
0.00%
A B C D E A B C D E
(B–M) ChIP assays were performed with anti-
D ChIP-H4K16ac on Jra Soro
Sorobitol
r bitol (-) E ChIP-H4K16ac on Jra Sorbitol (+) bodies against Atac2, H4K16ac, Active JNK,
dsRNA-Cont dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN

0.60%
dsRNA-Cont

6.00%
dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
phospho-MKK4 (S257T261) (pMKK4), MSN, and
0.50% 5.00% phospho-H3S10 (H3S10P) from ATAC and MSN
ChIP/Input

0.40%
knockdown S2 cells (as in Figure 4D) without
ChIP/Input 4.00%

0.30%
3.00%

0.20%
2.00% sorbitol stimulation (B, D, F, H, J, and L) (Figures
0.10%

0.00%
1.00%
6B, 6D, 6F, 6H, and 6J) or with sorbitol stimulation
0.00%
A B C D E
A B C D E (C, E, G, I, K, M) (Figures 6C, 6E, 6G, 6I, and 6K)
F ChIP-ActiveJNK
ChIP-Active
v JNK on Jra Sorbitol (-) G ChIP-ActiveJNK
ChIP-Active
v JNK on Jra Sorbitol (+)
on the probed region of the Jra gene as indicated
dsRNA-Cont dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN dsRNA-Cont dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
in (A), and input chromatin was measured by
1.20% 70.00%

1.00% 60.00%
qRT-PCR and normalized to input. The ratios of
50.00%
measured quantities are represented (see Figures
ChIP/Input

ChIP/Input

0.80%
40.00%
0.60%

0.40%
30.00%
S5A and S6). The average of three independent
20.00%
0.20% 10.00% experiments is graphed. Error bars represent
0.00% 0.00%
A B C D E A B C D E standard deviation.
See also Figures S2, S5, and S6.
H dsRNA-Contt
ChIP-pMKK4 on Jra Sorbitol (-)
dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
I dsRNA-Cont
ChIP-pMKK4 on Jra Sorbitol (+)
dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
0.30% 0.45%
0.40%
0.25%
0.35%
ChIP/Input

0.20% 0.30%
ChIP/Input

0.25%
0.15%
0.20%

0.10% 0.15%
0.10%
0.05% 0.05%
0.00%
the promoter and strongly suppressed
0.00%

A B C D E
A B C D E
downstream of the coding region
J dsRNA-Contt
ChIP-Misshapen on Jra Sorbitol (-)
dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
K dsRNA-Cont
ChIP-Missahpen on Jar Sorbitol (+)
dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN
(Figure 5L). Osmotic stress substantially
0.35% 0.30% increased H3S10 phosphorylation levels
0.30%
0.25%
0.25%
across the Jra gene except for the
ChIP/Input

ChIP/Input

0.20%
0.20%

0.15%

0.10%
0.15%

0.10%
upstream enhancer (note the different
0.05%

0.00%
0.05%
scales in Figures 5L and 5M). Induction
0.00%
A B C D E
A B C D E
of H3S10 phosphorylation at the
L ChIP-H3S10P on Jra Sorbitol (-) M ChIP-H3S10P on Jra Sorbitol (+) promoter, the coding region (primer D),
dsRNA-Contt dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN dsRNA-Cont dsRNA-Atac2
dsRNA-At
A ac2 dsRNA-CG10238 dsRNA-MSN

2.50% 20.00%
18.00%
and downstream of the coding region de-
pended on ATAC (especially CG10238);
2.00% 16.00%
14.00%
ChIP/Input

ChIP/Input

1.50% 12.00%

1.00%
10.00%
8.00% however, MSN suppressed H3S10 phos-
6.00%
0.50%

0.00%
4.00%
2.00% phorylation across the gene (Figure 5M).
0.00%
0 00%

A B
Finally, binding of MSN to the promoter is suppressed by ATAC
in the absence of osmotic stress (Figure 5J); however, when it
moves to the coding region where Jra is bound (primer D),
upon osmotic stress its binding is independent of ATAC
(Figure 5K).
A previous study linked the acetylation of H4K16 with
H3S10 phosphorylation in the activation of the FOSL1 gene
(Zippo et al., 2009). At this gene, H3S10 phosphorylation was
reported to recruit MOF-dependent H4K16 acetylation, which
then recruited the elongation factor P-TEFb to release paused
polymerase (Zippo et al., 2009). As ATAC is also an H4K16 ace-
tyltransferase, we tested for a relationship between H3S10
phosphorylation and H4K16 acetylation, ATAC, or MSN at Jra.
H3S10P was enriched in the enhancer and promoter regions
(primer B) in the absence of osmotic stress (Figure 5L). H3S10
phosphorylation levels were suppressed by ATAC and MSN at
C D E A B C D E
These results suggest that the relation-
ship between H3S10 phosphorylation
and H4K16 acetylation by ATAC at Jra
differs from what was observed with H3S10 phosphorylation
and H4K16 acetylation by MOF at the FOSL1 gene (Zippo
et al., 2009). Acetylation of H4K16 at the enhancer of FOSL1
by MOF required H3S10 phosphorylation; however, ATAC is
required for H3S10 phosphorylation in the presence of osmotic
stress. Moreover, ATAC-dependent H4K16 acetylation at the
Jra enhancer in the absence of osmotic stress (Figure 5D)
occurred independently of significant H3S10 phosphorylation
(Figure 5L).
ATAC Acetylates Nucleosomes and Coordinates
the JNK Upstream Kinases with Jra on chickadee
The most striking aspect of our ChIP analysis of the Jra gene is the
movement upon osmotic stress of ATAC and MAPKs to the
coding region (primer D), which contains an NF-jun binding
sequence and is occupied by the Jra transcription factor. This

732 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.

A Figure 6. Occupancy of ATAC, H4K16
Chickadee gene span Acetylation, MAPKs, and H3S10P on the
chic CDS chickadee Gene with and without Osmotic
Probes ChicA ChicB ChicC Stress
1K bp
Enhancer (A) Schematic representation of the chickadee
B ChIP-Atac2 on Chickadee Sorbitol (-) C ChIP-Atac2 on Chickadee Sorbitol (+) AP-1 gene including 7 kb upstream of the transcription
dsRNA-Cont dsRNA-Atac2 dsRNA-Cont dsRNA-Atac2
dsRNA-CG10238
dsRNA-
A CG10238
dsRNA-Jra
dsRNA-
A Jra
dsRNA-MSN
dsRNA-
A MSN dsRNA-CG10238
dsRNA-
A CG10238
dsRNA-
A Jra
dsRNA-Jra
dsRNA-MSN
dsRNA-
A MSN
start site indicating the position of different probes
0.14%
4% 1.00%
0.90%
on the gene (ChicA-C on x axis for B–K). The gray
0.12%
2%

0.10%
0%
0.80%
0.70%
box indicates the coding sequence (CDS), the gray
ChIP/Input

ChIP/Input
0.08%
8% 0.60%

0.06%
6%
0.50% circles indicate the enhancer, and the white circles
0.40%
4%
0.04% 0.30%
0.20%
indicate predicted AP-1 sites.
2%
0.02%

0.00%
0%
0.10%
0.00%
(B–K) ChIP assays were performed with antibodies
ChicA ChicB ChicC ChicA ChicB ChicC
against Atac2, H4K16ac, pMKK4 (S257T261),
D ChIP-H4K16ac on Chickadee Sorbitol (-) E ChIP-H4K16ac on Chickadee Sorbitol (+) MSN, and phospho-H3S10 (H3S10P) from ATAC
dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-Atac2
dsRNA-
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN
dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-Atac2
dsRNA-
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN and MSN knockdown S2 cells with or without
dsRNA-
A Jra
dsRNA-Jra dsRNA-Jra
dsRNA-
A Jra
.90%
0.90% 0.90% sorbitol stimulation (as in Figure 4D) on the probed
.80%
0.80% 0.80%
.70%
0.70% 0.70% region of the chickadee gene and upstream region
ChIP/Input

0.60%
.60% 0.60%
ChIP/Input

.50%
0.50% 0.50% as indicated in (A), and input chromatin was
0.40%
.40% 0.40%

.30%
0.30% 0.30%
0.20%
measured by qRT-PCR and normalized to input
.20%
0.20%
.10%
0.10%
0.00%
.00%
0.10%
0.00%
(see Figure S7A). The ratios of the measured
ChicA ChicB ChicC ChicA ChicB ChicC quantities are represented. The average of three
F ChIP-pMKK4 on Chicadee Sorbitol (-) G ChIP-pMKK4 on Chickadee Sorbitol (+)
independent experiments is graphed. Error bars
dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-Atac2
dsRNA-
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN
dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-Atac2
dsRNA-
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN represent standard deviation.
dsRNA-Jra dsRNA-
A Jra
dsRNA-Jra
0.60%
.60% 1.20% See also Figures S2, S5, S6, and S7.
0.50%
.50% 1.00%

0.80%
ChIP/Input

0.40%
.40%
ChIP/Input

0.30%
.30% 0.60%

0.20%
.20% 0.40%

0.10%
.10% 0.20%

0.00%
0.00%
.00%
ChicA ChicB ChicC ChicA ChicB ChicC MSN knockdown cells but not in Atac2
H ChIP-Misshapen on Chickadee Sorbitol (-) I ChIP-Misshapen on Chickadee Sorbitol (+)
knockdown cells (Figure S7A). These
dsRNA-Cont
dsRNA-
A Cont
dsRNA-CG10238
dsRNA-
A CG10238
dsRNA-
A Jra
dsRNA-Jra
dsRNA-
dsRNA-Atac2
A At
A ac2
dsRNA-MSN
dsRNA-
A MSN
dsRNA-
dsRNA-Cont
A Cont
dsRNA-CG10238
dsRNA-
A CG10238
dsRNA-Jra
dsRNA-
A Jra
dsRNA-
dsRNA-Atac2
A At
A ac2
dsRNA-MSN
dsRNA-
A MSN two regions were co-occupied by ATAC
0.30% 0.25%
and H4K16 acetylation (Figures 6B and
0.25% 0.20%
6D). The presence of Atac2 and H4K16
ChIP/Input

0.20%
ChIP/Input

0.15%
0.15%

0.10%
0.10% acetylation at these regions was not
0.05%
0.05%
affected by the knockdown of Jra, sug-
0.00% 0.00%

ChicA ChicB ChicC ChicA ChicB ChicC gesting ATAC does not require Jra to
J K bind the chickadee gene in the absence
ChIP-H3S10P on Chickadee Sorbitol (-) ChIP-H3S10P on Chickadee Sorbitol (+)

dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-
dsRNA-Atac2
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN
dsRNA-Cont
dsRNA-
A Cont
dsRNA-
dsRNA-CG10238
A CG10238
dsRNA-Atac2
dsRNA-
A At
dsRNA-
A ac2
dsRNA-MSN
A MSN
of osmotic stress. Interestingly, in the
dsRNA-
A Jra
dsRNA-Jra dsRNA-Jra
dsRNA-
A Jra
2.00% 18.00%
presence of osmotic stress, ATAC occu-
1.80% 16.00%
1.60%
1.40%
14.00% pancy and H4K16 acetylation were
ChIP/Input

ChIP/Input

12.00%
1.20%
1.00%
0.80%
10.00%
8.00%
increased at the ChicA region, and this
6.00%
0.60%
0.40% 4.00%
2.00%
increased occupancy was reduced in
0.20%
0.00%
0 00%

ChicA ChicB ChicC

0.00%
0 00%

ChicA ChicB ChicC

Jra knockdown cells (Figures 6C and 6E).
Phospho-MKK4 and MSN also occupied
the ChicA and ChicB regions, and these
raised the question as to whether ATAC and MAPKs move to kinases were also found in the open reading frame (ChicC) in
this location because Jra is bound there or because they need the absence of osmotic stress (Figures 6F and 6H). In contrast
to associate with the coding region. Thus, we performed a ChIP to what was seen at the Jra gene, on chickadee the kinases
assay on another JNK target gene, chickadee. Chickadee does move to the upstream ChicA region upon osmotic stress and their
not have an apparent NF-jun binding site in the coding region. It occupancy is largely (MKK4) or slightly (MSN) dependent on Jra
does, however, have several AP-1-like sequences upstream of (Figures 6G and 6I). Occupancy of ChicA by the kinases under
the promoter and at a far-upstream enhancer (Figure 6A). We these conditions is slightly suppressed by Atac2 but completely
analyzed 7 kb upstream of the chickadee gene and found five dependent on CG10238. Indeed, CG10238 was crucial for the
AP-1-like sequences. We probed a region containing an AP-1- occupancy of ATAC, MKK4, and MSN on the far-upstream
like site, which was also near two CCAAT motifs that are 6.4 kb ChicA region under conditions of osmotic stress (Figures 6C,
upstream from the coding sequence (ChicA primer). We also 6G, and 6I). At the Jra gene, occupancy of the primer D region
probed adjacent to the AP-1-like site 2 kb upstream from the containing the NF-Jun binding site by ATAC, Active JNK, and
coding sequence close to the predicted promoter (ChicB primer) phospho-MKK4 upon osmotic stress was dependent on
and the middle of the coding region (ChicC primer) (Figure 6A). CG10238. As seen on Jra, H3S10P on chickadee was sup-
ChIP analysis revealed that Jra was localized at the ChicA and pressed by ATAC and MSN in the absence of osmotic stress
ChicB regions, suggesting Jra binds AP-1 sites on chickadee and, although H3S10P increased upon osmotic stress, these
(Figure S7A). Jra occupancy was reduced in CG10238 and levels were suppressed by MSN specifically (Figures 6J and 6K).

Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc. 733

 A The models presented in Figure 7 summarize the molecular
ATAC events occurring on Jra and chickadee as revealed by the ChIP
CG10238
JNK data. Although differing in details having to do with the architec-
JNK
Atac2 Jra Jra-P MKK4 ture of each gene, comparison of the expression and ChIP data
Jra K16ac
S10P
Osmotic stress from Jra and chickadee reveals strikingly common features
E AP-1 NF-jun TATA-like
about the role of ATAC at these genes. First, in the absence of
MSN
MKK4
MSN osmotic stress, ATAC is required for the basal expression of
these genes, which is most highly dependent on the Atac2 ace-
B MKK4-P
tyltransferase subunit. Second, upon osmotic stress, ATAC
suppresses induced transcription levels of these genes, which
CG10238 JNK-P
is most highly dependent on the CG10238 subunit. Third, upon
ATAC
Atac2 osmotic stress, ATAC and the MAPKs move to the region of
Jra-P
Jra
K16ac K16ac the gene where Jra is bound. And fourth, the relocalization of
S10P
E AP-1 NF-jun TATA-like ATAC and MAPKs is dependent on CG10238, the MAP kinase
C MSN upstream kinase binding inhibitory protein. Thus, ATAC coordi-
P-1 nates the occupancy of Jra and upstream kinases while it
E E A 16ac
K controls the levels of JNK activation and target gene induction.
Jra

ATAC MSN
CG10238
MKK4
DISCUSSION
Osmotic stress
JNK
Jra
K16ac
ATAC shares four subunits with the SAGA transcription coactiva-
MSN
S10P
Chickadee AP-1 tor complex and was anticipated to function as a coactivator.
MKK4-P However, preliminary studies indicated that ATAC did not
AP-1c interact with the activation domains of VP16 or p53 under
E E K160aP JNK-P
D S10P S10P S1
conditions in which SAGA bound (Kusch et al., 2003). Prompted
Jra-P
by the identification of peptides from the Jra (Drosophila c-Jun)
ATAC MSN
CG10238
MKK4
transcription factor in ATAC affinity purifications, we demon-
strated that Jra coimmunoprecipitates with ATAC subunits. We
Jra JNK
K16ac further demonstrate that ATAC is localized to Jra target genes
S10P
Chickadee AP-1 (Jra and chickadee) and is required for H4K16 acetylation at
the Jra and chickadee enhancer and promoter. ATAC was
required for basal levels of expression of these genes, which
Figure 7. Models Summarizing ATAC Functions and the Molecular depended strongly on the Atac2 acetyltransferase subunit.
Events Occurring on Jra and chickadee as Revealed by the ChIP Thus, to our knowledge, this is the first report to demonstrate
Data
that ATAC functions as a transcription cofactor.
Summaries of ChIP data on the Jra gene without and with osmotic stress are
shown in (A) and (B), respectively. Summaries of ChIP data on the chickadee
Our pursuit of the functions of the CG10238 subunit of ATAC
gene without and with osmotic stress are shown in (C) and (D), respectively. and the potential role of its MoaE domain led us to discover
(A) In the absence of osmotic stress, ATAC occupies and acetylates H4K16 that ATAC is intimately integrated into MAP kinase signaling
(K16ac) at the enhancer (E) and promoter regions, containing AP-1 sites and target gene expression. CG10238 was found to inhibit
(AP-1), of the Jra gene and is required for basal levels of Jra transcription (small JNK activation by osmotic stress, an activity mediated through
arrow). ATAC blocks the recruitment of MKK4 to the enhancer and MSN to the
the MoaE domain. Thus, structural features of this protein that
promoter while inhibiting the activation of JNK. The phosphorylation of H3S10
(S10P) on the promoter region is suppressed by MSN.
are utilized for molybdopterin synthesis have been conscripted
(B) When cells are exposed to osmotic stress, the stress-activated kinases in in ATAC to function in the regulation of MAP kinase signaling.
the JNK cascade are recruited to the Jra binding motif (NF-jun) on the Jra gene The MSN kinase is the most likely direct target of CG10238, as
by ATAC, especially dependent on CG10238. In addition to interacting with the it coimmunoprecipitates with ATAC subunits. However, we
promoter region, ATAC also interacts with the NF-jun site by interaction with cannot rule out potential interactions with other kinases; for
Jra and further acetylates H4K16 at this site. ATAC continues to limit the extent example, peptides from MKK4 have been found in ATAC purifi-
of JNK activation. Phosphorylation of H3S10 is increased across the Jra gene
cations (Figure 1). Under conditions of osmotic stress, ATAC
by osmotic stress but is still suppressed by MSN. Jra transcription is rapidly
and transiently induced (large arrow). suppresses the level of target gene expression in a manner
(C) The chickadee gene does not have an apparent NF-jun site. In the absence strongly dependent on CG10238. We believe this is due to inhi-
of osmotic stress, Jra, ATAC, and acetyl-H4K16 are found at two AP-1 sites, bition of JNK activation rather than a direct negative effect of
which are 2 and 7 kb upstream from the coding sequence. Upstream kinases ATAC on transcription. In fact, ATAC may still function as a posi-
occupy these sites and the downstream coding region. tive cofactor for induced Jra expression, as the levels of H3S10P
(D) In the presence of osmotic stress, ATAC preferentially relocalizes to the
on the gene, considered a positive mark for transcription
far-upstream AP-1 site and recruits Jra to this location. Phospho-MKK4
(MKK4-P) and MSN are also relocalized to this far-upstream AP-1 site in
a CG10238- and Jra-dependent manner. The CG10238 subunit of ATAC is the levels of phosphorylated H3S10 are suppressed by MSN across the
required for ATAC, Jra, MKK4-P, and MSN relocalization to the far-upstream chickadee gene.
enhancer. Transcription of chickadee is activated (large arrow) whereas See also Figure S7.

734 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.

(Zippo et al., 2009), are dependent on ATAC (Figure 5M). In
a manner strongly dependent on CG10238, ATAC plays a striking
role in the binding of MAPKs to the Jra and chickadee genes.
Upon osmotic stress, ATAC was required for the recruitment of
the JNK, MKK4, and MSN kinases to the site of Jra binding.
The only exception was the binding of MSN to the Jra gene
(Figure 5K). However, MSN required CG10238 for binding chick-
adee under inducing conditions (Figure 6I). The crucial role of
CG10238 in both the recruitment of the kinases and the inhibition
of signaling is most consistent with a model by which ATAC
inhibits JNK activation at the downstream target genes. The
fact that ATAC and the MAPKs all move to the site of Jra binding
suggests that a localized signaling network is organized at the
site occupied by the downstream transcription factor.
Our data indicate that a positive cofactor of JNK target gene
transcription, ATAC, also serves to negatively feed back for inhi-
bition of the JNK signaling pathway. Why would a downstream
activator also repress upstream signaling? The JNK pathway
and c-Jun have been shown to suppress p53-dependent
apoptosis and cellular senescence (Alexaki et al., 2008; Butter-
field et al., 1997; Das et al., 2007; Fan et al., 2010; Ham et al.,
2000; Takekawa and Saito, 1998; Yang et al., 1997). Moreover,
c-Jun has been shown to repress transcription of the p53 gene
and hence its downstream targets (Schreiber et al., 1999).
Thus, a delicate balance of JNK signaling and p53 expression
controls cell growth, proliferation, and apoptosis. If ATAC
suppression of JNK is involved in this process, we might expect
loss of ATAC to result in decreased expression of p53 due to
increased JNK activity (Schreiber et al., 1999). Indeed, whereas
knockdown of Jra increased p53 expression in S2 cells, knock-
down of ATAC subunits resulted in a decrease in p53 expression
consistent with ATAC suppression of JNK activation (Figure S7B).
Our results reveal a mechanism for governing the transcrip-
tional response to signaling pathways. By serving as both a
downstream transcriptional cofactor and as an inhibitor of
upstream signaling, ATAC serves as a master regulator which
administers the appropriate level of response to the inducing
signals.
EXPERIMENTAL PROCEDURES
Cell Lines, Extract Preparation, and Complex Purification
The nuclear extracts from S2 cell lines were generated as described in
Extended Experimental Procedures and were used for FLAG affinity purifica-
tion as previously described (Suganuma et al., 2008).
Chromatin Immunoprecipitation Assay
The crosslinked S2 cell pellets were washed, resuspended, and sonicated,
and the DNA was immunoprecipitated with antibodies using Dynabeads
(Invitrogen). The bound DNA and input DNA were incubated with RNaseA
and the crosslinking was reversed. The ethanol-precipitated DNA pellets
were resuspended and then were analyzed by quantitative real-time RT-PCR.
Antibodies, buffer content, and primers for ChIP-qRT-PCR are described in
the Extended Experimental Procedures.
Osmotic Stress Response in dsRNAi Knockdown S2 Cells
and siRNA Knockdown 293T Cells
The dsRNAi was transfected in S2 cells. The siRNA was transfected in 293T
cells as described in the Extended Experimental Procedures. The cells were
incubated with 500 mM sorbitol for 12 min before harvest.
Multidimensional Protein Identification Technology Analysis
MudPIT analysis was performed as described in the Extended Experimental
Procedures.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures and
seven figures and can be found with this article online at doi:10.1016/j.cell.
2010.07.045.
ACKNOWLEDGMENTS
We thank the Workman Lab members, Bjoern Gaertner in the Zeitlinger Lab,
and Hidehisa Takahashi in the Conaway Lab for advice and support during
this project. This research was supported by the Stowers Institute for Medical
Research.
Received: December 30, 2009
Revised: May 14, 2010
Accepted: July 1, 2010
Published: September 2, 2010
REFERENCES
Alepuz, P.M., Jovanovic, A., Reiser, V., and Ammerer, G. (2001). Stress-
induced MAP kinase Hog1 is part of transcription activation complexes. Mol.
Cell 7, 767–777.
Alepuz, P.M., de Nadal, E., Zapater, M., Ammerer, G., and Posas, F. (2003).
Osmostress-induced transcription by Hot1 depends on a Hog1-mediated
recruitment of the RNA Pol II. EMBO J. 22, 2433–2442.
Alexaki, V.I., Javelaud, D., and Mauviel, A. (2008). JNK supports survival in
melanoma cells by controlling cell cycle arrest and apoptosis. Pigment Cell
Melanoma Res 21, 429–438.
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W.,
and Lipman, D.J. (1997). Gapped BLAST and PSI-BLAST: a new generation
of protein database search programs. Nucleic Acids Res. 25, 3389–3402.
Angel, P., and Karin, M. (1991). The role of Jun, Fos and the AP-1 complex in
cell-proliferation and transformation. Biochim. Biophys. Acta 1072, 129–157.
Angel, P., Hattori, K., Smeal, T., and Karin, M. (1988). The jun proto-oncogene
is positively autoregulated by its product, Jun/AP-1. Cell 55, 875–885.
Butterfield, L., Storey, B., Maas, L., and Heasley, L.E. (1997). c-Jun
NH2-terminal kinase regulation of the apoptotic response of small cell lung
cancer cells to ultraviolet radiation. J. Biol. Chem. 272, 10110–10116.
Ciurciu, A., Komonyi, O., Pankotai, T., and Boros, I.M. (2006). The Drosophila
histone acetyltransferase Gcn5 and transcriptional adaptor Ada2a are
involved in nucleosomal histone H4 acetylation. Mol. Cell. Biol. 26, 9413–9423.
Das, M., Jiang, F., Sluss, H.K., Zhang, C., Shokat, K.M., Flavell, R.A., and
Davis, R.J. (2007). Suppression of p53-dependent senescence by the JNK
signal transduction pathway. Proc. Natl. Acad. Sci. USA 104, 15759–15764.
de Nadal, E., and Posas, F. (2010). Multilayered control of gene expression by
stress-activated protein kinases. EMBO J. 29, 4–13.
de Nadal, E., Alepuz, P.M., and Posas, F. (2002). Dealing with osmostress
through MAP kinase activation. EMBO Rep. 3, 735–740.
de Nadal, E., Zapater, M., Alepuz, P.M., Sumoy, L., Mas, G., and Posas, F.
(2004). The MAPK Hog1 recruits Rpd3 histone deacetylase to activate
osmoresponsive genes. Nature 427, 370–374.
Devary, Y., Gottlieb, R.A., Lau, L.F., and Karin, M. (1991). Rapid and
preferential activation of the c-jun gene during the mammalian UV response.
Mol. Cell. Biol. 11, 2804–2811.
Dioum, E.M., Wauson, E.M., and Cobb, M.H. (2009). MAP-ping unconven-
tional protein-DNA interactions. Cell 139, 462–463.
Edmunds, J.W., and Mahadevan, L.C. (2004). MAP kinases as structural
adaptors and enzymatic activators in transcription complexes. J. Cell Sci.
117, 3715–3723.
Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc. 735
Fan, Y., Lee, T.V., Xu, D., Chen, Z., Lamblin, A.F., Steller, H., and Bergmann, A.
(2010). Dual roles of Drosophila p53 in cell death and cell differentiation. Cell
Death Differ. 17, 912–921.
Fukuyama, K., Yoshida, M., Yamashita, A., Deyama, T., Baba, M., Suzuki, A.,
Mohri, H., Ikezawa, Z., Nakajima, H., Hirai, S., et al. (2000). MAPK upstream
kinase (MUK)-binding inhibitory protein, a negative regulator of MUK/dual
leucine zipper-bearing kinase/leucine zipper protein kinase. J. Biol. Chem.
275, 21247–21254.
Guelman, S., Suganuma, T., Florens, L., Swanson, S.K., Kiesecker, C.L.,
Kusch, T., Anderson, S., Yates, J.R., III, Washburn, M.P., Abmayr, S.M.,
et al. (2006). Host cell factor and an uncharacterized SANT domain protein
are stable components of ATAC, a novel dAda2A/dGcn5-containing histone
acetyltransferase complex in Drosophila. Mol. Cell. Biol. 26, 871–882.
Ham, J., Eilers, A., Whitfield, J., Neame, S.J., and Shah, B. (2000). c-Jun and
the transcriptional control of neuronal apoptosis. Biochem. Pharmacol. 60,
1015–1021.
Hu, S., Xie, Z., Onishi, A., Yu, X., Jiang, L., Lin, J., Rho, H.S., Woodard, C.,
Wang, H., Jeong, J.S., et al. (2009). Profiling the human protein-DNA interac-
tome reveals ERK2 as a transcriptional repressor of interferon signaling. Cell
139, 610–622.
Iyer, L.M., Burroughs, A.M., and Aravind, L. (2006). The prokaryotic anteced-
ents of the ubiquitin-signaling system and the early evolution of ubiquitin-like
b-grasp domains. Genome Biol. 7, R60.
Jasper, H., Benes, V., Schwager, C., Sauer, S., Clauder-Munster, S., Ansorge,
W., and Bohmann, D. (2001). The genomic response of the Drosophila embryo
to JNK signaling. Dev. Cell 1, 579–586.
Kayali, A.G., Austin, D.A., and Webster, N.J. (2000). Stimulation of MAPK
cascades by insulin and osmotic shock: lack of an involvement of p38
mitogen-activated protein kinase in glucose transport in 3T3-L1 adipocytes.
Diabetes 49, 1783–1793.
Kimura, A., Matsubara, K., and Horikoshi, M. (2005). A decade of histone acet-
ylation: marking eukaryotic chromosomes with specific codes. J. Biochem.
138, 647–662.
Kusch, T., Guelman, S., Abmayr, S.M., and Workman, J.L. (2003). Two
Drosophila Ada2 homologues function in different multiprotein complexes.
Mol. Cell. Biol. 23, 3305–3319.
Kyriakis, J.M., and Avruch, J. (2001). Mammalian mitogen-activated protein
kinase signal transduction pathways activated by stress and inflammation.
Physiol. Rev. 81, 807–869.
Lawrence, M.C., Shao, C., McGlynn, K., Naziruddin, B., Levy, M.F., and Cobb,
M.H. (2009). Multiple chromatin-bound protein kinases assemble factors that
regulate insulin gene transcription. Proc Natl Acad Sci USA 106, 22181–22186.
Lee, K.K., and Workman, J.L. (2007). Histone acetyltransferase complexes:
one size doesn’t fit all. Nat. Rev. Mol. Cell Biol. 8, 284–295.
Leimkuhler, S., Freuer, A., Araujo, J.A., Rajagopalan, K.V., and Mendel, R.R.
(2003). Mechanistic studies of human molybdopterin synthase reaction and
characterization of mutants identified in group B patients of molybdenum
cofactor deficiency. J. Biol. Chem. 278, 26127–26134.
Mas, G., de Nadal, E., Dechant, R., Rodriguez de la Concepcion, M.L., Logie,
C., Jimeno-Gonzalez, S., Chavez, S., Ammerer, G., and Posas, F. (2009).
Recruitment of a chromatin remodelling complex by the Hog1 MAP kinase
to stress genes. EMBO J. 28, 326–336.
Morrison, D.K., Murakami, M.S., and Cleghon, V. (2000). Protein kinases and
phosphatases in the Drosophila genome. J. Cell Biol. 150, F57–F62.
736 Cell 142, 726–736, September 3, 2010 ª2010 Elsevier Inc.
Pokholok, D.K., Zeitlinger, J., Hannett, N.M., Reynolds, D.B., and Young, R.A.
(2006). Activated signal transduction kinases frequently occupy target genes.
Science 313, 533–536.
Proft, M., Mas, G., de Nadal, E., Vendrell, A., Noriega, N., Struhl, K., and Posas,
F. (2006). The stress-activated Hog1 kinase is a selective transcriptional elon-
gation factor for genes responding to osmotic stress. Mol. Cell 23, 241–250.
Rozek, D., and Pfeifer, G.P. (1993). In vivo protein-DNA interactions at the c-jun
promoter: preformed complexes mediate the UV response. Mol. Cell. Biol. 13,
5490–5499.
Rozek, D., and Pfeifer, G.P. (1995). In vivo protein-DNA interactions at the c-jun
promoter in quiescent and serum-stimulated fibroblasts. J. Cell. Biochem. 57,
479–487.
Rudolph, M.J., Wuebbens, M.M., Rajagopalan, K.V., and Schindelin, H. (2001).
Crystal structure of molybdopterin synthase and its evolutionary relationship to
ubiquitin activation. Nat. Struct. Biol. 8, 42–46.
Schreiber, M., Kolbus, A., Piu, F., Szabowski, A., Mohle-Steinlein, U., Tian, J.,
Karin, M., Angel, P., and Wagner, E.F. (1999). Control of cell cycle progression
by c-Jun is p53 dependent. Genes Dev. 13, 607–619.
Schwarz, G., and Mendel, R.R. (2006). Molybdenum cofactor biosynthesis and
molybdenum enzymes. Annu. Rev. Plant Biol. 57, 623–647.
Su, Y.C., Treisman, J.E., and Skolnik, E.Y. (1998). The Drosophila Ste20-
related kinase misshapen is required for embryonic dorsal closure and acts
through a JNK MAPK module on an evolutionarily conserved signaling
pathway. Genes Dev. 12, 2371–2380.
Suganuma, T., Gutierrez, J.L., Li, B., Florens, L., Swanson, S.K., Washburn,
M.P., Abmayr, S.M., and Workman, J.L. (2008). ATAC is a double histone
acetyltransferase complex that stimulates nucleosome sliding. Nat. Struct.
Mol. Biol. 15, 364–372.
Takekawa, M., and Saito, H. (1998). A family of stress-inducible GADD45-like
proteins mediate activation of the stress-responsive MTK1/MEKK4 MAPKKK.
Cell 95, 521–530.
Thomson, S., Mahadevan, L.C., and Clayton, A.L. (1999). MAP kinase-
mediated signalling to nucleosomes and immediate-early gene induction.
Semin. Cell Dev. Biol. 10, 205–214.
Treisman, J.E., Ito, N., and Rubin, G.M. (1997). misshapen encodes a protein
kinase involved in cell shape control in Drosophila. Gene 186, 119–125.
Wang, Y.L., Faiola, F., Xu, M., Pan, S., and Martinez, E. (2008). Human ATAC is
a GCN5/PCAF-containing acetylase complex with a novel NC2-like histone
fold module that interacts with the TATA-binding protein. J. Biol. Chem. 283,
33808–33815.
Yang, D.D., Kuan, C.Y., Whitmarsh, A.J., Rincon, M., Zheng, T.S., Davis, R.J.,
Rakic, P., and Flavell, R.A. (1997). Absence of excitotoxicity-induced
apoptosis in the hippocampus of mice lacking the Jnk3 gene. Nature 389,
865–870.
Yang, S.H., Sharrocks, A.D., and Whitmarsh, A.J. (2003). Transcriptional
regulation by the MAP kinase signaling cascades. Gene 320, 3–21.
Zapater, M., Sohrmann, M., Peter, M., Posas, F., and de Nadal, E. (2007).
Selective requirement for SAGA in Hog1-mediated gene expression depend-
ing on the severity of the external osmostress conditions. Mol. Cell. Biol. 27,
3900–3910.
Zippo, A., Serafini, R., Rocchigiani, M., Pennacchini, S., Krepelova, A., and
Oliviero, S. (2009). Histone crosstalk between H3S10ph and H4K16ac
generates a histone code that mediates transcription elongation. Cell 138,
1122–1136.
A Bacterial mRNA Leader that Employs
Different Mechanisms to Sense
Disparate Intracellular Signals
Sun-Yang Park,1,3,4 Michael J. Cromie,1,2,3,5 Eun-Jin Lee,1,2,5 and Eduardo A. Groisman1,2,5,*
1Department of Molecular Microbiology
2Howard Hughes Medical Institute
Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8230, St. Louis, MO 63110, USA
3These authors contributed equally to this work
4Present address: Section of Microbial Pathogenesis, Yale University School of Medicine, Boyer Center for Molecular Medicine,
295 Congress Avenue, New Haven, CT 06536-0812, USA
5Present address: Section of Microbial Pathogenesis, Howard Hughes Medical Institute, Yale University School of Medicine,
Boyer Center for Molecular Medicine, 295 Congress Avenue, New Haven, CT 06536-0812, USA
*Correspondence: eduardo.groisman@yale.edu
DOI 10.1016/j.cell.2010.07.046
SUMMARY
Bacterial mRNAs often contain leader sequences
that respond to specific metabolites or ions by
altering expression of the associated downstream
protein-coding sequences. Here we report that the
leader RNA of the Mg2+ transporter gene mgtA of
Salmonella enterica, which was previously known
to function as a Mg2+-sensing riboswitch, harbors
an 18 codon proline-rich open reading frame—
termed mgtL—that permits intracellular proline to
regulate mgtA expression. Interfering with mgtL
translation by genetic, pharmacological, or environ-
mental means was observed to increase the mRNA
levels from the mgtA coding region. Substitution of
the mgtL proline codons by other codons abolished
the response to proline and to hyperosmotic stress
but not to Mg2+. Our findings show that mRNA leader
sequences can consist of complex regulatory
elements that utilize different mechanisms to sense
separate signals and mediate an appropriate cellular
response.
INTRODUCTION
The leader region (LR) of many bacterial mRNAs has the ability
to form mutually exclusive secondary structures that determine
whether transcription will continue into the adjacent coding
region (CR). Which secondary structure forms is governed
by the binding of specific metabolites, ions, tRNAs, or proteins
to the LR, by pausing of RNA polymerase at particular sequences
within the LR and by translation of short open reading frames
(ORFs) located in the leader RNA (Grundy and Henkin, 2006;
Henkin, 2008; Henkin and Yanofsky, 2002; Landick et al., 1996;
Merino and Yanofsky, 2005; Turnbough and Switzer, 2008).
These mRNA leaders typically control the expression of the
proteins that synthesize and/or transport the metabolites or
ions to which they respond. Whereas leaders known as ribos-
witches sense metabolites or ions directly (Henkin, 2008; Winkler
and Breaker, 2005), the presence of stretches of certain nucleo-
tides in a leader sequence or particular codons in a leader ORF
enables cells to probe the cellular levels of the corresponding
nucleotides and amino acids indirectly. Here we identify a leader
mRNA that controls gene expression by responding to Mg2+
directly as a riboswitch and to proline levels indirectly, via trans-
lation of a proline-codon-rich open reading frame.
The mRNA for the Mg2+ transporter gene mgtA from Salmo-
nella enterica serovar Typhimurium includes a 264 nucleotide
long leader sequence that functions as a Mg2+-responding ribos-
witch determining whether transcription proceeds into the mgtA
CR (Cromie et al., 2006). In low Mg2+, the mgtA leader RNA
adopts a conformation (i.e., stem loop C) that favors transcrip-
tion of the full-length mgtA mRNA (Cromie et al., 2006) (Figure 1)
and production of the MgtA protein (Cromie and Groisman,
2010). MgtA-mediated Mg2+ uptake restores cytoplasmic Mg2+
to prestress levels, which promotes a different conformation in
the mgtA leader RNA (i.e., stem loop B) that hinders transcription
elongation into the mgtA CR (Cromie et al., 2006) (Figure 1).
In addition to being regulated by a Mg2+-sensing riboswitch,
mgtA transcription is controlled at the initiation step by two
DNA binding proteins: PhoP, which activates transcription
when its cognate sensor PhoQ detects low extracytoplasmic
Mg2+ (Garcia Vescovi et al., 1996), acid pH (Prost et al., 2007),
or antimicrobial peptides (Bader et al., 2005); and Rob, the over-
expression of which stimulates mgtA transcription from a site
located downstream of the PhoP-dependent start site (Barchiesi
et al., 2008) (Figure 1). Thus, there is PhoP/PhoQ-dependent
production of the mgtA leader mRNA following bacterial growth
in media with <1 mM Mg2+ (Cromie et al., 2006) or pH 5.7 (Choi
et al., 2009); however, due to the riboswitch action, the mRNA
corresponding to the mgtA CR is observed primarily upon growth
in media containing very low (i.e., 10 mM) Mg2+ (Choi et al., 2009;
Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 737

Cromie and Groisman, 2010; Cromie et al., 2006). This raises the
possibility of additional signals and/or genetic elements acting on
the mgtA leader mRNA synthesized when Salmonella experi-
ences inducing conditions for PhoP/PhoQ and/or Rob to
promote transcription elongation into the mgtA CR.
Here we demonstrate that the mgtA LR harbors an ORF rich in
proline codons that enables Salmonella to control transcription
elongation into the mgtA CR in response to changes in cytosolic
proline levels. We establish that high osmolarity promotes tran-
scription of the mgtA CR in a fashion dependent on the proline
codons present in the identified ORF. Also, we show that the
low Mg2+ signal and the low proline signal act synergistically to
increase the mRNA levels for the mgtA CR. Our findings highlight
the complexity of RNA regulatory sequences and provide
a singular example of a leader RNA that utilizes two distinct
mechanisms to detect two different signals.
RESULTS
Identification of a Translated ORF in the mgtA LR
In a search for new factors that might regulate mgtA expression
by acting upon the mgtA leader mRNA, we introduced a plasmid
library made in the multicopy number vector pUC19 into strain
YS774 (see Table S1 available online), which harbors the
PhoP-independent plac1-6 promoter driving transcription of
a lac fusion in the mgtA CR from the normal (i.e., PhoP-depen-
dent) start site (Cromie et al., 2006). Ampicillin-resistant trans-
738 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.
Figure 1. Expression of the Salmonella
Mg2+ Transporter Gene mgtA Is Regulated
by the PhoP/PhoQ System, the Rob Protein,
and the mgtA LR
When the sensor PhoQ detects low extracytoplas-
mic Mg2+, acid pH, or antimicrobial peptides,
it promotes phosphorylation of the PhoP protein,
which binds to the mgtA promoter resulting in
transcription initiation. Rob promotes mgtA tran-
scription in response to a yet unidentified signal.
Transcription elongation into the mgtA CR is regu-
lated by the mgtA leader via a Mg2+-sensing
riboswitch and by translation of an 18 codon
proline-rich ORF designated mgtL. The mgtL
ribosome binding site is denoted by RBS adjacent
to a black line. Positions and sequences of stop
codon mutations in the strains used in the experi-
ments presented in Figure 4 are denoted below the
linear mgtL RNA sequence. High-Mg2+ and high-
proline conditions promote formation of stem
loop B, hindering transcription elongation into the
mgtA CR. Low Mg2+ and/or low proline favor
formation of stem loop C, resulting in transcription
of the mgtA CR. See also Figure S1.
formants recovered on X-Gal-containing
Luria-Bertani (LB) ampicillin agar plates
that were of a darker or lighter shade of
blue than those that received a plasmid
control were purified and their plasmid
DNA extracted and used to transform
strain YS809, a derivative of strain
YS774 with the same mgtA-lac transcriptional fusion. One of
the plasmids—designated pSL55—promoted higher levels of
mgtA-lac expression than the plasmid control (Figure 2A), like
it did in strain YS774, indicative that the observed phenotype
was plasmid linked. Plasmid pSL55 appears to promote mgtA-
lac expression specifically because it had no effect on the Lac
phenotype of strains harboring lac transcriptional fusions in the
CRs of the PhoP-dependent mgtC gene or the PhoP-indepen-
dent corA gene (data not shown). The mgtC and corA transcripts
also include long leaders (Lejona et al., 2003) (T. Latifi and E.A.G.,
unpublished results) and specify proteins participating in Mg2+
homeostasis (Blanc-Potard and Groisman, 1997; Maguire,
2006; Soncini et al., 1996).
Sequencing of the insert in pSL55 revealed the presence
of 1579 nucleotides corresponding to position 78347–79925 in
the chromosome of strain 14028s, which is a portion of the
3228 nucleotide long carB gene (see Table S2 for primers).
The insert operates in an orientation-dependent manner
because it did not alter mgtA-lac expression when cloned into
pUC18 where the opposite strand is transcribed (data not
shown). Although this suggested that pSL55 might promote
mgtA-lac expression by functioning as an antisense RNA for
the carB transcript, we ruled out this possibility because
pSL55 upregulated mgtA transcription even in a strain deleted
for the carB gene (Figure 2A). Given that a transcript correspond-
ing to the strand opposite the carB gene was not detected when
RNA was harvested from wild-type Salmonella grown in LB or in
 A ing at the ninth codon, produced high levels of b-galactosidase,
B of b-galactosidase following growth in low and high Mg2+
Figure 2. Plasmid pSL55 Promotes mgtA-lac Transcription by Inter-
fering with mgtL Translation
(A) b-galactosidase activity (Miller units) from a chromosomal mgtA-lac tran-
scriptional fusion driven by the plac1-6 promoter was determined in wild-type
(YS809), carB (SP31), and hfq (SP54) strains harboring plasmids pSL55,
pSL55-Mut, or a control plasmid. Bacteria were grown in LB medium for
4.5 hr. Shown are the mean and standard deviation (SD) from two independent
experiments.
(B) Nucleotide sequences corresponding to position 61–75 from the mgtA LR
including the mgtL ribosome binding sequence (RBS; horizontal black line) and
start codon (bold AUG) (top); the sequence of the pSL55-derived transcript
that is complementary to the mgtA leader mRNA (middle); and the sequence
of the pSL55-Mut-derived transcript that is no longer complementary to
the mgtA leader mRNA (bottom).
N-minimal media (data not shown), and that mgtA-lac was not
derepressed in the DcarB strain (Figure 2A), we concluded that
pSL55 upregulates mgtA-lac expression by a mechanism that
does not involve the carB gene.
We then examined the nucleotide sequence of the insert in
pSL55 and identified a 15 nucleotide segment that is comple-
mentary to nucleotides 61–75 in the mgtA leader RNA
(Figure 2B). Interestingly, this region corresponds to the potential
start codon and ribosome binding site for a previously unidenti-
fied 18 codon ORF—hereafter designated mgtL—that begins at
position 71 and ends at position 124 (Figure 1). The presence of
an ORF and ribosome binding site is conserved in the mgtA LRs
from Escherichia coli, Shigella flexneri, Klebsiella pneumoniae,
Citrobacter koseri, Enterobacter sp. 638, Dickeya zeae, Dickeya
dadantii, and Serratia proteamaculans (Figure 3A). The mgtA
leader sequences from all these species have the potential ability
to form the alternative stem loop structures B and C identified in
the Salmonella mgtA leader (Figure 1) (Cromie et al., 2006) (data
not shown).
We determined that mgtL is translated in vivo because wild-
type Salmonella carrying the medium-copy-number plasmid
pmgtL-0 lacZ, with the full mgtL coding sequence and its putative
ribosome binding site fused in frame to the E. coli lacZ gene start-
whereas an isogenic strain harboring a modified plasmid with
a stop codon after the 17th mgtL codon did not (Figure S1). (There
is a potential start codon at position 26–28 in the Salmonella mgtA
leader sequence that is in frame with the mgtL ORF and is
preceded by a possible ribosome binding site [Figure 1]. However,
the AUG at this position does not appear to be a true translation
start site because first, the sequence between nucleotides 26 and
70 is not conserved in the other examined species, and second,
wild-type Salmonella harboring plasmid pmgtLD2nt-0 lacZ, a
derivative of plasmid pmgtL-0 lacZ with two nucleotides deleted
at position 29–30, which creates a stop codon at position
79–81 for the ORF starting at position 26, still expressed high
levels of b-galactosidase [Figure S1].) The strains harboring plas-
mids pmgtL-0 lacZ and pmgtLD2nt-0 lacZ displayed similar levels
(Figure S1). These data indicate that mgtL translation is not regu-
lated by changes in the Mg2+ concentration in the media. More-
over, they reflect that in the pmgtL-0 lacZ and pmgtLD2nt-0 lacZ
plasmids, mgtL-lac transcription is initiated from a Mg2+-blind
promoter and the Mg2+-responding riboswitch is disrupted.
Inhibiting mgtL Translation Promotes Expression
of the mgtA CR
Plasmid pSL55 appears to upregulate mgtA-lac expression by
producing a trans-acting RNA that interferes with mgtL transla-
tion because first, the region of complementarity with the mgtA
leader corresponds to the ribosome binding site and start codon
of mgtL (Figure 2B). Second, inactivation of the hfq gene, which
codes for the RNA chaperone usually required for pairing
between trans-acting regulatory RNAs and their targets
(Brennan and Link, 2007), eliminated pSL55’s regulatory effect
(Figure 2A). And third, pSL55-Mut, a pSL55 derivative with nucle-
otide substitutions in the region of complementarity with the
mgtA LR (Figure 2B), could not upregulate mgtA expression
(Figure 2A).
If plasmid pSL55 derepresses mgtA-lac transcription by
inhibiting mgtL translation, we reasoned that disrupting mgtL
translation by other means might have the same effect. To test
this hypothesis, we introduced mutations in plasmid pYS1010,
which harbors the plac1-6 derivative of the lac promoter driving
transcription of the full-length mgtA LR fused to a promoterless
lacZ gene and confers Mg2+-regulated synthesis of b-galactosi-
dase (Cromie et al., 2006). Wild-type Salmonella harboring
pYS1010 derivatives with stop codons at positions 80–82,
89–91, or 98–100 (Figure 1) produced high levels of b-galactosi-
dase when grown in high Mg2+ (Figure 4A). These levels were 46-
to 89-fold higher than those displayed by the strain carrying the
original plasmid pYS1010.
The high expression levels exhibited by strains harboring
pYS1010 derivatives with stop codon mutations in mgtL are
probably due to lack of translation of the full-length mgtL
(as opposed to resulting from major alterations in the mgtA
leader RNA structure that lock the riboswitch in an ON state)
because first, a plasmid expressing the amber suppressor
supF restored normal levels of mgtA-lac transcription to the
strain harboring a pYS1010 derivative with an amber stop codon
at position 98–100 whereas the plasmid vector had no effect
Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 739
A

Figure 3. The Presence of an Open Reading Frame and Ribosome-Binding Site Is Conserved in the mgtA LR from Salmonella enterica,
Escherichia coli, Shigella flexneri, Citrobacter koseri, Klebsiella pneumoniae, Enterobacter sp. 638, Dickeya zeae, Dickeya dadantii, and
Serratia proteamaculans
(A) Alignment of the RNA sequences corresponding to the ribosome binding site and mgtL from the species listed above. Sequences in blue correspond to mgtL.
Asterisks correspond to nucleotides conserved in all species. The number of nucleotides shown in each line is indicated to the right of the nucleotide sequences.
(B) Alignment of the deduced amino acid sequences corresponding to mgtL from the species listed above. Sequences in yellow correspond to proline residues.
Asterisks correspond to positions conserved in all species. The number of sense codons in the mgtL ORFs is indicated to the right of the deduced amino acid
sequences.

(Figure 4C). Furthermore, the supF-promoted suppression was
specific because the supF-carrying plasmid did not reduce
expression in a strain carrying a pYS1010 derivative with an
opal stop codon at the same position (Figure 4C), nor did
it modify it in a strain with the original pYS1010 plasmid carrying
the wild-type mgtA leader sequence (Figure 4C). Second,
the structures of the wild-type and stop codon mutant mgtA
leaders were comparable when examined by in-line probing
(Figure S2A). And third, in vitro transcription assays demon-
strated that Mg2+ regulated transcription elongation beyond
the mgtA LR similarly in DNA templates corresponding to the
wild-type and stop codon mutant mgtA leaders (Figure S2B).
We next tested whether inhibiting mgtL translation by means
other than mutation of the mgtA LR affected the mRNA levels
for the mgtA CR. The protein synthesis inhibitor tetracycline
promoted an 28-fold increase in the mgtA CR mRNA 15 min
after its addition to wild-type Salmonella (Figure 4D). Thus, tetra-
cycline could overcome the transcriptional silencing of the mgtA
CR that normally takes place when wild-type Salmonella experi-
ences 500 mM Mg2+ (Cromie and Groisman, 2010; Cromie et al.,
2006). Tetracycline appears to affect the mRNA levels produced
from the mgtA CR specifically because there was little change in
the mRNA levels of the mgtA LR or the phoP CR, which were
examined as controls (Figure 4D). Addition of chloramphenicol
also increased the mRNA levels of the mgtA CR in wild-type
Salmonella (data not shown), implying that the derepression
promoted by protein synthesis inhibitors is not limited to a partic-
ular mechanism of action. Tetracycline exerts its effect on the
mgtA LR (as opposed to the mgtA promoter) because
it increased lacZ mRNA levels in wild-type Salmonella harboring
plasmid pYS1010 (data not shown). Furthermore, tetracycline
acts by inhibiting mgtL translation because it did not promote
a significant increase in lacZ mRNA levels in wild-type Salmo-
nella carrying the pYS1010 derivative with a stop codon at posi-
tion 98–100 (data not shown). Taken together, our results
indicate that interfering with mgtL translation by genetic or phar-
macological means increases the mRNA levels for the mgtA CR.
The mgtL-encoded peptide does not appear to act in trans
because a plasmid carrying mgtL DNA failed to restore normal
mgtA-lac expression in strains harboring pYS1010 derivatives
with stop codons at different positions in mgtL (Figure S2C).
This suggested that mgtL exerts its regulatory effect in cis and
raised the question as to the physiological signal controlling
mgtL translation.
Proline Limitation Enhances Transcription
of the mgtA CR
The Salmonella mgtL sequence includes four proline codons
(Figure 1), which is a disproportionately high frequency for an
18 codon ORF. The number and location of the proline codons
(at the third, fifth, seventh, and ninth positions) are largely
conserved in the nine examined species (Figures 3A and 3B).

740 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.

A B Figure 4. Translation of mgtL Governs
10000 10000 Transcription Elongation beyond the
mgtA LR
-galactosidase activity

-galactosidase activity
8000 8000
(A) b-galactosidase activity (Miller units) produced
(Miller units)

(Miller units)
6000 by wild-type Salmonella (14028s) harboring 6000
plasmid pYS1010 with the wild-type mgtA leader
4000 4000 (Cromie et al., 2006), or derivatives with stop
2000 2000
codons at different positions in mgtL and/or
mutations that hinder stem loop C formation:
0 0 pstop 80–82; pstop 89–91; pstop 98–100; pstop
2

12
2
-91

0
0

09

12
0C

5G
5G
-82

0C

5G
-91

5G
0C
09

0C
-82
-11

107–109; pstop 110–112; pstop 107–109 D110–

-11
-10
01

01

-1 0

0-1
7-1

0-1
7-1
12

12

14
14

12

14
12
89

14
80

89
80
S1

S1
0
10
98

98
11

-G

-G

11
0-C

0-C

-G
0-C
0-G

0-C
10

11
10
top

112; pYS1010-G120C; pstop 80–82-G120C;

top
pY

top
pY

top
1
top

top
10

-82

-82
top
top

to p
to p
-10
01

01

-10
01
ps
ps

09

ps
ps

09
0

pYS1010-C145G; and pstop 98–100-C145G.

ps

80

ps

80
S1

S1

S1
S1
ps
ps

ps
98
7-1

98
ps

7-1
top
pY

pY

to p

pY
pY
top

top
10

Bacteria were grown in N-minimal medium with

10
ps

ps
top

ps

top

ps
10 mM Mg2+ for 4 hr. Shown are the mean and
ps

ps
SD from at least three independent experiments.
C D (B) b-galactosidase activity produced by the
pstop 98-100 pstop 98-100 pYS1010
strains listed in (A). Bacteria were grown in
40
-galactosidase activity

5000
(amber) (opal) N-minimal medium with 10 mM Mg2+ for 4 hr.
Shown are the mean and SD from at least three
(Miller units)

Fold change

30
(T15 / T0 )

4000
independent experiments.
3000
20 (C) b-galactosidase activity produced by wild-type
2000 Salmonella (14028s) harboring plasmid pUHsupF
1000
10
or the plasmid vector pUH21-2lacIq, and
0
pYS1010 or its derivatives with UAG (pstop
0
98–100; amber) or UGA (pstop 98–100; opal)
or

or

ng
or

er

ng
F
F

F
up
up

up

ad
ct

ct
ct

di

di

stop codons at position 98–100 in the mgtA leader.

ve

ve
ve

co
le

co
Hs
Hs

Hs

oP
pU

pU
pU

gt

gt

Bacteria were grown in N-minimal medium with

ph
m

500 mM Mg2+ for 3 hr and with 1 mM IPTG for

E F 1 hr. Shown are the mean and SD from three inde-
12
2.0 pendent experiments.
10
(no proline / proline)

(no proline / proline)

(D) Fold change in the mRNA levels of the mgtA LR

1.5
Fold change

Fold change

8 and the mgtA and phoP CRs produced by wild-

6 1.0
type Salmonella (14028s) treated with the protein
synthesis inhibitor tetracycline (25 mg/ml). Expres-
4
0.5 sion levels of target genes were normalized to that
2
of the 16S ribosomal RNA rrs gene. Fold change
0 0 was calculated by dividing the mRNA levels from
ng
ng

ng
er

er

ng

samples taken 15 min after treatment with tetracy-

ad

ad
di
di

di

di
co
co

co
le

le

co

cline (T150 ) by that of samples taken before addi-

A

A
A

oP

oP
gt

gt
gt

gt
ph

tion of the antibiotic (T00 ). Shown are the mean

m

ph
m
m

and SD from three independent experiments.

(E) Fold change in the mRNA levels of the mgtA LR and the mgtA and phoP CRs produced by a proline auxotrophic strain (EG19886) grown in modified N-minimal
media with 500 mM Mg2+ in the presence of 1 mM proline for 1 hr, and then grown for 15 min in media containing or lacking proline. Expression levels of target
genes were normalized as described in (D). Fold change was calculated by dividing the mRNA levels of cells grown in the absence of proline by that of cells grown
in the presence of proline. Shown are the mean and SD from three independent experiments.
(F) Fold change in the mRNA levels of the mgtA LR and the mgtA and phoP CRs produced by wild-type Salmonella (14028s) grown and analyzed as described in
(D). Shown are the mean and SD from two independent experiments.
See also Figure S2.

Because the proline codons are located in the mgtL region where The mRNA level for the mgtA CR was 8-fold higher in organisms
introduction of stop codons heightens expression downstream of grown in the absence of proline than in those grown in its pres-
the mgtA leader (Figure 1), we hypothesized that a drop in the ence (Figure 4E). This effect was unique to the mgtA CR, as no
cytosolic proline levels might decrease the availability of proline- differences were detected in the mRNA levels corresponding
charged tRNAs, causing ribosome stalling at mgtL proline to the mgtA LR or the phoP CR (Figure 4E). Furthermore,
codons, leading to formation of stem loop C and an increase in it was specific for proline, as limitation for histidine, which does
the levels of the mgtA CR mRNA. By contrast, when cytosolic not have codons in the mgtL sequence (Figure 1), failed to
proline is abundant, coupling of mgtA leader transcription and increase the mRNA level for the mgtA CR (data not shown).
mgtL translation would be restored, thereby favoring formation Indeed, despite the presence of three arginine codons in mgtL
of stem loop B and reducing the production of mgtA CR mRNA. (Figure 1), arginine limitation did not promote an increase in the
To explore this hypothesis, we grew a proline auxotroph in the mRNA levels of the mgtA CR in an arginine auxotroph (data not
presence of 1 mM proline for 1 hr, washed the cells, split shown). Perhaps this reflects the location of the arginine codons,
the culture into two different media—one containing and one one of which is in stem loop C, and that the remaining arginine
lacking proline—and then harvested the mRNA 15 min later. codons may not be sufficient to mediate mgtA derepression.

Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 741

 A Proline limitation promoted a 2-fold increase in the mRNA
mgtA leader mgtA coding phoP coding
levels of the mgtA CR in wild-type (i.e., prototrophic) Salmonella
14
(Figure 4F). Although this is lower than the increase observed in
12 the proline auxotroph (Figure 4E), it is similar to the gene expres-
(no proline / proline)

10 sion changes promoted by branched-chain amino acids in the

Fold change

8
attenuation-regulated ilvGMEDA operon (Chen et al., 1991).
Moreover, it is higher than what has been reported for the atten-
6
uation-regulated trp operon, where starvation for tryptophan
4 does not alter trp expression in a tryptophan prototroph
2 (Yanofsky and Horn, 1994).
0
L+

L+

L+
Pr o3

7,9

7,9
o3
9

9
7 ,9
Pr o3
P r 5 ,7 ,

Pr 5,7,
Pr 5,7,

The mgtL Proline Codons Are Necessary

gt

gt

gt
Pr

Pr

Pr
,5,

,5 ,

,5,
m

m
o

o
o3

o3

o3
Pr

for the Response to Proline Limitation

We determined that proline limitation failed to promote an
B increase in the mRNA levels for the mgtA CR in a derivative of
80 mgtA leader mgtA coding phoP coding the proline auxotroph with all four mgtL proline codons replaced
by codons specifying other amino acids (Figure 5A). As found
(10 µ M Mg2+ / 500 µM Mg2+)

70

60 with the isogenic mgtL+ strain, proline limitation had little effect
on the mRNA levels for the mgtA leader and phoP CRs
Fold change

50

40
(Figure 5A). Importantly, the mutant still responded to changes
in the Mg2+ concentration (Figure 5B) (albeit not as well as the
30
isogenic strain with a wild-type mgtA leader), indicating that
20
the proline codon substitutions did not lock the mutant into an
10 inactive conformation. These data indicate that the mgtL proline
0 codons are essential for the response to proline limitation, and
L+

L+

L+
Pr o3

7,9

7 ,9
Pr o3
9

o 3 ,9

9
7 ,9
Pr o3

they suggest that this ability is distinct from the mgtA leader’s
Pr 5,7,

Pr 5,7,
,7
gt

gt

gt
Pr

Pr

Pr
,5 ,
,5,

,5,
o5
m

m
o

role as a Mg2+-sensing device.

o3

o3
Pr

We hypothesized that the proline codon at the third position of

mgtL might not be required for regulation of mgtA expression by
C proline because D. dadantii does not have a proline codon at this
80
mgtA leader mgtA coding phoP coding position (Figures 3A and 3B). Indeed, a Salmonella strain in which
70
the mgtL proline codon at the third position was replaced by
Relative mRNA levels

60 a leucine codon derepressed the mgtA CR mRNA when limited

50
40
30
20
10
0 by codons specifying other amino acids might be sufficient to
H H L L
+ - + -
Figure 5. The mgtA Leader Responds to Proline and Mg2+ Indepen-
dently and Requires the mgtL Proline Codons for Regulation of the
mgtA CR by Changes in the Proline Concentration
(A) Fold change in the mRNA levels of the mgtA LR and the mgtA and phoP
CRs produced by a proline auxotroph (SP1; mgtL+), or derivatives in which
the mgtL proline codon at the third position was substituted (SP2; Pro3), the
last three mgtL proline codons were substituted (SP61; Pro5,7,9), or in which
all mgtL proline codons were substituted (SP8; Pro3,5,7,9). Bacteria were
grown and mRNA was analyzed as described in Figure 4E. Shown are the
mean and SD from three independent experiments.
(B) Fold change in the mRNA levels of the mgtA LR and the mgtA and phoP
CRs produced by the strains described in (A). Bacteria were grown in modified
N-minimal medium with 10 mM or 500 mM Mg2+ for 3.5 hr. Expression levels of
target genes were normalized as described in (A). Fold change was calculated
by dividing the mRNA levels of cells grown in N-minimal medium with 10 mM
Mg2+ by those present in cells grown in 500 mM Mg2+. Shown are the mean
and SD from three independent experiments.
for proline, like the strain with the wild-type mgtL (Figure 5A).
However, this mutant still responded to changes in the Mg2+
concentration, but to a smaller degree than the mutant
substituted in all four mgtL proline codons (Figure 5B).
The results presented above suggested that replacement of
the mgtL proline codons at the fifth, seventh, and ninth positions
H H L L H H L L [Mg2+]
+ - + - + - + - Proline eliminate the response to proline. As predicted, there was little
derepression of the mgtA CR mRNA upon proline limitation in
a derivative of the proline auxotroph with the last three mgtL
proline codons substituted by the same codons as in the mutant
with substitutions in all four mgtL proline codons (Figure 5A). As
with the other isogenic strains, proline limitation did not affect the
mRNA levels of the mgtA leader and phoP CRs (Figure 5A).
The response to Mg2+ of the mgtL derivative with substitutions
in the last three proline codons was intermediate to that
(C) Relative mRNA levels of the mgtA LR and the mgtA and phoP CRs
produced by a proline auxotroph (EG19886) that experienced 500 mM Mg2+
and 1 mM proline, 500 mM Mg2+ and no proline, 5 mM Mg2+ and 1 mM proline,
or 5 mM Mg2+ and no proline for 15 min. Relative mRNA levels were calculated
by dividing the mRNA levels of cells grown in a given condition by those
present in cells grown in 500 mM Mg2+ and 1 mM proline. Shown are the means
and SD from three independent experiments.

742 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.

exhibited by the one with the third proline codon substituted and A mgtL+
the one with all four proline codons substituted (Figure 5B).
mgtA leader mgtA coding phoP coding
7
Synergism between the Low Mg2+ and the Proline
6

Relative mRNA levels

Limitation Signals
To examine whether direct Mg2+ sensing by the mgtA leader RNA 5

affects indirect proline sensing by the mgtL ORF and vice versa, 4
we grew the mgtL+ proline auxotrophic strain in the presence of 3
500 mM Mg2+ and 1 mM proline for 1 hr, washed the cells, and split
2
the culture into four different media containing either 500 mM
Mg2+ and 1 mM proline, 500 mM Mg2+ and no proline, 5 mM 1
Mg2+ and 1 mM proline, or 5 mM Mg2+ and no proline. Fifteen 0
minutes later, we determined the mRNA levels corresponding - G N NG NP - G N NG NP - G N NG NP
to the mgtA and phoP CRs as well as to the mgtA LR. We found
that the mRNA levels corresponding to the mgtA CR were higher
B
in cells experiencing both low Mg2+ and low proline than in cells Pro3,5,7,9
limited only for Mg2+ or for proline (Figure 5C). As expected, the mgtA leader mgtA coding phoP coding
mRNA levels corresponding to the mgtA LR and the phoP CR 7

were higher in cells experiencing 5 mM Mg2+ than in those
exposed to 500 mM Mg2+ (Figure 5C), reflecting that the amount
of activated PhoP protein increases as the Mg2+ concentration
in the media decreases (Shin et al., 2006). Our results support
the notion that Mg2+ and proline are sensed independently
by the mgtA leader RNA. Furthermore, they suggest synergism
between the two inducing signals because the mRNA levels for
the mgtA CR were higher in cells experiencing both low Mg2+
and low proline than the sum of the mRNA levels produced by
cells experiencing only low Mg2+ or low proline (Figure 5C).
In agreement with this notion, wild-type Salmonella harboring
plasmid pYS1010 derivatives with early stop codons in the
mgtL sequence produced more b-galactosidase when grown in
low Mg2+ than the sum of the b-galactosidase activities produced
by the same strains grown in high Mg2+ plus the b-galactosidase
activity produced by wild-type Salmonella carrying pYS1010
grown in low Mg2+ (Figures 4A and 4B).
Hyperosmotic Shock Promotes an Increase in the mgtA
CR mRNA Levels
Proline plays two major functions in bacterial cells: it is a compo-
nent of proteins and it can function as an osmoprotectant
(Csonka and Leisinger, 2007). Thus, we hypothesized that
when bacteria experience hyperosmotic shock, the increased
requirement for proline in osmoprotection might decrease its
availability to charge proline tRNAs. This could potentially lead
to ribosome stalling at the mgtL proline codons and result in
derepression of the mgtA CR. We tested this hypothesis by
comparing the mRNA levels produced by wild-type Salmonella
experiencing hyperosmotic shock in the presence or absence
of osmoprotectants.
We determined that the mRNA levels corresponding to the
mgtA CR were 6-fold higher when Salmonella experienced
500 mM Mg2+ + 0.3 M NaCl for 1 hr than in organisms grown in
500 mM Mg2+ (Figure 6A). By contrast, the mRNA levels for the
mgtA LR and the phoP CR were similar under the two growth
conditions (Figure 6A), indicative that the induction of the mgtA
CR mRNA levels promoted by high osmolarity is not mediated
by the PhoP-dependent mgtA promoter. Addition of the osmo-
protectant glycine betaine together with NaCl compromised
Relative mRNA levels
6
5
4
3
2
1
0
- G N NG NP - G N NG NP - G N NG NP
Figure 6. Hyperosmotic Shock Promotes Transcription of the mgtA
CR in a Process that Requires the mgtL Proline Codons
Relative mRNA levels of the mgtA LR and the mgtA and phoP CRs produced
by wild-type Salmonella (YS957) (A), or a derivative in which all mgtL proline
codons were substituted (EG19870) (B). Bacteria were grown for 1 hr in modi-
fied N-minimal medium without casamino acids containing 500 mM Mg2+ and
either no additional supplements (-), 1 mM glycine betaine (G), 0.3 M NaCl (N),
0.3 M NaCl and 1 mM glycine betaine (NG), or 0.3 M NaCl and 1 mM proline
(NP). Relative mRNA levels were calculated by dividing the mRNA levels of
cells grown under the specified condition by the mRNA levels present in cells
grown in 500 mM Mg2+ (i.e., with no additional supplement). Shown are the
mean and SD from three independent experiments.
the induction of the mgtA CR mRNA promoted by NaCl but
had negligible effects when added in the absence of NaCl
(Figure 6A). Proline had a similar (albeit not as strong) effect as
glycine betaine (Figure 6A), presumably because it is not as
effective as glycine betaine in osmoprotection (Cayley et al.,
1992). The increase in the mgtA CR mRNA levels provoked by
hyperosmotic shock requires an intact mgtL ORF, because it
was not observed in an isogenic strain substituted in all four
mgtL proline codons (Figure 6B). Cumulatively, these data
demonstrate that hyperosmotic shock promotes transcription
of the mgtA CR in an mgtL-dependent manner.
Formation of Stem Loop C Is Necessary
for Transcription of the mgtA CR
Because transcription and translation are coupled in bacteria,
when cytosolic proline is abundant, a ribosome translating the
complete mgtL sequence is likely to occlude the left arm of

Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 743

 A mgtA leader region B trp leader region Figure 7. Regulation of Transcription Elon-
gation into the mgtA and trp CRs by ORFs
Located within Their Respective LRs
Antiterminator
(A) Regulation of transcription elongation by the
Stem-loop C mgtA leader RNA. Top: when proline levels are
limiting in the cytosol, a ribosome translating
Ribosome is stalled at Ribosome Ribosome mgtL stalls at proline codons, which favors forma-
STOP tion of stem loop C and results in transcription of
proline/tryptophan
AUG
G AUG STOP
the mgtA CR. Middle: when proline levels are not
codons RBS RBS UUUUUUU
limiting in the cytosol, a ribosome can translate
the complete mgtL ORF, which favors formation
ON state ON state of stem loop B and hinders transcription elonga-
tion into the mgtA CR by an unknown mechanism.
Stem-loop B Bottom: a ribosome loads but cannot translate
a mutant mgtA leader with nucleotide substitutions
Terminator
in the mgtL start codon, which favors formation of
stem loop C and results in transcription of the
Leader ORF is mgtA CR.
AUG STOP AUG STOP
completely translated RBS RBS
(B) Regulation of transcription elongation by the trp
UUUUUUU
leader RNA. Top: when tryptophan levels are limiting
in the cytosol, a ribosome translating trpL stalls at
OFF OFF state
tryptophan codons, which favors formation of an
state
antiterminator structure and results in transcription
of the trp operon. Middle: when tryptophan levels
Anti-antiterminator
Stem-loop C
are not limiting in the cytosol, a ribosome can trans-
STOP
Terminator
late the complete trpL ORF, which favors formation
of an intrinsic transcription terminator and there is no
Ribosome loads but transcription of the trp operon. Bottom: a ribosome
STOP
there is no mgtL or loads but cannot translate a mutant trp leader with
AUG
AU AU
AUG
trpL translation RBS RBS UUUUUUU nucleotide substitutions in the trpL start codon,
which favors formation of both the anti-antitermina-
ON state OFF state tor and terminator structures and there is no tran-
scription of the trp operon. Note the different posi-
tion of the mgtL and trpL ORFs relative to the
sequences forming stem loop C and the anti-antiter-
minator structures, respectively, which determines
the different phenotypes resulting from mutation of
the mgtL and trpL start codons.

stem loop C (Figure 1 and Figure 7A). This would favor formation
of stem loop B, which has been shown to hamper transcription
elongation beyond the mgtA LR (Cromie et al., 2006).
By contrast, conditions that reduce the levels of free cytosolic
proline would promote ribosome stalling at the mgtL proline
codons, thereby advancing formation of stem loop C and result-
ing in transcription elongation into the mgtA CR (Figure 1 and
Figure 7A). Therefore, the position that a translating ribosome
reaches in the mgtL ORF should determine whether transcription
continues into the mgtA CR (Figure 1 and Figure 7A).
We tested our model by investigating the phenotype of wild-
type Salmonella harboring pYS1010 derivatives with stop
codons at different positions within mgtL. A derivative with
a stop codon at position 110–112, which is only six nucleotides
upstream of the left arm of stem loop C (Figure 1), produced
very low levels of b-galactosidase when grown in 10 mM Mg2+
(Figure 4A), like the isogenic strain with plasmid pYS1010
harboring the wild-type mgtA leader (Figure 4A). By contrast,
there were high levels of b-galactosidase in a pYS1010 derivative
with a stop codon at position 107–109 (Figure 4A), which is nine
nucleotides from the left arm of stem loop C (Figure 1), similar to
the behavior of strains with stop codons at positions 80–82,
89–91, or 98–100 (Figure 4A). Furthermore, the pYS1010 deriva-
tive with a stop codon at position 107–109 no longer conferred
high levels of b-galactosidase upon wild-type Salmonella when
nucleotides 110–112 were deleted (Figure 4A), which brought
the mgtL stop codon only six nucleotides away from the left
arm of stem loop C. Cumulatively, these results indicate that
the distance between the ribosome translating mgtL and the
nucleotides forming the left arm of stem loop C (as opposed to
the size of the translated mgtL product) is critical for mgtL-medi-
ated gene control.
If the lack of translation of the full-length mgtL stimulates tran-
scription elongation beyond the mgtA LR by favoring formation of
stem loop C (Figure 1 and Figure 7A), hindering formation of stem
loop C should abolish this stimulation. As predicted, the G120C
and C145G substitutions in the mgtA leader (Figure 1), which
were previously shown to impede formation of stem loop C
(Cromie et al., 2006), thwarted derepression in constructs
harboring stop codons at either of two mgtL positions (Figure 4A).
DISCUSSION
The LR of many mRNAs can respond to specific nutritional
and/or physical signals by modifying expression of the associ-
ated downstream coding sequences. Some of these leader

744 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.

RNAs rely on the same mechanism to detect more than one
signal whereas others utilize different means of controlling
gene expression in response to one signal. For example, two
riboswitches located in tandem, one responding to S-adenosyl-
methionine and the other to coenzyme B12, control transcription
of the metE gene in Bacillus clausii (Sudarsan et al., 2006). In
other cases, a cis-acting riboswitch can also function as
a trans-acting regulatory RNA to modulate the expression of
genes located somewhere else in the genome (Loh et al.,
2009). The mgtA leader described here constitutes a singular
example of an mRNA leader that utilizes different mechanisms
to sense different signals.
We determined that the mRNA leader corresponding to the
Mg2+ transporter gene mgtA harbors a translated ORF rich in
proline codons designated mgtL (Figure 1; Figure S1), which is
conserved in other bacterial species (Figures 3A and 3B) and
enables Salmonella to regulate transcription elongation into the
mgtA CR in response to the levels of cytoplasmic proline.
Whereas the mgtA leader RNA senses Mg2+ directly (Cromie
et al., 2006), it monitors proline levels in an indirect fashion, via
mgtL translation (Figure 4A), using an attenuation-like mecha-
nism (Landick et al., 1996; Henkin and Yanofsky, 2002).
The dual sensing ability of the mgtA leader allows Salmonella
to transcribe the mgtA CR not only when Mg2+ levels drop below
a certain threshold (Cromie and Groisman, 2010; Cromie et al.,
2006) but also under conditions resulting in a decrease in cyto-
solic proline levels (Figures 4E and 4F).
The discovery of a translatable ORF in the mgtA leader
(Figure 1; Figure S1) suggests a possible explanation for the
recent observation that the mRNA levels corresponding to the
50 -most 127 nucleotides of the mgtA leader are much higher
than those for the 30 -most 94 nucleotides of the mgtA leader
when Salmonella is grown in high-Mg2+ media (Spinelli et al.,
2008). This is because the RNA-processing enzyme RNase E,
which has been implicated in mgtA mRNA degradation during
growth in high Mg2+ (Spinelli et al., 2008), can target nascent
transcripts as well as complete messages (Hammarlof and
Hughes, 2008). Therefore, the ribosome translating mgtL, which
extends to position 124 in the mgtA leader (Figure 1), could
protect the 50 region of the mgtA leader from RNase E action, re-
sulting in higher levels of this portion of the RNA.
mgtL Translation Regulates Transcription
of the mgtA CR
We determined that interfering with mgtL translation increased
the mRNA levels for the mgtA CR (Figures 4A and 4D–4F). This
appears to result from the formation of a particular secondary
structure (i.e., stem loop C) in the mgtA leader (Figure 1 and
Figure 7A) that promotes transcription elongation into the mgtA
CR, as nucleotide substitutions impeding formation of stem
loop C (Cromie et al., 2006) prevented derepression provoked
by stop codon mutations in mgtL (Figure 4A). Because transcrip-
tion and translation are coupled in bacteria and because the
ribosome occupies 30 nucleotides, covering 12–15 nucleo-
tides from the P site (Laursen et al., 2005), the position that
a translating ribosome reaches in mgtL relative to the sequences
that make up stem loop C would determine whether transcription
continues into the mgtA CR (Figure 1 and Figure 7A). If the ribo-
some is too close to the sequences that make up stem loop C,
then the alternative stem loop B would form (Figure 1 and
Figure 7A), and transcription would not continue into the mgtA
CR. For instance, stop codons at positions 80–82, 89–91,
98–100, or 107–109 in the mgtA leader resulted in derepression
whereas a stop codon at position 110–112 did not (Figure 4A).
Yet, the stop codon at position 107–109 no longer derepressed
expression if nucleotides 110–112 were also deleted (Figure 4A),
which moved the mgtL stop codon only six nucleotides away
from the left arm of stem loop C (Figure 1).
It was recently reported that a chromosomal C98T mutation in
the leader RNA causes constitutive high expression of the mgtA
CR (O’Connor et al., 2009). In the original description of this
mutation, it was not clear why this nucleotide substitution should
alter mgtA expression, but we now see that it created a stop
codon in mgtL that is identical to the mutation we had previously
constructed, which resulted in constitutive high expression of
mgtA (Figures 4A and 4B).
The mgtL proline codons are located at positions 77–79,
83–85, 89–91, and 95–97 (Figure 1), which correspond to the
region where placing a stop codon results in derepression (Fig-
ure 4A). Therefore, we propose that when cytosolic proline levels
are low, there would be less proline-charged tRNAs, resulting in
ribosome stalling at mgtL proline codons. This would favor
formation of stem loop C and transcription elongation into the
mgtA CR (Figure 7A). In contrast, when cytosolic proline levels
are high, the ribosome would translate the complete mgtL
sequence, thereby occluding the left arm of stem loop C and,
in this manner, favor formation of stem-loop B (Figure 7A) and
hinder transcription of the mgtA CR. Interestingly, despite
utilizing different mechanisms to sense proline and Mg2+, the
mgtA leader relies on formation of the same structure—stem
loop C—to promote transcription elongation into the mgtA CR,
because mutations that hinder formation of stem loop C pre-
vented the derepression caused by stop codons in mgtL
(Figure 4A) and by low Mg2+ (Figure 4B) (Cromie et al., 2006).
The regulation of mgtA by changes in the cytosolic proline
concentration is reminiscent of class I transcription attenuation,
where certain amino acid biosynthetic operons are regulated by
translation of a short ORF in the LR. These ORFs are rich in
codons specifying the amino acid(s) synthesized by the enzymes
encoded in the operon (Landick et al., 1996). Yet, there are signif-
icant differences between the two systems. First, mgtA codes for
a transporter rather than for an amino acid biosynthetic enzyme.
This provides experimental support for the proposal, based on
genomic analysis, that classical transcriptional attenuators
might direct the expression of operons mediating functions other
than nutrient biosynthesis (Merino and Yanofsky, 2005). Second,
the position of the leader ORF relative to the potential RNA
secondary structures that can be adopted differs between the
mgtA leader (Figure 7A) and the leaders of attenuation-regulated
amino acid biosynthetic operons such as the trp operon
(Figure 7B). This provides a plausible explanation for the oppo-
site effects that mutation of the leader ORF start codon has in
these two classes of leaders. It results in superattenuation for
the trp operon (Landick and Yanofsky, 1987) because there
would be no ribosomes to stall at Trp codons, which would result
in formation of a terminator structure (Figure 7B). However,
Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 745
it gives rise to derepression of the mgtA CR (our unpublished
results), because formation of stem loop C would be favored
when mgtL is not translated (Figure 7A). Third, an intrinsic
terminator structure forms in the trp leader when the levels of
charged tRNATrp are high and the ribosome does not pause at
the consecutive Trp codons in the leader ORF (Landick and
Yanofsky, 1987) (Figure 7B). This is in contrast to the absence
of an intrinsic transcription terminator in the mgtA leader, where
sequences located downstream of the stem loop B structure
(Figure 1) are necessary to control transcription elongation into
the mgtA CR (Cromie et al., 2006) by an unknown mechanism.
And fourth, apart from utilizing a classical transcription attenua-
tion-like mechanism to mediate the response to proline, the
mgtA leader senses Mg2+ directly functioning as a riboswitch
(Cromie et al., 2006). This raises the intriguing possibility of clas-
sical transcriptional attenuators also sensing physical and/or
chemical signals directly.
2+
High Osmolarity Enhances the mRNA Levels of the Mg
Transporter Gene mgtA
Why does proline limitation promote transcription of the Mg2+
transporter gene mgtA? We contemplated the possibility of the
MgtA protein being a conduit for proline whereby cells experi-
encing low proline would derepress expression of a proline
import system. However, we found that [14C]proline uptake
was similar between wild-type and mgtA Salmonella, and
between a triple mutant lacking all known proline uptake
systems (ProP, ProU, and PutP) (Csonka and Leisinger, 2007)
and the isogenic mgtA derivative (our unpublished results). We
also considered that MgtA-mediated Mg2+ uptake might be
necessary for proline biosynthesis, because the activity of the
proline biosynthetic enzyme glutamate 5-kinase is Mg2+ depen-
dent (Perez-Arellano et al., 2005). Yet, inactivation of the mgtA
gene did not render Salmonella auxotrophic for proline (our
unpublished results).
We determined that hyperosmotic shock promoted an
increase in the mRNA levels corresponding to the mgtA CR
(Figure 6A). This appears to result from a transient decrease in
the levels of charged-proline tRNAs taking place when proline
is used as an osmoprotectant, as there would be less cytosolic
proline available to charge the tRNAPros. Indeed, the increase
in mgtA mRNA levels was severely compromised if the osmotic
shock was alleviated by the osmoprotectant glycine betaine,
which did not influence mgtA expression when added in the
absence of NaCl (Figure 6A). The osmotic shock-promoted
increase in the mRNA levels for the mgtA CR requires the mgtL
proline codons, because a mutant with substitutions in all four
proline codons in mgtL lost the ability to derepress the mgtA
CR mRNA in response to hyperosmotic shock (Figure 6B).
Why does hyperosmotic shock induce transcription of the
Mg2+ transporter gene mgtA when Salmonella also harbors
the constitutively expressed Mg2+ transporter CorA? On the
one hand, high osmolarity promotes excretion of putrescine
(Schiller et al., 2000), which constitutes the major organic diva-
lent cation in bacterial cells and is normally bound to nucleic
acids (Wortham et al., 2007). This might create an increased
need for Mg2+ in order to neutralize the negatively charged
DNA and RNA, and to stabilize structures such as membranes
746 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.
and ribosomes, which could be affected during hyperosmotic
stress. On the other hand, CorA-mediated Mg2+ uptake may
be compromised because it is driven by the membrane potential
(Froschauer et al., 2004), which decreases in cells experiencing
high osmolarity (Csonka, 1989). Yet, this stress would not affect
Mg2+ uptake by the P-type ATPase MgtA protein because it is
energized by ATP hydrolysis (Maguire, 1992).
That proline limitation promotes expression of the MgtA Mg2+
transporter suggests that there is a physiological connection
between proline and Mg2+. In agreement with this notion, tran-
scription of the proline transporter gene proP is promoted during
growth in low Mg2+ by the PhoP protein (Eguchi et al., 2004),
which directs transcription initiation of the mgtA gene (Garcia
Vescovi et al., 1996).
EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, Primers, and Growth Conditions
Bacterial strains and plasmids used in this study are listed in Table S1. Primers
used in this study are listed in Table S2. Unless otherwise stated, bacteria were
grown at 37 C in LB broth or in N-minimal medium (pH 7.4) (Snavely et al.,
1991) supplemented with 0.1% casamino acids, 38 mM glycerol, and the indi-
cated concentration of MgCl2.
Construction of Strains with Chromosomal Mutations
Deletion strains were constructed using the one-step inactivation method
(Datsenko and Wanner, 2000). See Extended Experimental Procedures for
detailed protocols.
Treatment with Bacteriostatic Protein Synthesis Inhibitors
An overnight bacterial culture grown in N-minimal medium with 10 mM MgCl2
was used to inoculate 125 ml flasks containing 10 ml of the same medium
(1:50 dilution) and grown for 3 hr at 37 C with shaking. To promote PhoP-
dependent mgtA transcription initiation, bacteria were washed and resus-
pended in 10 ml of N-minimal medium with 500 mM MgCl2. Following growth
for 1 hr, a 500 ml aliquot was removed to determine the mRNA levels before
treatment. Subsequently, tetracycline (to 25 mg/ml final concentration) or
chloramphenicol (to 200 mg/ml final concentration) was added and bacteria
were grown for 15 min, when they were harvested and their RNA was isolated
for analysis.
Effect of Proline Limitation
Bacteria were grown overnight in modified N-minimal medium containing
0.2% glucose, 10 mM MgCl2, and 1 mM proline. The overnight culture was
used to inoculate 125 ml flasks containing 10 ml of the same medium (1:50 dilu-
tion) and grown for 3 hr at 37 C with shaking. The harvested bacteria were
washed in the modified N-minimal medium containing 500 mM MgCl2, and
grown in 10 ml of the same medium with 1 mM proline for 1 hr. After removing
a 250 ml aliquot to determine the mRNA levels before treatment, the harvested
bacteria were washed with the modified N-minimal medium containing 500 mM
MgCl2, and suspended in 100 ml of the same medium. To see the effect of
proline limitation, the suspended bacterial cells were split into two flasks con-
taining the modified N-minimal media with 500 mM MgCl2 and a mixture of
each of the 19 essential amino acids (50 mM final concentration for each of
them): one flask contained 1 mM proline and the other had no proline added.
Bacteria were grown for 15 min, when they were harvested and their RNA was
isolated for analysis.
Testing Effect of Growth in High versus Low Mg2+
Bacteria were grown overnight in modified N-minimal medium containing
0.2% glucose, 10 mM MgCl2, and 1 mM proline. The harvested bacteria
were washed in the modified N-minimal medium containing 500 mM MgCl2,
and this culture was used to inoculate 125 ml flasks containing 10 ml of the
modified N-minimal medium containing either 500 mM MgCl2 or 10 mM
MgCl2 (1:50 dilution) and grown for 3.5 hr at 37 C with shaking. The bacteria
were then harvested and their RNA was isolated for analysis.
2+
Testing Effect of Mg and/or Proline Limitation Treatment
Bacteria were grown overnight in modified N-minimal medium containing
0.2% glucose, 10 mM MgCl2, and 1 mM proline. The overnight culture
was used to inoculate 125 ml flasks containing 20 ml of the same medium
(1:50 dilution) and grown for 3 hr at 37 C with shaking. The harvested bacteria
were washed in the modified N-minimal medium containing 500 mM MgCl2,
and grown in 20 ml of the same medium with 1 mM proline for 1 hr. The
harvested bacteria were washed with the modified N-minimal medium
containing 500 mM MgCl2, and suspended in 200 ml of the same medium.
This cell suspension was used to inoculate 125 ml flasks containing 5 ml of
media containing either 500 mM MgCl2 and 1 mM proline, 500 mM MgCl2
and no proline, no MgCl2 and 1 mM proline, or no MgCl2 and no proline
(1:100 dilution). Bacteria were grown for 15 min, when they were harvested
and their RNA was isolated for analysis.
Testing Effect of Hyperosmotic Shock
Bacteria were grown overnight in modified N-minimal medium containing
0.2% glucose, 0.1% casamino acids, and 10 mM MgCl2. The overnight culture
was used to inoculate 125 ml flasks containing 20 ml of the same medium
(1:50 dilution) and grown for 3.5 hr at 37 C with shaking. The harvested
bacteria were washed in modified N-minimal medium containing 500 mM
MgCl2 without casamino acids and suspended in 250 ml of the same medium.
This cell suspension was used to inoculate 125 ml flasks containing 5 ml of the
modified N-minimal medium without casamino acids containing 500 mM MgCl2
and either no additional supplements, 1 mM glycine betaine, 0.3 M NaCl, 0.3 M
NaCl and 1 mM glycine betaine, or 0.3 M NaCl and 1 mM proline. Bacteria were
grown for 1 hr at 37 C with shaking, when they were harvested and their RNA
was isolated for analysis.
RNA Isolation and Determination of Transcript Levels
Total RNA was extracted using an RNeasy Mini Kit (QIAGEN). cDNA was
synthesized using TaqMan reverse-transcription reagents (Applied Biosys-
tems) following the manufacturer’s instructions. Quantification of transcripts
was performed by real-time PCR using Fast SYBR Green Master Mix (Applied
Biosystems) in a 7500 Sequence Detection System (Applied Biosystems). A list
of primers used for quantitative RT-PCR is presented in Extended Experi-
mental Procedures. See Extended Experimental Procedures for a detailed
protocol.
b-Galactosidase Assays
b-galactosidase activity was determined as described (Cromie et al., 2006).
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, two
figures, and two tables and can be found with this article online at doi:10.
1016/j.cell.2010.07.046.
ACKNOWLEDGMENTS
We thank Laszlo Csonka, Kerry Hollands, Robert Landick, and Charles
Yanofsky for comments on the manuscript; Henry Huang for the plasmid
harboring supF; and John Roth for strains. This work was supported, in part,
by grant AI49561 from the NIH to E.A.G., who is an Investigator of the Howard
Hughes Medical Institute.
Received: March 19, 2010
Revised: May 24, 2010
Accepted: July 14, 2010
Published: September 2, 2010
REFERENCES
Bader, M.W., Sanowar, S., Daley, M.E., Schneider, A.R., Cho, U., Xu, W.,
Klevit, R.E., Le Moual, H., and Miller, S.I. (2005). Recognition of antimicrobial
peptides by a bacterial sensor kinase. Cell 122, 461–472.
Barchiesi, J., Castelli, M.E., Soncini, F.C., and Vescovi, E.G. (2008). mgtA
expression is induced by rob overexpression and mediates a Salmonella en-
terica resistance phenotype. J. Bacteriol. 190, 4951–4958.
Blanc-Potard, A.B., and Groisman, E.A. (1997). The Salmonella selC locus
contains a pathogenicity island mediating intramacrophage survival. EMBO
J. 16, 5376–5385.
Brennan, R.G., and Link, T.M. (2007). Hfq structure, function and ligand
binding. Curr. Opin. Microbiol. 10, 125–133.
Cayley, S., Lewis, B.A., and Record, M.T., Jr. (1992). Origins of the osmopro-
tective properties of betaine and proline in Escherichia coli K-12. J. Bacteriol.
174, 1586–1595.
Chen, J.W., Bennett, D.C., and Umbarger, H.E. (1991). Specificity of attenua-
tion control in the ilvGMEDA operon of Escherichia coli K-12. J. Bacteriol. 173,
2328–2340.
Choi, E., Groisman, E.A., and Shin, D. (2009). Activated by different signals, the
PhoP/PhoQ two-component system differentially regulates metal uptake.
J. Bacteriol. 191, 7174–7181.
Cromie, M.J., and Groisman, E.A. (2010). Promoter and riboswitch control of
the Mg2+ transporter MgtA from Salmonella enterica. J. Bacteriol. 192,
604–607.
Cromie, M.J., Shi, Y., Latifi, T., and Groisman, E.A. (2006). An RNA sensor for
intracellular Mg2+. Cell 125, 71–84.
Csonka, L.N. (1989). Physiological and genetic responses of bacteria to
osmotic stress. Microbiol. Rev. 53, 121–147.
Csonka, L.N., and Leisinger, T. (2007). Biosynthesis of proline. In EcoSal—Es-
cherichia coli and Salmonella: Cellular and Molecular Biology, G. Cohen, ed.
(Washington, DC: ASM Press), http://www.ecosal.org.
Datsenko, K.A., and Wanner, B.L. (2000). One-step inactivation of chromo-
somal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad.
Sci. USA 97, 6640–6645.
Eguchi, Y., Okada, T., Minagawa, S., Oshima, T., Mori, H., Yamamoto, K., Ish-
ihama, A., and Utsumi, R. (2004). Signal transduction cascade between EvgA/
EvgS and PhoP/PhoQ two-component systems of Escherichia coli. J. Bacter-
iol. 186, 3006–3014.
Froschauer, E.M., Kolisek, M., Dieterich, F., Schweigel, M., and Schweyen,
R.J. (2004). Fluorescence measurements of free [Mg2+] by use of mag-fura 2
in Salmonella enterica. FEMS Microbiol. Lett. 237, 49–55.
Garcia Vescovi, E., Soncini, F.C., and Groisman, E.A. (1996). Mg2+ as an extra-
cellular signal: environmental regulation of Salmonella virulence. Cell 84,
165–174.
Grundy, F.J., and Henkin, T.M. (2006). From ribosome to riboswitch: control of
gene expression in bacteria by RNA structural rearrangements. Crit. Rev. Bio-
chem. Mol. Biol. 41, 329–338.
Hammarlof, D.L., and Hughes, D. (2008). Mutants of the RNA-processing
enzyme RNase E reverse the extreme slow-growth phenotype caused by
a mutant translation factor EF-Tu. Mol. Microbiol. 70, 1194–1209.
Henkin, T.M. (2008). Riboswitch RNAs: using RNA to sense cellular metabo-
lism. Genes Dev. 22, 3383–3390.
Henkin, T.M., and Yanofsky, C. (2002). Regulation by transcription attenuation
in bacteria: how RNA provides instructions for transcription termination/anti-
termination decisions. Bioessays 24, 700–707.
Landick, R., and Yanofsky, C. (1987). Transcription attenuation. In Escherichia
coli and Salmonella Cellular and Molecular Biology, J.L. Ingraham, ed. (Wash-
ington, DC: ASM Press).
Landick, R., Turnbough, C.L., and Yanofsky, C. (1996). Transcription attenua-
tion. In Escherichia coli and Salmonella Cellular and Molecular Biology, F.C.
Neidhardt, ed. (Washington, DC: ASM Press).
Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc. 747
Laursen, B.S., Sorensen, H.P., Mortensen, K.K., and Sperling-Petersen, H.U.
(2005). Initiation of protein synthesis in bacteria. Microbiol. Mol. Biol. Rev. 69,
101–123.
Lejona, S., Aguirre, A., Cabeza, M.L., Garcia Vescovi, E., and Soncini, F.C.
(2003). Molecular characterization of the Mg2+-responsive PhoP-PhoQ regu-
lon in Salmonella enterica. J. Bacteriol. 185, 6287–6294.
Loh, E., Dussurget, O., Gripenland, J., Vaitkevicius, K., Tiensuu, T., Mandin, P.,
Repoila, F., Buchrieser, C., Cossart, P., and Johansson, J. (2009). A trans-
acting riboswitch controls expression of the virulence regulator PrfA in Listeria
monocytogenes. Cell 139, 770–779.
Maguire, M.E. (1992). MgtA and MgtB: prokaryotic P-type ATPases that
mediate Mg2+ influx. J. Bioenerg. Biomembr. 24, 319–328.
Maguire, M.E. (2006). Magnesium transporters: properties, regulation and
structure. Front. Biosci. 11, 3149–3163.
Merino, E., and Yanofsky, C. (2005). Transcription attenuation: a highly
conserved regulatory strategy used by bacteria. Trends Genet. 21, 260–264.
O’Connor, K., Fletcher, S.A., and Csonka, L.N. (2009). Increased expression of
Mg2+ transport proteins enhances the survival of Salmonella enterica at high
temperature. Proc. Natl. Acad. Sci. USA 106, 17522–17527.
Perez-Arellano, I., Rubio, V., and Cervera, J. (2005). Dissection of Escherichia
coli glutamate 5-kinase: functional impact of the deletion of the PUA domain.
FEBS Lett. 579, 6903–6908.
Prost, L.R., Daley, M.E., Le Sage, V., Bader, M.W., Le Moual, H., Klevit, R.E.,
and Miller, S.I. (2007). Activation of the bacterial sensor kinase PhoQ by acidic
pH. Mol. Cell 26, 165–174.
Schiller, D., Kruse, D., Kneifel, H., Kramer, R., and Burkovski, A. (2000). Poly-
amine transport and role of potE in response to osmotic stress in Escherichia
coli. J. Bacteriol. 182, 6247–6249.
748 Cell 142, 737–748, September 3, 2010 ª2010 Elsevier Inc.
Shin, D., Lee, E.J., Huang, H., and Groisman, E.A. (2006). A positive feedback
loop promotes transcription surge that jump-starts Salmonella virulence
circuit. Science 314, 1607–1609.
Snavely, M.D., Miller, C.G., and Maguire, M.E. (1991). The mgtB Mg2+ trans-
port locus of Salmonella typhimurium encodes a P-type ATPase. J. Biol.
Chem. 266, 815–823.
Soncini, F.C., Garcia Vescovi, E., Solomon, F., and Groisman, E.A. (1996).
Molecular basis of the magnesium deprivation response in Salmonella typhi-
murium: identification of PhoP-regulated genes. J. Bacteriol. 178, 5092–5099.
Spinelli, S.V., Pontel, L.B., Garcia Vescovi, E., and Soncini, F.C. (2008). Regu-
lation of magnesium homeostasis in Salmonella: Mg2+ targets the mgtA tran-
script for degradation by RNase E. FEMS Microbiol. Lett. 280, 226–234.
Sudarsan, N., Hammond, M.C., Block, K.F., Welz, R., Barrick, J.E., Roth, A.,
and Breaker, R.R. (2006). Tandem riboswitch architectures exhibit complex
gene control functions. Science 314, 300–304.
Turnbough, C.L., Jr., and Switzer, R.L. (2008). Regulation of pyrimidine biosyn-
thetic gene expression in bacteria: repression without repressors. Microbiol.
Mol. Biol. Rev. 72, 266–300.
Winkler, W.C., and Breaker, R.R. (2005). Regulation of bacterial gene expres-
sion by riboswitches. Annu. Rev. Microbiol. 59, 487–517.
Wortham, B.W., Patel, C.N., and Oliveira, M.A. (2007). Polyamines in bacteria:
pleiotropic effects yet specific mechanisms. Adv. Exp. Med. Biol. 603,
106–115.
Yanofsky, C., and Horn, V. (1994). Role of regulatory features of the trp operon
of Escherichia coli in mediating a response to a nutritional shift. J. Bacteriol.
176, 6245–6254.
Structural Basis of Semaphorin-Plexin
Recognition and Viral Mimicry from
Sema7A and A39R Complexes with PlexinC1
Heli Liu,1 Z. Sean Juo,2,3,4 Ann Hye-Ryong Shim,1 Pamela J. Focia,1 Xiaoyan Chen,1 K. Christopher Garcia,2,3,4
and Xiaolin He1,*
1Department of Molecular Pharmacology and Biological Chemistry, Northwestern University Feinberg School of Medicine, Searle 8-417,
303 East Chicago Avenue, Chicago, IL 60611, USA
2Howard Hughes Medical Institute
3Department of Molecular and Cellular Physiology
4Department of Structural Biology
Stanford University School of Medicine, Beckman B171B, 279 Campus Drive, Stanford, CA 94305, USA
*Correspondence: x-he@northwestern.edu
DOI 10.1016/j.cell.2010.07.040
SUMMARY
Repulsive signaling by Semaphorins and Plexins is
crucial for the development and homeostasis of the
nervous, immune, and cardiovascular systems.
Sema7A acts as both an immune and a neural Sem-
aphorin through PlexinC1, and A39R is a Sema7A
mimic secreted by smallpox virus. We report the
structures of Sema7A and A39R complexed with
the Semaphorin-binding module of PlexinC1. Both
structures show two PlexinC1 molecules symmetri-
cally bridged by Semaphorin dimers, in which the
Semaphorin and PlexinC1 b propellers interact in
an edge-on, orthogonal orientation. Both binding
interfaces are dominated by the insertion of the Sem-
aphorin’s 4c-4d loop into a deep groove in blade 3 of
the PlexinC1 propeller. A39R appears to achieve
Sema7A mimicry by preserving key Plexin-binding
determinants seen in the mammalian Sema7A
complex that have evolved to achieve higher affinity
binding to the host-derived PlexinC1. The complex
structures support a conserved Semaphorin-Plexin
recognition mode and suggest that Plexins are acti-
vated by dimerization.
INTRODUCTION
Semaphorins and their receptors, Plexins, are two families of
widely expressed proteins whose respective structures and
functions are conserved across the animal kingdom. Originally
identified as ligand-receptor pairs that control axon guidance
by repulsion during central nervous system (CNS) development
(Kolodkin et al., 1993; Luo et al., 1993; Tamagnone et al.,
1999; Tessier-Lavigne and Goodman, 1996), Semaphorins and
Plexins have been shown to serve as path-finding controls for
a diverse array of additional functions in physiology, including
vascularization and angiogenesis (normal and pathological),
organogenesis, and immune responses (Kruger et al., 2005;
Suzuki et al., 2008). Semaphorins are also associated with tumor
progression and other diseases (reviewed in Kruger et al., 2005;
Tamagnone and Comoglio, 2000). Within the neural system,
Semaphorin-Plexin signaling is implicated well beyond axon
guidance, ranging from axon pruning to synaptic formation,
specificity, and plasticity (reviewed in He et al., 2002; Waimey
and Cheng, 2006). While Plexins are the predominant receptors
for Semaphorins, the alternative Semaphorin receptors, Neuro-
pilin-1 and -2, differ from Plexins in structure and function
and appear to serve as obligate coreceptors for either Sema-
phorin signaling or VEGF signaling in a variety of specific cellular
(neuronal, vascular, etc.) contexts (Pellet-Many et al., 2008;
Uniewicz and Fernig, 2008).
Semaphorins have been grouped into eight classes on the
basis of primary sequence and source, where classes 3–7 are
vertebrate Semaphorins and class V are viral Semaphorins
(Semaphorin Nomenclature Committee, 1999). The Semaphor-
ins are characterized by an N-terminal 500 amino acid (aa)
Sema domain that is essential for signaling through Plexins
(Koppel et al., 1997). The crystal structures of the Sema domains
of Sema3A and Sema4D (also known as CD100) have been
determined, showing that the Sema domain is a seven-bladed
b propeller, and appear to exist as a homodimer by virtue of
conserved structural elements (Antipenko et al., 2003; Love
et al., 2003). There are four subfamilies of vertebrate Plexins
(A, B, C, and D), which are all type I transmembrane glycopro-
teins featuring an extracellular segment containing an N-terminal
Sema domain, followed by variable numbers of PSI (‘‘found in
Plexins, Semaphorins and Integrins’’) domains, immunoglob-
ulin-like (Ig) domains (Bork et al., 1999), and an intracellular
GTPase-activating (GAP) domain that regulates Rho family
GTPases (Oinuma et al., 2004). The structures of the intracellular
domains of two Plexins have been solved (He et al., 2009; Tong
et al., 2009), but three-dimensional structural information does
not exist for the Plexin extracellular segment or for the Sema-
phorin-Plexin interaction.
In the immune system, a subset of Semaphorins are active
and are designated as ‘‘immune Semaphorins.’’ These include
Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 749
Sema4D, Sema4A, and Sema7A, which appear to modulate
a variety of immune responses ranging from thymic selection
to B cell homing (Suzuki et al., 2008). Sema7A is a GPI-anchored
cell-surface glycoprotein expressed on activated lymphocytes
and thymocytes (Xu et al., 1998; Yamada et al., 1999), and its
receptor is PlexinC1 (also named VESPR or CD232) (Tamagnone
et al., 1999). The Sema7A/PlexinC1 interaction was originally
shown to induce the activation of monocytes (Holmes et al.,
2002), but recently this receptor-ligand pair has been shown to
regulate melanocyte adhesion, and silencing of PlexinC1 is
seen during the development and progression of melanoma
(Scott et al., 2009). Sema7A also has been suggested to have
neuronal functions, by promoting axon growth through integrins
(Pasterkamp et al., 2003). Underscoring the importance of Sem-
aphorin/Plexin interactions in the immune system, viruses have
evolved proteins that engage the immune Semaphorin/Plexin
system, presumably to enhance virus survival in the host. The
most well-characterized examples derive from Vaccinia
(smallpox) virus and Alcelaphine herpesvirus, which encode
secreted Semaphorin homologs A39R and AHVsema (Comeau
et al., 1998; Kolodkin et al., 1993). These viral Semaphorins
show sequence similarity to Sema7A and can also bind to and
activate PlexinC1 (Comeau et al., 1998). The viral Semaphorins
probably play an immunomodulatory role by mimicking Sema7A;
an antibody against PlexinC1 inhibits A39R-induced induction of
inflammatory cytokines by monocytes (Comeau et al., 1998).
We present here the crystal structures of both the Sema7A/
PlexinC1 complex and the A39R/PlexinC1 complex, revealing
the basic architecture of a Semaphorin/Plexin recognition com-
plex. Our data point to a conserved mode of recognition of
Plexins by Semaphorins and provide insights into Plexin activa-
tion and viral mimicry.
RESULTS
Biochemical Analysis of PlexinC1 Binding
by the Mammalian Semaphorin Sema7A
and the Viral Semaphorin A39R
The extracellular domains of Semaphorins and Plexins are large
and highly glycosylated; thus we expressed recombinant forms
of PlexinC1, Sema7A and A39R using baculovirus-mediated
mammalian cell gene transduction (BacMam) (Dukkipati et al.,
2008) and purified the secreted proteins from HEK293H cells
or N-acetylglucosaminyltransferase I-deficient HEK293 (GnTI-
HEK293) cells (Reeves et al., 2002). Our construct designs
were informed by prior structural analysis of Sema4D (Love
et al., 2003), which showed that the first membrane-distal PSI
domain is integrally associated with the Sema domain. Deletion
of the PSI domains in either PlexinC1 or Sema7A destabilized the
proteins. Ultimately, after testing a variety of constructs, for
PlexinC1, we expressed the Sema domain plus the first PSI
domain (SemaPSI), and the entire extracellular segment (ECD).
For Sema7A, we expressed the entire extracellular segment
excluding the GPI anchor; for A39R, we expressed the full-length
protein, which is a ‘‘minimal’’ secreted Semaphorin that does not
contain a PSI domain (Figure 1A).
To analyze the solution binding of PlexinC1 to its A39R and
Sema7A ligands, we used isothermal titration calorimetry (ITC)
750 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.
(Figures 1B–1E). In HEPES-buffered saline (HBS), Sema7A
bound PlexinC1-SemaPSI with a KD of 290 nM (Figure 1B).
The ITC measurements for Sema7A/PlexinC1 were complicated
by the small enthalpy change ( 2 kcal/mol), which resulted in
shallow titration curves with poor signal-to-noise ratio. There-
fore, we used high concentrations of the samples and injected
large amounts of protein so as to maximize the heat per injection,
and also to fully saturate the titration curve. The Sema7A/Plex-
inC1 KD measured here by calorimetry using soluble proteins,
albeit in the nanomolar range, is weaker than the reported KD
of 2.1nM from Scatchard analysis using intact receptors in the
cell membrane (Tamagnone et al., 1999). Previously reported
Semaphorin/Plexin affinity measurements are primarily cell
based and generally in the low nanomolar KD range. We attribute
the lower-affinity values of the soluble recombinant proteins to
the lack of membrane confinement in the ITC format, not to
structural differences in the recombinant proteins. Similar differ-
ences between solution-based and cell-based affinities have
been seen for ligand binding to other receptors, such as SCF/
KIT (55 nM versus 2 nM) (Lemmon et al., 1997; Lev et al.,
1992), CNP/NPRB (2.7 nM versus 0.03 nM) (He et al., 2006; Kol-
ler and Goeddel, 1992), and MHC/TCR (micromolar versus nano-
molar) (Huppa et al., 2010; Krogsgaard et al., 2003). We also
determined that the affinity of Sema7A for the truncated Plex-
inC1-SemaPSI and full-length PlexinC1 ECD were nearly iden-
tical (Figures 1B and 1C), indicating the Ig and PSI domains C-
terminal to the PlexinC1-SemaPSI module are not significantly
involved in Semaphorin recognition. Thus, for structural studies
we proceeded with crystallization of the Sema7A/PlexinC1-
SemaPSI complex.
The viral-derived A39R bound PlexinC1-SemaPSI with a KD of
9.4 nM (Figure 1D). The large enthalpy change ( 14.6 kcal/mol)
facilitated the obtainment of high-quality binding curves. We also
characterized A39R for the effect of various PlexinC1 ECD trun-
cations on affinity. A39R affinity and thermodynamic parameters
for full-length PlexinC1-ECD (KD 8.9 nM) were nearly identical
to that of A39R binding to PlexinC1-SemaPSI (Figures 1D and
1E). Thus, similar to Sema7A/PlexinC1 binding, this indicated
that the N-terminal Sema-PSI domains of PlexinC1 were suffi-
cient for Semaphorin binding and represented an appropriate
form for cocrystallization of both the Sema7A/PlexinC1 and
A39R/PlexinC1 complexes.
Structures of the Sema7A/PlexinC1 Complex, Free
A39R, and the A39R/PlexinC1 Complex
To obtain diffraction quality crystals of the heavily glycosylated
Sema7A/PlexinC1-SemaPSI complex (four N-linked glycosy-
lated sites on Sema7A and seven on PlexinC1-SemaPSI), we
used Endo-H to trim the N-linked glycans attached to the GnTI
HEK293-expressed Sema7A and PlexinC1-SemaPSI proteins.
This treatment leaves a single N-acetylglucosamine (GlcNAc)
residue remaining at each occupied N-linked glycan site,
preserving the core glycosylation of the proteins. The glycan-
trimmed proteins exhibited the same solution behavior to wild-
type proteins, e.g., the Semaphorins existed exclusively as
dimers after glycan-trimming, as assessed by gel filtration chro-
matography (Figure S1 available online). The Sema7A/PlexinC1-
SemaPSI complex structure was determined to a resolution of
Figure 1. Characterization of the Binding of Sema7A and A39R to PlexinC1
(A) Domain diagram of PlexinC1, Sema7A, and A39R and the constructs used in characterization. The cartoon key for the diagrams is shown to the right.
(B–E) Profiles of the binding of PlexinC1 to Sema7A and A39R as measured by isothermal titration calorimetry. The binding parameters are indicated below the
traces.
See also Figure S1 for gel filtration profiles showing that Sema7A and A39R exist as dimers.

2.4 Å via the method of single isomorphous replacement with
anomalous scattering (SIRAS) (Table S1, Figure S2). The asym-
metric unit of the crystal contains one dimeric complex in a 2:2
stoichiometry, comprised of a central Sema7A dimer and two
monomeric PlexinC1-SemaPSI molecules (Figure 2A).
Overall, the shape of the complex resembles a crab where
Sema7A comprises the body, and the two PlexinC1 molecules
resemble pincers emanating radially from the body at four and
eight o’clock. The approximate dimensions of the complex are
80 Å 3 140 Å 3 160 Å (Figure 2A). The head-to-head docking
architecture of the complex is in an ideal orientation for interac-
tion of two cell surface-associated proteins across a cell-cell
junction. In the complex, both Sema7A and PlexinC1 contain
large, disk-shaped Sema domains composed of seven-bladed
b propellers, each of which is intimately associated with its
respective PSI domain and, in the case of Sema7A, a single Ig
domain that would lead to the membrane. In the complex, the
Sema domains mediate the vast majority of the receptor-ligand
contact. The center-to-center distances between the Sema
propellers are 50 Å for Sema7A-Sema7A, 55 Å for Sema7A-Plex-
inC1, and 80 Å for PlexinC1-PlexinC1.
Interestingly, the Sema domains of Sema7A and PlexinC1
interact ‘‘edge-on’’ using their sides to contact one another,
rather than the top or bottom faces, as was generally expected.
The planes of the Sema7A and PlexinC1 b propellers are orthog-
onally related. The orientation shown in Figure 2 places the
C-termini of Sema7A close to the cell membrane, leading to
the GPI anchor. The C terminus of the PSI domain of PlexinC1,
which is 160 Å from the Semaphorin C terminus, leads to the
opposing cell membrane, albeit through one additional PSI and
four Ig domains not present in the current structure. Because
the remaining PSI and Ig domains of PlexinC1 would appear to
emanate away from the central dyad axis of the complex, they
are unlikely to contact Sema7A.
We expressed A39R in baculovirus and determined the
unliganded A39R structure, to a resolution of 2.0 Å, by SIRAS
(Table S1). The structure of A39R shows a dimer of minimal
Sema domains with no PSI or Ig domains (Figure 2D). The

Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 751

Figure 2. Structures of the Sema7A/PlexinC1-SemaPSI and A39R/PlexinC1-SemaPSI Complexes
(A) Ribbon models of the Sema7A/PlexinC1-SemaPSI complex in front view (left) and side view (right), with the Sema7A protomers colored in cyan and blue, and
the PlexinC1-SemaPSI protomers in pink and magenta. The N-linked glycans are depicted as sticks with carbon atoms colored in green. A cartoon of a membrane
is drawn above and below the complex to indicate where the respective proteins would be attached to the cell surfaces.
(B) The structure of an individual PlexinC1-SemaPSI molecule from the complex in two orthogonal views, with each of the seven b -propeller blades, the extrusion,
the flap, and the PSI domain individually colored.
(C) Ribbon model of the A39R/PlexinC1-Sema-PSI complex in front view (left) and side view (right), with the A39R protomers colored in yellow and wheat and the
other components colored similarly to (A).
(D) Ribbon model of an A39R protomer from the free A39R dimer, with the structural modules colored in the same format as PlexinC1 shown in (B).
See also Table S1 and Figure S2 for crystallographic statistics and structural comparisons.

Sema domains of A39R and Sema7A are clearly built on the
same scaffold, with a root-mean-square deviation (rmsd) of
1.5 Å for matching Ca atoms (Figure S3). Major structural differ-
ences are located only at the N-terminal segment and several
loops remote from the Plexin-binding site of Sema7A. Sema7A,
like Sema3A and Sema4D, has a long N-terminal segment that
provides an additional, fifth b strand on the outer edge of blade
6, whereas A39R lacks this segment (Figure S3).
We followed the same approach as Sema7A/PlexinC1-
SemaPSI to obtain the crystals of the A39R/PlexinC1-SemaPSI
complex, which we solved by molecular replacement (Table
S1). The architecture of the A39R/PlexinC1 complex is similar
to that of Sema7A/PlexinC1, except that as a result of the lack
of PSI and Ig domains in A39R, it is shorter in its longest dimen-
sion (Figure 2C). As in the Sema7A/PlexinC1 complex, the Sema
domains mediate all the A39R-PlexinC1 contact. The orienta-
tions of the Sema propellers in the complex are nearly identical
to those in Sema7A/PlexinC1, as is the ‘‘edge-on’’ interaction
mode between the ligand and receptor b propellers.
Dimerization of Sema7A and A39R Are Mediated by
Conserved Structural Elements with Varied Chemistry
Dimerization appears to be a general and probably important
property of Semaphorins (Antipenko et al., 2003; Love et al.,
2003) that we can now examine within the context of a receptor
complex. The Sema7A molecules in the complex, consisting of
Sema, PSI, and Ig domains, form a dimer roughly similar to the
Sema4D dimer (Love et al., 2003) and the Sema3A dimers that
only contain the Sema domains (Antipenko et al., 2003)
(Figure S2C). This is consistent with our gel filtration analysis

752 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.

(Figure S1). The Sema7A dimer interface buries 2860 Å2 of
solvent-accessible surface area. Two sets of three loops on
the top face of the Sema domain, 4b-4c, 4d-5a, and 5b-5c are
intertwined at the dimer interface. In comparison, both Sema4D
and Sema3A use an additional loop for dimerization
(Figure S2C). The Sema7A Sema-Sema interface is primarily
hydrophobic. The Ig domain of Sema7A is also involved in the
dimer interface but only contributes 15% to the total buried
surface area. The Ig domain of Sema7A, unlike the Sema4D Ig
domain, which is canonical, is an unusual variation of the Ig
fold, with a 5-on-2 topology in which strand D switches from
the BED b sheet to join the AFGC b sheet (Figure S2D).
The viral Semaphorin A39R also exists as a dimer both in solu-
tion and in the crystal (Figure S1). Dimerization of A39R is exclu-
sively mediated by its Sema domain, since this protein does not
have PSI or Ig domains, and the dimer interface is considerably
smaller (1990 Å2) than that seen on the canonical Semaphorins
(Figure S2C). A39R uses the same set of loops as the Sema7A
Sema for dimerization, but in contrast to the hydrophobic
Sema7A dimer interface, the A39R dimer interface is composed
almost exclusively of hydrophilic interactions involving six salt
bridges.
The Sema7A and A39R structures, together with the previous
Sema3A and Sema4D structures, indicate that although the
dimer interfaces of Semaphorins can vary significantly in
sequence and chemical nature, the general structural mode of
domain dimerization is conserved, and this is likely to be impor-
tant for the appropriate dimerization geometry of the bound
Plexin receptors for signaling. The preservation of the dimeriza-
tion geometry despite vastly different interface chemistries
suggests that there is evolutionary pressure to achieve this
dimerization mode for Semaphorin function, presumably to
orient bound Plexins for signaling.
The Structure of PlexinC1-SemaPSI Has Unique
Features
Given that the extracellular segments of Plexins have not been
structurally characterized, the fold of the PlexinC1 Sema-PSI
domains merits some description. The Sema domain of PlexinC1
in the complex is generally similar to other seven-bladed
b propeller domains in topology but has unique features that
are distinct from Semaphorins or MET (Stamos et al., 2004)
(Figure 2B). After the common nomenclature used by other
propeller proteins, blade 7 of the PlexinC1 Sema domain is
C-terminally adjacent to blade 1; each of the seven blades is
formed by four antiparallel b strands with strands a-d from the
inside to the outside of the b propeller (Figure 2B). The surface
bears the loops linking strands b and c (e.g., loop 4b-4c) and link-
ing strands d and a (e.g., loop 5d-6a) on the top face. Loop 1d-2a
traverses the top of the propeller like a flap, sealing the central
channel of the propeller, which is hollow in Semaphorins, MET,
and integrins (Figure S2E) (Antipenko et al., 2003; Love et al.,
2003; Stamos et al., 2004; Xiao et al., 2004; Xiong et al., 2002).
A long insertion, termed the ‘‘extrusion’’ (Love et al., 2003), is
located between strands 5c and 5d. A major structural difference
is that the extrusion of PlexinC1 is shorter than that of Semaphor-
ins. The structure of the complex shows that this extrusion in
Semaphorins is critical for binding Plexins, but the extrusion in
Plexins is not a central part of the interface. It appears, then,
that Plexins have lost the prominence of this structural element
as the relative functions of the Sema domain in Semaphorins
versus Plexins specialized over the course of evolution
(Figure S2E) (Antipenko et al., 2003; Love et al., 2003; Stamos
et al., 2004).
The PSI domain of PlexinC1 is a small cysteine-rich domain
similar to that of Semaphorins and integrins, but its orientation
relative to the Sema domain is different from that in Semaphorins
by an 20 rotation (Figure S2E). The PSI domain is intimately
packed against the Sema domain by a broad array of primarily
hydrophobic interactions, which would facilitate the sensitive
structural transmission of Sema binding from the membrane-
distal to the membrane-proximal regions of PlexinC1.
The Sema7A-PlexinC1 Interaction Features
a ‘‘Loop-in-Groove’’ Recognition Mode
The edge-on, orthogonal stacking of the respective b propellers
in the Sema7A-PlexinC1 interface (Figure 3A) buries a total of
2100 Å2 solvent-accessible surface area. The extensive inter-
face can be divided into three principal regions: (1) The 4c-4d
loop of Sema7A inserting into the groove on PlexinC1 that is
bounded by walls composed of the PlexinC1 loop 3b-3c and
the bulged strand 3d—this PlexinC1 groove has an open and
a closed (obstructed) end (Figure 3B); (2) the extrusion helix 2
of Sema7A contacting the loop 3b-3c of PlexinC1 at one side
of the groove; and (3) a small area of contact between the
Sema7A blade 3, and the PlexinC1 strand 3d, that forms the
opposing wall of the groove (Figures 3C–3E).
The central ‘‘loop-in-groove’’ interaction (Figure 3D) features
ten hydrogen bonds (all hydrogen bonds discussed are pre-
dicted from geometry) between the protruding 4c-4d loop of
Sema7A and the PlexinC1 groove (Table S2). The polar interac-
tions are supplemented by two hydrophobic residues (Leu276
and Val278) in the Sema7A 4c-4d loop, which contact several
hydrophobic residues in the PlexinC1 groove through van der
Waals interactions. Overall, the PlexinC1/Sema7A loop-in-
groove interaction includes a mixture of hydrophobic residues
lining the groove wall, interspersed with hydrophilic residues,
to presumably provide ligand specificity through polar interac-
tions. Importantly, at the edge of this groove in PlexinC1,
Sema7A Lys280 forms a salt bridge with PlexinC1 Asp200, and
this interaction appears to be a key component of A39R mimicry
(discussed below).
At the obstructed (closed) end of the groove (Figure 3B), the
Sema7A extrusion helix 2 interacts in a roughly parallel manner
with the PlexinC1 3b-3c loop, also showing several ancillary
long-range interactions with the tips of the PlexinC1 2b-2c and
2d-3a loops (Figure 3C). The interaction is intimate only at the
PlexinC1 Ala197-Ala198-Ser199 bulge. The obstructed end of
the PlexinC1 groove is largely due to the steric bulk sidechain
of Arg131, which forms a salt bridge with Sema7A Glu376
(Figure 3C).
In the third region of the interface, at the open end of the Plex-
inC1 groove (Figure 3B), the outer edge of Sema7A blade 3 forms
extended interactions with the bulged strand 3d of PlexinC1
(Figure 3E). There are two spatially separate clusters of salt
bridges in this region of the interface. The first cluster involves
Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 753
Figure 3. The Interface between Sema7A and PlexinC1-SemaPSI
(A) The overall structure of an interacting pair of Sema7A and PlexinC1-SemaPSI protomers, highlighting the structural elements involved in their binding, colored
cyan for Sema7A and pink for PlexinC1-SemaPSI.
(B) Close-up view of the interface with Sema7A depicted as ribbons and PlexinC1-SemaPSI in a surface representation, showing how the protruding loop 4c-4d
of Sema7A inserts into a groove in PlexinC1 blade 3, and is flanked by the contacts between the Sema7A extrusion helix 2 and the PlexinC1 loop 3b-3c, and the
contacts between the blade 3 of Sema7A and the bulged strand 3d of PlexinC1.
(C) The interactions of residues near the obstructed end of the PlexinC1 groove, with Sema7A in chocolate and PlexinC1 in pink.
(D) The interaction between residues of the 4c-4d loop of Sema7A (cyan) and the PlexinC1 groove (pink).
(E) The interaction between the residues of blade 3 of Sema7A (green) and strand 3d of PlexinC1 (pink). Note that residues 218–220 of PlexinC1 comprise a bulge
from strand 3d.
See also Table S2 and Figure S4 for a list of the interactions and for mutagenesis/binding data on this interface, respectively.

Sema7A residues Arg204 and Arg202 both forming salt bridges
with PlexinC1 residue Glu219. Sema7A Tyr213 occupies the
space between this salt bridge and a salt bridge involving the
important Sema7A residue Lys280 and PlexinC1 Asp200 (Fig-
ure S2A). The second cluster of charged interactions involve
Sema7A Asp216, which forms bifurcated salt bridges with
Arg222 and Lys224 of PlexinC1. Mutagenesis data support that
the interactions at this region are important for Sema7A-PlexinC1
binding (Figure S4). In the complex structure, the location of the
ridges flanking the groove in the Semaphorin binding site is
consistent with a previous mutagenesis study implicating
the region between residues 166–235 of the Sema domain of
Sema3A in its Plexin-binding specificity (Koppel et al., 1997). In
summary, the Sema7A/PlexinC1 interface is extensive and varied
in chemical and structural character and is dominated by the
insertion of a long loop in Sema7A into a deep groove in PlexinC1.

754 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.

Figure 4. The Interface between A39R and PlexinC1-SemaPSI
(A) Structural superposition of the A39R/PlexinC1-SemaPSI and Sema7A/PlexinC1-SemaPSI complexes indicating the closely related docking modes of the viral
and mammalian Semaphorins to the PlexinC1 receptor. The complexes have been aligned on the PlexinC1 component in order to visualize the respective overlap
of the two different Semaphorins. The A39R is in yellow, Sema7A in cyan.
(B) Close-up view of the interface with A39R depicted as ribbons and PlexinC1-SemaPSI in a surface representation.
(C) The interactions of residues near the obstructed end of the PlexinC1 groove, with A39R in chocolate and PlexinC1-SemaPSI in pink.
(D) The interaction between residues of the 4c-4d loop of A39R (blue) and the PlexinC1 groove (pink).
(E) The interaction between the residues of blade 3 of A39R (green) and strand 3d of PlexinC1-SemaPSI (pink).
See also Table S3 for a list of the interactions.

The A39R-PlexinC1 Interaction Globally Resembles
the Sema7A-PlexinC1 Interaction
With strikingly similar orientations of the respective b propellers
in the two complexes, the A39R-PlexinC1 interface involves
the same set of structural elements as the Sema7A-PlexinC1
interface (Figure 4). The A39R-PlexinC1 interface buries a
total of 1890 Å2 solvent-accessible surface area, slightly smaller
than the Sema7A-PlexinC1 interface. The protruding loop 4c-4d
of A39R is in an almost identical conformation as that of Sema7A,
inserting into the groove of the blade 3 surface of PlexinC1
(Figures 4A, 4B, and 5). The neighboring segments of this loop,
including the extrusion helix 2 of A39R that contacts the loop
3b-3c of PlexinC1, and the A39R blade 3 that contacts the Plex-
inC1 strand 3d, have undergone some relatively minor structural
accommodations at the periphery of the interface (Figure 4A).
These small movements result in remodeled pairwise interac-
tions by these surrounding structural elements, relative to the
Sema7A/PlexinC1 interface, but still preserving several key
contacts that presumably are important for the cross-reactivity
(discussed below).
The ‘‘loop-in-groove’’ interaction in A39R/PlexinC1 (Fig-
ure 4D) has six hydrogen bonds between the A39R 4c-4d loop
and the PlexinC1 groove, four less than in Sema7A/PlexinC1
(Table S3). Only one of these hydrogen bonds is conserved
between Sema7A/PlexinC1 and A39R/PlexinC1. The diversity
of amino acid contacts with PlexinC1 formed by the Sema7A
versus A39R 4c-4d loop indicates that the PlexinC1 pocket
has the capacity for highly degenerate interactions, which

Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 755

Figure 5. The Conformation of the 4c-4d Loop of Semaphorins Is Central for Plexin Recognition
(A–D) Stick models of the isolated 4c-4d loops of Sema7A (cyan), A39R (yellow), Sema4D (green), and Sema3A (pink), showing its conserved conformation at the
bases (top) and the midpoints of the loops, and variable conformation at the tips of the loops (bottom).
(E) Sequence comparison of the 4c-4d loop of Semaphorins, with the key conserved structural determinants highlighted.
See also Figure S5 for the RGD motif at the base of the Sema7A 4c-4d loop.

facilitates cross-reactivity. At the edge of this groove in PlexinC1,
A39R presents an arginine (Arg207), as opposed to a lysine
(Lys280) in Sema7A, to form a salt bridge with PlexinC1
Asp200 (discussed below).
Peripheral to the 4c-4d loop, the structural chemistry of the
A39R/PlexinC1 interactions at the obstructed (closed) end of
the PlexinC1 groove (Figures 4B and 4C) are quite different
than what is seen for Sema7A/PlexinC1 as a result of a slight
rigid-body repositioning (Figure 4A). Compared to Sema7A, the
A39R extrusion helix 2 is shifted outwards relative to the center
of the interface and tilted away from PlexinC1 (Figures 4A
and 4C). Consequently, the A39R/PlexinC1 interaction is not as
intimate as the Sema7A/PlexinC1 interaction at this region.
However, the shifting of the helix also allows A39R Asp300,
corresponding to Sema7A Gln379 (Figures 3C and 4C), to
move into the position to form a salt bridge with PlexinC1
Arg131, which is not present in Sema7A/PlexinC1.
The A39R/PlexinC1 interactions at the open end of the
PlexinC1 groove (Figures 4B and 4E) involve the same set of
PlexinC1 residues as in Sema7A/PlexinC1, but the pattern
of interaction is altered. While there are four salt bridges in
two clusters in Sema7A/PlexinC1, there are only two (A39R
Arg132/Arg134 to PlexinC1 Glu219) in A39R/PlexinC1 at this
region (Figure 4E). In A39R/Sema7A, the loss of the other cluster
of salt bridges seen in Sema7A/PlexinC1 (PlexinC1 Arg222/
Lys224 to Sema7A Asp216) is due to the raised A39R blade 3.
While PlexinC1 Lys224 is not directly bonded to A39R in the
A39R/PlexinC1 complex, PlexinC1 Arg222 manages to form
hydrogen bonds with the hydroxyl of A39R Tyr145. The loss
of these two salt bridges from the mammalian complex may
be further compensated by the strengthening of the A39R
Arg132-PlexinC1 Glu219 salt bridge, which is more deeply
buried than in Sema7A/PlexinC1, because of the neighboring
hydrophobic interaction between A39R Ile125 and PlexinC1
Leu220 (Figure 4E). Collectively, The A39R/PlexinC1 interface
shows variations from the Sema7A/PlexinC1 interface due to
some intermolecular repositioning around the central 4c-4d
loop. Nevertheless, the general scheme of inserting the long

756 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.

4c-4d loop into the blade 3 groove in PlexinC1 is identical for both Sema7A and A39R 4c-4d loops (Figure 5) ensures similar
Sema7A and A39R. loop conformations presented to PlexinC1. There is mimicry of
an array of interactions peripheral to the tip of the 4c-4d loop,
as is clear from comparison of the respective binding surfaces.
The Similarities and Variations of the Central
For example, the Lys280 in Sema7A that salt bridges to
Recognition Loops of Semaphorins
Asp200 in PlexinC1 is mimicked by Arg207 in A39R that also
The structural features of the 4c-4d interaction loop of Sema7A
salt bridges to Asp200 on PlexinC1. Also found in both Sema-
and A39R are similar in all Semaphorin structures determined
phorins is a cluster of salt bridges on blade 3 that interact with
to date (Figure 5). The conformation of this loop is stabilized
residues on strand d of PlexinC1. Here, Arg202 of Sema7A salt
by an extensive intraloop hydrogen-bonding network, which
bridges with Glu219 in PlexinC1, and this interaction is mimicked
involves residues highly conserved in most Semaphorins (Fig-
by Arg134 of A39R salt bridging to PlexinC1 Glu219. Arg202
ures 5A–5E). At the base of the loop, Asp269 (Sema7A
and Asp216 in Sema7A are spatially mimicked by Arg132 and
numbering), a buried aspartate conserved in all Semaphorins
Asp148 in A39R, but the hydrogen bonding network with the
(highlighted in Figures 5A–5E), forms a network of four hydrogen
same PlexinC1 residues is remodeled. Key to this cluster is
bonds with main chain amides, which play a central role in orga-
a Tyrosine residue (Tyr213 in Sema7A, Tyr145 in A39R) (Fig-
nizing the loop conformation. At the midpoint, a serine (274 in
ure 6F) that is in a nearly identical position in both complexes,
Sema7A and 201 in A39R) hydrogen bonds with two main-chain
in the center of the cluster of salt bridges.
amides on the opposing strand to narrow the loop. The kinking
In order to ask whether the residues A39R uses to ‘‘mimic’’
of the loop is also facilitated by the presence of two consecutive
Sema7A were energetically important for the viral Semaphorin
glycines (Gly271-Gly272) conserved in all Semaphorins (Fig-
binding, we mutated several of them and tested binding by ITC
ure 5E). At the tip of the loop, the Sema7A Ser274 main-chain
(Figure 6). Mutations of A39R Arg132 or Tyr145 (corresponding
carbonyl forms hydrogen bonds with the main-chain amides
to Sema7A Arg202 and Tyr213) to glutamate or serine abolished
of Ser277 and Val278, a pattern reminiscent of reverse-turns,
PlexinC1 binding (Figures 6C and 6D), whereas mutation of A39R
which may help to maintain a rigid and protruding conformation.
Arg207 (corresponding to Sema7A Lys280) to glutamate
Similar conformation is observed for the tip of this A39R loop
reduced PlexinC1 binding by >60-fold (Figure 6E). The A39R
(Figures 5A and 5B).
Arg207 has a stronger charge and a larger head group than
In Sema3A and Sema4D, however, the tip of this loop has
a Lys residue, and its aliphatic stem is kinked, contacting the
a variation from Sema7A/A39R (Figures 5C and 5D). Class 3–5
neighboring hydrophobic moieties more intimately. Arg207 is a
Semaphorins, including Sema3A and Sema4D, have a one-
more complementary fit in the A39R/PlexinC1 interface com-
residue deletion at the Ser274 position. Sema4D and Sema3A
pared to the Lys280 in the Sema7A-PlexinC1 interface, where
also lack the main-chain hydrogen bonds at the tip seen in
there is more solvent-occupied space adjacent to the side chain
Sema7A and A39R. Consequently, the tips of the 4c-4d loops
(Figures S2A and S2B). Our data (Figure 6E) suggest that the
of these Semaphorins can adopt different main chain conforma-
difference at this position could be an important contributing
tions, probably reflecting the structural requirements of their
factor to the higher-affinity binding to PlexinC1 achieved by
specific Semaphorin-Plexin interactions, yet all appear to
A39R, although additional interactions no doubt also contribute.
contain a polar residue at the loop tips for hydrogen bonding.
We also mutated PlexinC1 Arg222, which is hydrogen bonded to
Because the residues that reside at the apex of the loop are
A39R Tyr145 and serves as shared contact with both Semaphor-
the most deeply embedded in the base of the Plexin groove,
ins. We found a 9-fold affinity loss in A39R-PlexinC1 binding
as exemplified by Sema7A/PlexinC1, it is likely that these are
(Figure 6G). While Sema7A and A39R have exactly the same resi-
major determinants for Semaphorin-Plexin binding specificity
dues at the local region contacting PlexinC1, PlexinC1 Arg222
between classes. Indeed, there is extensive sequence diver-
adopts very different rotamers in the two complexes (Figure 6F).
gence between different classes of Semaphorins at these
The 9-fold affinity loss upon PlexinC1 Arg222Ser mutation
positions, but few within the same class of Semaphorins
suggests that this is an important determinant in A39R/PlexinC1
(Figure 5E).
binding (Figure 6G). Therefore, the residues we mutated, which
are strikingly coincident in the viral and mammalian Sema
The Mimicry of the Mammalian Semaphorin Sema7A complexes, while certainly not giving us a comprehensive ener-
by the Viral Semaphorin A39R getic map of the A39R/PlexinC1 interface landscape, are indeed
The mammalian and viral Semaphorins have arrived at nearly energetically important in ligand-receptor binding.
identical binding modes despite substantial variations in Collectively, our structures suggest that A39R, built on a
sequence (31% identity between Sema7A and A39R). Visualiza- smaller scaffold, has used a limited set of key, highly coincident
tion of the PlexinC1 binding surfaces of Sema7A and A39R amino acid contacts to evolve a more efficient binding interac-
(Figures 6A and 6B) indicates that there are several key corre- tion with PlexinC1 through both the enhancement of particularly
sponding regions of structural mimicry that likely serve as the important polar interactions, such as the substitution of A39R
foundation for their cross-reactivity with a common receptor. Arg207 for the Sema7A Lys280, and slight adjustments of side
There is obvious structural conservation of the centrally located rotamer and main chain positions throughout the binding
4c-4d loop and the presence of identical residues at the tips of surface. These adjustments sum to a substantial optimiza-
the loop (Ser-Leu) that engage the deepest portion of the tion of binding energetics and a concomitant gain in binding
PlexinC1 groove. The preservation of the di-Glycine pair in affinity.

Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 757

Figure 6. Viral Mimicry of Sema7A by A39R
(A and B) The respective PlexinC1-binding surfaces of Sema7A (cyan, left) and A39R (yellow, right). The residues mapped at the interface are colored red for acidic
residues, blue for basic residues, gray for polar, noncharged residues, and green for apolar residues.
(C–E) Mutations of the A39R residues mapped to the Sema7A-PlexinC1 interface abolished or dramatically reduced A39R-PlexinC1 binding.
(F) The different conformation of PlexinC1 Arg222 in binding A39R and Sema7A.
(G) Calorimetric measurement of the binding between A39R and the Arg222Ser mutant of PlexinC1-SemaPSI.
See also Figure S3 for a structural and sequence comparison of Sema7A and A39R.

758 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.

DISCUSSION
We present here the structures of two Semaphorin-Plexin
complexes, establishing the general recognition paradigm
between two protein families with important and wide-ranging
physiological functions involving deliverance of repulsive guid-
ance cues. Unexpectedly, the two b propellers of the Sema-
phorin and Plexin Sema domains contact each other edge-on.
In comparison, MET uses the bottom face of the propeller for
HGF-b chain interaction (Stamos et al., 2004), whereas integrins
bind their ligands with the top face of the propeller (Xiao et al.,
2004; Xiong et al., 2002). Thus, as for other recognition modules
in biology, such as leucine-rich repeats, immunoglobulin folds,
and even MHC-folds, the b propeller reveals itself to be capable
of a diverse array of ligand interaction modes. The moderate-to-
low sequence identity between different classes of Semaphorins
and Plexins, especially at the binding segments, suggests that
there is variability in the specific interatomic contacts in Sema-
phorin/Plexin interfaces between families that is buttressed by
a framework of conserved interactions. This variability may be
necessary for coding the complex pattern of specificity in the
growth- or motility-guiding activities. Nevertheless, the orienta-
tion of each component in the complex, the Sema-Sema dock-
ing geometry, and the global usage of structural elements should
be generally conserved for Semaphorins and Plexins, as has
been found in other ligand-receptor families with similar extents
of sequence identity.
The poxvirus-derived A39R has evolved a more efficient
PlexinC1 binding surface, yet it is smaller than that of Sema7A,
with fewer overall contacts. Our structural analysis showed
that the central binding elements of Sema7A are rigorously main-
tained in the viral complex; therefore, how does A39R achieve
higher affinity binding? This is very difficult to clearly answer,
as the sources of free energy differences between interacting
proteins are not obviously identified by inspection of structures
and are contributed by many subtle effects such as intramolec-
ular side chain cooperativity, solvent restructuring at the binding
surface, and enthalpy-entropy compensation. Thus, we suspect
that it is due to a collective reduction of energetic penalties
across the entire Semaphorin/Plexin interface, perhaps in
combination with an optimization of protein dynamics, such as
lowering overall temperature factors (Table S1), to gain favor in
entropic change. The high similarity of Sema7A and A39R
suggests the likelihood that the poxviruses have acquired the
A39R gene by hijacking and mutating the Sema7A gene, rather
than through convergent evolution where the virus would have
evolved a structurally distinct scaffold for PlexinC1 binding
(Gewurz et al., 2001). Thus, this does not seem to be a case of
molecular ‘‘mimicry’’ as much as simply modifying an existing
binding scaffold with subtle improvements. Given the lack of
involvement of the PSI and Ig domains in Plexin interactions,
these domains may have been lost in A39R over the course of
evolution.
The dimerized Sema7A/PlexinC1 and A39R/PlexinC1 com-
plexes support a model of PlexinC1 activation via dimerization
akin to many other cell-surface receptors. This, in itself, is an
important advance since the downstream mechanisms of Sem-
aphorin/Plexin repulsive signaling are not well understood, and
the knowledge that Plexins are active as dimers should help
refine our thinking about possible downstream effectors. The
quiescent PlexinC1 possibly exists as preformed oligomers,
given the evidence that some other Plexins, e.g., plexin B, exist
as stable homodimers in the absence of ligands (Tong et al.,
2007). We have also found that unliganded PlexinC1 ECD exists
as a mixture of oligomeric states, as assessed by gel filtration
chromatography and crosslinking, although we have not quanti-
tated the exact stochiometries of these states. We propose that
constitutively dimeric ligands such as Sema7A and A39R orient
the inactive PlexinC1 into a dimerized state, in which the
distance, geometry and conformation enables the activation of
the intracellular GAP activity. This PlexinC1 activation scheme
is simpler than a previously proposed model for Sema3A-
induced PlexinA1 activation. That model proposed that the
Plexin Sema domain associates intramolecularly with its addi-
tional extracellular domains in an autoinhibited conformation,
analogous to the LDL receptor (Rudenko et al., 2002), and that
Sema3A disrupts this intramolecular association (Antipenko
et al., 2003; Takahashi and Strittmatter, 2001). Although we
have no unequivocal data to support or refute such a model,
our ITC experiments of A39R binding to full-length versus trun-
cated PlexinC1 ECD show identical binding energetics and
monophasic association, indirectly implying that PlexinC1 is in
the same conformation in the full-length ECD and truncated
forms (Figures 1D and 1E). An important caveat to our experi-
ments is that thermodynamic measurements would not detect
a potential biphasic binding kinetics indicative of a two-step
binding event involving a Plexin conformational change. Never-
theless, whether Plexin undergoes a conformational change
prior to Semaphorin binding, clearly a Semaphorin-induced
arrangement of Plexin into a specifically orientated dimer is the
means of receptor activation. Future structural and mechanistic
studies can now address whether the preactivated Plexin is
involved in an intramolecular association that contributes to
the maintenance of an autoinhibited state.
The Sema7A/PlexinC1 and A39R/PlexinC1 complexes evoke
several important questions. First, where is the Neuropilin-
docking site in a class 3 Semaphorin-Plexin complex? Although
predictions could be made on the basis of our structures
and the previous mutagenesis mapping data (Antipenko et al.,
2003), accurate insights may require a complex structure
encompassing Neuropilin. Second, does Sema7A bind integrins
directly? It has been suggested Sema7A promotes axon growth
and initiates T cell-mediated inflammation responses through
integrins, but not PlexinC1, using an RGD-dependent mechanism
(Pasterkamp et al., 2003; Suzuki et al., 2007). The RGD motif of
Sema7A (Arg267-Gly268-Asp269) is buried (Figure S5) and is
unlikely to be recognized by integrins. Finally, the complex struc-
tures can now be used as a basis to interrogate Sema/Plexin
interactions in vivo with high precision. While mutation of the
key determinants of both Sema7A-PlexinC1 and A39R-PlexinC1
interfaces clearly compromise binding (Figure 6, Figure S4), fur-
ther work will be necessary to demonstrate the consequences
of such perturbation in a cellular or organismal context. Such
functional validation should determine whether modulation of
Semaphorin-Plexin recognition holds clinical promise for applica-
tions involving cell guidance, e.g., directional nerve regeneration.
Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 759
EXPERIMENTAL PROCEDURES
Cloning, Cell Culture, and Baculovirus Generation
Human Sema7A, PlexinC1-SemaPSI, and PlexinC1-ECD, were expressed in
the BacMam system as described previously (Dukkipati et al., 2008), with
a 7-His tag directly attached to the C terminus of each expression construct.
Full-length Vaccinia virus A39R was expressed in both the BacMam system
and the baculovirus-insect cell system. Sf9 and hi5 cells were maintained in
the HyQ-SFX (HyClone) and InsectXpress (Lonza) media. HEK293H cells
(Invitrogen) and GnTI- HEK293 cells (Reeves et al., 2002) were maintained in
CDM4HEK293 media (HyClone). For generation of recombinant viruses, the
expression constructs and the BacVector-3000 baculovirus DNA (EMD
Chemicals) were used to cotransfect sf9 cells in 6-well plates. After 5 days,
the resulted low titer virus stock was harvested and was used to infect Sf9 cells
at 2 3 106 cell/ml for amplification.
Protein Preparation
The amplified baculoviruses were used to infect 1-6 liters of HEK293 cells
or hi5 cells at a density of 1.0–1.5 3 106 cells/ml. After 72 hr, the cells were
pelleted and the supernatants were concentrated. The recombinant proteins
in the supernatant were captured by Ni-NTA metal affinity resin (QIAGEN)
and eluted with 300 mM imidazole (pH 7.5). The eluted proteins were further
purified with gel filtration.
Glycan Trimming, Crystallization, and Data Collection
and Processing
For the Sema7A-PlexinC1 complex, GnTI HEK293-expressed proteins were
trimmed with carboxypeptidase-A and Endo-H. The digested products were
further purified with size exclusion columns and concentrated to 10 mg/ml.
Crystallization was performed via the sitting-drop vapor-diffusion method.
The Sema7A and PlexinC1-SemaPSI proteins, at 1:1 molar ratio, were mixed
with an equal volume of reservoir solution and equilibrated against the reser-
voir solution containing 12% PEG1000, 0.1 M Tris (pH 7.5), and 0.2 M Li2SO4.
The A39R used for crystallization was expressed from hi5 cells using the
pAcGP67A vector. A39R was His-tagged at the N-terminus and expressed
in the presence of 10 mM kifunensine in the expression media (InsectXpress).
The His-tag was removed from A39R with TEV protease, the glycans were
trimmed with Endo-H, and the protein was further purified by gel filtration.
A39R crystallized in the following condition: 50 mM sodium citrate (pH 5.0),
0.2 M Li2SO4, 15% PEG3350, and 20% glycerol.
For the A39R-PlexinC1-SemaPSI complex, the A39R protein was expressed
from the BacMam system following the same procedure for PlexinC1-Sema-
PSI. The complex was crystallized in the following condition: 6% PEG8000,
0.1 M CaCl2, 0.1 M HEPES (pH 7.0), and 0.2 M NaCl.
The Sema7A/PlexinC1 and A39R/PlexinC1 data were measured at the
Life Sciences Collaborative Access Team (LS-CAT) beamlines 21-ID-D and
21-ID-G at the Advanced Photon Source (APS) and were processed with
HKL2000 (Otwinowski and Minor, 1997) (Table S1). The A39R data were
measured at both the Stanford Synchrotron Research Lab (SSRL) and the
Advanced Light Source (ALS) and were processed with Mosflm and scaled
with Scala of the CCP4 program suite (Collaborative Computational Project,
Number 4, 1994). The statistics on the data are presented in Table S1.
Structure Determination and Model Refinement
For structure determinations of Sema7A/PlexinC1-SemaPSI and free A39R,
experimental phases were calculated using heavy atom derivatives by the
SIRAS method, as implemented in the program SOLVE (Terwilliger and
Berendzen, 1999) and SHARP (Bricogne et al., 2003). For structure determina-
tion of the A39R/PlexinC1-SemaPSI complex, the components from the above
two structures were used as search models in molecular replacement. The
models were refined with CNS (Brünger et al., 1998). The final models include
seven N-linked glycans per PlexinC1-SemaPSI molecule (Asn86, Asn141,
Asn149, Asn241, Asn252, Asn386, and Asn407), four per Sema7A (Asn105,
Asn157, Asn258, and Asn330), and one per A39R (Asn51). Each glycan
was modeled with a single GlcNAc residue, as expected, due to the Endo-H
treatment of the proteins. A summary of the refinement statistics is given in
Table S1.
760 Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc.
Isothermal Titration Calorimetry
The proteins used for calorimetry were all expressed from HEK293H cells. ITC
was carried out on a VP-ITC calorimeter (MicroCal, Northhampton, MA) at
30 C. All samples were thoroughly degassed before titration. The titration
data were processed with MicroCal Origin software, Version 5.0.
ACCESSION NUMBERS
The structure factors and coordinates for the Sema7A/PlexinC1-SemaPSI
complex, free A39R and the A39R/PlexinC1-SemaPSI complex have been
deposited in the Protein Data Bank (accession codes 3NVQ, 3NVX, and
3NVN).
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, five
figures, and three tables and can be found with this article online at doi:
10.1016/j.cell.2010.07.040.
ACKNOWLEDGMENTS
We thank the staff at the LS-CAT beamlines at APS and the staffs of ALS and
SSRL for the support in X-ray data collection. We also thank L. Colf, and
N. Goriatcheva for help with the expression of A39R. X.H. is supported by
the NIH grant 1R01GM07805, and K.C.G. is an Investigator of the Howard
Hughes Medical Institute and is supported by NIH-R01-AI51321. The Struc-
tural Biology Facility is supported by the R.H. Lurie Comprehensive Cancer
Center of Northwestern University.
Received: March 19, 2010
Revised: June 8, 2010
Accepted: July 20, 2010
Published online: August 19, 2010
REFERENCES
Antipenko, A., Himanen, J.P., van Leyen, K., Nardi-Dei, V., Lesniak, J., Barton,
W.A., Rajashankar, K.R., Lu, M., Hoemme, C., Püschel, A.W., and Nikolov,
D.B. (2003). Structure of the semaphorin-3A receptor binding module. Neuron
39, 589–598.
Bork, P., Doerks, T., Springer, T.A., and Snel, B. (1999). Domains in plexins:
links to integrins and transcription factors. Trends Biochem. Sci. 24, 261–263.
Bricogne, G., Vonrhein, C., Flensburg, C., Schiltz, M., and Paciorek, W. (2003).
Generation, representation and flow of phase information in structure determi-
nation: recent developments in and around SHARP 2.0. Acta Crystallogr. D
Biol. Crystallogr. 59, 2023–2030.
Brünger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,
Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S.,
et al. (1998). Crystallography & NMR system: a new software suite for macro-
molecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54,
905–921.
Collaborative Computational Project, Number 4. (1994). The CCP4 suite:
programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr.
50, 760–763.
Comeau, M.R., Johnson, R., DuBose, R.F., Petersen, M., Gearing, P.,
VandenBos, T., Park, L., Farrah, T., Buller, R.M., Cohen, J.I., et al. (1998). A
poxvirus-encoded semaphorin induces cytokine production from monocytes
and binds to a novel cellular semaphorin receptor, VESPR. Immunity 8,
473–482.
Dukkipati, A., Park, H.H., Waghray, D., Fischer, S., and Garcia, K.C. (2008).
BacMam system for high-level expression of recombinant soluble and
membrane glycoproteins for structural studies. Protein Expr. Purif. 62,
160–170.
Gewurz, B.E., Gaudet, R., Tortorella, D., Wang, E.W., and Ploegh, H.L. (2001).
Virus subversion of immunity: a structural perspective. Curr. Opin. Immunol.
13, 442–450.
He, Z., Wang, K.C., Koprivica, V., Ming, G., and Song, H.J. (2002). Knowing
how to navigate: mechanisms of semaphorin signaling in the nervous system.
Sci. STKE 2002, re1.
He, X.L., Dukkipati, A., and Garcia, K.C. (2006). Structural determinants of
natriuretic peptide receptor specificity and degeneracy. J. Mol. Biol. 361,
698–714.
He, H., Yang, T., Terman, J.R., and Zhang, X. (2009). Crystal structure of the
plexin A3 intracellular region reveals an autoinhibited conformation through
active site sequestration. Proc. Natl. Acad. Sci. USA 106, 15610–15615.
Holmes, S., Downs, A.M., Fosberry, A., Hayes, P.D., Michalovich, D.,
Murdoch, P., Moores, K., Fox, J., Deen, K., Pettman, G., et al. (2002). Sema7A
is a potent monocyte stimulator. Scand. J. Immunol. 56, 270–275.
Huppa, J.B., Axmann, M., Mörtelmaier, M.A., Lillemeier, B.F., Newell, E.W.,
Brameshuber, M., Klein, L.O., Schütz, G.J., and Davis, M.M. (2010). TCR-
peptide-MHC interactions in situ show accelerated kinetics and increased
affinity. Nature 463, 963–967.
Koller, K.J., and Goeddel, D.V. (1992). Molecular biology of the natriuretic
peptides and their receptors. Circulation 86, 1081–1088.
Kolodkin, A.L., Matthes, D.J., and Goodman, C.S. (1993). The semaphorin
genes encode a family of transmembrane and secreted growth cone guidance
molecules. Cell 75, 1389–1399.
Koppel, A.M., Feiner, L., Kobayashi, H., and Raper, J.A. (1997). A 70 amino
acid region within the semaphorin domain activates specific cellular response
of semaphorin family members. Neuron 19, 531–537.
Krogsgaard, M., Prado, N., Adams, E.J., He, X.L., Chow, D.C., Wilson, D.B.,
Garcia, K.C., and Davis, M.M. (2003). Evidence that structural rearrangements
and/or flexibility during TCR binding can contribute to T cell activation. Mol.
Cell 12, 1367–1378.
Kruger, R.P., Aurandt, J., and Guan, K.L. (2005). Semaphorins command cells
to move. Nat. Rev. Mol. Cell Biol. 6, 789–800.
Lemmon, M.A., Pinchasi, D., Zhou, M., Lax, I., and Schlessinger, J. (1997). Kit
receptor dimerization is driven by bivalent binding of stem cell factor. J. Biol.
Chem. 272, 6311–6317.
Lev, S., Yarden, Y., and Givol, D. (1992). Dimerization and activation of the kit
receptor by monovalent and bivalent binding of the stem cell factor. J. Biol.
Chem. 267, 15970–15977.
Love, C.A., Harlos, K., Mavaddat, N., Davis, S.J., Stuart, D.I., Jones, E.Y., and
Esnouf, R.M. (2003). The ligand-binding face of the semaphorins revealed
by the high-resolution crystal structure of SEMA4D. Nat. Struct. Biol. 10,
843–848.
Luo, Y., Raible, D., and Raper, J.A. (1993). Collapsin: a protein in brain that
induces the collapse and paralysis of neuronal growth cones. Cell 75,
217–227.
Oinuma, I., Ishikawa, Y., Katoh, H., and Negishi, M. (2004). The Semaphorin 4D
receptor Plexin-B1 is a GTPase activating protein for R-Ras. Science 305,
862–865.
Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data
collected in oscillation mode. Methods Enzymol. 276, 307–326.
Pasterkamp, R.J., Peschon, J.J., Spriggs, M.K., and Kolodkin, A.L. (2003).
Semaphorin 7A promotes axon outgrowth through integrins and MAPKs.
Nature 424, 398–405.
Pellet-Many, C., Frankel, P., Jia, H., and Zachary, I. (2008). Neuropilins: struc-
ture, function and role in disease. Biochem. J. 411, 211–226.
Reeves, P.J., Callewaert, N., Contreras, R., and Khorana, H.G. (2002).
Structure and function in rhodopsin: high-level expression of rhodopsin with
restricted and homogeneous N-glycosylation by a tetracycline-inducible
N-acetylglucosaminyltransferase I-negative HEK293S stable mammalian cell
line. Proc. Natl. Acad. Sci. USA 99, 13419–13424.
Rudenko, G., Henry, L., Henderson, K., Ichtchenko, K., Brown, M.S.,
Goldstein, J.L., and Deisenhofer, J. (2002). Structure of the LDL receptor extra-
cellular domain at endosomal pH. Science 298, 2353–2358.
Scott, G.A., McClelland, L.A., Fricke, A.F., and Fender, A. (2009). Plexin C1,
a receptor for semaphorin 7a, inactivates cofilin and is a potential tumor
suppressor for melanoma progression. J. Invest. Dermatol. 129, 954–963.
Semaphorin Nomenclature Committee. (1999). Unified nomenclature for the
semaphorins/collapsins. Cell 97, 551–552.
Stamos, J., Lazarus, R.A., Yao, X., Kirchhofer, D., and Wiesmann, C. (2004).
Crystal structure of the HGF beta-chain in complex with the Sema domain of
the Met receptor. EMBO J. 23, 2325–2335.
Suzuki, K., Okuno, T., Yamamoto, M., Pasterkamp, R.J., Takegahara, N.,
Takamatsu, H., Kitao, T., Takagi, J., Rennert, P.D., Kolodkin, A.L., et al.
(2007). Semaphorin 7A initiates T-cell-mediated inflammatory responses
through alpha1beta1 integrin. Nature 446, 680–684.
Suzuki, K., Kumanogoh, A., and Kikutani, H. (2008). Semaphorins and their
receptors in immune cell interactions. Nat. Immunol. 9, 17–23.
Takahashi, T., and Strittmatter, S.M. (2001). Plexina1 autoinhibition by the
plexin sema domain. Neuron 29, 429–439.
Tamagnone, L., and Comoglio, P.M. (2000). Signalling by semaphorin recep-
tors: cell guidance and beyond. Trends Cell Biol. 10, 377–383.
Tamagnone, L., Artigiani, S., Chen, H., He, Z., Ming, G.I., Song, H., Chedotal,
A., Winberg, M.L., Goodman, C.S., Poo, M., et al. (1999). Plexins are a large
family of receptors for transmembrane, secreted, and GPI-anchored sema-
phorins in vertebrates. Cell 99, 71–80.
Terwilliger, T.C., and Berendzen, J. (1999). Automated MAD and MIR structure
solution. Acta Crystallogr. D Biol. Crystallogr. 55, 849–861.
Tessier-Lavigne, M., and Goodman, C.S. (1996). The molecular biology of
axon guidance. Science 274, 1123–1133.
Tong, Y., Chugha, P., Hota, P.K., Alviani, R.S., Li, M., Tempel, W., Shen, L.,
Park, H.W., and Buck, M. (2007). Binding of Rac1, Rnd1, and RhoD to a novel
Rho GTPase interaction motif destabilizes dimerization of the plexin-B1
effector domain. J. Biol. Chem. 282, 37215–37224.
Tong, Y., Hota, P.K., Penachioni, J.Y., Hamaneh, M.B., Kim, S., Alviani, R.S.,
Shen, L., He, H., Tempel, W., Tamagnone, L., et al. (2009). Structure and
function of the intracellular region of the plexin-b1 transmembrane receptor.
J. Biol. Chem. 284, 35962–35972.
Uniewicz, K.A., and Fernig, D.G. (2008). Neuropilins: a versatile partner of
extracellular molecules that regulate development and disease. Front. Biosci.
13, 4339–4360.
Waimey, K.E., and Cheng, H.J. (2006). Axon pruning and synaptic develop-
ment: how are they per-plexin? Neuroscientist 12, 398–409.
Xiao, T., Takagi, J., Coller, B.S., Wang, J.H., and Springer, T.A. (2004).
Structural basis for allostery in integrins and binding to fibrinogen-mimetic
therapeutics. Nature 432, 59–67.
Xiong, J.P., Stehle, T., Zhang, R., Joachimiak, A., Frech, M., Goodman, S.L.,
and Arnaout, M.A. (2002). Crystal structure of the extracellular segment of
integrin alpha Vbeta3 in complex with an Arg-Gly-Asp ligand. Science 296,
151–155.
Xu, X., Ng, S., Wu, Z.L., Nguyen, D., Homburger, S., Seidel-Dugan, C., Ebens,
A., and Luo, Y. (1998). Human semaphorin K1 is glycosylphosphatidylinositol-
linked and defines a new subfamily of viral-related semaphorins. J. Biol. Chem.
273, 22428–22434.
Yamada, A., Kubo, K., Takeshita, T., Harashima, N., Kawano, K., Mine, T.,
Sagawa, K., Sugamura, K., and Itoh, K. (1999). Molecular cloning of a
glycosylphosphatidylinositol-anchored molecule CDw108. J. Immunol. 162,
4094–4100.
Cell 142, 749–761, September 3, 2010 ª2010 Elsevier Inc. 761
Photoadaptation in Neurospora by
Competitive Interaction of Activating
and Inhibitory LOV Domains
Erik Malzahn,1,3 Stilianos Ciprianidis,1,3 Krisztina Káldi,2,3 Tobias Schafmeier,1 and Michael Brunner1,*
1University
of Heidelberg Biochemistry Center, Im Neuenheimer Feld 328, 69120 Heidelberg, Germany
2Department of Physiology, Semmelweis University, POB 259 H-1444 Budapest, Hungary
3These authors contributed equally to this work
*Correspondence: michael.brunner@bzh.uni-heidelberg.de
DOI 10.1016/j.cell.2010.08.010
SUMMARY
Light responses and photoadaptation of Neurospora
depend on the photosensory light-oxygen-voltage
(LOV) domains of the circadian transcription factor
White Collar Complex (WCC) and its negative regu-
lator VIVID (VVD). We found that light triggers LOV-
mediated dimerization of the WCC. The activated
WCC induces expression of VVD, which then dis-
rupts and inactivates the WCC homodimers by the
competitive formation of WCC-VVD heterodimers,
leading to photoadaptation. During the day, expres-
sion levels of VVD correlate with light intensity, allow-
ing photoadaptation over several orders of magni-
tude. At night, previously synthesized VVD serves
as a molecular memory of the brightness of the
preceding day and suppresses responses to light
cues of lower intensity. We show that VVD is essen-
tial to discriminate between day and night, even in
naturally ambiguous photoperiods with moonlight.
INTRODUCTION
Molecular clocks are circadian timing systems that control
rhythmic transcription of a large number of genes (Dibner
et al., 2010; McClung, 2008; McDonald and Rosbash, 2001;
Siepka et al., 2007; Vitalini et al., 2006; Woelfle and Johnson,
2006). To ensure robust undisturbed circadian oscillation,
clock controlled transcription is in principal independent of light
and compensated for temperature and other environmental
stimuli. However, exogenous stimuli, such as light, are needed
as resetting cues to synchronize the clock with the environ-
mental cycles (Doyle and Menaker, 2007; Foster et al., 2007;
Helfrich-Förster, 2002; Merrow and Roenneberg, 2007; Nayak
et al., 2007). Hence, circadian systems must be able to sense
and properly interpret light intensity signals under a variety of
natural conditions.
Molecular mechanisms that allow circadian systems to
respond reliably to natural photoperiods with noisy and ambig-
uous light intensity characteristics are not known. In nature,
762 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
photoperiodic entrainment is further complicated by light
reflected from the moon since the photoreceptors generally
detect even low light levels (Bachleitner et al., 2007; Merrow
et al., 1999).
The response of Neurospora to light is mediated by WHITE
COLLAR-1 (WC-1) and VIVID (VVD). Together with WC-2,
WC-1 forms the circadian transcription factor White Collar
Complex (WCC), and VVD is expressed under control of light-
activated WCC. WC-1 and VVD contain a flavin-binding light-
oxygen-voltage (LOV) domain (He et al., 2002; Schwerdtfeger
and Linden, 2003). Flavin-binding LOV domains are blue-light
sensors in light receptors of plants and fungi (Krauss et al.,
2009). Transduction of the light signal depends on formation of
a covalent photoadduct between a conserved cysteinyl residue
and the bound flavin cofactor (Salomon et al., 2000; Swartz et al.,
2001; Zoltowski et al., 2007, 2009). The photoadduct interferes
with the interaction of the LOV core and an adjacent helix (Corch-
noy et al., 2003; Harper et al., 2003), but it is not clear how the
signal is further transduced to downstream effectors and tar-
gets. In particular, the molecular basis of light activation of
WCC and adaptation by VVD is not known. VVD has a propensity
to form rapidly exchanging dimers in response to light (Zoltowski
and Crane, 2008), but the physiological relevance of the dynamic
dimerization is obscure.
Once activated by light, LOV domains cannot sense additional
light cues. Spontaneous decay of the flavin-cysteinyl photoad-
duct resets the LOV domain to the dark state and restores its
light-sensing capacity. The photoadduct of VVD is extremely
stable (t1/2
4.5 hr), but generally the kinetics of photoadduct
decay varies from seconds to hours in different LOV domains
(Corchnoy et al., 2003; Swartz et al., 2001; Zoltowski et al.,
2007, 2009).
WCC directly controls expression of a large number of genes
(Smith et al., 2010). The products of two of these genes,
frequency (frq) and vvd, inhibit WCC in negative feedback loops
(Brunner and Schafmeier, 2006; Loros et al., 2007). The clock
protein FRQ inhibits WCC in a light-independent fashion and
generates a daily rhythm of WCC activity that is essential for
self-sustained circadian rhythms (Schafmeier et al., 2008; Schaf-
meier et al., 2005). In contrast, VVD inhibits WCC in a light-
dependent fashion, and the oscillating feedback loop dampens
rapidly in constant darkness (Elvin et al., 2005; Heintzen et al.,
Figure 1. Neurospora Responds to Light Intensities over Several Orders of Magnitude
(A) Photoadaptation and reactivation of vvd transcription. Dark-grown Neurospora was exposed for 4 hr to a light intensity of 5 mmol (photons) 3 m2 3 s1 and
then transferred to 130 mmol 3 m2 3 s1 (red line). vvd RNA was measured by qRT-PCR and normalized to actin RNA (black line).
(B) In the photoadapted state, levels of vvd RNA correlate with light intensity. Neurospora was grown for 48 hr in constant light of the indicated intensity, and vvd
RNA was quantified. Average of six RNA measurements of two independent experiments is shown; error bars indicate the standard deviation (SD).
(C) VVD protein levels depend on light intensity. Mycelia were grown for 48 hr at the indicated light conditions and VVD protein levels were measured. A repre-
sentative western blot is shown (n = 3).
(D) Entrainment of the Neurospora clock during cycles of low light intensities. Race tubes were inoculated with WT conidia in light and subsequently entrained to
13/13 hr light-dark (LD) cycles with the indicated light intensities during the light phase. Conidial bands were analyzed densitometrically. Light and dark periods
are indicated by white and black boxes, respectively. Conidial densities of consecutive cycles (26 hr) are plotted below each other. Plots show representative
individual race tubes (n R 9).
(E) Subcellular localization of VVD-GFP. Mycelia of Dvvd strains expressing GFP or VVD-GFP under control of the cpc-1 promoter were analyzed by fluorescence
microscopy. DAPI staining reveals the nuclei.
See also Figure S1.

2001). VVD is not essential for clock function. Deletion of vvd has
a mild circadian phenotype, resulting in a slight phase delay of
the circadian clock (Heintzen et al., 2001). This raises a question
about the physiologically relevant function of VVD.
When dark-grown Neurospora is abruptly exposed to light,
WCC-dependent transcript levels increase rapidly and adapt to
a low steady state within 1–2 hr (Schwerdtfeger and Linden,
2001). In the photoadapted state, a second transcription spike
can be evoked by increasing the light intensity (Schwerdtfeger
and Linden, 2001). Photoadaptation and light-sensing in the
adapted state depend on VVD.
In nature, mycelia growing deeply concealed behind the
bark of a tree may receive the same light intensity level during
the day as openly exposed mycelia receive by moonlight in the
course of a clear night. How the photoreceptor system adapts
to different light intensities and facilitates photoperiodic entrain-
ment over a broad range of conditions (Tan et al., 2004) is not
known.
In the present study, we elucidated the molecular mechanism
of photoactivation and adaptation of WCC. We show that WCC
and VVD form highly dynamic homo- and heterodimers via their
activated LOV domains. In response to light, WCC initially forms
homodimers that efficiently activate transcription of vvd. VVD
accumulates with a delay after light induction and acts as
a competitive inhibitor of WCC homodimerization, leading to
an attenuation of light-induced transcription. Interaction with
VVD protects light-activated WCC from rapid degradation and
thus allows a sizable fraction of WCC to equilibrate with the
WCC dark form. In the photoadapted state, VVD synthesis trig-
gered by light-activated WCC is balanced by VVD-dependent
inhibition of WCC. The VVD-mediated desensitization and stabi-
lization of the light activated WCC explain on a molecular level
how Neurospora can robustly entrain to artificial and natural
photoperiods.
RESULTS
Neurospora Responds to Light over Several Orders
of Magnitude
To characterize the light-sensing system of Neurospora,
we exposed dark-grown mycelia to increasing light intensities
(5 mmol 3 m2 3 s1 and 130 mmol 3 m2 3 s1) for two consec-
utive 4 hr periods, and expression of vvd RNA was analyzed by
qPCR (Figure 1A). In dark-grown mycelia, vvd RNA could not

Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc. 763

be detected. After the mycelia were transferred to the light
(5 mmol 3 m2 3 s1), RNA levels increased rapidly, reached
a peak after
20 min, and subsequently decreased to a lower
steady-state level. When the light intensity was increased after
4 hr to 130 mmol 3 m2 3 s1, a second spike of vvd RNA induc-
tion and adaptation was observed. The data confirm that
Neurospora adapts to constant light but can still sense light in
the adapted state (Schwerdtfeger and Linden, 2001). However,
we noted that levels of vvd RNA in the adapted state were higher
at the higher light intensity.
To determine whether steady-state levels of vvd RNA correlate
with the light intensity, we grew mycelia for 48 hr at a number of
different light intensities, and vvd RNA was quantified. Surpris-
ingly, vvd RNA levels correlated with light intensity over more
than two orders of magnitude (Figure 1B). Above
130 mmol 3
m2 3 s1, the light-dependent increase in transcript levels
was less pronounced than at lower fluence levels, suggesting
that the system approached saturation. VVD protein levels
increased in a similar fashion (Figure 1C). Transcript levels of
frq, al-2, and wc-1, which are also controlled by the WCC initially,
also rise with increasing light intensity. However, light induction
of these genes saturates readily and therefore does not reflect
light intensities over a wide range (Figure S1A available online).
At light intensities below 0.8 mmol 3 m2 3 s1, vvd RNA and
VVD protein levels were hardly above background and hence
difficult to quantify. To address whether the Neurospora photo-
receptor system can sense light of lower intensity, we examined
entrainment to light cues of different intensities in the race-tube
assay (Figure 1D). Entrainment of the conidiation rhythm of
Neurospora to light:dark (LD) cycles depends strictly on func-
tional WCC (Liu, 2003). Neurospora entrained to LD cycles with
light intensities as low as 2.8 nmol 3 m2 3 s1, the lowest level
of light tested. Assuming that the molecular mechanism of light
sensing and adaptation is the same at high and low light, the
data suggest that the photoreceptors of Neurospora can
discriminate light intensities over five orders of magnitude.
(2.8 nmol 3 m2 3 s1 to 750 mmol 3 m2 3 s1).
Subcellular Localization of VVD
The molecular mechanism by which VVD attenuates the activity
of the WCC is not known, and interaction partners of VVD have
not been found. VVD might directly inhibit WCC in the nucleus
or indirectly affect its activity via signaling molecules, such as
kinases or phosphatases. To determine whether a nuclear
pool of VVD could directly interfere with transcription activation
by the WCC, we constructed a strain, Dvvd cpc-1-vvd-gfp,
which constitutively expressed GFP-tagged VVD under control
of the cpc-1 promoter. Expression of VVD-GFP suppressed
the intense coloring of Dvvd mycelia and restored the wild-
type (WT) phase of conidiation (not shown) suggesting that
VVD-GFP was functional. Analysis of living mycelia by fluores-
cence microscopy revealed that VVD-GFP was located in
the cytosol and nuclei but displayed slightly greater nuclear
enrichment than GFP alone, which was uniformly distributed
(Figure 1E).
The observed nuclear localization of VVD-GFP appears to
contradict previous reports that VVD was not detected in the
nuclear fraction when mycelia were fractionated (Schwerdtfeger
764 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
and Linden, 2003). However, VVD is a small protein (21 kDa) and
might freely equilibrate between nuclei and cytosol. Hence, it
might readily be lost from nuclei during fractionation. To address
this issue, we fractionated mycelia that were untreated or treated
with 1% formaldehyde to induce low levels of nonspecific cross-
linking (Figure S1B). In untreated mycelia, VVD was found exclu-
sively in the cytosolic fraction, in agreement with previous
reports (Schwerdtfeger and Linden, 2003). In treated mycelia,
a substantial portion of VVD was recovered in the nuclear frac-
tion, in agreement with the localization of VVD-GFP in living
mycelia. These data suggest that nuclear VVD is readily lost
during subcellular fractionation procedures.
VVD Competes with Light-Induced Dimerization of WCC
VVD has been reported to form light-dependent dimers that are,
however, highly unstable and dynamic (Zoltowski and Crane,
2008). We reasoned that WCC might dimerize in corresponding
fashion via the LOV domain of WC-1 and that VVD could
compete with this process by interacting with the LOV domain
of WC-1. Attempts to coimmunoprecipitate light or dark states
of VVD and WCC from total cell extracts failed in all possible
combinations (data not shown), indicating that VVD and WCC
do not interact in a stable manner. We constructed a strain
expressing a strep-tagged version of VVD (cpc-1-vvdStrep).
When VVDStrep was purified via a streptavidin affinity resin,
WCC copurified in low but significant amounts (Figure S2A).
To detect interactions of low affinity by independent means,
we examined the interaction of the LOV domains of WC-1 and
VVD in the yeast two-hybrid system. The DNA binding domain
(BD) of the Gal4 transcription factor was fused to the LOV
domain of WC-1, and the transactivation domain of Gal4 (AD)
was fused to the LOV domain of WC-1 and VVD, respectively.
Pairs of LOV domain fusion proteins with BD (‘‘bait’’) and with
the AD (‘‘prey’’) were expressed in a yeast reporter strain that
allows selection for the interaction of ‘‘bait’’ and ‘‘prey’’ via the
fused LOV domains. Since LOV domains are light receptors,
the yeast transformants were grown in light as well as in dark-
ness. The LOV domains of WC-1 displayed homodimerization
as well as heterodimeric interaction with VVD when the corre-
sponding yeast strains were grown on selective medium in light.
However, the strains did not grow in darkness, indicating that the
LOV domains did not interact in the absence of light (Figure 2A).
A positive control strain grew in light and in darkness while
a negative control did not grow at all (Figure 2A). The data
demonstrate that the LOV-domain of WC-1 has the capacity to
interact with itself and with VVD in a light-dependent fashion.
We then asked whether the interaction of VVD with WC-1
affects WCC function. WCC is a GATA-type transcription factor
that binds to light-responsive elements, such as the proximal
light regulatory element (LRE) of the frq promoter (pLRE).
EMSA analysis can discriminate between light and dark states
of the WCC (Froehlich et al., 2002; He and Liu, 2005). Light-acti-
vated and dark forms of WCC display predominantly low or high
electrophoretic mobility shifts (Figure 2B, lanes 1 and 2). Since
we have shown above that the LOV domains of WC-1 dimerize
in light-dependent fashion, the light-activated WCC should
represent a dimer of the WCC monomer that is the prevalent
form in the dark.

(1 ml/assay) is shown to compare the ratio of monomeric and dimeric WCC to a high concentration of extract (9 ml/assay).
(D) VVD interferes with dimerization of the light activated WCC functionally mimicking the dark form. Nuclear extract (DD) was light pulsed (LP) and preincubated
with and without recombinant His6-VVD36. When indicated, His antibody was added to neutralize the His6-VVD36, and EMSAs were performed. All EMSAs were
performed at least three times with identical results.
See also Figure S2.
To demonstrate this directly in an EMSA, we generated an
antibody against the LOV domain of WC-1 that should interfere
with dimerization of WCC protomers. Nuclear extracts of dark-
grown mycelia were preincubated with an excess of LOV anti-
body over WCC to saturate each WC-1 molecule with one anti-
body. Subsequently, one aliquot was exposed to a saturating
light pulse, while the other was kept in the dark. By EMSA anal-
ysis, the dark form of WCC was super-shifted to a single discrete
band of lower mobility, indicating that one LOV antibody had
bound to the WC-1 subunit of the WCC (Figure 2B, lane 3).
However, a subsequent light pulse did not induce a further
super-shift (lane 4). This finding indicates that binding of LOV
antibodies to the dark form of WCC interfered with light-depen-
dent dimerization. In a further assay, nuclear extract was first
preincubated in light with the pLRE oligonucleotide to allow
Figure 2. VVD Interferes with Dimerization
of the Light-Activated WCC and Thereby
Mimics the Monomeric Dark State of the
WCC
(A) Light-dependent homo- and heterodimeriza-
tion of the LOV domains of WC-1 and VVD
analyzed by yeast two-hybrid assay. The LOV
domain of WC-1 was fused to the DNA binding
(BD) of the Gal4 transcription factor. The transac-
tivation domain of Gal4 (AD) was fused to the LOV
domains of WC-1 and VVD, respectively as indi-
cated. The yeast strain PJ69-4A expressing HIS3
under control of a Gal4-dependent promoter was
transformed with the indicated combination of
constructs. Positive control, BD-p53/AD-SV40;
negative control, BD and AD vectors without
insert. Corresponding yeast transformants were
plated on either selective (SDC -Leu -Trp -His;
Two Hybrid) or nonselective medium (SDC -Leu
-Trp; Plating) in dilution series as indicated
(1 corresponds to 2.5 OD). Growth of transform-
ants after 3 day incubation is shown.
(B) Light-activated WCC forms dimers. Nuclear
extract of dark-grown mycelia (DD) or light-pulsed
extract (LP) was preincubated without or with
excess of affinity-purified antibody against the
LOV-domain of WC-1. Subsequently samples
were either kept in the dark (DD) or exposed to
a saturating light pulse (LP) when indicated, and
electrophoretic mobility shift assays (EMSA) of
radiolabeled pLRE oligonucleotide were per-
formed. The LOV antibody interferes with dimer-
ization of the light-activated WCC. Monomeric
and dimeric WCC with and without bound LOV
antibody are schematically depicted above the
corresponding lanes. The dark form of the WCC
is shown in gray, and the light-activated WCC is
shown in white.
(C) Concentration dependence of the monomer-
dimer equilibrium of light-activated WCC. EMSA
was performed with 1 and 9 ml light-pulsed nuclear
extract (total volume 20 ml). A long exposure (l.e.)
of the EMSA with low concentration of extract
binding of the light-activated WCC dimer to its DNA binding
site. Subsequently, LOV antibody was added and the sample
was then analyzed by EMSA (lane 5). Under these conditions,
the light-activated WCC was super-shifted, suggesting that the
WCC dimer was crosslinked by one antibody bound to two
LOV domains, a reaction that should be kinetically favored,
when WCC is a dimer. Together, the data demonstrate that
WCC dimerizes in a light-dependent fashion.
If WCC dimers are highly dynamic, the monomer-dimer ratio
should be sensitive to the WCC concentration. To vary the
concentration of WCC, we performed EMSA in a fixed volume
with increasing amounts of light-exposed nuclear extract. As
expected, the intensity of the EMSA signals increased due to
the increasing amounts of WCC (Figure S2B). At high concentra-
tions of extract, only the mobility shift corresponding to the
Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc. 765

dimeric WCC was detected. However, a long exposure of the
X-ray film revealed monomeric and dimeric WCC at low con-
centrations of light-exposed nuclear extract (Figure 2C and
Figure S2B). This finding indicates that the light-induced dimer-
ization of WCC is dynamic and depends on the concentration of
WCC.
For functional characterization of the interaction of VVD and
WCC, nuclear extract was first exposed to a saturating light-
pulse, then incubated with recombinant light-activated His6-
VVD36 (Zoltowski et al., 2007), and finally, examined by EMSA
(Figure 2D). In the presence of His6-VVD36, the light-activated
WCC displayed a mobility shift characteristic for the dark form.
In its absence, WCC displayed the dimer-specific mobility shift.
These observations indicate that His6-VVD36 disrupted the WCC
dimers. In further characterizations, the nuclear extract was
exposed to light and preincubated with recombinant light-acti-
vated His6-VVD36 as described above. The sample was trans-
ferred to darkness, and anti-His antibody was added to oligo-
merize His6-VVD36 and thereby functionally neutralize the
protein. Although the sample did not receive additional light,
subsequent analysis by EMSA revealed that the dimer-specific
766 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
Figure 3. Formation and Decay of the Light Form of
the WCC
(A) Kinetics of formation of the light-activated WCC.
Nuclear extracts of dark grown mycelia (DD, time point
‘‘0’’) were irradiated with the indicated intensities and for
the indicated times and examined by EMSA.
(B) Upper (light complex) and lower (dark complex) bands
were quantified by densitometry and the ratio of upper/
(upper + lower) was calculated for each time point and
the halftime for conversion was determined.
(C) Decay kinetics of the WCC light form. Light-treated
nuclear extracts were incubated in the dark for the indi-
cated times. Left: Samples were analyzed by EMSA.
Open and filled arrowheads indicate the light-activated
and the dark form of the WCC, respectively. EMSAs
were performed at least three times with identical results.
Right: Densitometric analysis of shown EMSA. So that the
functionality of the WCC could be controlled, an aliquot of
each sample was light pulsed to recover the light form
(data not shown).
See also Figure S3.
mobility shift of the light-activated WCC was
fully restored. The data demonstrate that VVD
interacts with WCC and interferes with dimeriza-
tion of the WCC protomers. It thereby function-
ally resets the light-activated WCC to the dark
form.
Photochemical Properties of WCC
and VVD
Activation of WCC protomers by light-induced
dimerization and subsequent competitive inhibi-
tion of dimerization by newly synthesized VVD
readily explains the phenomenon of induction
and adaptation of WCC-dependent transcrip-
tion in response to light on a molecular level.
However, the model does not explain how the
system maintains its light-sensing capacity over several orders
of magnitude, since a sizable pool of the WCC must be in the
dark state to sense a further increase in light intensity in constant
light conditions.
To estimate the equilibrium between WCC light and dark
states, we measured the kinetics of formation and decay of the
light complex. Nuclear extracts of dark-grown mycelia were
exposed to light of different intensities, and the kinetics of
WCC dimerization was analyzed by EMSA (Figures 3A and 3B).
At high light intensity (420 mmol 3 m2 3 s1), an exposure of
15 s was sufficient for complete light activation of WCC. At inter-
mediate intensity (80 mmol 3 m2 3 s1), an exposure of about
5 min was required, and at low light (3.6 and 6 mmol 3 m2 3
s1), 20–30 min was needed. When light-exposed nuclear
extract was transferred to the dark, WCC dimers were stable
for several hours (t1/2
4 hr) (Figure 3C).
The ratio of forward versus reverse rates indicates that the
fraction of WCC in the dark state depends on light intensity.
In bright light (420 mmol 3 m2 3 s1), the fraction of WCC in
the dark form is rather small (6 s/14,400 s = 4 3 104). In constant
intermediate light (6 mmol 3 m2 3 s1), about 12.5% of the

Figure 4. VVD Stabilizes the Light-Activated WCC
(A) Light-induced degradation of WC-1 in the WT and Dvvd, cpc-1-vvd-gfp
strain. Mycelia were grown in constant darkness. Cycloheximide (CHX)
(10 mg/ml) was added, and cultures were transferred to the light. Samples
were harvested after the indicated times, and WC-1 was analyzed by western
blotting.
(B) Positive feedback of FRQ and VVD on levels of light-activated WCC.
Steady-state expression levels of WC-1 in frq- and vvd-deficient strains grown
in light are shown in comparison to the WT. A long exposure is shown to illus-
trate differences in WC-1 levels in the low-expressing mutant strains. Repre-
sentative western blots are shown (n = 3).
See also Figure S4.
WCC should be in the dark state (30 min/240 min). Extrapolating
to low light, we estimate that the fraction of WCC in the dark form
is substantial in dim light (>30% below 1 mmol 3 m2 3 s1).
Thus, the pool of WCC in the dark form is detectable and there-
fore of physiological relevance for light-sensing in the photoa-
dapted state. Recombinant His6-VVD36 displayed similar light
sensitivity and photoadduct stability (Figure S3).
VVD Stabilizes the Light-Activated WCC
DNA-binding-competent, activated WCC is unstable (Schafme-
ier et al., 2008). Since VVD competes with dimerization of the
WCC and thereby reduces its activity, we asked whether VVD
also affects the degradation kinetics of WCC. WC-1 was rapidly
degraded when dark-grown mycelia were treated with cyclohex-
imide (CHX) and then exposed to saturating light (Figure 4A)
(Schafmeier et al., 2008). To assess the turnover of light-
activated WCC in the presence of VVD under corresponding
conditions, we analyzed the stability of WC-1 in the Dvvd, cpc-
1-vvd-gfp strain that constitutively expresses VVD-GFP in a
light-independent manner. WC-1 was stable when dark-grown
mycelia of Dvvd, cpc-1-vvd-gfp were treated with CHX and
exposed to light (Figure 4A), indicating that the previously
synthesized VVD-GFP interfered with the light-induced turnover
of the WCC. The acute induction of transcription by light was
attenuated in the Dvvd, cpc-1-vvd-gfp strain, demonstrating
that VVD-GFP was immediately active as a repressor of the
light-activated WCC (Figure S4A). In a vvd deletion mutant,
light-dependent gene expression is induced to a higher level
and adapts slower (Schwerdtfeger and Linden, 2001) (Fig-
ure S2B). In summary, VVD reduces the transcriptional activity
of the WCC, which results in a stabilization of the transcription
factor, since only the active WCC is subjected to rapid degrada-
tion (Schafmeier et al., 2008).
WCC levels and activity do not depend solely on VVD but are
regulated in a complex manner. On the posttranscriptional level,
FRQ-dependent phosphorylation inactivates and stabilizes
WCC (Schafmeier et al., 2008; Schafmeier et al., 2005). On the
transcriptional level, WCC supports expression of WC-1, the
limiting component of WCC (Káldi et al., 2006).
To assess the contribution of FRQ and VVD to stabilizing the
light-activated WCC, we analyzed WC-1 levels in frq- and vvd-
deficient strains grown in light (Figure 4B). WC-1 levels were
distinctly lower in frq9 (nonfunctional allele of frq) and Dvvd
than in WT cells, demonstrating that FRQ and VVD indepen-
dently stabilize the light-activated WCC to a similar extent.
WC-1 levels were even lower in the double deficient frq9, Dvvd
strain. This observation indicates that the positive feedback of
WCC on wc-1 transcription cannot compensate its light-induced
degradation.
VVD Synthesized during the Day Suppresses Moonlight
during the Night
We reasoned that overexpression of VVD above levels expected
by the feedback system on the basis of the prevalent light inten-
sity should result in blindness of Neurospora. To test this predic-
tion, we compared the light-entrainment characteristics of Dvvd,
cpc-1-vvd-gfp, and Dvvd (Figure 5A). After transfer from bright
light to darkness (DD), both strains displayed a free-running con-
idiation rhythm, indicating that the circadian clock was functional
and synchronized by the LD transfer. Dvvd entrained to LD
cycles at low and high light intensities. In contrast, Dvvd, cpc-
1-vvd-gfp entrained to LD cycles with a light intensity of
180 mmol 3 m2 3 s1 but displayed a free-running rhythm at
lower light intensities (2.8 and 180 nmol 3 m2 3 s1). Hence,
overexpression of VVD-GFP desensitizes the WCC and renders
the strain blind to the lower levels of light intensity.
What is the physiological role of desensitization of the WCC
photoreceptor by VVD? Photoperiods in nature are quite noisy.
In particular, nights are not always entirely dark, and thus, abso-
lute light intensity is an ambiguous cue that does not allow
discrimination between day and night. On a clear night with
a full moon, the light intensity (
20 nmol 3 m2 3 s1) is well
above the sensitivity of the WCC photoreceptor (<2.8 nmol 3
m2 3 s1), and hence, moonlight should reset the clock.
However, light intensities during the day are more than 103-
fold higher than moonlight. VVD that is expressed at high levels
during the day is turned over in the dark with a half-time of about
2.5 hr (Figure 5B). Accordingly, VVD levels should remain
Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc. 767

Figure 5. VVD Is a Critical Element for Controlling Neurospora
Response to Light
(A) Desensitization of Neurospora to light depends on VVD expression levels.
Densitometric analysis of race-tube assays of Dvvd and Dvvd, cpc-1-vvd-gfp
strains as shown in Figure 1D. Strains were grown in high light and then trans-
ferred to the dark and exposed to a 13/13 hr LD regime with the indicated light
intensity during the light phase. Overexpression of VVD-GFP caused blindness
to entraining light cues of lower intensity resulting in a free-running conidiation
rhythm. Plots from representative race tubes are shown (n R 9).
(B) VVD synthesized during the light phase is progressively degraded in the
dark. Light-grown cultures were transferred to darkness, and VVD levels
were determined after the indicated time periods by western blotting with an
antibody against the N terminus of VVD. Asterisk indicates unspecific reaction
of the VVD antibody. Representative western blot is shown (n = 3).
(C) VVD is essential for the entrainment of Neurospora to natural photoperiods
with moonlight. Analysis of race-tube assays of the WT and Dvvd. Race tubes
were exposed to natural day-night cycles for 6 days beginning at a dark new-
moon night and ending at almost half moon. Incubation was performed in an
isolated room at 21 C. Conidial density traces of the race-tubes of the WT
and Dvvd are shown. The moon phases are indicated below the traces.
Dvvd became arrhythmic with upcoming moonlight.
See also Figure S5.
768 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
sufficiently high throughout the night to suppress responses
evoked by the considerably dim moonlight and thus allow
entrainment of Neurospora to natural photoperiods.
To test this hypothesis, we exposed race tubes inoculated
with the WT and Dvvd to natural photocycles. Incubation was
started on a dark new moon night and terminated 6 days later
at almost half moon (Figure 5C). The WT entrained to the natural
photoperiod throughout the entire time period. In contrast, Dvvd
entrained for 3 or 4 days, but the rhythm was then obviously
disturbed by the increasing moonlight, and the strain became
arrhythmic. Similarly, the WT entrained to artificial moonlight
(240 nmol 3 m2 3 s1) in a 12 hr high-light:12 hr low-light cycle
while entrainment of Dvvd was disturbed (Figure S5). Both
strains became arrhythmic when the intensity of the artificial
moonlight was raised to 800 nmol 3 m2 3 s1. The data
demonstrate that VVD is required for robust entrainment of
Neurospora to ambiguous natural photoperiods.
DISCUSSION
In this study, we showed that Neurospora has evolved an elabo-
rate molecular mechanism of light activation and adaptation of
transcription that allows discrimination between night and day
on the basis of relative differences in light intensity levels. The
system requires the interaction of two LOV domain photorecep-
tors, the activating photoreceptor WCC, and the inhibitory
photoreceptor VVD that constitute a negative feedback loop
(Elvin et al., 2005; Heintzen et al., 2001). We present here the
molecular mechanism of the feedback and describe properties
that are crucial for the dynamic behavior of this system that
allows light sensing in the photoadapted state over several
orders of light intensity (see model in Figure 6). Light activation
of the LOV domain of the WC-1 facilitates dimerization of proto-
mers of the WCC. Dimers of WCC display a higher activity than
the monomeric WCC protomers, which are prevalent in dark-
ness. Hence, abrupt light exposure of dark grown Neurospora
triggers a burst of transcription of vvd, frq, and other light-
induced genes. Subsequent photoadaptation is a complex pro-
cess dependent on one hand on rapid degradation of activated
WCC and on the other hand on inactivation and stabilization of
WCC by VVD and FRQ. Rapid degradation of activated WCC
would, in principle, lead to an efficient attenuation of light-
induced transcription. However, it would be detrimental if light-
induced degradation of WCC were the sole mechanism of adap-
tation, since this would result in severe depletion of the pool of
WCC. Although transcription of wc-1 is light induced by the
WCC (Káldi et al., 2006), this positive feedback loop cannot
compensate for the light-induced degradation. Hence, WCC
levels are extremely low in a frq9, Dvvd strain.
VVD, which is rapidly synthesized in light, acts on the light-
activated WCC by inhibiting, in a competitive manner, the forma-
tion of highly active WCC dimers. This causes a rapid and effi-
cient attenuation of light-induced transcription. Since inactive
WCC is not subjected to rapid degradation (Schafmeier et al.,
2008) VVD promotes accumulation of the light-activated but
transcriptionally inactive WCC. Hence, VVD resets the light-acti-
vated WCC to a monomeric, inactive, and stable form that is
functionally equivalent to the dark state (Figure 6).

In addition to VVD, the circadian clock protein FRQ also
supports inactivation and stabilization of WCC (Schafmeier
et al., 2008; Schafmeier et al., 2005). FRQ inactivates WCC in
a light-independent fashion by facilitating its phosphorylation
by casein kinase 1a. However, since FRQ expression is sup-
ported by WCC, FRQ gradually accumulates to an elevated level
in constant light, resulting in an enhanced feedback of FRQ that
partially compensates for the increased activity of light-activated
WCC. Thus, vvd-deficient but frq-proficient strains display
a slow and partial attenuation of light-induced transcription
(Schwerdtfeger and Linden, 2001).
Notably, the circadian clock is slightly phase delayed but
otherwise fully functional in vvd-deficient (frq-proficient) strains.
This mild phenotype provokes the question of whether VVD
has a physiologically relevant specific function that cannot
be provided by FRQ. While FRQ is a general inhibitor of the
WCC, VVD inhibits specifically the light-activated WCC. Under
entrained conditions VVD accumulates to a high level during
the day and efficiently attenuates transcription induced by the
light-activated WCC. Although VVD is degraded during the night,
the levels remain sufficiently high to suppress activation of the
WCC by moonlight (>1000-fold lower light intensity than
daylight). In the absence of VVD (Dvvd), the highly sensitive
WCC is activated by moonlight, and the clock cannot entrain
because of the ambiguous light cues. Hence, in natural photope-
riods, VVD is an essential component of the circadian clock.
It modulates the sensitivity and responsiveness of the system
Figure 6. Model of the Molecular Interactions of WCC
and VVD upon Light Induction and Adaptation
W and V denote the dark forms of the WCC and VVD. Accord-
ingly, W* and V* denote the light-activated forms of the photo-
receptors. Upon light exposure, W is converted to W*, which
dimerizes (W*W*) and efficiently activates transcription of
vvd. The DNA-bound W*W* complex is unstable and rapidly
degraded (Deg.). As V* accumulates, it interacts with W* and
thereby competes with the homodimerization of W*. The
W*V* complex is transcriptionally inactive and stable, leading
to light adaptation of vvd transcription. The stable W*V* accu-
mulates gradually and equilibrates with the dark forms of W
and V. According to law of mass action, the equilibrium is
established by light-dependent conversion of W to W* and
light-independent decay of the W* photoadduct. Stabilization
of W*V* and equilibration with the dark forms lead to a substan-
tial population of W, which allows light sensing in the photoa-
dapted state.
to light cues and thus allows proper interpretation
of ambiguous light signals. Desensitization of
WCC correlates with the expression level of VVD,
which accumulates as a function of duration and
intensity of light exposure. The accumulated VVD
provides a molecular memory of the previous light
history, which is gradually fading in darkness. Since
the extent of desensitization is gradually lost during
the night, the system remains blind for levels of light
that are low in relation to the previous day and can
thus ignore moonlight as a resetting cue. However,
the adapted system remains responsive to stimuli
of higher light intensity. The capacity to respond to light cues
of higher intensity depends on accumulation of a significant
pool of the dark form of WCC. This is facilitated by VVD, which
inactivates the light-activated WCC and thereby blocks its
degradation. Since the light and dark forms of the WCC are in
equilibrium, stabilization of the light-activated WCC allows accu-
mulation of WCC in the dark form. According to law of mass
action, the equilibrium is established by formation of the photo-
adduct, which depends on light intensity, and its spontaneous,
light independent decay. When this equilibrium is disturbed by
a sudden increase in light intensity, the accumulated pool of
WCC in the dark state is activated. It transiently outcompetes
the inhibitory capacity of VVD until sufficient additional VVD is
synthesized to establish a new steady state that reflects adapta-
tion to the higher light intensity. Since the WCC/VVD system can
discriminate light intensities over more than five orders of magni-
tude, Neurospora readily adapts to habitats that differ in lighting
conditions.
In constant darkness, rhythmic expression of VVD dampens
rapidly and levels of VVD drop below the limit of detection (Elvin
et al., 2005). The absence of VVD after several days in darkness
renders the WCC highly sensitive to light cues. This can
be viewed as dark adaptation of the light-sensing system of
Neurospora.
Finally, the flavin-cysteinyl photoadducts of WCC and VVD are
stable for several hours while the photoadducts of other LOV
domains decay in the range of minutes. This extraordinary
Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc. 769
photoadduct stability should smoothen the activity profile of the
WCC such that noisy fluctuations in light intensity (e.g., as a result
of moving clouds concealing the sun) are not transduced by the
photoreceptor system. Rather, the photoreceptors integrate the
light signals and keep a slowly decaying molecular memory of
the maximal light level of the previous time period.
We showed that light-sensing and adaptation (desensitization)
in Neurospora depend on homo- and heterodimerization of the
activating and inhibitory LOV domains of WC-1 and VVD,
respectively. Flavin-binding LOV domains are also found in plant
phototropins (phot1 and phot2). These are membrane-associ-
ated receptor kinases that dimerize in response to blue light
and activate a number of different processes, such as phototro-
pism and chloroplast movement (Christie, 2007). Phototropins
always contain a pair of LOV domains, LOV1 and LOV2. LOV2
is a dark-state inhibitor of receptor kinase autophosphorylation,
and its activation by light is essential for signaling (Cho et al.,
2007; Christie et al., 2002; Kaiserli et al., 2009). The role of
LOV1 is less clear. It is not essential for light signaling. Photoac-
tivation of LOV1 appears to have an inhibitory role at high light
intensities (Kaiserli et al., 2009), but the underlying molecular
mechanism is not understood. The photochemical properties
of phot1 suggest an interaction between LOV1 and LOV2
(Kaiserli et al., 2009). It is tempting to speculate that, like the
WCC-VVD system, LOV2 might act by facilitating light-depen-
dent dimerization of two phototropins, while heterodimerization
of LOV2 with LOV1 could be inhibitory by interfering with
receptor dimerization. In principle, heterodimerization of LOV1
and LOV2 should be kinetically favored since both domains
occur in the same molecule. However, the quantum yield for
photoadduct formation (Kasahara et al., 2002) and the rapid
decay of the photoadducts of LOV1 and LOV2 (Salomon et al.,
2000) suggest that light of considerable intensity is required to
simultaneously activate both LOV domains in one phototropin
molecule. Hence, an intramolecular inhibitory dimerization of
LOV1 and LOV2 could desensitize the system at high light inten-
sities and attenuate the strength of downstream signaling. Thus,
competitive homo- and heterodimerization of activating and
inhibitory LOV domains might be a general mechanism for
light-sensing and adaptation.
In summary, we report a molecular mechanism of photo-acti-
vation and adaptation in Neurospora. In response to light, WCC
forms homodimers that induce transcription of vvd. VVD, in turn,
competes for binding with WCC, disrupts the WCC dimers, and
attenuates light-induced transcription. In the photoadapted
state, VVD synthesis by light activated WCC is balanced by
VVD-dependent inhibition of WCC. The VVD-mediated desensi-
tization of the light-activated WCC provides a molecular mecha-
nism for how Neurospora robustly entrains to photoperiods.
EXPERIMENTAL PROCEDURES
Neurospora Strains and Culture Conditions
Neurospora strains used in this study (WT; Dvvd; frq9, Dvvd; Dvvd, cpc-1-gfp;
Dvvd, cpc-1-vvd-gfp; and Dvvd, cpc-1-vvdStrep) carried the ras-1bd mutation
(Belden et al., 2007). Standard growth medium contained 2% glucose, 0.5%
L-arginine, 1 3 Vogel’s and 10 ng/ml biotin. Cultures were incubated 48 hr
at 25 C if not indicated otherwise. In the race-tube medium, glucose was
omitted, and arginine was reduced to 0.17%. Race tubes contained 2.2%
770 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
agar. Analysis of race tubes was performed by densitometry with the ChronoII
software (T. Roenneberg, Ludwig-Maximilians-Universität Munich).
Plasmid Construction
cpc-1-gfp was constructed by replacement of the ccg-1 promoter with the
cpc-1 promoter in pMF272 (generous gift of Michael Freitag) and insertion of
an AscI site between the promoter and the SpeI site. For construction of
cpc-1-vvd-gfp, a genomic PCR product, including the coding region of vvd
and 46 bp upstream of the start codon, was inserted as an AscI-SpeI fragment
into cpc-1-gfp. For construction of cpc-1-vvd-Strep, a PCR fragment contain-
ing a genomic region of vvd from position 46 to +691, a short linker and nucle-
otides coding for a C-terminal Strep-tag (Herzberg et al., 2007), followed by
a stop codon, was amplified and inserted between the AscI and SpeI sites
of pcpc-1-gfp.
pET28a-vvd36 for the expression of a soluble N-terminal truncated version of
VVD fragment (Zoltowski et al., 2007) was kindly provided by Brian R. Crane.
Plasmids for the yeast two-hybrid assay were based on pGBD-C1 and
pGAD-C1 (James et al., 1996). The VVD36 fragment and the LOV domain of
WC-1 were amplified by PCR from Neurospora total complementary DNA
(cDNA) and inserted in the EcoRI and BamHI sites of the respective vectors.
Neurospora Transformation
Neurospora conidia (5–7 days old) were transformed by electroporation as
described (Schafmeier et al., 2006). The his-3 locus was used for homologous
recombination. Insertion of the target gene was verified by PCR and
sequencing. Homokaryotic strains were generated by purification by several
plating steps and checked by quantitative PCR.
RNA Analysis
Total RNA was extracted with the peqGOLD TriFast reagent (Peqlab). cDNA
was synthesized from 1 mg total RNA with the QuantiTect Reverse Transcrip-
tion Kit (QIAGEN). Transcript levels were quantified by RT-PCR with TaqMan-
Probes in a StepOne system (Applied Biosystems). Triplicate reactions (20 ml)
containing cDNA equivalent to 0.05 mg RNA were analyzed.
Protein Analysis
Extraction of Neurospora protein, subcellular fractionation, and extraction of
nuclear proteins were performed as described (Schafmeier et al., 2005).
Protein concentration was determined by Bradford assay, and 350 mg (total
extract, cytosol) or 100 mg (nuclei or nuclear extract) were analyzed by SDS-
PAGE. Western blots were performed as described (Schafmeier et al.,
2005). Enhanced chemiluminescence signals were detected with X-ray films.
Series of exposures were generated and signals were quantified with Adobe
Photoshop as described previously (Schafmeier et al., 2008). The VVD anti-
body used for Western blotting was directed against the N terminus.
Fluorescence Microscopy
Neurospora cultures were grown in liquid medium for 1–2 days. Pieces of
mycelia were placed on an object slide and embedded in mounting medium
(13% Mowiol [Hoechst], 30% glycerol [pH 8.5]) containing 0.25 mg/ml DAPI.
Microscopy was performed with a Leica DMI 6000B device. GFP and DAPI
were excited at 470 and 360 nm, respectively.
Purification of Recombinant VVD
Affinity purification of VVDStrep was performed with the One-STrEP System
(IBA). Protein extracts of light grown Dvvd and Dvvd, cpc-1-vvdStrep corre-
sponding to 150 mg protein were applied to 400 ml Strep-Tactin Superflow
and incubated for 1 hr at 4 C under constant rotation. After several washing
steps, the resin was packed into a column and proteins were eluted with
desthiobiotin in 250 ml fractions.
His6-VVD36 was expressed in Eschericia coli BL21 (DE3) cells under control
of the T7 promoter. T7 RNA polymerase was induced by addition of 0.1 mM
IPTG for 18 hr at 16 C. His6-VVD36 was purified by affinity chromatography with
HisTrap HP columns (GE Healthcare) as recommended by the manufacturer.
Yeast Transformation and Two-Hybrid Assay
The yeast strain PJ69-4A was transformed with 500 ng of plasmid DNA (Gietz
and Woods, 2002; Ito et al., 1983). Transformed cells were plated on SDC solid
medium lacking leucin and tryptophane (SDC -Leu -Trp). Single colonies were
picked, plated again, and checked for positive transformation by analysis of
protein expression.
For the yeast two-hybrid assay, cells were transformed with WC-1LOV-
GAL4-BD (bait) and either GAL4-AD-VVD or GAL4-AD- WC-1LOV (prey). Plas-
mids containing GAL4-BD-P53 and GAL4-AD-SV40 were used as positive
control, and GAL4-BD and GAL4-AD were used as negative control. Single
clones were inoculated on a sector of plates containing SDC -Leu -Trp and
SDC -Leu -Trp -His medium. Dilution series of positive clones were spotted
on plates for direct comparison of growth rate.
Electrophoretic Mobility Shift Assay
EMSA with g32P-ATP-labeled double-stranded oligonucleotide corresponding
to the proximal LRE (50 -CGCAGAGGACCCTGAACTTTTCGATCCGCTCGA
TCCCCTG GAA-30 ) was performed as described (Froehlich et al., 2002). Puri-
fied oligonucleotide (180 fmol) and 10–20 mg nuclear extract were incubated in
binding buffer (20 mM HEPES, 40 mM KCl, 1 mM EDTA, 10% glycerol, 0.2 mM
DTT, 10 mM MgCl2, 0.1% N-p40) on ice for 30 min. Poly dIdC (1 mg) was used
as unspecific competitor. Samples were loaded on a gel containing 4% acryl-
amide in TBE buffer. Gels were fixed in 10% acetic acid, dried, and analyzed
by autoradiography. Antibodies used for EMSA specifically were a-His6
(QIAGEN) and a-LOV, which was generated against a peptide obtained from
the WC-1 LOV-domain (RFLQAPDGNVEAGTKREF).
Analysis of Photoadduct Formation and Decay
Purified His6-VVD36 or Neurospora nuclear extract were prepared as
described above. For analysis of photoadduct formation, samples were
prepared in red light. His6-VVD36 and nuclear extracts were irradiated with
the indicated light intensities. Conditions with different light intensities were
established through the use of a variable number of neon tubes in combination
with light absorber foil. Light intensity was measured with a LI-250A Light
Meter (LI-COR Biosciences). For decay analysis, His6-VVD36 or nuclear
extracts were irradiated with saturating light (10 min, 130 mmol 3 m2 3 s1)
and transferred to darkness. WC-1 photoadduct formation and decay were
analyzed by EMSA, and the His6-VVD36 photoadduct was analyzed photomet-
rically by measurementof the absorbance spectrum of the FAD cofactor as
described (Zoltowski et al., 2007).
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures and
five figures and can be found with this article online at doi:10.1016/j.cell.
2010.08.010.
ACKNOWLEDGMENTS
We thank Johanna Scholz and Krisztina Makara for excellent technical
assistance. We are grateful to Cornelia Ulbrich for her experienced help with
setting up the yeast two-hybrid system. This work was supported by the
DFG grant BR 1375-3-1 to M.B. and T.S., SFB 638 and the FCI to M.B.,
GP1188 to E.M. and by an EMBO-HHMI startup grant by the Hungarian
OTKA (K 68960) and ETT (477-05) to K.K., M.B. is a member of CellNetworks
and S.C. is a member of HBIGS.
Received: February 16, 2010
Revised: June 16, 2010
Accepted: August 7, 2010
Published: September 2, 2010
REFERENCES
Bachleitner, W., Kempinger, L., Wülbeck, C., Rieger, D., and Helfrich-Förster,
C. (2007). Moonlight shifts the endogenous clock of Drosophila melanogaster.
Proc. Natl. Acad. Sci. USA 104, 3538–3543.
Belden, W.J., Larrondo, L.F., Froehlich, A.C., Shi, M., Chen, C.H., Loros, J.J.,
and Dunlap, J.C. (2007). The band mutation in Neurospora crassa is a domi-
nant allele of ras-1 implicating RAS signaling in circadian output. Genes Dev.
21, 1494–1505.
Brunner, M., and Schafmeier, T. (2006). Transcriptional and post-transcrip-
tional regulation of the circadian clock of cyanobacteria and Neurospora.
Genes Dev. 20, 1061–1074.
Cho, H.Y., Tseng, T.S., Kaiserli, E., Sullivan, S., Christie, J.M., and Briggs, W.R.
(2007). Physiological roles of the light, oxygen, or voltage domains of phototro-
pin 1 and phototropin 2 in Arabidopsis. Plant Physiol. 143, 517–529.
Christie, J.M. (2007). Phototropin blue-light receptors. Annu. Rev. Plant Biol.
58, 21–45.
Christie, J.M., Swartz, T.E., Bogomolni, R.A., and Briggs, W.R. (2002). Photo-
tropin LOV domains exhibit distinct roles in regulating photoreceptor function.
Plant J. 32, 205–219.
Corchnoy, S.B., Swartz, T.E., Lewis, J.W., Szundi, I., Briggs, W.R., and Bogo-
molni, R.A. (2003). Intramolecular proton transfers and structural changes
during the photocycle of the LOV2 domain of phototropin 1. J. Biol. Chem.
278, 724–731.
Dibner, C., Schibler, U., and Albrecht, U. (2010). The mammalian circadian
timing system: organization and coordination of central and peripheral clocks.
Annu. Rev. Physiol. 72, 517–549.
Doyle, S., and Menaker, M. (2007). Circadian photoreception in vertebrates.
Cold Spring Harb. Symp. Quant. Biol. 72, 499–508.
Elvin, M., Loros, J.J., Dunlap, J.C., and Heintzen, C. (2005). The PAS/LOV
protein VIVID supports a rapidly dampened daytime oscillator that facilitates
entrainment of the Neurospora circadian clock. Genes Dev. 19, 2593–2605.
Foster, R.G., Hankins, M.W., and Peirson, S.N. (2007). Light, photoreceptors,
and circadian clocks. Methods Mol. Biol. 362, 3–28.
Froehlich, A.C., Liu, Y., Loros, J.J., and Dunlap, J.C. (2002). White Collar-1,
a circadian blue light photoreceptor, binding to the frequency promoter.
Science 297, 815–819.
Gietz, R.D., and Woods, R.A. (2002). Transformation of yeast by lithium
acetate/single-stranded carrier DNA/polyethylene glycol method. Methods
Enzymol. 350, 87–96.
Harper, S.M., Neil, L.C., and Gardner, K.H. (2003). Structural basis of a photo-
tropin light switch. Science 301, 1541–1544.
He, Q., and Liu, Y. (2005). Molecular mechanism of light responses in Neuros-
pora: from light-induced transcription to photoadaptation. Genes Dev. 19,
2888–2899.
He, Q., Cheng, P., Yang, Y., Wang, L., Gardner, K.H., and Liu, Y. (2002). White
collar-1, a DNA binding transcription factor and a light sensor. Science 297,
840–843.
Heintzen, C., Loros, J.J., and Dunlap, J.C. (2001). The PAS protein VIVID
defines a clock-associated feedback loop that represses light input, modu-
lates gating, and regulates clock resetting. Cell 104, 453–464.
Helfrich-Förster, C. (2002). The circadian system of Drosophila melanogaster
and its light input pathways. Zoology (Jena) 105, 297–312.
Herzberg, C., Weidinger, L.A., Dörrbecker, B., Hübner, S., Stülke, J., and
Commichau, F.M. (2007). SPINE: a method for the rapid detection and analysis
of protein-protein interactions in vivo. Proteomics 7, 4032–4035.
Ito, H., Fukuda, Y., Murata, K., and Kimura, A. (1983). Transformation of intact
yeast cells treated with alkali cations. J. Bacteriol. 153, 163–168.
James, P., Halladay, J., and Craig, E.A. (1996). Genomic libraries and a host
strain designed for highly efficient two-hybrid selection in yeast. Genetics
144, 1425–1436.
Kaiserli, E., Sullivan, S., Jones, M.A., Feeney, K.A., and Christie, J.M. (2009).
Domain swapping to assess the mechanistic basis of Arabidopsis phototropin
1 receptor kinase activation and endocytosis by blue light. Plant Cell 21, 3226–
3244.
Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc. 771
Káldi, K., González, B.H., and Brunner, M. (2006). Transcriptional regulation of
the Neurospora circadian clock gene wc-1 affects the phase of circadian
output. EMBO Rep. 7, 199–204.
Kasahara, M., Swartz, T.E., Olney, M.A., Onodera, A., Mochizuki, N., Fuku-
zawa, H., Asamizu, E., Tabata, S., Kanegae, H., Takano, M., et al. (2002).
Photochemical properties of the flavin mononucleotide-binding domains of
the phototropins from Arabidopsis, rice, and Chlamydomonas reinhardtii.
Plant Physiol. 129, 762–773.
Krauss, U., Minh, B.Q., Losi, A., Gärtner, W., Eggert, T., von Haeseler, A., and
Jaeger, K.E. (2009). Distribution and phylogeny of light-oxygen-voltage-blue-
light-signaling proteins in the three kingdoms of life. J. Bacteriol. 191, 7234–
7242.
Liu, Y. (2003). Molecular mechanisms of entrainment in the Neurospora circa-
dian clock. J. Biol. Rhythms 18, 195–205.
Loros, J.J., Dunlap, J.C., Larrondo, L.F., Shi, M., Belden, W.J., Gooch, V.D.,
Chen, C.H., Baker, C.L., Mehra, A., Colot, H.V., et al. (2007). Circadian output,
input, and intracellular oscillators: insights into the circadian systems of single
cells. Cold Spring Harb. Symp. Quant. Biol. 72, 201–214.
McClung, C.R. (2008). Comes a time. Curr. Opin. Plant Biol. 11, 514–520.
McDonald, M.J., and Rosbash, M. (2001). Microarray analysis and organiza-
tion of circadian gene expression in Drosophila. Cell 107, 567–578.
Merrow, M., and Roenneberg, T. (2007). Circadian entrainment of Neurospora
crassa. Cold Spring Harb. Symp. Quant. Biol. 72, 279–285.
Merrow, M., Brunner, M., and Roenneberg, T. (1999). Assignment of circadian
function for the Neurospora clock gene frequency. Nature 399, 584–586.
Nayak, S.K., Jegla, T., and Panda, S. (2007). Role of a novel photopigment,
melanopsin, in behavioral adaptation to light. Cell. Mol. Life Sci. 64, 144–154.
Salomon, M., Christie, J.M., Knieb, E., Lempert, U., and Briggs, W.R. (2000).
Photochemical and mutational analysis of the FMN-binding domains of the
plant blue light receptor, phototropin. Biochemistry 39, 9401–9410.
Schafmeier, T., Haase, A., Káldi, K., Scholz, J., Fuchs, M., and Brunner, M.
(2005). Transcriptional feedback of Neurospora circadian clock gene by
phosphorylation-dependent inactivation of its transcription factor. Cell 122,
235–246.
Schafmeier, T., Káldi, K., Diernfellner, A., Mohr, C., and Brunner, M. (2006).
Phosphorylation-dependent maturation of Neurospora circadian clock
772 Cell 142, 762–772, September 3, 2010 ª2010 Elsevier Inc.
protein from a nuclear repressor toward a cytoplasmic activator. Genes Dev.
20, 297–306.
Schafmeier, T., Diernfellner, A., Schäfer, A., Dintsis, O., Neiss, A., and Brunner,
M. (2008). Circadian activity and abundance rhythms of the Neurospora clock
transcription factor WCC associated with rapid nucleo-cytoplasmic shuttling.
Genes Dev. 22, 3397–3402.
Schwerdtfeger, C., and Linden, H. (2001). Blue light adaptation and desensiti-
zation of light signal transduction in Neurospora crassa. Mol. Microbiol. 39,
1080–1087.
Schwerdtfeger, C., and Linden, H. (2003). VIVID is a flavoprotein and serves as
a fungal blue light photoreceptor for photoadaptation. EMBO J. 22, 4846–
4855.
Siepka, S.M., Yoo, S.H., Park, J., Lee, C., and Takahashi, J.S. (2007). Genetics
and neurobiology of circadian clocks in mammals. Cold Spring Harb. Symp.
Quant. Biol. 72, 251–259.
Smith, K.M., Sancar, G., Dekhang, R., Sullivan, C.M., Li, S., Tag, A.G., Sancar,
C., Bredeweg, E.L., Priest, H.D., McCormick, R.F., et al. (2010). Transcription
factors in light and circadian clock signaling networks revealed by genome-
wide mapping of direct targets for Neurospora WHITE COLLAR COMPLEX.
Eukaryot. Cell, in press. Published online July 30, 2010. 10.1128/EC.00154-10.
Swartz, T.E., Corchnoy, S.B., Christie, J.M., Lewis, J.W., Szundi, I., Briggs,
W.R., and Bogomolni, R.A. (2001). The photocycle of a flavin-binding domain
of the blue light photoreceptor phototropin. J. Biol. Chem. 276, 36493–36500.
Tan, Y., Merrow, M., and Roenneberg, T. (2004). Photoperiodism in Neuros-
pora crassa. J. Biol. Rhythms 19, 135–143.
Vitalini, M.W., de Paula, R.M., Park, W.D., and Bell-Pedersen, D. (2006). The
rhythms of life: circadian output pathways in Neurospora. J. Biol. Rhythms
21, 432–444.
Woelfle, M.A., and Johnson, C.H. (2006). No promoter left behind: global circa-
dian gene expression in cyanobacteria. J. Biol. Rhythms 21, 419–431.
Zoltowski, B.D., and Crane, B.R. (2008). Light activation of the LOV protein
vivid generates a rapidly exchanging dimer. Biochemistry 47, 7012–7019.
Zoltowski, B.D., Schwerdtfeger, C., Widom, J., Loros, J.J., Bilwes, A.M.,
Dunlap, J.C., and Crane, B.R. (2007). Conformational switching in the fungal
light sensor Vivid. Science 316, 1054–1057.
Zoltowski, B.D., Vaccaro, B., and Crane, B.R. (2009). Mechanism-based
tuning of a LOV domain photoreceptor. Nat. Chem. Biol. 5, 827–834.
Cell Flow Reorients the Axis
of Planar Polarity in the
Wing Epithelium of Drosophila
Benoı̂t Aigouy,1 Reza Farhadifar,2 Douglas B. Staple,2 Andreas Sagner,1 Jens-Christian Röper,1 Frank Jülicher,2,*
and Suzanne Eaton1,*
1Max Planck Institute of Molecular Cell Biology and Genetics, Pfotenhauerstrasse 108, Dresden 01307, Germany
2Max Planck Institute for the Physics of Complex Systems, Noethnitzer Strasse 38, Dresden 01187, Germany
*Correspondence: julicher@pks.mpg.de (F.J.), eaton@mpi-cbg.de (S.E.)
DOI 10.1016/j.cell.2010.07.042
SUMMARY
Planar cell polarity (PCP) proteins form polarized
cortical domains that govern polarity of external
structures such as hairs and cilia in both vertebrate
and invertebrate epithelia. The mechanisms that
globally orient planar polarity are not understood,
and are investigated here in the Drosophila wing
using a combination of experiment and theory.
Planar polarity arises during growth and PCP
domains are initially oriented toward the well-charac-
terized organizer regions that control growth and
patterning. At pupal stages, the wing hinge con-
tracts, subjecting wing-blade epithelial cells to
anisotropic tension in the proximal-distal axis. This
results in precise patterns of oriented cell elongation,
cell rearrangement and cell division that elongate the
blade proximo-distally and realign planar polarity
with the proximal-distal axis. Mutation of the atypical
Cadherin Dachsous perturbs the global polarity
pattern by altering epithelial dynamics. This mecha-
nism utilizes the cellular movements that sculpt
tissues to align planar polarity with tissue shape.
INTRODUCTION
The PCP pathway coordinates tissue planar polarity in vertebrate
and invertebrate epithelia (Simons and Mlodzik, 2008; Vladar
et al., 2009). PCP proteins localize to apical junctions and form
intracellularly polarized domains with different compositions.
In the Drosophila wing, these domains link proximal and distal
cell boundaries and orient wing hair outgrowth distally. Distal
complexes containing Flamingo (Fmi), a.k.a. Starry night (Stan),
Frizzled (Fz), Dishevelled (Dsh) and Diego (Dgo) interact across
cell boundaries with proximal complexes containing Fmi, Stra-
bismus (Stbm), a.k.a. Van Gogh (Vang) and Prickle (Pk). In the
absence of any single PCP protein, the others do not polarize
well and hair outgrowth is misoriented (Seifert and Mlodzik,
2007; Strutt and Strutt, 2005; Uemura and Shimada, 2003).
Feedback loops arising from interactions of PCP proteins within
and between cells may be sufficient for the local alignment of
PCP complexes between small groups of cells (Amonlirdviman
et al., 2005; Tree et al., 2002a; Tree et al., 2002b). Proposed
cellular mechanisms underlying local feedback loops include
preferential interactions between Fmi/Fz and Fmi/Stbm com-
plexes across cell boundaries, and the decreased susceptibility
of these complexes to endocytosis (Chen et al., 2008; Strutt
and Strutt, 2008; Wu and Mlodzik, 2008). Also, alignment of
microtubules may bias the delivery of Fz to the distal side of
the cell (Shimada et al., 2006).
While these mechanisms can polarize PCP domains intracellu-
larly and align them locally, less is understood about mecha-
nisms that specify global alignment of PCP domains with the
proximal-distal (PD) axis of the wing. The atypical Cadherins
Fat (Ft) and Dachsous (Ds), and a regulatory golgi kinase Four-
jointed (Fj) influence global orientation of PCP domains by
a mechanism that is unclear (Adler et al., 1998; Ma et al., 2003;
Matakatsu and Blair, 2004; Strutt and Strutt, 2002; Zeidler
et al., 2000). The Fat ligand Ds is expressed highly in proximal
regions that give rise to the hinge, but at lower levels in the
more distal wing blade, during both larval and pupal stages
(Cho and Irvine, 2004; Clark et al., 1995; Ma et al., 2003; Mata-
katsu and Blair, 2004; Strutt and Strutt, 2002). Fj, which regulates
their activity (Ishikawa et al., 2008) is expressed with an opposite
pattern (Villano and Katz, 1995; Zeidler et al., 2000). These
opposing patterns have been proposed to cause intracellular
asymmetries that directly bias accumulation of core PCP
proteins in wing cells (Ma et al., 2008, 2003; Strutt and Strutt,
2002; Tree et al., 2002a). However this is not the case in the
abdomen where Ft/Ds and PCP proteins act independently to
control trichome polarity (Casal et al., 2006). Even in the wing,
Ft and Ds do not act at the time that PCP domains actually align
with the PD axis, but much earlier (during larval or prepupal
stages), to influence the global PCP pattern (Matakatsu and
Blair, 2004; Strutt and Strutt, 2002). Ft and Ds regulate both
the amount and orientation of proliferation in the larval wing
epithelium (Baena-Lopez et al., 2005; Bryant et al., 1988; Clark
et al., 1995; Garoia et al., 2005, 2000). It is not yet clear whether
this activity is relevant to the PD alignment of PCP proteins
during pupal stages.
Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 773
 A B
50 µm
15 hAPF
C D
20 µm
16 hAPF 20h50 APF
Figure 1. Time Evolution of Planar Polarity
(A and B) The magnitude and axis of average nematic order (yellow bars) overlaid on Stbm:YFP-expressing wings, at 15 hAPF (A) and 30 hAPF (B). Red dots
indicate Fz:YFP domain polarity. Green ellipses indicate the anterior crossvein.
(C–E) Stbm:YFP at the indicated times between veins L1 and L3. Yellow bars show the PCP nematic for each cell.
(F) Magnitude (arbitrary units) and axis of average nematic order (left) and cell elongation (right) as a function of time (indicated on the y axis and by color) derived
from the wing shown in (C–E).
In all figures, anterior is top and proximal is left.
See also Figure S1.
Our previous experiments suggested that PCP domain
polarity does not develop de novo during pupal stages; PCP
domains are polarized in both larval wing discs and prepupal
wings, but polarity is not aligned with the PD axis. At early pupal
stages, the PCP axis is oriented at an angle to the PD axis
(Classen et al., 2005). Later, by the time wing hairs form, PCP
domains are oriented along the PD axis of the wing. Alignment
with the PD axis occurs during a phase in which wing epithelial
cells are exchanging their cell contacts (Classen et al., 2005).
One consequence of remodelling is an increase in hexagonal
order of the wing epithelium. Theoretical analysis suggested
that different types of fluctuations could guide epithelial cells
initially disordered by proliferation toward a hexagonal state
(Farhadifar et al., 2007). Here, we investigate the relationship
between these cell rearrangements and the temporal evolution
of PCP orientation patterns. We quantitatively analyze time-
lapses of pupal wing epithelia and combine these data with theo-
774 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.
30 hAPF
E F
Time [hAPF]
32h20 APF
PCP order elongation
retical analysis to extract key mechanisms that couple cell rear-
rangements to PCP reorientation.
RESULTS
Planar Polarity Points Initially toward the Wing Margin
and Reorients Distally
To investigate how PCP order evolves during pupal develop-
ment, we imaged wings expressing Stbm:YFP between 15 and
34 hr after puparium formation (hAPF). We developed a method
to quantify planar polarity based on the cell perimeter intensity of
Stbm:YFP. This method quantifies the axis and magnitude of
polarity, but not its vector orientation (Supplemental Theoretical
Procedures, 1.1 and Figures S1A–S1E available online). We
started by quantifying the axis of PCP at 15 hAPF (Figure 1A);
we also determined the polarity vectors in these wings by
creating Fz:YFP-expressing clones (Figures S1F–S1H). As
suggested previously (Classen et al., 2005), PCP at early stages
is correlated over long distances (Figure S1M) and is oriented
such that Fz:YFP domains face the wing margin (Figures 1A
and 1C and Figures S1F–S1I). Starting at about 18 hAPF, the
magnitude of the average alignment of groups of cells (local
nematic order, see Supplemental Theoretical Procedures, 1.2)
begins to decrease (Figures 1D and 1F and Figure S1J), and
the average axis of alignment begins to rotate toward the PD
axis (Figure 1F and Figures S1J and S1K). Reduced alignment
reflects both reduced polarity of individual cells and reduced
polarity correlation over distance (Figures S1L and S1M). After
reaching a minimum about 20 hAPF, the magnitude of average
nematic order increases; by 26 hAPF nematic order is again
maximal and aligned parallel to the PD axis (Figures 1B, 1E,
and 1F, Figures S1J and S1K, and Movie S1). Where we have
determined the direction of polarity, we quantify average polar
order by a measure ranging from zero to one (one corresponding
to perfect alignment) (Supplemental Theoretical Procedures,
1.3). At 15 hAPF, polar order averaged over the whole wing blade
is only 0.60, because polarity is oriented differently in anterior
and posterior wing regions (Figure 1A). By 32 hAPF average polar
order reaches 0.98 (Figure 1B). Thus, PCP domains do not
develop polarity de novo during pupal stages, but rather reorient
pre-existing polarity.
Polarity reorientation is reflected in the influence of fz mutant
clones on the polarity of adjacent wild-type tissue. At 15 hAPF,
fz clones perturb polarity of tissue lying between the clone and
the wing margin (Figures S1N and S1O) consistent with the early
polarity pattern. After reorientation, polarity is disturbed in wild-
type tissue lying distal to the clone (Vinson and Adler, 1987).
We noted that reorientation of PCP correlated with a transient
cell shape deformation that elongates cells in the PD axis
(Figure 1F). Quantifying elongation (Supplemental Theoretical
Procedures, 2) reveals that average polarity shrinks in magnitude
and reorients toward the PD axis as cells elongate. The magni-
tude of average polarity along the PD axis then increases as
cell elongation decreases (Figure 1F). The precise temporal
correlation of these events suggests that they are controlled by
the same underlying mechanisms.
Dramatic Morphogenetic Movements Reshape the Wing
during Pupal Development
To explore global changes occurring in the wing as cells change
shape and PCP reorients, we imaged ECadherin:GFP-express-
ing wings at low magnification from 15 to 32 hAPF. Dramatic
morphogenetic movements reshape the wing during this time
(Figures 2A–2C and Movie S2). At 15–16 hAPF, the hinge and
blade regions are roughly equal in size and the shape of the hinge
is not yet recognizable (Figure 2A). Between 15 and 24 hAPF, the
hinge undergoes patterned contractions that halve its area
and shape the allula and costa (Figures 2A–2C, Movie S2, and
Movie S3A). While wing-blade area remains constant during
this time, wing shape changes – elongating in the PD axis and
narrowing in the AP axis (Figures 2A–2C and Movie S2).
To quantify tissue movements underlying this shape change,
we tracked different regions and calculated local velocity vectors
(Supplemental Theoretical Procedures, 3). As the hinge con-
tracts, wing-blade cells flow proximally toward the hinge at
different velocities (Figures 2D–2F, Movie S3B). Between 15
and 18 hAPF, velocities have both inward and proximal compo-
nents (Figure 2D). Later, flow is proximally oriented and fastest in
the middle of the wing blade (Figure 2E). These inhomogeneous
velocities define local compression, shear and rotation as cells
move with respect to each other (Supplemental Theoretical
Procedures, 4). The shear axis in a particular region can be
thought of as the local axis of tissue deformation. In the
wing, local shear axes are oriented in a fan-like pattern that is
roughly symmetric about the third wing vein (Figures 2G–2I,
Figures S2J and S2K, and Movie S3C). Although the wing as
a whole does not rotate, local rotation does occur and is mainly
clockwise anterior to the third wing vein and anti-clockwise
posterior to it (Figures 2J–2L, Figure S2L, and Movie S3D). These
patterns of shear and rotation lengthen the wing blade in the
PD axis and narrow it in the anterior-posterior (AP) axis. Strik-
ingly, the observed antisymmetric rotation pattern would shift
early margin-oriented polarity toward the distal direction.
To investigate whether the tissue flows that reshape the wing
blade were driven by hinge contraction, we completely severed
the hinge from the blade before contraction occurred (Figure 3A).
After severing, wing-blade tissue flows distally rather than prox-
imally (Figure 3C and Movie S4A). Shear is reduced and is mainly
perpendicular to the PD axis (Figure 3D and Movie S4B), rather
than forming a fan-like pattern. Rotation is also reduced, and
its orientation is reversed compared to unperturbed wings
(compare Figure 3E with Figure 2K, and Movie S4C with Movie
S3D). As a result, the wing blade undergoes opposite shape
changes to those that normally occur. Nevertheless, wing hairs
(not shown) and longitudinal veins (Figure 3B) form normally.
These data suggest that hinge contraction exerts mechanical
stresses that contribute to the observed cell flow patterns that
change the shape of the wing blade.
Cell Boundary Tension Is Elevated along the PD Axis
during Hinge Contraction
To ask how cell boundary tension changes during hinge contrac-
tion, we used a pulsed UV laser to sever cell boundaries lying at
different angles to the PD axis, and monitored the movements
of adjacent vertices. The initial velocity with which vertices move
apart is a measure of cell boundary tension. Before hinge contrac-
tion, tension is similar on all cell boundaries. When the hinge
contracts, tension increases specifically on those cell boundaries
lying at angles close to that of the PD axis (Figures S3A and S3B).
This suggests that hinge contraction exerts forces on the blade
that direct cellular flows via an anisotropic stress profile.
Tissue Flow in the Wing Blade Results from Oriented
Cell Divisions, Cell Elongation, and Neighbor Exchanges
To investigate the cellular events underlying tissue flow,
shear, and rotation, we tracked groups of cells between veins 3
and 4, starting 15 hAPF. This revealed that local tissue shape
change occurs in two distinct phases. The first is dominated by
oriented cell division and PD cell elongation, and the second by
oriented cell rearrangements (Figure 4 and Movie S5A).
During phase I (about 15–24 hAPF), cell elongation increases
in the PD axis (Figures 4A–4C and 4G). Cell boundary loss
exceeds new boundary formation (Figure 4G), as confirmed by
Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 775
A B
* C
*
15h30' 80 µm
ubi-Ecad:GFP 24h00' 33h30'
D E F
50 µm/h

L3 L3 L3

80 µm
ubi-Ecad:GFP 24h00' 31h40'
G H I -1
0.3 h

J K L -1
0.1 h
-1
0.05 h
-1
0.01 h

anti-clockwise clockwise

Figure 2. Tissue Shape Changes during Hinge Contraction

(A–C) Wing images from a movie spanning roughly 15-33 hAPF. The hinge (blue) contracts, the wing blade (red) elongates. Allula (arrowhead) and costa (asterisk)
are indicated. The area of the wing hinge is 102 3 103 mm2 in (A) and 50 3 103 mm2 in (C). The area of the wing blade is 121 3 103 mm2 in (A) and 127 3 103mm2 in (C).
(D–L) Locally averaged features of the cellular flow field at indicated times.
(D–F) Flow velocity vectors. Speed corresponds to arrow length as indicated.
(G–I) Rate and axis of pure shear. Bar length indicates shear rate.
(J–L) Local rotation rate (indicated by circle size) in radians per hour.
L3 indicates the third longitudinal vein.
Results shown are representative of five separate experiments.
See also Figure S2.

the decrease in average neighbor number (Figure 4F). These new
cell boundaries have no preferred direction during phase I
(Figure 4I). Cells undergo between one and two rounds of
oriented cell division that reduce cell size. Between veins 3
and 4, these divisions are on average oriented 20
to the PD
axis (Figure 4H). As phase I ends, cell elongation peaks (Fig-
ure 4G) and cell division stops (Figure 4F). At the end of this
phase, tracked tissue patches have deformed in a way that
can be accounted for by the combination of cell elongation
and oriented cell division. During phase II (about 24–32 hAPF),
cells assemble new contacts parallel to the PD axis (Figures
4G and 4I). These oriented neighbor exchanges relieve cell elon-
gation (Figure 4G) and stabilize the tissue shape change caused
by cell elongation during phase I (Figures 4C–4E). They also
increase average neighbor number and the fraction of hexago-
nally packed cells (Figure 4F).
Severing the Hinge from the Wing Blade Blocks PD
Elongation and Misorients Cell Divisions and Neighbor
Exchanges
To investigate the cellular basis of altered shear and rotation in
severed wing blades, we quantified cell elongation and the
orientation of neighbor exchanges and cell divisions after
severing from the hinge (Figures S3C–S3G and Movie S5B).

776 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.

A Figure 3. Tissue Shape Changes in Severed Wings
B
15h30' Stbm:YFP 32h00'
D as indicated.
C 0.3 h
50 µm/h
L3
80 µm K ments.
22h30' 22h30'
-1 -1 -1
E 0.1 h 0.05 h 0.01 h F
22h30' anti-clockwise clockwise 32h00'
Cell divisions are strongly oriented in unperturbed wings. When
the hinge is severed from the blade, cell division continues
(Figure S3H), but the angle of cell division is more disperse
(Figure S3J). In addition, these cells do not elongate further along
the PD axis (Figure S3I), becoming instead more isotropic. While
neighbor exchanges still occur, their axis is shifted with respect
to unwounded wings, lying at ±45 degrees with respect to the PD
axis (compare Figure S3K to Figure 4I). Finally, wing severing
reduces the final number of hexagonally packed cells
(62.44% ± 3.5% in wounded wings [n = 3] versus 77% ± 4.8%
in unperturbed wings [n = 5]). Thus, severing the wing dramati-
cally changes the observed patterns of cell elongation, cell divi-
sion, and rearrangements. This suggests that anisotropic stress
caused by hinge contraction has a key role in guiding the tissue
flow that reshapes the wing blade. It also suggests that cell flow
increases hexagonal packing, similar to the annealing effect of
cell boundary fluctuations (Farhadifar et al., 2007).
Severing the Hinge Alters the Global Pattern
of Planar Polarity
Reorientation of planar polarity in the intact wing occurs at the
same time that the wing blade is reshaped by hinge contraction,
suggesting that these events depend on one another. The
observed pattern of clockwise and anti-clockwise tissue rotation
tends to reorient early margin-directed polarity toward the PD
axis. In addition, the shear pattern presages the final orientation
of planar polarity (Figure 2H). To ask whether tissue movements
induced by hinge contraction guided polarity reorientation, we
examined the distribution of Stbm:YFP in wing blades that had
been severed from the hinge before contraction. In severed
wings, the global PCP pattern differs from that of the intact
wing (compare Figure 1B with Figure 3F), suggesting that polarity
(A and B) Images taken at the indicated times from a movie
spanning 15–32 hAPF. The hinge was severed from the blade
along the line indicated.
(C–E) Locally averaged features of the cellular flow field at
22h300 APF.
80 µm (C) Flow velocity vectors. Speed corresponds to arrow length
-1
(D) Rate and axis of pure shear. Shear rate corresponds to bar
length as indicated.
(E) Local rotation rate (indicated by circle size) in radians
per hour.
(F) Magnitude and axis of average nematic order (yellow bars)
overlaid on a severed Stbm:YFP-expressing wing at 32 hAPF.
Green ellipse = anterior crossvein. L3, third longitudinal vein.
Results shown are representative of four separate experi-
is altered by the changed patterns of cell rearrange-
ments, cell division, and cell elongation. This
implies that cell flows guide reorientation of PCP.
Oriented Cell Boundary Assembly Increases
Stbm:YFP the Visibility of Planar Polarity
While the majority of polarity reorientation occurs
during phase I, average PCP order increases in
magnitude during phase II, as cells extend new contacts mostly
parallel to the PD axis and develop hexagonal packing geometry
(Figure 1F and Figures 4F and 4I). Could the increase in polar
order be related to these processes? To investigate this, we
imaged PCP protein localization as new cell boundaries formed
(Figures 5A–5F and Movie S5C). We noted that Stbm:YFP had
a clustered rather than a uniform distribution on cell boundaries
throughout pupal development. Even at early stages, these clus-
ters contain both Stbm:YFP and Fz:CFP, suggesting that they
constitute PCP protein complexes (Figures S4A–S4C). Consis-
tent with this, Stbm:YFP clustering is reduced in fz mutant tissue
(Figures S4D and S4E). Individual clusters are persistent and can
be tracked for several hours. However, proteins within them turn
over more rapidly (Figures S4G–S4I). Clusters do not form rapidly
on new cell boundaries derived either from cell division or
neighbor exchange (Figure 5 and Movie S5C). During phase II,
new cell boundaries, expanding parallel to the PD axis within
minutes, typically remain devoid of these clusters for hours
(Figures 5A–5H). Consequently, Stbm:YFP clusters become
restricted mainly to older cell interfaces on the proximal and
distal sides of the cell. Thus, oriented cell boundary expansion
and the concomitant increase in hexagonal packing order ensure
that PCP clusters become well separated within cells and better
aligned between cells; even though PCP has already reoriented
during phase I, oriented cell boundary expansion during phase II
tends to increase the average nematic order (compare
Figure S4F with Figures S1A—S1E).
A Theoretical Analysis of Polarity Reorientation
While local tissue rotation clearly implies local rotation of the
polarity axis, it is less simple to understand how tissue shear
affects the axis of planar polarity. Tissue shear is caused by
Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 777
 phase I phase II

A P D B P DC P D D P D E P D

20 µm
15 hAPF 19h30 APF 24 hAPF 28h30 APF 31h30 APF
persistent disappearing new boundary from neighbor exchange new boundary from cell division

F G
1 6 1
phase I phase II 0.32
5.9 cell number (fraction of maximum)
average neighbor number

0.8 0.8 0.28

average neighbor number
5.8

PD elongation
0.24
0.6 0.6

fraction
fraction hexagons
fraction

5.7 0.2
PD cell elongation 0.4
0.4 0.16
5.6 new cell contacts
(fraction of maximum) 0.12
0.2 0.2
5.5 cell contacts that will disappear
phase I phase II (fraction of maximum) 0.08
0 5.4 0
14 16 18 20 22 24 26 28 30 32 14 16 18 20 22 24 26 28 30 32
time (in hAPF) time (in hAPF)

H 90 90 I 90 90

0 180 0 180 0 180 0 180

270 270 270 270

phase I phase II phase I phase II
orientation of division axis orientation of new boundary assembly

Figure 4. Cell Elongation, Division, and Rearrangement

(A–E) A group of cells anterior to the posterior crossvein was tracked between 15h and 31h300 APF. Cell boundaries are color-coded depending on their fate:
red boundaries persist indefinitely, yellow boundaries disappear, green boundaries form during the movie as a result of neighbor exchange, and blue boundaries
form between daughter cells upon division.
(F and G) Quantification of cellular changes in the patch of tissue tracked in (A–E). Fraction of maximal cell number (dark blue), average neighbor number (light
blue, averaged over five frames), fraction of hexagonal cells (brown, averaged over five frames), PD cell elongation (magenta, average maximum value = 0.388 ±
0.053, n = 5), fraction of boundaries that will disappear (yellow), and the fraction of newly formed boundaries from neighbor exchange (green).
(H and I) Angular distribution of cell divisions (blue) (H) and new cell boundaries (green) (I) at the end of phase I and phase II. Yellow and magenta bars indicate
average angle of nematic order of cell division (yellow) and new cell boundary formation (magenta) (see Supplemental Theoretical Procedures). Bar length indi-
cates the degree of focus (magnitude of average nematic order, ranging from 0 to 1, see Supplemental Theoretical Procedures, 1.2). The outer circles in (H) and (I)
correspond to magnitude = 0.4. Average magnitude of phase I cell division order is 0.44+/0.12, n = 4. Average magnitude of new boundary formation order in
phase II = 0.35+/0.04, n = 4.
Data are presented as mean ± SD.
See also Figure S3.

oriented cell divisions, cell elongation, and oriented cell rear- PD axis at this time (Shimada et al., 2006). We turned to theoret-
rangements. Because microtubules have been shown to align ical analysis to explore how each of these processes might affect
with the axis of cell elongation (Cortes et al., 2006; Daga and the axis of planar polarity. Previously, we used a vertex model to
Nurse, 2008; Haase and Lew, 2007; Strauss et al., 2006), cell study the effects of proliferation on cell packing geometry
shape may directly influence the distribution of PCP within the (Farhadifar et al., 2007). Here, we add to this model a simplified
cell. Indeed, microtubules in wing epithelial cells align with the description of the dynamics of PCP order.

778 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.

 A 7 B Stbm:YFP A Physical Description of PCP in the Vertex Model
14
6
Cell shape, as defined by the network of adherens junctions, can
15 11
19 16 5 be described as force-balanced configurations in a vertex
8
20 17
12
3 1 model, in which an arrangement of cells is represented by a
18 13 9
4
set of cell bonds. This model takes into account cell mechanics,
2
21 hAPF 10 cell adhesion and cell division (Farhadifar et al., 2007). Our
C 7a
7b
D description of PCP order is based on attractive interactions
14
6
19 15
11 5 between proximal and distal PCP domains across cell bound-
8b
16 12 8a 3
1 aries and repulsive interactions within cells. This differs from
20 17 9 2 previous approaches (Amonlirdviman et al., 2005; Le Garrec
13 4
18 et al., 2006), in that we do not attempt to describe all details of
25 hAPF 10
14 interactions between PCP proteins, but rather consider effective
E 19 11
6 7b F
16
15 8b 5 interactions between proximal and distal domains. Each cell
8a
20
17
12
9 3
1
bond i, separating cells a and b, is assigned two variables, sai
18
13
10
4 2 and sbi , which describe the level of PCP proteins on either side
8 µm of this bond (Figure 6A). The variables sai can take continuously
29 hAPF varying values; sai = 1 corresponds to a high level of proximal
G 90 H 90 proteins (blue), while sai =  1 corresponds to a high level of
distal proteins (red). We introduce a potential function for interac-
tions of PCP domains:
0 180 0 180
X X X
E = J1 sai sbi  J2 sai saj  J3 3a $Qa ; (1)
i fi;jg a
270 270
I where J1 and J2 are PCP interaction parameters and J3
describes coupling of PCP to cell elongation (for J3 = 0, PCP
does not couple to cell elongation). For J1 >0 the potential favors
J K interactions between proximal and distal proteins across cell
bonds. For J2 >0, accumulation of proximal and distal proteins
on neighboring bonds within a given cell is disfavored. For
J3 >0, alignment of PCP with the long axis of the cell is favored:
L M this is motivated by the idea that aligned microtubules bias the
transport of PCP proteins (Shimada et al., 2006). The first sum
is over all bonds i, the second sum is over all pairs fi; jg of adja-
cent bonds in the network, and the third sum is over all cells a.
N O
The cell shape and PCP nematic order are described by the
tensors 3a and Qa , respectively (Supplemental Theoretical
Procedures, 5). We introduce a time evolution of PCP level on
P Q cell bonds: dsai =dt =  gvE=vsai , where t is time and g is a kinetic
coefficient that determines the PCP relaxation time t = 1=ðgJ1 Þ.
We solve these dynamic equations numerically, imposing two
R S constraints: (1) in each cell, amounts of proximal and distal
proteins are equal and (2) these amounts do not change with
time (Supplemental Theoretical Procedures, 5).
T U

Global Polar Order Generated by Network Growth

Figure 5. Cell Rearrangements Increase the Visibility of PCP
(A–F) Images from a 21–29 hAPF movie of a Stbm:YFP-expressing wing. In (A),
(C), and (E), cell identity is indicated by numbers and cell boundaries are
colored according to their fate as in Figure 4. (B), (D), and (F) show PCP
nematics for each cell (yellow).
(G and H) Angular distribution of newly formed (green, [G]) and permanent
boundaries (red, [H]).
(I) Each kymograph line depicts the linearized perimeter of a selected cell at
time t.
(J–O) Perimeter intensity kymographs for two cells of the movie (frame rate =
60 , movie length = 8 hr) shown in (A–F). In (L) and (M), each boundary is
identified by a random color. Colors in (N) and (O) indicate boundary fate
To examine how different types of shear reorient polarity, we
must start from a cellular configuration in the vertex model with
(see Figure 4). Permanent boundaries (red in [N] and [O]) carry PCP clusters
that persist for 8 hr. New boundaries (green in [N] and [O]) remain devoid of
Stbm:YFP clusters.
(P-U) Kymographs of a nondividing cell (P), (R), and (T) or a dividing cell (Q), (S),
and (U) between 20-21 hAPF (300 frame rate). In (R) and (S), each boundary is
identified by a random color. Colors in (T) and (U) indicate boundary fate (see
Figure 4). After cell division (Q), (S), and (U) depicts the joint perimeter of the
two daughter cells and the boundary created by cell division (blue in [U])
separately.
See also Figure S4.

Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 779


A B
α Rotation on the Reorientation of Planar
σ iα
σ iβ
β and type of PCP domains on bond i shared by cells
C D
E F
G H
globally aligned PCP variables. To generate such a configuration
without an external bias is nontrivial (Burak and Shraiman, 2009).
Surprisingly, a simple and general way to generate large
networks with long-range polarity is to start from a small group
of cells with an initially random PCP configuration (random
values of sai , Supplemental Theoretical Procedures, 6). This
network is then expanded by simulating stochastic cell divisions
at a rate kd . Simultaneously, the dynamic equations for the PCP
variables are solved. Interactions of PCP variables generate
local order, which aligns over the whole network when it is
still small. Order is maintained during growth and leads to glob-
ally aligned polarity in the resulting large networks if the PCP
relaxation time t is sufficiently fast (kd t<1, Figures 6B, Fig-
780 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.
Figure 6. Theoretical Analysis of Shear and
Polarity
(A) Schematic representation of PCP in the vertex
model. The variables, sai and sbi , represent the level
a and b.
(B) Cell packing with PCP order generated by
simulated proliferation in the vertex model with
PCP dynamics, starting from 36 randomly polar-
ized cells with kdt = 0.01, J2/J1 = 0.5 and J3 = 0.
Arrows in (B)–(D) show the direction of polarity.
(C and D) Reorientation of polar order by shear due
to external forces (C) or oriented cell division (D).
Starting from the network (B), and simulating shear
along the horizontal axis, leads to the networks
shown. Parameter values for (C) are kdt = 0.01
and J3/J1 = 0.05. For (D), kdt = 0.01, J3 = 0 and
3 rounds of cell division were simulated. The
shear generated corresponds to an aspect ratio
of 4 in (C) and 1.5 in (D).
(E and F) Calculations of polarity reorientation
based on Equation 2 and measured patterns of
shear and rotation. Starting with the observed
early polarity (E), the time evolution described by
Equation 2 with n =  3 generates a final PD
pattern (F). The wing is the same as that shown
in Figures 2D–2L.
(G and H) same procedure as in (E and F) applied to
the wounded wing shown in Figure 3 with n = 2. We
start with an initial pattern mimicking the early
polarity (G).
See also Figure S5.
ure S5A, and Movie S6A). This is consis-
tent with the observation of global PCP
patterns in the growing wing disc
(Classen et al., 2005), and suggests that
these patterns arise early during
development.
Shear Reorients the Polarity Axis in
the Vertex Model
Starting with a globally polarized
network, we used the vertex model to
examine how different types of shear
could influence the average polarity
orientation of groups of cells. We first studied the effects of
shear caused by cell elongation and oriented cell rearrange-
ments. In our simulations, we induced pure shear at a rate ks
by forcing the network to elongate along one axis at constant
total area (Supplemental Theoretical Procedures, 6). We
oriented the initial axis of average polarity at an angle of 45
degrees to this shear axis (Figure 6B). As the network is
deformed, cells elongate and undergo T1 transitions that are
oriented along the stretch axis. We find that shear reorients
polarity either parallel or perpendicular to the shear axis, de-
pending on the value of ks t and on the strength J3 of the
coupling of cell shape to PCP distribution (Figures 6C, Figures
S5B–S5D, and Movies S6B—S6C). Reorientation occurs during
a characteristic time 1=j2nks j, where n is a dimensionless coef-
ficient. We define n to be negative if polarity aligns with the
shear axis, and positive if it aligns perpendicular to this axis.
Positive values of J3 promote the alignment of the polarity
vector with the shear axis, as do faster relaxation times t of
PCP (Figures S5B–S5D). Network configurations generated by
this process are irregular and cells are stretched. We relaxed
these networks by introducing fluctuations of cell bond tension
(Supplemental Theoretical Procedures, 6). This reduced cell
stretch and increased both the fraction of hexagons and the
magnitude of average polar order, similar to what is observed
during phase II in vivo (Figures 4F and 4G, and Figure S5E).
We next studied whether shear generated by oriented cell divi-
sions reorients polarity. We imposed cell divisions in the vertex
model at rate kd oriented at an angle of about 45 degrees to
the initial polarity axis (Supplemental Theoretical Procedures,
6). Oriented cell division generates shear parallel to the cell divi-
sion axis (defined as the line connecting the centers of the two
daughter cells) at a rate ks = mkd , where m depends on network
mechanics. We find again that polarity is reoriented either
parallel or perpendicular to the shear axis and that the dynamics
of reorientation can be characterized by the coefficient n defined
above (Figure 6D, Figure S5F, and Movie S6D). Again, faster
relaxation times t of PCP promote alignment of the polarity
vector with the shear axis (Figure S5F).
An Effective Description of Large-Scale Polarity
Reorientation by Flow
The reorientation of average polarity by shear flow that we
observe in the vertex model is well known in complex fluids
such as liquid crystals. In these systems, polarity reorientation
is determined by local rotation and shear rates of the flow field
and can be described by the following equation (de Gennes
and Prost, 1993; Joanny et al., 2007)
vq
= nks sin2q + u: (2)
vt
Here, u is the local rotation rate, and ks is the rate of pure shear
(Supplemental Theoretical Procedures, 7). The orientation of
polarity is described by the angle q, relative to the axis of pure
shear. The coefficient n describes how pure shear influences
reorientation of polarity. In the absence of rotation, u = 0, pure
shear aligns polarity either with the shear axis ðq = 0Þ or perpen-
dicular to it ðq = p=2Þ depending on the sign of n. In liquid crystals,
the value of n depends on molecular geometries and intermolec-
ular interactions (de Gennes and Prost, 1993). In the absence of
pure shear, ks = 0, local tissue rotations reorient polarity as
vq=vt = u.
Equation 2 for u = 0 describes the main features of polarity
reorientation by pure shear in the vertex model described above.
The coefficient n that characterizes how shear affects reorienta-
tion in the vertex model is the same as the parameter n in Equa-
tion 2. We can thus use Equation 2 to determine quantitatively
how the observed patterns of rotation and shear reorient the
global pattern of planar polarity in the wing.
Starting with the observed polarity pattern at 15 hAPF, we
locally applied Equation 2, with n being the only free parameter,
to calculate the corresponding time-dependent reorientation of
polar order. Local rates of rotation and shear were obtained
from the experimentally determined flow field between 15 and
32 hAPF. To ask to what extent local rotation alone could repro-
duce the observed polarity reorientation, we first considered
n = 0, i.e., shear does not influence reorientation. In this case
polarity realigns toward the PD axis, however, the final average
polar order achieved is only 0.76 (77% of the experimental value
at 32 hAPF). Thus, local rotation can account for only part of the
observed reorientation of the polarity axis.
If n is nonzero, shear also influences reorientation. Repeating
this calculation for different values of n shows that the final
average polar order is largest for nz  3, reaching a value of
0.90 (92% of the experimental value, Figures S5G–S5J). For n
within the range 10<n<  1 the polarity field reorients robustly
to reach an average polar order of at least 0.85 (Figure S5K,
6E,F and Movie S6E), tolerating small deviations from the exper-
imentally determined pattern of initial conditions (Figure S5K).
Note that the effects of local shear on polarity orientation depend
on PCP domain interactions and dynamics, as captured in the
vertex model. In the calculations using Equation 2, these effects
are represented via the value of the parameter n. However, in
writing Equation 2 we choose to neglect interactions of average
polar order between neighboring patches of tissue to highlight
the role of shear and rotation. We expect such additional interac-
tions could account for the remaining deviation between the
solution to Equation 2 and the observed final polarity. Note
also that positive values of n produce a final polarity pattern
with lower polar order than rotation alone (n = 0) (Figure S5K).
This analysis suggests that shear caused by the combination
of oriented cell divisions, cell elongation and cell rearrangement
plays an important role in aligning polarity parallel to the shear
axis by a process that is characterized by a negative value of n.
The close agreement between the polarity pattern generated
by Equation 2 (Figure 6F) and experimental observations
(Figure 1B) provides strong evidence that tissue remodelling is
responsible for reorienting the axis of planar polarity.
To ask whether Equation 2 could describe the polarity defects
observed when the hinge and wing blade are severed, we used it
to calculate polarity reorientation given the shear and rotation
patterns measured for these wings. For wounded wings, we
find that Equation 2 describes the main features of polarity reor-
ientation provided that n is positive (Figures 6G and 6H, Figures
S5L and S5N, and Movie S6F). This implies that, in wounded
wings, shear tends to align polarity perpendicular to the shear
axis, rather than parallel to it, as it does in unperturbed wings.
Why should shear affect polarity differently in wounded wings?
In simulations of the vertex model, we observed that shear gener-
ated by oriented cell divisions and by external stresses reorients
polarity with different characteristics (Figures S5B–S5D and S5F).
The value of n thus depends on the types of cell shape changes
and rearrangements occurring during shear. In wounded wings,
where mechanical boundary conditions are perturbed, the re-
modeling processes that dominate shear differ from those in
unperturbed wings (compare Figures S3C–S3K with Figure 4).
In early pupal wings, fz clones nonautonomously reorient
polarity of adjacent wild-type tissue lying between the clone
and the wing margin (Figure S1N and S1O). But after remodeling,
Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 781
the location of affected wild-type tissue is distal to the fz clone
(Vinson and Adler, 1987). We wondered whether the shear and
rotation observed during remodeling could account for reposi-
tioning of affected wild-type tissue. To investigate this, we first
simulated growth in the vertex model mimicking the situation
of a fz mutant clone. Starting with a network of 36 cells, we chose
a single cell and set sai = 1 on all its bonds (corresponding to loss
of Fz domains). We then simulated growth of the network while
simultaneously solving the dynamic equation of the PCP network
(Supplemental Theoretical Procedures, 5), keeping sai = 1 in the
simulated clone (indicated in blue). In these simulations, the
polarity pattern becomes distorted on one side of the clone as
domains with sai <0 (red) are positioned to face the clone
(compare Figure S5O with S5Q). These simulations generate
PCP patterns that mimic domineering nonautonomy observed
in vivo for fz and stbm mutant clones. Such patterns suggest
the interesting possibility that domineering nonautonomy may
arise during growth of the wing and may explain why the polarity
of wild-type tissue is already reoriented by fz clones at 15 hAPF.
We use these patterns as initial conditions to study reorientation
during shear.
To investigate the effects of shear and rotation on the per-
turbed PCP polarity patterns, we solve Equation 2 with n =  3
for both networks with and without simulated clones, and an
initial average polarity axis lying 45
to the horizontal (mimicking
margin-oriented polarity). We impose shear and rotation at
constant rate, with the shear axis oriented horizontally (corre-
sponding to the PD axis of the wing) (Supplemental Theoretical
Procedures, 8). This reorients the global polarity of both net-
works and shifts the perturbed region toward the ‘‘distal’’ side
of the clone (compare Figure S5P and R) consistent with exper-
iments (Vinson and Adler, 1987).
Ds Guides Global Planar Cell Polarity by Influencing
Epithelial Dynamics
Perturbing the Ft/Ds pathway misaligns PCP domains, but the
underlying mechanism is controversial (Axelrod, 2009; Casal
et al., 2006; Lawrence et al., 2007; Ma et al., 2003; Tree et al.,
2002a). These mutations also cause shorter, broader wings –
at least in part by perturbing oriented cell divisions in larval discs
(Baena-Lopez et al., 2005). To investigate whether Ds also influ-
enced pupal epithelial remodelling, we analyzed time-lapses of
ds pupal wings expressing ECadherin:GFP. At 15 hAPF, ds
wings are already misproportioned; the hinge is overgrown and
the wing blade is short and broad (Figure 7A), consistent with
altered larval growth. Nevertheless, the hinge contracts as it
does in wild-type, reducing its area about twofold (Figures 7A
and 7B). Thus, Ds is not required for hinge contraction. Calcu-
lating shear and rotation over the whole blade is difficult in ds
mutants because the epithelium folds; in regions that can be
quantified, rotation rates are smaller than in wild-type (Figures
S2G–S2I0 and S2L and Movie S7A). However, differences in
the shear patterns could not be resolved (Figures S2D–S2F0 ).
A clearer picture emerged from analyzing ds mutants at the
cellular level (Figures 7D–7H and Movie S5D). PD cell elongation
is reduced in ds wings (compare Figure 4G and Figure 7J), and
the axis of cell division is less focused than in wild-type (compare
Figures 4H and 7K). Thus Ds is required to tightly focus cell divi-
782 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.
sion orientation in both larval and pupal wings. Simulations
suggest that both cell elongation and oriented cell divisions
can reorient polarity (Figures 6C and 6D). Interestingly, both
are perturbed in ds mutants.
Ds influences both larval and pupal epithelial dynamics—so
which is most important for development of PD polarity? To
investigate this, we altered Ds levels in the posterior compart-
ment of the wing either throughout development or only during
larval growth or after pupariation. Overexpressing Ds throughout
wing development causes strong polarity defects in both poste-
rior and anterior compartments (Figure S6A). Overexpressing Ds
during larval or pupal development alone causes much milder
polarity defects (Figures S6A–S6C). We obtain complementary
results by lowering Ds levels. Inducing RNAi against Ds in the
posterior compartment throughout development perturbs
polarity strongly in both anterior and posterior compartments
(Figure S6D). Ds knock-down during larval or pupal development
alone causes milder polarity defects (Figures S6E and S6F).
These data suggest that Ds is needed both during growth and
subsequent epithelial remodelling to ensure proper alignment
of PCP with the PD axis. Ds loss or overexpression throughout
development causes cumulative defects stronger than those
caused by stage-specific perturbation.
Loss of Ds during pupal stages may perturb PD polarity by
reducing cell elongation or misorienting cell divisions during
remodelling. How could loss of Ds during larval growth affect
evolution of PD polarity? One possibility is that Ds-dependent
oriented cell divisions in the wing disc guide development of
early margin-oriented polarity; complex global PCP patterns
are as already seen in the third instar, and prepupal wings
show a pattern consistent with margin-oriented polarity (Classen
et al., 2005). To test this, we quantified the Stbm:YFP pattern in
ds wings at 15 hAPF. The pattern of average nematic order in ds
wings deviates from that of wild-type over large regions,
although PCP domains are well aligned locally, (compare
Figure 7C and Figure 1A). These observations together with
simulations suggest that the larval pattern of oriented cell divi-
sions contributes to development of margin directed polarity.
Loss of Ds may perturb the early polarity pattern indirectly by
misaligning larval cell divisions.
To ask whether abnormal early polarity in combination with the
altered flow pattern could in principle produce normal PD
polarity, we started with the altered polarity observed in early
ds wings and applied the mutant velocity field using Equation
2. For all values of n, polarity fails to orient as it does in wild-
type (average polar order < 0.67, compare Figure S5M with
S5K). Taken together, these studies indicate that development
of PD polarity depends both on correct initial polarity and on
tissue rotation and shear produced during hinge contraction.
Although Ds is expressed at high levels in the hinge and at low
levels in the wing blade, this discontinuity along the PD axis is not
required to direct PD orientation of PCP domains. Uniform Ds
expression rescues polarity in a ds mutant background (Mata-
katsu and Blair, 2004 and Figure S6G-I). Furthermore, lowering
Ds levels in the wing blade perturbs polarity more than knock-
down in the hinge (Figure S6K,L). In contrast, discontinuities in
Ds levels along the AP axis do perturb polarity at a distance;
overexpression or loss of Ds in the posterior compartment
 A 15h30' APF B 30 hAPF C

2 2 2 2
212*103µm 123*103µm 123*103µm 146*103µm 15 hAPF
phase I phase II

D P D
E P D
F P D
G P D
H P D

20 µm
15h45 APF 19h30 APF 24 hAPF 28h30 APF 31h APF
persistent disappearing new boundary from neighbor exchange new boundary from cell division

I 6
J 1
1
phase I phase II 0.32
5.9 cell number (fraction of maximum)
average neighbor number

0.8 0.8 0.28

average neighbor number
5.8

PD elongation
0.24
0.6 0.6
fraction

fraction hexagons
fraction

5.7 0.2
PD cell elongation 0.4
0.4 0.16
5.6 new cell contacts
(fraction of maximum) 0.12
0.2 0.2
5.5 cell contacts that will disappear
phase I phase II (fraction of maximum) 0.08
0 5.4 0
14 16 18 20 22 24 26 28 30 32 14 16 18 20 22 24 26 28 30 32
time (in hAPF) time (in hAPF)

K 90 90 L 90 90

0 180 0 180 0 180 0 180

270 270 270 270

phase I phase II phase I phase II
orientation of division axis orientation of new boundary assembly

Figure 7. Early Polarity, Cell Elongation, Division, and Rearrangement Are Perturbed in ds05142 Wings
(A and B) A ds05142, ubi-ECad:GFP wing at 15h300 and 30 hAPF. The hinge is colored blue and the blade red. Numbers indicate blade and hinge areas in mm2.
(C) Average nematic order in a ds05142,act-stbm:YFP/ ds05142 wing at 15 hAPF (compare to Figure 1A).
(D–H) A group of cells anterior to the posterior crossvein in a ds05142, ubi-ECad:GFP wing was tracked between 15h450 and 31 hAPF. Cell boundaries are color-
coded as in Figure 4A.
(I and J) Quantification of cellular changes in the patch of tissue tracked in (D–H). Fraction of maximal cell number (dark blue), average neighbor number (light blue,
averaged over 8 frames), fraction of hexagonal cells (brown, averaged over 8 frames), PD cell elongation (magenta, average maximum values = 0.243, n = 2
(0.2245, 0.261), fraction of boundaries that will disappear (yellow), and fraction of new boundaries resulting from neighbor exchange (green).
(K and L) Angular distribution of cell divisions (blue) (K) and new boundaries (green) (L) at the end of Phase I and Phase II. Yellow and magenta bars indicate
average angle of nematic order of cell division (yellow) and new cell boundaries (magenta) (see Supplemental Theoretical Procedures and Figure 4). Average
magnitude of phase I cell division order is 0.145 (0.180, 0.109 n = 2). Average magnitude of new boundary formation order in phase II = 0.338 (0.278, 0.398;
n = 2).
See also Figure S6.

Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 783

nonautonomously perturbs polarity anteriorly (Figures S6A–
S6F). Altering Ds levels changes the response of wing epithelial
cells to hinge-dependent pulling forces (Figure 7). We therefore
considered the possibility that local changes in cell rearrange-
ment and elongation could alter large-scale cellular flows and
thus the patterns of shear and rotation in wild-type regions.
Indeed, Ds overexpression in the posterior compartment nonau-
tonomously reverses local tissue rotation in anterior cells near
the compartment boundary (compare Figures S6M–S6O, Fig-
ures 2J–2L, and Movie S7B). These observations suggest that,
to generate normal flow patterns in response to hinge contrac-
tion, cells in the anterior and posterior compartments must react
similarly to mechanical stress.
DISCUSSION
The mechanisms that couple tissue shape to the planar polarity
of constituent cells are not well understood. Here, we describe
a novel morphogenetic event that both shapes the wing of
Drosophila, and orients global planar polarity of wing cells. We
show that the hinge region contracts shortly after the prepupal
to pupal transition. This contraction not only shapes the wing
hinge, but also exerts anisotropic tension on the adjacent
wing-blade region, causing it to elongate and narrow. Reshaping
of the wing blade occurs as constituent cells flow proximally with
different velocities, generating reproducible patterns of flow,
shear and rotation. These patterns result from a combination of
oriented cell division, cell elongation and neighbor exchanges
that are guided by the anisotropic stresses.
This scenario, where externally generated forces have an
important role in epithelial remodeling differs from other con-
vergent-extension events where autonomous cell move-
ments actively drive tissue shape changes (Bertet et al., 2004;
Blankenship et al., 2006; Keller et al., 2008; Rauzi et al., 2008;
Zallen and Wieschaus, 2004). Externally induced epithelial
stretching has been proposed to provoke stress-relieving rear-
rangements during Drosophila germband extension (Butler
et al., 2009), and in the amphibian ectoderm (Keller et al.,
1992, 2008). But in these cases movements of the underlying
mesoderm, rather than contraction of adjacent epithelial cells,
provides the external force. The Drosophila pupal wing will
be a powerful system for uncovering molecular mechanisms
underlying stress-induced epithelial remodelling.
The mechanisms specifying global PCP domain alignment
have been elusive and controversial (Axelrod, 2009; Casal
et al., 2006; Lawrence et al., 2007; Ma et al., 2003; Tree et al.,
2002a). It had been thought that PCP proteins in the Drosophila
wing are initially randomly distributed, but become intracellularly
polarized along the PD axis starting approximately 10 hr before
hair formation (Strutt and Strutt, 2005; Vladar et al., 2009; Wu
and Mlodzik, 2009; Zallen, 2007). This led to the search for
a global polarizing signal that operated at this time. Here, we
show that PD orientation of PCP complexes does not arise
from a random distribution; rather it evolves from a different
global polarity pattern present at the prepupal-pupal transition.
At this stage, PCP domains are organized such that Fz-contain-
ing domains face the wing margin rather than distally. Our
previous work identified a similar global polarity pattern in
784 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.
prepupal wings, and showed that even larval wing discs show
a global pattern of PCP (Classen et al., 2005). Thus PCP domains
are always polarized, but their global orientation evolves dynam-
ically during development.
What causes global reorientation of PCP during phase I, and
the subsequent increase in the magnitude of polar order during
phase II of wing-blade remodeling? During phase I, margin-
oriented polarity is reoriented by specific patterns of local rotation
and shear due to cell flows caused by hinge contraction. Local
tissue rotation and shear reorient PCP domains by simple phys-
ical rules (see Equation [2] and Figure S7) similar to those that re-
orient molecular order in liquid crystal hydrodynamics. A signifi-
cant fraction of PCP reorientation is produced by local tissue
rotation alone. However, tissue shear also helps rotate polarity
toward the shear axis. Simulations suggest that shear caused
by oriented cell divisions, cell rearrangements and PD cell elonga-
tion can all contribute to polarity reorientation. One mechanism
that may underlie the effect of cell elongation on PCP is the
tendency of microtubules to align with the long axis of the cell.
Indeed microtubules are aligned with the PD axis of wing epithelial
cells and are essential for delivery of PCP proteins to the cortex
(Shimada et al., 2006). Significantly, under conditions where
normal PD polarity does not develop (e.g., in ds mutant wings
or in severed wings), oriented cell division, cell rearrangements
and cell elongation are disturbed. During phase II, oriented cell
boundary rearrangements increase hexagonal packing geometry
and improve PCP order. Simulations suggest this may occur via
an annealing process that relaxes irregular cell packing to a
more ordered hexagonal lattice with improved PCP order.
The atypical Cadherins Ft and Ds are needed to evolve the
global PD polarity pattern. Exactly how they influence the global
pattern has been controversial (Axelrod, 2009; Lawrence et al.,
2007). It was proposed that this pathway provides a PD polarity
cue that depends on higher relative levels of Ds expression in the
hinge. This inhomogeneity was thought to generate intracellular
asymmetry of Ft/Ds heterodimers throughout the wing, directly
generating a small bias in Fz activity within each cell that could
then be amplified to produce strong and stable alignment of
PCP domains with the PD axis (Ma et al., 2003; Tree et al.,
2002a). But the fact that uniform Ds expression suffices to
rescue polarity in ds wings argues against this view (Matakatsu
and Blair, 2004) (Figures S6G–S6I). Furthermore, Ft and Ds do
not appear to act directly on core PCP proteins in the Drosophila
abdomen (Casal et al., 2006). We propose that the Ft/Ds
pathway influences the global PCP pattern, not by generating
intracellular asymmetries that directly regulate PCP signaling,
but indirectly through its effects on epithelial dynamics during
both larval and pupal stages. Time-controlled loss of Ds shows
that it is required throughout development for evolution of PD
polarity (Figures S6D–S6F) (see also (Matakatsu and Blair,
2004)). During larval stages, Ds may guide the development of
early margin-oriented polarity through its influence on oriented
growth; our simulations suggest that oriented cell division
strongly influences the global axis of planar polarity. Later, during
hinge contraction and wing-blade remodeling, Ds is required
for the oriented cell division and cell elongation that occur in
response to anisotropic stresses. Both processes influence the
polarity axis in vertex model simulations.
 In future, it will be interesting to examine the roles of aniso-
tropic growth and local morphogen signaling in establishing
the global PCP pattern as wing discs grow. Our theoretical anal-
ysis shows that large scale planar polarity can be stably main-
tained during growth even in the absence of global polarizing
signals. This is possible because polarity easily aligns over
a few cell diameters without orientation defects, and subsequent
cell rearrangements due to cell division do not destroy this order
during growth. Thus, polarity that was initially established when
the tissue was small can be expanded during growth, and any
anterior-posterior or dorsal-ventral boundary-derived signal
would not need to act directly over long distances.
The coupling of planar polarity to epithelial dynamics is
a robust and flexible mechanism for coordinating tissue shape
with planar polarity of constituent cells and may generalize to
other systems. Indeed, mutations in the Wnt pathway perturb
both convergent extension and the planar orientation of sensory
hair cells in the mouse cochlea (Wang et al., 2005). Furthermore,
the planar polarity of sensory hair cells in the lateral line organ of
zebrafish is oriented by the direction of cell migration rather than
long-range secreted cues (Lopez-Schier et al., 2004). Our theo-
retical analysis suggests that shear caused by stretch-induced
rearrangements or by oriented cell divisions can reorient polarity
either perpendicular or parallel to the shear axis, depending on
the relative rates of shear and PCP turnover. Thus, this process
is also highly versatile, and could be a fundamental principle of
tissue organization.
EXPERIMENTAL PROCEDURES
Imaging
Pupae were prepared for imaging as described (Classen et al., 2008). Images
were acquired with a Leica TCS-SP2 or an Olympus FV-1000 microscope.
203 and 403 oil immersion objectives were used to follow cell flows and
global morphological changes occurring in the wing. A 633 oil immersion
objective was used to image pupal wings with cellular resolution. Imaging
was performed at 25
± 2
C, or at 29
C using a Bachhoffer chamber.
Wing Wounding
Wings were wounded by gently scratching the pupal cuticule in the hinge
region of the wing using forceps just before imaging (at around 15 hAPF).
Image Analysis and Figures
Z projections were created with ImageJ. For image analysis, we developed
‘‘packing analyzer v2.0’’, which measures cell areas, perimeter, packing, elon-
gation, polarity, cell divisions and tracks the cells and their boundaries. Briefly,
images are segmented using the watershed algorithm (Vincent and Soille,
1991). Each watershed catchment’s basin is defined as a cell. Vertices are
defined as pixels shared between three or more cells. Cell boundaries are
identified as pixels shared by exactly two cells. Cell boundaries smaller than
three pixels are reclassified as vertices. The polygon class is determined by
counting cell vertices. Cell area is defined as the number of pixels in each
cell. Cell perimeter is defined as the sum of the distance between all the pixels
surrounding a cell. To track cells, we assign a unique identity to each in the first
movie frame. Cells are then re-identified in the next images based on their
positions and neighborhoods. Boundaries are defined by the two cells that
share them. Division orientation is defined by a line connecting the centers
of the two daughter cells, and the angle it makes with the PD axis (which is
plotted as horizontal).
Movies created using our software were compressed using FFmpeg or
Quicktime Pro. Graphs were created using Microsoft Excel, Gnuplot and
Grace. Images were vectorized using Batik. Figures were composed with
Adobe Illustrator.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures,
Supplemental Theoretical Procedures, seven figures, and seven movies and
can be found with this article online at doi:10.1016/j.cell.2010.07.042.
ACKNOWLEDGMENTS
This work represents a truly collaborative effort. Each author has contributed
significantly to the findings and regular group discussions guided the develop-
ment of the ideas presented here. The manuscript was written jointly by all
authors. S.E., A.S., J.R., R.F., and F.J. were supported by the Max Planck
Gesellschaft. B.A. was funded by the Fondation pour la Recherche Médicale.
D.S. acknowledges the Natural Sciences and Engineering Research Council of
Canada and the German Academic Exchange Service. J.R. was supported by
a predoctoral fellowship from the Boehringer Ingelheim Fonds. We thank Tony
Hyman and Stephan Grill for use of the laser cutting microscope. We are grate-
ful to Stephan Grill, Ewa Paluch, Elisabeth Knust, and Carl-Philipp Heisenberg
for critical comments on the manuscript. We thank David Strutt, Seth S.Blair,
VDRC and the Bloomington Stock Center for providing fly stocks, Barry Dick-
son for DNA constructs, Julia Gäbel for help with cloning, Sven Ssykor for
transgenic injections.
Received: August 28, 2009
Revised: May 11, 2010
Accepted: July 23, 2010
Published: September 2, 2010
REFERENCES
Adler, P.N., Charlton, J., and Liu, J. (1998). Mutations in the cadherin super-
family member gene dachsous cause a tissue polarity phenotype by altering
frizzled signaling. Development 125, 959–968.
Amonlirdviman, K., Khare, N.A., Tree, D.N., Chen, W., Axelrod, J., and Tomlin,
C.J. (2005). Mathematical modeling of planar cell polarity to understand domi-
neering non-autonomy. Science 307, 423–426.
Axelrod, J.D. (2009). Progress and challenges in understanding planar cell
polarity signaling. Semin. Cell Dev. Biol. 20, 964–971.
Baena-Lopez, L.A., Baonza, A., and Garcia-Bellido, A. (2005). The orientation
of cell divisions determines the shape of Drosophila organs. Curr. Biol. 15,
1640–1644.
Bertet, C., Sulak, L., and Lecuit, T. (2004). Myosin-dependent junction remod-
elling controls planar cell intercalation and axis elongation. Nature 429,
667–671.
Blankenship, J.T., Backovic, S.T., Sanny, J.S., Weitz, O., and Zallen, J.A.
(2006). Multicellular rosette formation links planar cell polarity to tissue
morphogenesis. Dev. Cell 11, 459–470.
Bryant, P.J., Huettner, B., Held, L.I., Jr., Ryerse, J., and Szidonya, J. (1988).
Mutations at the fat locus interfere with cell proliferation control and epithelial
morphogenesis in Drosophila. Dev. Biol. 129, 541–554.
Burak, Y., and Shraiman, B.I. (2009). Order and stochastic dynamics in
Drosophila planar cell polarity. PLoS Comput. Biol. 5, e1000628.
Butler, L.C., Blanchard, G.B., Kabla, A.J., Lawrence, N.J., Welchman, D.P.,
Mahadevan, L., Adams, R.J., and Sanson, B. (2009). Cell shape changes indi-
cate a role for extrinsic tensile forces in Drosophila germ-band extension. Nat.
Cell Biol. 11, 859–864.
Casal, J., Lawrence, P.A., and Struhl, G. (2006). Two separate molecular
systems, Dachsous/Fat and Starry night/Frizzled, act independently to confer
planar cell polarity. Development 133, 4561–4572.
Chen, W.S., Antic, D., Matis, M., Logan, C.Y., Povelones, M., Anderson, G.A.,
Nusse, R., and Axelrod, J.D. (2008). Asymmetric homotypic interactions of the
Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc. 785
atypical cadherin flamingo mediate intercellular polarity signaling. Cell 133,
1093–1105.
Cho, E., and Irvine, K.D. (2004). Action of fat, four-jointed, dachsous and dachs
in distal-to-proximal wing signaling. Development 131, 4489–4500.
Clark, H.F., Brentrup, D., Schneitz, K., Bieber, A., Goodman, C., and Noll, M.
(1995). Dachsous encodes a member of the cadherin superfamily that controls
imaginal disc morphogenesis in Drosophila. Genes Dev. 9, 1530–1542.
Classen, A., Anderson, K., Marois, E., and Eaton, S. (2005). Hexagonal Packing
of Drosophila Wing Epithelial Cells by the Planar Cell Polarity Pathway. Dev.
Cell 9, 805–817.
Classen, A.K., Aigouy, B., Giangrande, A., and Eaton, S. (2008). Imaging
Drosophila pupal wing morphogenesis. Methods Mol. Biol. 420, 265–275.
Cortes, S., Glade, N., Chartier, I., and Tabony, J. (2006). Microtubule self-orga-
nisation by reaction-diffusion processes in miniature cell-sized containers and
phospholipid vesicles. Biophys. Chem. 120, 168–177.
Daga, R.R., and Nurse, P. (2008). Interphase microtubule bundles use global
cell shape to guide spindle alignment in fission yeast. J. Cell Sci. 121, 1973–
1980.
de Gennes, P.G., and Prost, J. (1993). The Physics of Liquid Crystals, Second
Edition (Gloucestershire, UK: Clarendon Press).
Farhadifar, R., Roper, J.C., Aigouy, B., Eaton, S., and Julicher, F. (2007). The
Influence of Cell Mechanics, Cell-Cell Interactions, and Proliferation on Epithe-
lial Packing. Curr. Biol. 17, 2095–2104.
Garoia, F., Grifoni, D., Trotta, V., Guerra, D., Pezzoli, M.C., and Cavicchi, S.
(2005). The tumor suppressor gene fat modulates the EGFR-mediated prolifer-
ation control in the imaginal tissues of Drosophila melanogaster. Mech. Dev.
122, 175–187.
Garoia, F., Guerra, D., Pezzoli, M.C., Lopez-Varea, A., Cavicchi, S., and Gar-
cia-Bellido, A. (2000). Cell behaviour of Drosophila fat cadherin mutations in
wing development. Mech. Dev. 94, 95–109.
Haase, S.B., and Lew, D.J. (2007). Microtubule organization: cell shape is
destiny. Curr. Biol. 17, R249–R251.
Ishikawa, H.O., Takeuchi, H., Haltiwanger, R.S., and Irvine, K.D. (2008). Four-
jointed is a Golgi kinase that phosphorylates a subset of cadherin domains.
Science 321, 401–404.
Joanny, J.F., Julicher, F., Kruse, K., and Prost, J. (2007). Hydrodynamic theory
for multi-component active polar gels. N. J. Phys. 9, 422.
Keller, R., Shih, J., and Sater, A. (1992). The cellular basis of the convergence
and extension of the Xenopus neural plate. Dev. Dyn. 193, 199–217.
Keller, R., Shook, D., and Skoglund, P. (2008). The forces that shape embryos:
physical aspects of convergent extension by cell intercalation. Phys. Biol. 5,
15007.
Lawrence, P.A., Struhl, G., and Casal, J. (2007). Planar cell polarity: one or two
pathways? Nat. Rev. Genet. 8, 555–563.
Le Garrec, J.F., Lopez, P., and Kerszberg, M. (2006). Establishment and main-
tenance of planar epithelial cell polarity by asymmetric cadherin bridges:
a computer model. Dev. Dyn. 235, 235–246.
Lopez-Schier, H., Starr, C.J., Kappler, J.A., Kollmar, R., and Hudspeth, A.J.
(2004). Directional cell migration establishes the axes of planar polarity in the
posterior lateral-line organ of the zebrafish. Dev. Cell 7, 401–412.
Ma, D., Amonlirdviman, K., Raffard, R.L., Abate, A., Tomlin, C.J., and Axelrod,
J.D. (2008). Cell packing influences planar cell polarity signaling. Proc. Natl.
Acad. Sci. USA 105, 18800–18805.
Ma, D., Yang, C.H., McNeill, H., Simon, M.A., and Axelrod, J.D. (2003). Fidelity
in planar cell polarity signalling. Nature 421, 543–547.
786 Cell 142, 773–786, September 3, 2010 ª2010 Elsevier Inc.
Matakatsu, H., and Blair, S.S. (2004). Interactions between Fat and Dachsous
and the regulation of planar cell polarity in the Drosophila wing. Development
131, 3785–3794.
Rauzi, M., Verant, P., Lecuit, T., and Lenne, P.F. (2008). Nature and anisotropy
of cortical forces orienting Drosophila tissue morphogenesis. Nat. Cell Biol. 10,
1401–1410.
Seifert, J.R., and Mlodzik, M. (2007). Frizzled/PCP signalling: a conserved
mechanism regulating cell polarity and directed motility. Nat. Rev. Genet. 8,
126–138.
Shimada, Y., Yonemura, S., Ohkura, H., Strutt, D., and Uemura, T. (2006).
Polarized transport of Frizzled along the planar microtubule arrays in
Drosophila wing epithelium. Dev. Cell 10, 209–222.
Simons, M., and Mlodzik, M. (2008). Planar cell polarity signaling: from fly
development to human disease. Annu. Rev. Genet. 42, 517–540.
Strauss, B., Adams, R.J., and Papalopulu, N. (2006). A default mechanism of
spindle orientation based on cell shape is sufficient to generate cell fate diver-
sity in polarised Xenopus blastomeres. Development 133, 3883–3893.
Strutt, H., and Strutt, D. (2002). Nonautonomous planar polarity patterning
in Drosophila: dishevelled-independent functions of frizzled. Dev. Cell 3,
851–863.
Strutt, H., and Strutt, D. (2005). Long-range coordination of planar polarity in
Drosophila. Bioessays 27, 1218–1227.
Strutt, H., and Strutt, D. (2008). Differential stability of flamingo protein
complexes underlies the establishment of planar polarity. Curr. Biol. 18,
1555–1564.
Tree, D.R., Ma, D., and Axelrod, J.D. (2002a). A three-tiered mechanism for
regulation of planar cell polarity. Semin. Cell Dev. Biol. 13, 217–224.
Tree, D.R., Shulman, J.M., Rousset, R., Scott, M.P., Gubb, D., and Axelrod,
J.D. (2002b). Prickle mediates feedback amplification to generate asymmetric
planar cell polarity signaling. Cell 109, 371–381.
Uemura, T., and Shimada, Y. (2003). Breaking cellular symmetry along planar
axes in Drosophila and vertebrates. J. Biochem. 134, 625–630.
Villano, J.L., and Katz, F.N. (1995). four-jointed is required for intermediate
growth in the proximal-distal axis in Drosophila. Development 121, 2767–
2777.
Vincent, L., and Soille, P. (1991). Watersheds in digital spaces: an efficient
algorithm based on immersion simulations. IEEE 13, 583–598.
Vinson, C., and Adler, P.N. (1987). Directional non-cell autonomy and the
transmission of polarity information by the frizzled gene of Drosophila. Nature
329, 549–551.
Vladar, E.K., Antic, D., and Axelrod, J.D. (2009). Planar cell polarity signaling:
The developing cell’s compass. Cold Spring Harb. Perspect. Biol. 1, a002964.
Wang, J., Mark, S., Zhang, X., Qian, D., Yoo, S.J., Radde-Gallwitz, K., Zhang,
Y., Lin, X., Collazo, A., Wynshaw-Boris, A., et al. (2005). Regulation of polarized
extension and planar cell polarity in the cochlea by the vertebrate PCP
pathway. Nat. Genet. 37, 980–985.
Wu, J., and Mlodzik, M. (2008). The frizzled extracellular domain is a ligand for
Van Gogh/Stbm during nonautonomous planar cell polarity signaling. Dev. Cell
15, 462–469.
Wu, J., and Mlodzik, M. (2009). A quest for the mechanism regulating global
planar cell polarity of tissues. Trends Cell Biol. 19, 295–305.
Zallen, J.A. (2007). Planar polarity and tissue morphogenesis. Cell 129, 1051–
1063.
Zallen, J.A., and Wieschaus, E. (2004). Patterned gene expression directs
bipolar planar polarity in Drosophila. Dev. Cell 6, 343–355.
Zeidler, M.P., Perrimon, N., and Strutt, D.I. (2000). Multiple roles for four-
jointed in planar polarity and limb patterning. Dev. Biol. 228, 181–196.
Generation of Rat Pancreas in Mouse
by Interspecific Blastocyst Injection
of Pluripotent Stem Cells
Toshihiro Kobayashi,1,2 Tomoyuki Yamaguchi,1,2 Sanae Hamanaka,1,2 Megumi Kato-Itoh,2,3 Yuji Yamazaki,1,2
Makoto Ibata,2 Hideyuki Sato,1,2 Youn-Su Lee,1,2 Jo-ichi Usui,1,6 A.S. Knisely,5 Masumi Hirabayashi,3,4
and Hiromitsu Nakauchi1,2,*
1Division of Stem Cell Therapy, Center for Stem Cell Biology and Regenerative Medicine, Institute of Medical Science, University of Tokyo,

4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639, Japan

2Japan Science Technology Agency, ERATO, Nakauchi Stem Cell and Organ Regeneration Project, 4-6-1 Shirokanedai, Minato-ku,

Tokyo 108-8639, Japan

3Center for Genetic Analysis of Behavior, National Institute for Physiological Sciences, Okazaki, Aichi 444-8585, Japan
4School of Life Science, The Graduate University for Advanced Studies, Okazaki, Aichi 444-8585, Japan
5Institute of Liver Studies, King’s College Hospital, London SE5 9RS, UK
6Present address: Department of Nephrology, Graduate School of Comprehensive Human Sciences, University of Tsukuba, 1-1-1 Tennoudai,

Tsukuba, Ibaraki 305-8575, Japan

*Correspondence: nakauchi@ims.u-tokyo.ac.jp
DOI 10.1016/j.cell.2010.07.039

SUMMARY
The complexity of organogenesis hinders in vitro
generation of organs derived from a patient’s pluripo-
tent stem cells (PSCs), an ultimate goal of regenera-
tive medicine. Mouse wild-type PSCs injected into
Pdx1 / (pancreatogenesis-disabled) mouse blasto-
cysts developmentally compensated vacancy of
the pancreatic ‘‘developmental niche,’’ generating
almost entirely PSC-derived pancreas. To examine
the potential for xenogenic approaches in blastocyst
complementation, we injected mouse or rat PSCs into
rat or mouse blastocysts, respectively, generating
interspecific chimeras and thus confirming that
PSCs can contribute to xenogenic development
between mouse and rat. The development of these
mouse/rat chimeras was primarily influenced by
host blastocyst and/or foster mother, evident by
body size and species-specific organogenesis. We
further injected rat wild-type PSCs into Pdx1 /
mouse blastocysts, generating normally functioning
rat pancreas in Pdx1 / mice. These data constitute
proof of principle for interspecific blastocyst comple-
mentation and for generation in vivo of organs derived
from donor PSCs using a xenogenic environment.
INTRODUCTION
Current stem cell therapy mainly targets diseases that can be
treated by cell replacement, such as Parkinson’s disease or dia-
betes mellitus. One of the ultimate goals of regenerative medi-
cine, however, is to grow organs using the patient’s own stem
cells and to transplant those organs into the patient. With the
development of induced pluripotent stem cell (iPSC) technology,
we are now able to obtain patient-derived PSCs (Takahashi
et al., 2007; Takahashi and Yamanaka, 2006), although actual
developmental potentials remain to be defined, as do risks asso-
ciated with somatic cell reprogramming. The real challenge is to
create a reproductive system for generation of PSC-derived
organs. The interactions among cells and tissues during devel-
opment and organogenesis are so complex that the recapitula-
tion of these interactions to generate organs in vitro is essentially
impractical. We have challenged this goal using the biology of
blastocyst complementation.
Blastocyst complementation was first reported by Chen et al.
They demonstrated that deficiency of T and B lymphocyte
lineages in Rag2-deficient (Rag2 / ) mice was complemented
by injecting normal mouse embryonic stem cells (mESCs) into
Rag2 / mouse-derived blastocysts (Chen et al., 1993).
Because Rag2 is an indispensable enzyme for rearrangement
of immunoglobulin and T cell receptor genes, the T and B cells
generated in the complemented animals were mESC derived;
there were no host T or B lymphocytes. We assumed that this
complementation was possible because the Rag2 / host, inca-
pable of generating mature T and B cells, provided a ‘‘develop-
mental niche’’ for ESC-derived T and B cells.
We hypothesized that with blastocysts derived from a mutant
mouse strain in which the gene necessary to form a particular
organ is deficient, the same principle might apply. To test
this hypothesis, we used, in this study, blastocyst complementa-
tion to generate functional pancreas from donor PSCs. The
pancreas, consisting of endocrine and exocrine glands, is formed
by early embryonic interactions of mesenchyme and epithelium
(Slack, 1995). Pdx1 (pancreatic and duodenal homeobox1) is
a Hox-type transcription factor that plays a critical role in pancre-
atic development and b cell maturation. Homozygous deficiency
of Pdx1 in the mouse results in death soon after birth due to

Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 787

pancreatic insufficiency (Offield et al., 1996). Targeted disruption
of Pdx1 thus should empty a pancreatic ‘‘developmental niche’’
in embryos derived from Pdx1 / blastocysts. Therefore, injec-
tion of mouse PSCs (mPSCs) into Pdx1 / blastocysts should
result in generation of pancreata almost entirely derived from
injected mPSCs.
Although our results verified our initial hypothesis, this system
cannot be applied to generate human organs. A xenogenic, but
not allogenic, blastocyst complementation system must there-
fore be established. However, little is known about the nature
of the xenogenic barrier, how organogenesis by donor PSCs is
influenced, or how intrinsic developmental programs can be
regulated in xenogenic environments. To address these issues,
we attempted to generate interspecific chimeras between
mouse and rat using a blastocyst injection technique with mouse
and rat PSCs.
Generation of interspecific chimeras in livestock animals using
preimplantation embryos of each species is described; the
‘‘geep,’’ or chimera between goat and sheep, is famous as an
interspecific live chimera (Fehilly et al., 1984). In rodents,
however, methods like those producing the geep succeeded
only between mouse subspecies, such as Mus musculus and
Mus caroli, that are closely related but not capable of inter-
breeding (Rossant and Frels, 1980). Many groups have sought
to generate interspecific chimeras between mouse and rat,
successful with chimeric preimplantation embryos in vitro but
failing with live chimeric animals (Stern, 1973; Zeilmaker,
1973). Extraembryonic lineage cells like trophectoderm or prim-
itive endoderm, derived from xenogenic embryos, might suffer
inhibition of implantation on exposure to host uterus; inhibition
of further intrauterine development also is possible (Rossant
et al., 1982; Tarkowski, 1962). Only cells of pre-blastocyst origin
can contribute to extraembryonic lineage cells, and mESCs/
iPSCs do not thus contribute (Rossant, 2007). Therefore, we
rechallenged this old issue with new technology, using iPSCs
or ESCs that are not capable of contributing to extraembryonic
tissues. Using recently established culture conditions with a
combination of signaling inhibitors (Buehr et al., 2008; Li et al.,
2008; Ying et al., 2008), we generated rat-ESCs and -iPSCs
(rESCs, riPSCs). Our work has now confirmed the existence of
interspecific chimeras generated with PSCs, using injection not
only of mouse PSCs into rat embryos but also of rat PSCs into
mouse embryos.
Finally, by combining the principle of blastocyst complemen-
tation with the production of interspecific chimeras, we suc-
ceeded in generating rat pancreas in Pdx1 / mice. These sets
of experiments provide proof-of-principle data for donor iPSC-
derived organ generation in a xenogenic environment.
RESULTS
Generation of Donor Mouse iPSC-Derived Pancreas
in Pdx1-Deficient Neonates
Our first goal was to generate pancreas from mPSCs. We per-
formed a blastocyst complementation experiment using
Pdx1 / blastocysts that would provide a niche for pancreatic
organogenesis. These mice exhibit pancreatic agenesis due to
the absence of Pdx1-specified interactions between prepancre-
788 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.
atic epithelium and mesenchyme, a key step in pancreatic devel-
opment. Homozygote Pdx1-LacZ knockin mice (Pdx1 / ) are
born alive but die within 1 week after birth, presumably due to
pancreatic insufficiency (Offield et al., 1996).
For complementation donor cells, we used GT3.2 mouse
iPSCs (miPSCs) that had been generated from tail tip fibroblasts
(TTFs) of an adult EGFP-transgenic (TG) C57BL/6 mouse (Okabe
et al., 1997) using three factors (Oct3/4, Sox2, Klf4) in retroviral
vectors (Figure S1 available online). These GT3.2 miPSCs
were injected into blastocysts obtained by an intercross of
heterozygote Pdx1-LacZ knockin mice (Pdx1+/ ), offspring of
C57BL/6(Pdx1+/ ) 3 DBA(Pdx1+/+) or C57BL/6(Pdx1+/ ) 3
BDF1(Pdx+/+) mice. G4.2 mESCs were also injected for compar-
ison. Neonates were assessed macroscopically and histologi-
cally for pancreatic development and were genotyped for
Pdx1. These neonates were highly chimeric: Due to silencing
of EGFP and to contamination by donor cells, accurate determi-
nation of the genotypes required analysis at the single-cell level.
To identify the genotypes of these neonates accurately, EGFP-
negative, c-Kit-positive, and Sca1-positive, lineage marker-
negative (EGFP KSL) bone marrow cells were clone-sorted by
fluorescence-activated cell sorting (FACS) to obtain single-cell-
derived hematopoietic colonies (Figure S2A). After culture for
2 weeks, cells collected from each colony were subjected to
PCR analysis (Figure S2B). This ‘‘colony PCR’’ method unambig-
uously determined the host embryo genotype and proved that
pancreas had been formed in Pdx1 / mice by blastocyst
complementation (Figure S2C).
In all postnatal chimeric mice derived from blastocysts
injected with either miPSCs or mESCs, pancreas was present
regardless of host genotype (including Pdx1 / ). As expected,
the pancreas in Pdx1+/ or Pdx1+/+ chimeric mice was a
composite of host-derived cells and EGFP-miPSC- or mESC-
derived cells, as in whole-body chimerism (Pdx1+/ + miPSCs
or mESCs in Figure 1A). In contrast, the pancreatic epithelium
in Pdx1 / chimeric mice was almost entirely composed of
EGFP-marked miPSC- or mESC-derived cells (Pdx1 / +
miPSCs or mESCs in Figure 1A). The pancreas of these mice
was grossly and histologically normal. We examined these
pancreatic tissues further for contributions of donor mPSCs to
different pancreatic lineages. Both miPSCs and mESCs supplied
all pancreatic cell lineages (exocrine and endocrine tissues,
including ductal epithelia), but pancreatic stromal elements—
vessels, nerves, and fibrocytes—were composites of host- and
mPSC-derived cells in all mice (Figure 1B).
miPSCs Rescued Pdx1 / Mice by Blastocyst
Complementation
We next addressed whether mPSCs can rescue Pdx1 /
lethality via blastocyst complementation. As we predicted based
on neonatal analysis (Figure 1), both mESCs and miPSCs
contributed to pancreatic organogenesis; Pdx1 / chimeric
mice survived to adulthood (mESCs injection: n = 4, miPSCs
injection: n = 15 in Figure 2A). They even served as Pdx1 /
founders, transmitting their genotype to the next generation.
Mating Pdx1 / founder male mice with Pdx1+/ female mice
increased to 50% the proportion of pups of homozygously dis-
rupted genotype (Figure 2A). These data indicate that blastocyst
 A Figure 1. Mouse PSC-Derived Pancreas Gener-
+ miPSCs + mESCs
Pdx1-/- Pdx1+/- Pdx1-/- Pdx1+/-
Phase
EGFP
B
Pancreatic tissues
Endocrine Exocrine
Insulin Glucagon Somatostatin α-Amylase
complementation can be used to generate a functional organ
and to yield homozygous founder mice even when uncomple-
mented homozygosity is fatal. The method described here thus
can be useful for efficient production of gene-targeted mice
that are embryonic lethal without introducing a conditional
gene targeting system.
As expected, in both Pdx1 / and Pdx1+/ mice injected with
EGFP-miPSC, miPSC-derived cells contributed to all the non-
pancreatic tissues of the body, including lung, heart, liver,
muscle, testis, and brain (data not shown). The extent of contri-
bution varied from tissue to tissue and also varied depending on
individual chimera. However, the pancreas in adult Pdx1 / mice
was entirely derived from donor miPSCs (Pdx1 / in Figure 2B).
Detailed analysis of EGFP distributions in adult miPSC-derived
pancreas demonstrated that pancreatic islets, exocrine tissues,
and duct epithelia were entirely derived from donor miPSCs, as
already shown in neonates (Pdx1 / in Figures 2C and 2D). In
expected contrast, pancreas from Pdx1+/ mice was a com-
posite of host and donor derivatives (Pdx1+/ in Figures 2B
and 2C). Quantitative analysis of these sections by image J soft-
ware revealed percentages of EGFP-positive cells to be 27.4% ±
27.8% in the miPSC + Pdx1+/ setting, 95.6% ± 4.6% in the
miPSC + Pdx1 / setting. Of note is that most individual islets
ated by Way of Blastocyst Complementation
Pdx1-/- (A) mPSC-derived pancreas at neonatal stage. In
Pdx1 / mice complemented with mPSCs, the pan-
creas was almost entirely positive for EGFP, indicating
mPSC origin (left panels in +miPSCs and +mESCs). On
the other hand, in Pdx1+/ mice complemented with
mPSCs, the pancreas was partially positive for EGFP,
indicating a composite of both mPSC- and host-derived
cells (right panels in +miPSCs and +mESCs). Right panel
shows absence of pancreas in a nonchimeric Pdx1 /
mouse at the same developmental stage. See also
Figure S1 for detailed characteristics of miPSCs and
Figure S2 for genotyping of Pdx1 status.
(B) Histological analysis of the distribution of mESC-
derived cells in neonatal mouse pancreas. Sections of
pancreas generated in a Pdx1 / mouse complemented
with mESCs were immunostained for EGFP; for pancre-
Blood vessel atic endocrine markers (insulin, glucagon, and somato-
Ductal cell Endothelial cell statin); for the pancreatic exocrine marker a-amylase;
DBA-Lectin PECAM1 for the ductal marker rhodamine-conjugated DBA-
lectin; and for blood vessels with PECAM-1, with DAPI
nuclear counterstaining.
Sections were observed under confocal laser scanning
microscopy. Scale bars in (B), 20 mm.
were composed of both host-derived and
miPSC-derived cells (Pdx1+/ in Figure 2D),
indicating, as previously reported (Deltour
et al., 1991), nonclonal origin of pancreatic
islets.
To test whether miPSC-derived pancreata
are functional in Pdx1 / chimeric mice, we
performed glucose tolerance testing (GTT).
Whereas streptozotocin (STZ)-induced dia-
betic mice failed to respond to GTT, Pdx1 /
chimeric mice responded well, indicating that Pdx1 / mice
with miPSC-derived pancreas secreted insulin in response to
glucose loading and maintained normal serum glucose levels
(Figure 2E). Histology and function of exogenously derived
pancreas were essentially the same for mice derived from
miPSC- and from mESC-complemented blastocysts (Figure S3).
These data demonstrate that the vacant ‘‘pancreatic niche’’
provided in Pdx1 / mice can be occupied, with developmental
compensation, by miPSC- or mESC-derived cells after intraspe-
cific blastocyst complementation, generating functionally intact
pancreas.
Transplantation of miPSC-Derived Islets Corrected
Hyperglycemia in Diabetic Mice
To assess the functionality of miPSC-derived pancreas, islets
from miPSC-derived pancreas were transplanted into mice of
the original strain (C57BL/6) in which STZ administration had
induced diabetes. As a control, islets from pancreas in Pdx1+/
chimeric mice injected with miPSCs were transplanted. Blasto-
cysts were obtained by an intercross of Pdx1+/ mice (C57BL/
6 3 DBA2 or C57BL/6 3 BDF1 F1 strains) that were semi-
allogenic to miPSCs used in this study. EGFP-expressing
miPSC-derived islets (Figure 3A) were isolated conventionally
Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 789
A Figure 2. Analysis of miPSC-Derived
Injected
Parents
No. of No. of Pdx1 genotype Pancreas in Adult Mice: Derivation of
cell adult mice chimeras +/+ +/- -/- Pancreatic Functional Cells and Their
G4.2 Pdx1+/- 27 18
5 9 4 Response to Glucose Load
+/-
(mESC) x Pdx1 ( 1 : 2 : 1 )
(A) Results of embryo manipulation and genotypes
GT3.2
-/-
(Pdx1 + mESCs) 0 13 15 of adult mice. Note that Pdx1 / mice comple-
29 28
(miPSC) x Pdx1+/- ( 0 : 1 : 1 ) mented with mESCs served as founders and
generated knockout mice more efficiently.
(B) Macroscopic images at adult stage
B Pdx1-/- + miPSCs Pdx1+/- + miPSCs C Pdx1+/- + miPSCs (8–12 weeks) of miPSC-derived pancreas gener-
Pdx1-/- + miPSCs
ated in a Pdx1 / mouse (Pa in left panel) and of
Li
composite pancreas generated in a Pdx1+/
Phase

mouse (right). In left panel, Li = liver, Pa = pancreas,

HE
Pa St
Sp
St = stomach, Sp = spleen, and Ki = kidney.
Ki
(C) Immunohistological studies of sections ob-
tained from pancreas revealed clear differences
in the distributions of miPSC-derived cells.
DAPI / EGFP
EGFP

Sections were stained with HE or for EGFP with

DAPI nuclear counterstaining.
(D) Immunohistological analysis of a generated
islet, showing the distribution of miPSC-derived
cells. Sections were stained for EGFP, with anti-
insulin antibodies, and with DAPI nuclear counter-
D E (mg/dl)
staining. Yellow lines show borders of endocrine
DAPI / EGFP / Insulin DAPI EGFP Insulin
600 and exocrine tissues. See also Figure S3 for anal-
+ miPSCs
Pdx1-/-

ysis of mESC-derived pancreas in adult mice.

400
(E) Results of GTT in Pdx1 / (-) and Pdx1+/ (:)
STZ-diabetic model
Pdx1-/-+ miPSCs mice (n = 6 each) complemented with miPSCs.
200
Pdx1+/-+ miPSCs Mice with STZ-induced diabetes (d and A) served
+ miPSCs

as controls. Fasting after 20 hr, blood was sampled

Pdx1+/-

0
via tail vein at intraperitoneal glucose administra-
0 30 60 90 120 tion (1 g/kg; 0 min) and 15, 30, 60, and 120 min
Minutes after glucose administration thereafter.
Sections in (C) were observed under fluorescence
microscopy and in (D) were observed under
confocal laser scanning microscopy. Scale bars
in (C), 100 mm; in (D), 50 mm. Error bars in (E) indi-
cate ± SD.

based on a method developed previously (Gotoh et al., 1985) Blastocyst complementation thus permits demonstration of
and were transplanted beneath the renal capsule of recipient proof-of-principle both for pancreas generation from PSCs and
mice. Nonfasting blood glucose levels were then monitored. As for diabetic therapy using donor iPSC-derived syngenic islets.
these islets were of donor origin (i.e., C57BL/6 strain), an immu-
nosuppressive regimen was not used. Generation of Interspecific Chimeras between Mouse
To prevent nonspecific loss of islets due to inflammation, and Rat
anti-inflammatory cytokine monoclonal antibody (mAb) cocktails Our second goal was to generate interspecific chimeras
were given at transplantation and 2 and 4 days thereafter between mouse and rat. To achieve this goal, we generated
(arrows, Figure 3E), as described (Satoh et al., 2007). Two not only miPSCs but also riPSCs and rESCs using an established
months after transplant, EGFP-expressing miPSC-derived islets protocol (Hirabayashi et al., 2009; Ying et al., 2008). These
were still detectable at the graft site (Figures 3B and 3C). Produc- mouse and rat PSCs enabled us bidirectionally to generate inter-
tion of insulin by the transplanted islets was confirmed immuno- specific chimeras.
histologically (Figure 3D). The induced-diabetic recipients no We injected EGFP-marked GT3.2 miPSCs into rat blastocysts
longer exhibited hyperglycemia; they maintained normal blood (r-blastocysts) or EGFP-marked riPSCs (Figures S4A and S4B)
glucose levels and responded normally to GTT (Figures 3E into mouse blastocysts (m-blastocysts). Because post-implan-
and 3F). This is in contrast to the therapeutic effect conferred tation development reportedly is severely hampered after intra-
by islets, composites of blastocyst- and miPSC-derived cells, uterine transfer of xenogenic blastocysts (Rossant et al., 1982;
obtained from pancreas of Pdx1+/ chimeric mice (C57BL/6 3 Tarkowski, 1962), injected r-blastocysts or m-blastocysts were
DBA2 or C57BL/6 3 BDF1 F1 origin). This effect lasted only for transferred, respectively, into the uteri of pseudopregnant rats
a short time, presumably due to immune rejection by the host or mice. After intrauterine transfer of injected r-blastocysts or
C57BL/6 diabetic mice (Figure 3E). These data strongly indicate m-blastocysts with development to the fetal stage, we evaluated
that the miPSC-derived pancreas, with islets, formed in an allo- EGFP expression by fluorescence microscopy for each trans-
genic host is functional and that the ‘‘autologous’’ islets thus ferred embryo. EGFP-expressing cells were found in the body
formed can be used to treat diabetes, without rejection. of each injected conceptus, but never in placenta (Figure 4A).

790 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.

 A B C D
DAPI / EGFP / Insulin
Phase Phase HE

EGFP EGFP EGFP

DAPI

E F
(mg/dl) (mg/dl)
600 600
Blood glucose

400 Blood glucose 400

200 200

0 0
0 10 20 30 40 50 60 (days) 0 30 60 90 120 (min)
Days after islet transplantation Minutes after glucose administration

Pdx1-/- + miPSCs derived islet transplantation Syngenic islet transplantation (PC)

Pdx1+/- + miPSC derived islet transplantation Sham transplantation (NC)

Figure 3. Transplantation of ‘‘Autologous’’ Islets from iPSC-Derived Pancreas: A Therapeutic Model

(A) EGFP-positive islets isolated from miPSC-derived pancreas. Via the common bile duct and pancreatic duct, miPSC-derived pancreas was perfused by colla-
genase. Density gradient centrifugation isolated a fraction enriched in EGFP-expressing islets.
(B) Kidney capsule 2 months after islet transplantation. At the site of transplantation (arrow), miPSC-derived islets still expressed EGFP.
(C) Renal subcapsular transplantation site (B, arrow); sections were stained with HE or immunostained for EGFP with DAPI nuclear counterstaining.
(D) Transplanted miPSC-derived islets in (C), immunostained for EGFP and insulin with DAPI nuclear counterstaining.
(E) Transplantation of miPSC-derived islets into mice with STZ-induced diabetes. Each mouse received 150 islets. Arrows indicate time points at which an mAb
cocktail (anti-INF-g, anti-TNF-a, anti-IL-1b) was administered. Nonfasting blood glucose levels were measured weekly for 2 months after transplantation.
Glucose levels are shown for STZ-induced diabetic mice transplanted with miPSC-derived islets (green; n = 3), with islets derived from Pdx1+/ chimeric
mice (orange; n = 2), with islets derived from syngenic strain of host strain (C57BL/6) as positive control (blue; n = 2), and with sham transplantation control
mice as negative control (purple; n = 3).
(F) Results of GTT 2 months after islet transplantation. Donors were Pdx1 / chimeric mice and Pdx1+/ chimeric mice, with blood sampling at the same time
points as in Figure 2E.
Sections in (C) were observed under fluorescence microscopy and in (D) were observed under confocal laser scanning microscopy. Scale bar in (D), 50 mm. Error
bars in (E) and (F) indicate ± SD.

This finding indicates that injected mouse or rat iPSCs can
contribute to xenogenic development, with generation of inter-
specific chimeras.
Next, we tried to quantitate contribution of mouse or rat iPSCs
to these interspecific chimeras. Chimerism in interspecific
embryos appeared to vary individual-to-individual and organ-
to-organ. Since quantitation of PSC-derived cells was difficult
in organs, we analyzed embryonic fibroblasts and hematopoietic
cells. FACS analysis of embryonic fibroblasts revealed that
donor-derived EGFP+ cell percentages of about 28.0% and
26.5%, respectively, were detected in mouse and rat inter-
specific chimeras (representative FACS data shown in Fig-
ure 4B). We also examined chimerism in hematopoietic cells
by staining cells from livers of interspecific chimera fetuses
with antibodies specific for mouse and rat CD45 antigens.
Cells that expressed mouse or rat CD45 represented distinct
populations in interspecific chimeras, with only cells derived
from injected iPSCs expressing EGFP (Figure 4B). Whereas
a high proportion (28.3%) of mouse blood cells was detected
in r-blastocyst-derived chimeric fetal liver, rat blood cells were

Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 791

A B Figure 4. Generation of Interspecific
Embryonic fIbroblast Fetal liver
rBL + miPSCs mBL + riPSCs Chimeras between Mouse and Rat
104
(A) Interspecific chimera fetuses generated by
19.3 1.7

rBL + miPSCs

mCD45-APC
103
Phase

injection of miPSCs into r-blastocyst (rBL: left

30.0 49.0
28.0 102
98.8
panels) and by injection of riPSCs into m-blasto-
101
cyst (mBL: right panels). With rat embryo manipu-
100
100 101 102 103 10
4
lation, fetuses were analyzed 12 days after embryo
100 101 102 103 104 100 101 102 103 104

EGFP rCD45-PE EGFP

transfer into uteri of 3.5 dpc pseudopregnant rats
4
10
EGFP

6.68 0.02 (embryonic day (E) 15.5). With mouse embryo

mBL + riPSCs

mCD45-APC
103
93.1 0.23
26.5 87.0 manipulation, fetuses were analyzed 11 days after
102

embryo transfer into uteri of 2.5 dpc pseudopreg-

101

nant mice (E13.5). See also Figure S4 for charac-

100
4
100 101 102 103 10
teristics of riPSCs and interspecific embryos after
100 101 102 103 104 100 101 102 103 104

EGFP rCD45-PE EGFP

injection of riPSCs into mouse 8-cell/morula stage
embryos.
C D (B) Representative data of FACS analysis on cells
rBL mBL
+ miPSCs +riPSCs Control
from fetal liver and on embryonic fibroblasts
Exon 2 Exon 3 Exon 4
5+

5+
derived from interspecific chimeras. Right panels

in D45 +
5+

5+

te
in 45 +

tem No
r

pla
B
D4

D4
rke

B
D4

D4

mP

rP
D
show cells from fetal liver immunostained with

mC
mC

mC
Ma

rC
rC

rC
rOct3/4 locus = 966bp
mOct3/4 locus = 861bp
rOct3/4 locus
mOct3/4 locus
anti-mCD45 and -rCD45 antibodies. Note that
anti-mouse or -rat monoclonal antibodies (mAb)
against CD45 can distinguish CD45-expressing
E F G hematopoietic cells in a species-specific manner.
No. of
Almost all CD45-expressing cells derived from
(total transferred embryos = 100%)

n = 16 n = 19 24 26 109 30 transferred n=6 n = 12 n = 23 n = 19

(%) (%) (%)
embryoa
100 100 100 injected cells express EGFP, indicating iPSC
Relative frequency

75 75 75 origin.
(C) Schema of mouse and rat Oct3/4 loci (mOct3/4
50 50 50
and rOct3/4). Of note is that in the rat Oct3/4 locus
Aborted
25 25 25
Non-chimera
Chimera
the 2 introns flanking exon 3 are longer than in the
0 0 0 mouse Oct3/4 locus. This difference in length can
+ +
EF CD45 EF CD45 Embryo Rat Mouse Mouse Rat Embryo Rat Mouse Mouse Rat
distinguish origins of otherwise similar cells.
in FL in FL
Cell miPSC riPSC Cell miPSC riPSC
mBL+riPSCs rBL+riPSCs Arrowheads indicate common primers of each
species for PCR. PCR product sizes for both
species are shown below.
(D) Results of genotyping to identify origin of hematopoietic cells expressing mouse or rat CD45 in fetal liver of interspecific chimeras. Peripheral-blood CD45-
expressing cells from each species served as positive controls.
(E) Correlation of chimerism between embryonic fibroblasts and CD45+ hematopoietic cells in fetal liver. Cells were prepared from interspecific chimeras gener-
ated by injection of riPSCs into m-blastocysts or from intraspecific chimeras generated by injection of riPSCs into r-blastocysts.
(F) Relative frequencies of aborted, nonchimeras, and chimeras embryonic development (E13.5 and 15.5).
(G) Chimerism analysis of embryonic fibroblasts from chimeras generated by injection of miPSCs or riPSCs into mouse or rat blastocysts. Fibroblasts were
obtained from chimeras and analyzed for EGFP intensity by FACS. Plotted dots show degrees of chimerism for individual embryos. See also Figure S5 for inter-
specific chimeras using ESCs.
Scale bars in (A), 5 mm.

only rarely present (less than 3.3%) in m-blastocyst-derived chimeras (Figures 4F and 4G). In addition, high contributions
chimeric fetal liver (Figure 4B). This tendency was specific to by xenogenic cells appeared to be associated with morpholog-
interspecific chimeras and not observed in intraspecific ical abnormalities and embryonic lethality (data not shown). To
chimeras (Figure 4E). It is not clear why the difference in contri- exclude the possibility that these abnormalities were caused by
bution of iPSCs to hematopoietic cells between mouse and rat donor iPSCs, we also attempted to generate interspecific
interspecific chimeras was so marked. chimeras using mouse or rat ESCs. DsRed-marked EB3DR
To further confirm interspecific chimerism, genomic DNA mESCs could also generate interspecific rat chimeras (top
extracted from FACS-sorted cells expressing CD45 was PCR- panels in Figure S5A), with embryonic development rate and
amplified using primers common to the mouse and rat Oct3/4 degree of chimerism similar to those generated by miPSC
loci (Figure 4C). PCR products of different lengths, indicating injection (Figure S5B). The Venus-marked WIv3i-1 and -5
origin in each species, were clearly present (Figure 4D). These rESC lines, with high contribution to rat embryo development
results strongly indicated that the animals harboring these cells and germline competency (Hirabayashi et al., 2009), could
were mouse/rat interspecific chimeras. also generate interspecific chimeras after injection into m-blas-
To investigate the influence of iPSC contribution to xenogenic tocyst (middle and bottom panels in Figure S5A), but embry-
development at the fetal stage, we assessed embryonic devel- onic development rate and degree of chimerism were lower
opment rate of interspecific chimeras, and the extent of chime- than reported for intraspecific chimeras (Figure S5C). These
rism, by FACS analysis using established embryonic fibroblasts. results suggest that generation of interspecific chimeras
Both embryonic development rate and degree of chimerism between mouse and rat is less efficient than generation of intra-
were lower in interspecific chimeras than in intraspecific specific chimeras.

792 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.

A B Figure 5. Postnatal Analysis of Interspecific
rBL + miPSCs mBL + riPSCs
(g) Chimeras
7.5 n=4 n=4 n=8 n=7 (A) Photograph of newborn interspecific chimeras
Phase
derived from r-blastocyst injected with miPSCs
Phase

and m-blastocyst injected with riPSCs.

Body weight
5.0 (B) Neonatal body weights of interspecific
Non-chimera chimeras were measured and plotted. One
Non-chimera
EGFP chimera obtained after injection of miPSCs into
2.5
r-blastocyst showed a high contribution of mouse
cells as shown in the insert, with body weight and
EGFP

0 size equivalent to those of newborn mouse (right

rBL+ mBL+ Non-chimera
miPSCs Rat riPSCs Mouse
panel).
(C) Photograph of interspecific chimera at 8 weeks
of age.
(D) Analysis of chimerism in neonates by organ or
rBL + miPSCs mBL + riPSCs tissue. Interspecific chimeras generated by miPSC
C
injection into r-blastocyst (left panels) and by
riPSC injection into m-blastocyst (right panels)
are shown. Sections of representative organs
(heart, liver, pancreas, and kidney) after immunos-
taining for EGFP antibody with DAPI nuclear coun-
terstaining are shown.
(E) Neonatal pancreas and adult liver of inter-
specific chimera generated from r-blastocyst
injected with miPSC, immunostained for EGFP
and insulin or for EGFP and albumin or CK19
with DAPI nuclear counterstaining. In pancreas,
an islet-like cell cluster contains cells producing
D rBL + miPSCs mBL + riPSCs insulin composed of both miPSC-derived cells
that express EGFP and r-blastocyst-derived cells
Heart Pancreas Heart Pancreas that do not. In liver, albumin-positive hepatocytes
and CK19-positive cholangiocytes (arrowhead)
also express EGFP, indicating miPSC origin.
DAPI / EGFP

(F) Distribution of xenogenic cells in neonatal testis

and in developing gonad. MVH-positive germ cells
Liver Kidney Liver Kidney did not express EGFP, indicating nonxenogenic
cell origin.
Sections in (D) were observed under fluorescence
microscopy and in (E) and (F) were observed under
confocal laser scanning microscopy. See also
Figure S6 for macroscopic images of neonates
E F rBL + miPSCs mBL + rESCs and results of peripheral blood analysis of adults.
rBL + miPSCs Scale bars in (A) and (B), 10 mm; in (D), 100 mm;
Pancreas (P1) Liver (Adult) Testis (P1) Gonad (♀ E13.5) (E) and (F), 50 mm.
DAPI / EGFP / Insulin DAPI / EGFP / Albumin DAPI / EGFP / CK19 DAPI / EGFP / MVH DAPI / EGFP / MVH

Interspecific Chimeras Were Live-born; Some Grew black coat) (Figure 5C, r-blastocyst + miPSCs: n = 8, m-blasto-
into Adulthood cyst + riPSCs: n = 4). The full-term development rate of inter-
To investigate the developmental potential of generated specific chimeras, either with miPSCs into r-blastocyst or with
chimeras and to assess the functionality of the cells, tissues, or riPSCs into m-blastocyst, was 20%. In both settings it was
organs derived from injected cells, we analyzed interspecific lower than that for intraspecific chimeras (50%).
chimeras at neonatal and adult stages. Mouse- or rat-iPSC-
injected interspecific chimeras survived after birth and expressed Determination of Body Size in Interspecific Chimeras
EGFP ubiquitously (as did intraspecific chimeras; Figure 5A, Adult rats typically are ten times bigger than adult mice, whereas
r-blastocyst + miPSCs: n = 5, m-blastocyst + riPSCs: n = 10). newborn rats are three times bigger than newborn mice.
As these chimeras developed into adulthood, chimerism could Because mouse and rat gestations are of similar length (19 and
be judged by coat color because miPSCs (C57BL/6, black 21 days, respectively), organogenesis requires more cell prolifer-
coat) were injected into r-blastocyst (Wistar, white coat) or riPSCs ation and differentiation during rat development than during
(Wistar) were injected into m-blastocyst (BDF11 3 C57BL/6, mouse development. What determines the size of interspecific

Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 793

A Figure 6. Correlation between Body Weight
Chimera #1 Chimera #2 Chimera #3 Chimera #4
and Chimerism in Interspecific Chimeras
(A) Photographs of interspecific chimeras derived
from r-blastocyst injected with miPSCs, with FACS
analysis of their chimerisms in peripheral blood
mononuclear cells.
(B) Body weight change after weaning for inter-
specific chimeras (green) and their nonchimeric
105 105 105 105 littermates (orange).
mCD45-APC

104 104 104 104

103 103 103 103

5.35 0.29 15.6 2.06 17.8 2.62 55.5 1.3

102 102 102 102
0
0.24 94.1 0
0.05 82.3 0
1.28 78.3
positive cells (Figure S6B). In sections
0
0.27 42.9
0 10 10 10 10 2 3
0 10 4
10 5
10 10 2 3
0 10 104
10 5
10 0 10 10 102 10 3 4 immunostained with anti-EGFP antibody,
5 2 3 4 5

rCD45-PE representative images demonstrated that

EGFP-positive cells become various
types of tissues (Figure 5D). Of particular
B note was a pancreatic islet generated by
(g) (g)
injection of miPSCs into r-blastocyst.
♀
600 600
Cells in this islet marked immuno-
Body weight

Body weight

histochemically for insulin; the islet was

400 Littermate #1 400 Littermate #3
Littermate #2 Littermate #4
a composite of EGFP-positive and -nega-
Chimera #2 Chimera #1 tive cells, indicating that the islet con-
Chimera #4 Chimera #3
200 200 sisted of host rat cells and exogenous
donor mouse cells (Figure 5E). Islet poly-
0 0 clonality also was seen with intraspecific
4 9 14 (wk) 4 9 14 (wk)
chimeras (Figure 2E). The expression of
Weeks after birth Weeks after birth
other functional molecules in xenogenic
cells (i.e., hepatocytes or cholangiocytes
chimeras is an intriguing biological question. Interestingly, body in the liver or leukocytes in the peripheral blood) was also
size and weight of interspecific chimeras born from rat foster detected (Figure 5E and Figure S6C).
mothers were similar to those of normal newborn rats, whereas It is known that rats do not have gallbladders, whereas mice
for those born from mouse foster mothers they were similar to do. To date, when mPSCs were injected into r-blastocysts, the
those of newborn mice (Figures 5A and 5C). However, high interspecific chimeras produced (generally the size of rats)
contributions by xenogenic cells may also affect interspecific have not had gallbladders (n = 8). In contrast, interspecific
chimera size. One chimera obtained after injection of miPSCs chimeras generated from m-blastocysts complemented with
into r-blastocysts showed body weight and size equivalent to rPSCs (more like mice in size) have had gallbladders (n = 4).
those of a newborn mouse, although its birth mother was a rat To see the xenogenic contribution to germ cells, testis and
(Figure 5B). This particular chimera’s donor miPSC-derived cell developing gonad of the interspecific chimeras were examined.
contribution was extremely high (Figure 5B, inlet). However, Cells were not found that coexpressed EGFP or Venus and the
chimeras with high contributions of xenogenic iPSC-derived germ cell marker mouse vasa homolog (MVH) (Figure 5F). As
cells generally were not identified, suggesting an association the injected miPSCs or rESCs were both confirmed as germ-
with embryonic lethality. In most chimeras, the size of newborns line-competent pluripotent stem cells in intraspecific settings
seemed to conform with that in the species from which the blas- (data not shown), germ cell development may be impaired or
tocyst originated. A correlation between body weight and contri- more limited in the xenogenic environment.
bution of donor iPSC-derived cells as estimated from hair color
and peripheral blood cells is shown in Figure 6. The origins of Generation of Rat Pancreas in Mouse via Interspecific
placenta and uterus may have a key role in size determination. Blastocyst Complementation
Injected xenogenic PSCs never developed into placenta, which Our last goal was to generate xenogenic rat pancreas in Pdx1 /
was always of host blastocyst origin (Figure 4A, Figure S5A). mice by interspecific blastocyst complementation. For efficient
Therefore it is difficult to determine whether it is placenta or production of Pdx1 / mice, embryos were generated by inter-
uterine environment that influences the size of embryos. cross of Pdx1 / founder male mice (in which pancreas arose
principally from exogenous miPSCs) with Pdx1+/ female mice.
Distribution of Donor iPSC-Derived Cells With this founder system, half the mice born were Pdx1 /
in the Xenogenic Environment (in contrast to intercross of Pdx1+/ heterozygotes in which
We determined the distribution of mouse- or rat-iPSC-derived only 25% of offspring were Pdx1 / ). For donor riPSCs, we
EGFP-positive cells in neonatal interspecific chimeras. With used the riPSC#3 line (Figure S4A). This line was selected among
miPSC injection into r-blastocyst (Figure S6A) and with riPSC 11 established riPSC clones for embryonic developmental rate
injection into m-blastocyst, almost all organs contained EGFP- and degree of chimerism after injection into mouse embryos

794 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.

A Phase EGFP
Pdx1-/- + riPSCs
Pdx1+/- + riPSCs
C D (mg/dl)
Pdx1-/- + riPSCs 600
400
200
0
0 30
Minutes after glucose administration
E Phase EGFP
Li
Pdx1-/- + riPSCs
St
Du
Sp
Pa
Ki
Pdx1+/- + riPSCs
(data not shown). After injection, 139 embryos were transferred
into uteri of pseudopregnant mice and 34 mice were born.
They were analyzed at neonatal and adult stages.
The contribution of EGFP-marked riPSC-derived cells in the
pancreas of neonatal Pdx1+/ interspecific chimeras was small
relative to that of host-derived cells, as seen in whole-body
chimerism (Pdx1+/ + riPSCs in Figure 7A bottom panel; n = 5).
In contrast, the pancreatic epithelia in Pdx1 / interspecific
chimeras were entirely composed of EGFP-marked riPSC-
derived cells (Pdx1 / + riPSCs in Figure 7A top panel; n = 10).
Each genotype was confirmed by PCR using genomic DNA
extracted from FACS-sorted mouse CD45 (mCD45)-positive
splenocytes (Figure S7A). Neonates with entirely EGFP-positive
pancreata thus were clearly identified as of Pdx1 / genotype
(Figure S7B). The existence of interspecific chimeras between
mouse and rat was also confirmed by FACS patterning, which
demonstrated distinct populations of mCD45- and rat CD45
(rCD45)-positive cells, with only rCD45-positive cells expressing
EGFP after riPSC injection (Figure S7A). FACS-sorted mCD45-
and rCD45-positive cells were also confirmed as, respectively,
mouse or rat in origin by genomic PCR testing using Oct3/4
locus primers (Figure S7C), which clearly identified cell origin.
On immunostaining, riPSC-derived pancreas expressed EGFP
almost universally (Figure 7B) and also expressed a-amylase
B Figure 7. Generation of Rat Pancreas in
EGFP α-Amylase Pdx1-Deficient Mouse by Interspecific Blas-
DAPI DAPI
tocyst Complementation
(A) riPSC-derived pancreas in neonatal Pdx1 /
mouse. In Pdx1 / mice complemented with
riPSCs, almost all pancreata expressed EGFP,
Glucagon Somatostatin Insulin indicating riPSC origin (top panel). On the other
DAPI DAPI DAPI
hand, in Pdx1+/ mice complemented with riPSCs,
pancreata only partially expressed EGFP, indi-
cating a composite of both riPSC- and mouse-
derived cells (bottom panel).
(B) Immunohistological studies of sections ob-
tained from riPSC-derived pancreas. Serial frozen
sections were immunostained for EGFP, a-
amylase, insulin, glucagon, and somatostatin
with DAPI nuclear counterstaining. See also
Pdx1-/-+ riPSCs
Figure S7 for genotyping to identify Pdx1 status
Pdx1+/-+ riPSCs and origin of each species in interspecific chimera.
(C) Photograph of adult Pdx1 / mouse comple-
mented with riPSCs.
(D) Results of GTT in Pdx1 / (-) and Pdx1+/
(:, A) mice complemented with riPSCs.
60 90 120 (min) (E) riPSC-derived pancreas in adult Pdx1 /
mouse. Compare similar results of neonatal anal-
ysis (Figure 5D).
F (F) Histological studies of sections obtained from
HE EGFP riPSC-derived adult pancreas revealed clear
DAPI
differences in the distributions of riPSC-derived
cells after staining by anti-EGFP antibody with
DAPI nuclear counterstaining (right panel).
Sections were also stained with HE.
Sections in (F) were observed under light or fluo-
rescence microscopy and in (B) were observed
under confocal laser scanning microscopy. Scale
bars in (B) and (F), 100 mm.
(an exocrine tissue marker) and insulin, glucagon, and somato-
statin (endocrine tissue markers; Figure 7B).
As in wild-type experiments, successful maturation into adult-
hood (8 weeks) was uncommon in Pdx1 / mice complemented
with riPSCs; however, adult mice with riPSC-derived pancreas
(Figure 7C; n = 2) had intact pancreas expressing EGFP (Figures
7E and 7F). Quantitative analysis of the sections by image J soft-
ware revealed percentages of EGFP-positive cells to be 81.9% ±
3.4%. Additionally, on GTT in adulthood, insulin was secreted in
response to glucose loading and normal serum glucose levels
were maintained (Figure 7D). These results indicate that genera-
tion of a xenogenic iPSC-derived organ is possible via interspe-
cific blastocyst complementation.
DISCUSSION
We report three innovative observations, using proof-of-prin-
ciple approaches. (1) If an empty developmental niche for an
organ is provided (as with the Pdx1 / mouse and the pancreatic
niche), PSC-derived cellular progeny can occupy that niche
and developmentally compensate for the missing contents of
the niche, forming an organ almost entirely composed of
cells derived from donor PSCs. (2) Generation of interspecific
chimeras between mouse and rat is possible with injection of
Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 795
mouse or rat PSCs into embryos from the other species; injected
PSC-derived cells are distributed throughout the body and
appear to function normally. (3) The combination of (1) and (2)
successfully generates rat pancreas in mouse with injection of
riPSCs into Pdx1 / mouse embryos, a technique that we term
‘‘interspecific blastocyst complementation.’’
We demonstrated that the Pdx1 / mice derived from blasto-
cysts complemented with mPSCs were born with functional
pancreas almost entirely derived from donor PSCs and grew
into adulthood without showing any evidence of pancreatic
insufficiency. Several groups have used the same technique to
study the development of thymic epithelium (Muller et al.,
2005), to compensate for cardiac defects (Fraidenraich et al.,
2004), or to determine if yolk sac hematopoiesis and germ cell
development are of clonal or nonclonal origin (Ueno et al.,
2009; Ueno and Weissman, 2006). Although Stanger et al. tested
development of pancreas and liver in embryos to define organ
size determinants (Stanger et al., 2007), no study has exploited
this technique to produce donor-derived functional organs and
rescued a lethal phenotype to adulthood.
Direct in vitro differentiation of insulin-producing cells from
PSCs has been a major focus of stem cell therapy, as recently
demonstrated by a sophisticated protocol to generate pancre-
atic endoderm efficiently via stepwise endodermal differentia-
tion (Kroon et al., 2008). However, the in vitro generation
of insulin-producing cells still needs further improvement in
differentiation efficiency, in insulin production levels, or in speed
of insulin response to glucose changes. In addition, the risk
of tumor development due to contamination with undifferenti-
ated PSCs must be rigorously assessed before clinical use.
Compared with those generated in vitro, insulin-producing cells
obtained from pancreas that is formed in vivo by blastocyst
complementation must have gone through near-normal differ-
entiation processes with proper epigenetic changes. The
tissues obtained, such as insulin-producing cells, thus are
presumed to be fully functional and the risk of teratoma devel-
opment due to contamination of undifferentiated PSCs to be
negligible. Nonetheless, the oncogenicity of iPSC-derived cells
due to reactivation of introduced genes (Miura et al., 2009;
Nakagawa et al., 2008) or to genome abnormality due to long-
term culture remains to be assessed. To establish iPSCs without
genomic integration of a retroviral sequence should further
reduce the risk associated with the use of iPSCs (Okita et al.,
2008).
We assumed that aggregation of early embryos would lead
to the presence in trophectoderm of xenogenic cells that are
reported to be harmful to embryonic development after uterine
implantation. We therefore injected mESCs/iPSCs, but not blas-
tomere cells, into r-blastocysts. As predicted, EGFP-positive
PSC-derived cells were not detected in placentas (Figure 4A
and Figure S5A), and we succeeded in generating interspecific
chimeras. To confirm this further, we then attempted to generate
interspecific chimeras by injecting rESCs/iPSCs into m-blasto-
cysts. After injection into 8-cell/morula stage mouse embryos,
the riPSCs were eventually enclosed within the inner cell mass
of the m-blastocyst and were never detected in the m-blastocyst
trophectoderm (Figures S4C and S4D). Our study, consistent
with others, indicates that the presence of xenogenic cells
796 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.
among extraembryonic lineage cells, with exposure of xeno-
genic cells to the uterine environment, is inhibitory to implanta-
tion and/or to further intrauterine development of interspecific
chimeras.
It is of prime interest that body size and weight of interspecific
chimeras conformed with those of the species of the foster
mother (Figures 5A–5C). What xenogenic components contribute
to the phenotypic determination of interspecific chimeras?
It seems that placenta and/or uterine environment are respon-
sible for size determination of adult interspecific chimeras
both as embryos and as adults. Given that placenta must be of
the same origin as the foster mother for successful generation
of interspecific chimeras, it is not clear which is primarily respon-
sible for this determination. Degree of chimerism may also
influence the phenotype. As observed in intraspecific chimeras,
contribution and distribution of xenogenic cells in interspecific
chimeras vary organ-to-organ. There is a negative correlation
between contribution of donor (mouse) PSC-derived cells and
body weight (Figure 6). Although we could obtain interspecific
chimeras consistently (Figure 4 and Figure 5), embryonic lethality
was high and postnatal development was poor. Some interac-
tions between mouse- and rat-derived cells indispensable
for organism survival may not work across species, resulting in
death during embryonic development or in retarded postnatal
development.
Another example is the formation of gallbladders in inter-
specific chimeras. Those derived from m-blastocysts have
gallbladders but those from r-blastocysts do not. It is conceiv-
able that the temporo-spatial development of donor PSC-
derived cells is regulated by the xenogenic host microenviron-
ment, which governs morphogenesis and organogenesis.
However, the data also indicate that the intrinsic develop-
mental program imprinted in PSCs may create competition
between host blastocyst-derived and donor PSC-derived cells
to form chimeric organs. The balance between host and donor
cells at certain critical points during embryonic development
thus may be important. Xenogenic developmental systems
may be useful to elucidate these and other developmental
questions and to help in generation of chimeras between
species evolutionarily more distant from one another than
are mouse and rat.
Most importantly, we succeeded in generating functional rat
pancreas in Pdx1 / mice via interspecific blastocyst comple-
mentation. In all interspecific neonates derived from Pdx1 /
blastocysts injected with riPSCs, pancreas was present.
Although full maturation into adulthood was not common, once
the mice matured into adulthood, the generated riPSC-derived
pancreas was morphologically and histologically normal and
was not associated with any sign of diabetes or other abnormal-
ities; GTT results strongly indicated normal function. Generation
of functional cells (sperm, hepatocytes) in xenogenic environ-
ments has been reported (Mercer et al., 2001; Shinohara et al.,
2006). In addition, hematopoietic xeno-chimeras have been
used commonly as a method to study functionality of hematopoi-
etic stem cells (Kamel-Reid and Dick, 1988; McCune et al.,
1988). No study, however, has demonstrated generation in
a xenogenic environment of a PSC-derived functional organ
that can rescue embryonic lethality to adulthood.
 The organ generation system described may be applied to
treat organ failure in humans if pigs or other large animals are
used. There are, however, several issues that need to be
addressed to bring this principle into the clinic. For example,
though we were able to generate interspecific chimeras between
mouse and rat, their embryonic lethality is high and maturation
into adulthood is uncommon. The nature of this xenogenic barrier
is not clear, but it is evident that the evolutionary distance
accounts for this, as we do not see these problems in intraspecific
chimeras. Livestock animals such as pigs or sheep may be too
distant evolutionarily for successful complementation. In addi-
tion, as described in the allogenic system, vessels, nerves, and
some interstitial elements that are not under the influence of
Pdx1 expression were composites of host- and miPSC- or
mESC-derived cells. Although we showed in this study that islets
prepared from miPSC-derived pancreas generated in allogenic
hosts indeed were successfully transplanted into ‘‘autologous’’
mice with STZ-induced diabetes without rejection, whether the
same principle applies to transplantation of islets obtained from
pancreas generated in xenogenic animals remains to be seen.
Transplantation of islets from rat pancreas generated in Pdx1 /
mice into diabetic rats should answer this question. However,
due to the size difference between mouse and rat, and to high
embryonic lethality and poor postnatal maturation of interspecific
chimeras, it is not possible to obtain sufficient numbers of islets to
treat diabetic rats. Generation of mouse pancreas in Pdx1 / rats
will be necessary to do such experiments.
Production of organ-deficient livestock animals and genera-
tion of chimeras is another issue, but nuclear transfer technology
available for livestock animals may permit establishment of
organ-deficient pig lines, for example (Lai et al., 2002). The
successful generation of pig chimeras using blastocyst injection
has also been reported (Nagashima et al., 2004). The major diffi-
culty seems to be that primate and rodent PSCs differ (Nichols
and Smith, 2009); limits of primate PSCs in contributing to
embryo development have been suspected. Poor contribution
of human ESCs to embryo development after injection into
mouse blastocysts or chick embryo has been demonstrated
(Goldstein et al., 2002; James et al., 2006). Generation of inter-
specific chimera technology may prove a tool useful in assess-
ment of pluripotent stem cell potential, thereby addressing this
issue.
Another issue of concern is the fact that PSC-derived cells
are found not only in pancreas but in all organs and tissues,
including brain and gonads. Therefore, without proper control
of the differentiation potential of PSCs, generation of human
organs in livestock animals will face an ethical issue. There are
several approaches to address this. One is use of committed
stem or progenitor cells in place of PSCs. If they are introduced
into an appropriate microenvironment at an appropriate devel-
opmental time point, to restrict differentiation toward a particular
organ may be possible. An alternative is to use genetically modi-
fied PSCs whose differentiation potential is restricted to certain
tissues or organs.
In conclusion, the approach described here will be of use not
only for better understanding of the mechanism of organogen-
esis but also as an initial step toward the ultimate regenerative
medicine of the future.
EXPERIMENTAL PROCEDURES
Animals
C57BL/6NCrSlc, BDF1, DBA/2CrSlc, ICR mice, and Wistar rats were
purchased from SLC Japan (Shizuoka, Japan). Pdx1-LacZ heterozygous
mice (Offield et al., 1996), kindly provided by Dr. Y. Kawaguchi (Kyoto
University) and Dr. C.V. Wright (Vanderbilt University), were crossed with
C57BL/6, DBA2, or BDF1 strain mice. C57BL/6 mice were given STZ (Sigma,
St. Louis, MO, USA) to induce diabetes. Mice with nonfasting blood glucose
levels > 400 mg/dL 1 week after STZ administration (200 mg/kg) were regarded
as hyperglycemic and thus as diabetic mice. All experiments were performed
in accordance with the animal care and use committee guidelines of the Insti-
tute of Medical Science, University of Tokyo.
Culture of ESCs/iPSCs
Undifferentiated mESCs were maintained on gelatin-coated dishes without
feeder cells in Glasgow’s modified Eagle’s medium (Sigma) supplemented
with 10% fetal bovine serum (FBS; Nichirei Bioscience, Tokyo, Japan),
0.1 mM 2-mercaptoethanol (Invitrogen, San Diego, CA, USA), 0.1 mM nones-
sential amino acids (Invitrogen), 1 mM sodium pyruvate (Invitrogen), 1%
L-glutamine penicillin streptomycin (Sigma), and 1000 U/ml of mouse leukemia
inhibitory factor (LIF; Millipore, Bedford, MA, USA). The G4.2 mESCs and
EB3DR mESCs, kindly provided by Dr. H. Niwa (Center for Developmental
Biology, RIKEN), were derived from EB3 mESCs (Niwa et al., 2000) and carried
the CAG promoter-driven EGFP or DsRed gene. Undifferentiated miPSCs
were maintained on mitomycin-c treated mouse embryonic fibroblasts
(MEFs) in Dulbecco’s modified Eagle’s medium (Invitrogen) supplemented
with 15% knockout serum replacement (Invitrogen), 0.1 mM 2-mercaptoetha-
nol, 0.1 mM nonessential amino acids, 1 mM HEPES buffer solution (Invitro-
gen), 1% L-glutamine penicillin streptomycin, and 1000 U/ml of mouse LIF.
The GT3.2 miPSCs were generated from TTFs of a male EGFP Tg mouse
(kindly provided by Dr. M. Okabe, Osaka University) by introducing three
factors (Klf4, Sox2, Oct3/4) in retroviral vectors (Okabe et al., 2009). The
GT3.2 miPSCs ubiquitously express EGFP under the control of the CAG
promoter.
Undifferentiated rESCs/iPSCs were maintained on mitomycin-c treated
MEFs in N2B27 medium (Ying et al., 2003) containing 1 mM MEK inhibitor
PD0325901 (Axon Groeningen, The Netherlands), 3 mM GSK3 inhibitor
CHIR99021 (Axon), with or without FGF receptor inhibitor SU5402 (Calbio-
chem, La Jolla, CA, USA), and 1000 U/ml of rat LIF (Millipore). The riPSCs
were generated from Wistar rat embryonic fibroblast by introducing three
mouse factors (Oct3/4, Klf4, Sox2) in lentiviral vectors as a doxycycline-induc-
ible expression unit using N2B27 medium containing the three inhibitors
described above. Expression of representative ESC marker genes has been
confirmed and teratoma formation by these riPSCs after injection into immu-
nodeficient mice has also been confirmed (our unpublished data). The riPSCs
ubiquitously express EGFP under the control of the Ubiquitin-C promoter.
The WIv3i-1 or WIv3i-5 rESCs were generated from Venus Tg rat blastocyst
(Hirabayashi et al., 2009). These rESCs ubiquitously express Venus under
the control of the CAG promoter.
Embryo Culture and Manipulation
Preparation of wild-type or Pdx1 heterozygous intercrossing embryos was
carried out according to published protocols (Nagy et al., 2003). In brief,
mouse 8-cell/morula stage embryos were collected in Medium 2 (Millipore)
from oviduct and uterus of mice 2.5 days postcoitum (dpc). These embryos
were transferred into potassium simplex optimized medium with amino acids
(Millipore) and were cultured for 24 hr for blastocyst injection.
Rat blastocysts were collected in a bicarbonate-buffered medium
composed of Roswell Park Memorial Institute medium (RPMI) 1640, Eagle’s
solution, and Ham’s F12 containing 18% FBS medium (Ogawa et al., 1971)
from oviduct and uterus of rats 4.5 dpc. These embryos were transferred
into modified rat 1-cell embryo culture medium (Oh et al., 1998) containing
80 mM NaCl (Wako Pure Chemical Industries, Osaka, Japan) and 0.1% poly-
vinyl alcohol (Sigma) and were cultured for about 1 hr until injection.
For micromanipulation, ESCs or iPSCs were trypsinized and suspended in
ESC or iPSC culture medium. A piezo-driven micromanipulator (Prime Tech,
Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 797
Tokyo, Japan) was used to drill zona pellucida and trophectoderm under the
microscope and 10–15 ESCs or iPSCs were introduced into blastocyst cavities
near the inner cell mass. After blastocyst injection, embryos underwent follow-
up culture for 1–2 hr. Mouse blastocysts then were transferred into the uteri
of pseudopregnant recipient ICR female mice (2.5 dpc) and rat blastocysts
were transferred into the uteri of pseudopregnant recipient Wistar female
rats (3.5 dpc).
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures and
seven figures and can be found with this article online at doi:10.1016/j.cell.
2010.07.039.
ACKNOWLEDGMENTS
We thank Ryo Sumazaki for critical advice on this work, Kazuya Ise and
Hirofumi Noguchi for technical advice on islet isolation and islet transplanta-
tion, and Naoki Iwamori for technical advice on embryo manipulation. This
work was supported by grants from JST, the Ministry of Education, Culture,
Sport, Science, and Technology, Japan. H.N. is a founder and shareholder
of ReproCELL Inc.
Received: January 25, 2010
Revised: May 5, 2010
Accepted: June 30, 2010
Published: September 2, 2010
REFERENCES
Buehr, M., Meek, S., Blair, K., Yang, J., Ure, J., Silva, J., McLay, R., Hall, J.,
Ying, Q.L., and Smith, A. (2008). Capture of authentic embryonic stem cells
from rat blastocysts. Cell 135, 1287–1298.
Chen, J., Lansford, R., Stewart, V., Young, F., and Alt, F.W. (1993). RAG-2-
deficient blastocyst complementation: an assay of gene function in lympho-
cyte development. Proc. Natl. Acad. Sci. USA 90, 4528–4532.
Deltour, L., Leduque, P., Paldi, A., Ripoche, M.A., Dubois, P., and Jami, J.
(1991). Polyclonal origin of pancreatic islets in aggregation mouse chimaeras.
Development 112, 1115–1121.
Fehilly, C.B., Willadsen, S.M., and Tucker, E.M. (1984). Interspecific chimae-
rism between sheep and goat. Nature 307, 634–636.
Fraidenraich, D., Stillwell, E., Romero, E., Wilkes, D., Manova, K., Basson,
C.T., and Benezra, R. (2004). Rescue of cardiac defects in id knockout
embryos by injection of embryonic stem cells. Science 306, 247–252.
Goldstein, R.S., Drukker, M., Reubinoff, B.E., and Benvenisty, N. (2002). Inte-
gration and differentiation of human embryonic stem cells transplanted to the
chick embryo. Dev. Dyn. 225, 80–86.
Gotoh, M., Maki, T., Kiyoizumi, T., Satomi, S., and Monaco, A.P. (1985). An
improved method for isolation of mouse pancreatic islets. Transplantation
40, 437–438.
Hirabayashi, M., Kato, M., Kobayashi, T., Sanbo, M., Yagi, T., Hochi, S., and
Nakauchi, H. (2009). Establishment of rat embryonic stem cell lines that can
participate in germline chimeras at high efficiency. Mol. Reprod. Dev. 77, 94.
James, D., Noggle, S.A., Swigut, T., and Brivanlou, A.H. (2006). Contribution of
human embryonic stem cells to mouse blastocysts. Dev. Biol. 295, 90–102.
Kamel-Reid, S., and Dick, J.E. (1988). Engraftment of immune-deficient mice
with human hematopoietic stem cells. Science 242, 1706–1709.
Kroon, E., Martinson, L.A., Kadoya, K., Bang, A.G., Kelly, O.G., Eliazer, S.,
Young, H., Richardson, M., Smart, N.G., Cunningham, J., et al. (2008). Pancre-
atic endoderm derived from human embryonic stem cells generates glucose-
responsive insulin-secreting cells in vivo. Nat. Biotechnol. 26, 443–452.
Lai, L., Kolber-Simonds, D., Park, K.W., Cheong, H.T., Greenstein, J.L., Im,
G.S., Samuel, M., Bonk, A., Rieke, A., Day, B.N., et al. (2002). Production of
798 Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc.
alpha-1,3-galactosyltransferase knockout pigs by nuclear transfer cloning.
Science 295, 1089–1092.
Li, P., Tong, C., Mehrian-Shai, R., Jia, L., Wu, N., Yan, Y., Maxson, R.E.,
Schulze, E.N., Song, H., Hsieh, C.L., et al. (2008). Germline competent embry-
onic stem cells derived from rat blastocysts. Cell 135, 1299–1310.
McCune, J.M., Namikawa, R., Kaneshima, H., Shultz, L.D., Lieberman, M., and
Weissman, I.L. (1988). The SCID-hu mouse: murine model for the analysis of
human hematolymphoid differentiation and function. Science 241, 1632–1639.
Mercer, D.F., Schiller, D.E., Elliott, J.F., Douglas, D.N., Hao, C., Rinfret, A.,
Addison, W.R., Fischer, K.P., Churchill, T.A., Lakey, J.R., et al. (2001). Hepatitis
C virus replication in mice with chimeric human livers. Nat. Med. 7, 927–933.
Miura, K., Okada, Y., Aoi, T., Okada, A., Takahashi, K., Okita, K., Nakagawa,
M., Koyanagi, M., Tanabe, K., Ohnuki, M., et al. (2009). Variation in the safety
of induced pluripotent stem cell lines. Nat. Biotechnol. 27, 743–745.
Muller, S.M., Terszowski, G., Blum, C., Haller, C., Anquez, V., Kuschert, S.,
Carmeliet, P., Augustin, H.G., and Rodewald, H.R. (2005). Gene targeting of
VEGF-A in thymus epithelium disrupts thymus blood vessel architecture.
Proc. Natl. Acad. Sci. USA 102, 10587–10592.
Nagashima, H., Giannakis, C., Ashman, R.J., and Nottle, M.B. (2004). Sex
differentiation and germ cell production in chimeric pigs produced by inner
cell mass injection into blastocysts. Biol. Reprod. 70, 702–707.
Nagy, A., Gertsenstein, M., Vintersten, K., and Behringer, R. (2003). Manipu-
lating the Mouse Embryo: A Laboratory Manual, Third Edition (Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory Press).
Nakagawa, M., Koyanagi, M., Tanabe, K., Takahashi, K., Ichisaka, T., Aoi, T.,
Okita, K., Mochiduki, Y., Takizawa, N., and Yamanaka, S. (2008). Generation of
induced pluripotent stem cells without Myc from mouse and human fibro-
blasts. Nat. Biotechnol. 26, 101–106.
Nichols, J., and Smith, A. (2009). Naive and primed pluripotent states. Cell
Stem Cell 4, 487–492.
Niwa, H., Miyazaki, J., and Smith, A.G. (2000). Quantitative expression of Oct-
3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat.
Genet. 24, 372–376.
Offield, M.F., Jetton, T.L., Labosky, P.A., Ray, M., Stein, R.W., Magnuson,
M.A., Hogan, B.L., and Wright, C.V. (1996). PDX-1 is required for pancreatic
outgrowth and differentiation of the rostral duodenum. Development 122,
983–995.
Ogawa, S., Sato, K., and Hashimoto, H. (1971). In vitro culture of rabbit ova
from the single cell to the blastocyst stage. Nature 233, 422–424.
Oh, S.H., Miyoshi, K., and Funahashi, H. (1998). Rat oocytes fertilized in modi-
fied rat 1-cell embryo culture medium containing a high sodium chloride
concentration and bovine serum albumin maintain developmental ability to
the blastocyst stage. Biol. Reprod. 59, 884–889.
Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T., and Nishimune, Y. (1997).
‘Green mice’ as a source of ubiquitous green cells. FEBS Lett. 407, 313–319.
Okabe, M., Otsu, M., Ahn, D.H., Kobayashi, T., Morita, Y., Wakiyama, Y., Ono-
dera, M., Eto, K., Ema, H., and Nakauchi, H. (2009). Definitive proof for direct
reprogramming of hematopoietic cells to pluripotency. Blood 114, 1764–1767.
Okita, K., Nakagawa, M., Hyenjong, H., Ichisaka, T., and Yamanaka, S. (2008).
Generation of mouse induced pluripotent stem cells without viral vectors.
Science 322, 949–953.
Rossant, J. (2007). Stem cells and lineage development in the mammalian
blastocyst. Reprod. Fertil. Dev. 19, 111–118.
Rossant, J., and Frels, W.I. (1980). Interspecific chimeras in mammals:
successful production of live chimeras between Mus musculus and Mus caroli.
Science 208, 419–421.
Rossant, J., Mauro, V.M., and Croy, B.A. (1982). Importance of trophoblast
genotype for survival of interspecific murine chimaeras. J. Embryol. Exp.
Morphol. 69, 141–149.
Satoh, M., Yasunami, Y., Matsuoka, N., Nakano, M., Itoh, T., Nitta, T., Anzai,
K., Ono, J., Taniguchi, M., and Ikeda, S. (2007). Successful islet transplantation
to two recipients from a single donor by targeting proinflammatory cytokines in
mice. Transplantation 83, 1085–1092.
Shinohara, T., Kato, M., Takehashi, M., Lee, J., Chuma, S., Nakatsuji, N.,
Kanatsu-Shinohara, M., and Hirabayashi, M. (2006). Rats produced by inter-
species spermatogonial transplantation in mice and in vitro microinsemination.
Proc. Natl. Acad. Sci. USA 103, 13624–13628.
Slack, J.M. (1995). Developmental biology of the pancreas. Development 121,
1569–1580.
Stanger, B.Z., Tanaka, A.J., and Melton, D.A. (2007). Organ size is limited by
the number of embryonic progenitor cells in the pancreas but not the liver.
Nature 445, 886–891.
Stern, M.S. (1973). Letter: Chimaeras obtained by aggregation of mouse eggs
with rat eggs. Nature 243, 472–473.
Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K.,
and Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human
fibroblasts by defined factors. Cell 131, 861–872.
Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells
from mouse embryonic and adult fibroblast cultures by defined factors. Cell
126, 663–676.
Tarkowski, A.K. (1962). Inter-specific transfers of eggs between rat and
mouse. J. Embryol. Exp. Morphol. 10, 476–495.
Ueno, H., and Weissman, I.L. (2006). Clonal analysis of mouse development
reveals a polyclonal origin for yolk sac blood islands. Dev. Cell 11, 519–533.
Ueno, H., Turnbull, B.B., and Weissman, I.L. (2009). Two-step oligoclonal
development of male germ cells. Proc. Natl. Acad. Sci. USA 106, 175–180.
Ying, Q.L., Stavridis, M., Griffiths, D., Li, M., and Smith, A. (2003). Conversion
of embryonic stem cells into neuroectodermal precursors in adherent mono-
culture. Nat. Biotechnol. 21, 183–186.
Ying, Q.L., Wray, J., Nichols, J., Batlle-Morera, L., Doble, B., Woodgett, J.,
Cohen, P., and Smith, A. (2008). The ground state of embryonic stem cell
self-renewal. Nature 453, 519–523.
Zeilmaker, G.H. (1973). Fusion of rat and mouse morulae and formation of
chimaeric blastocysts. Nature 242, 115–116.
Cell 142, 787–799, September 3, 2010 ª2010 Elsevier Inc. 799
Profiling by Image Registration
Reveals Common Origin of Annelid
Mushroom Bodies and Vertebrate Pallium
Raju Tomer,1,* Alexandru S. Denes,1,2 Kristin Tessmar-Raible,1,3 and Detlev Arendt1,*
1Developmental Biology Unit, European Molecular Biology Laboratory, D-69117 Heidelberg, Germany
2Present address: Biozentrum, University of Basel, Klingelbergstrasse 50/70, CH-4056 Basel, Switzerland
3Present address: Max F. Perutz Laboratories, Campus Vienna Biocenter, Dr. Bohr-Gasse 9/4, A-1030 Vienna, Austria
*Correspondence: tomer@embl.de (R.T.), arendt@embl.de (D.A.)
DOI 10.1016/j.cell.2010.07.043
SUMMARY
The evolution of the highest-order human brain
center, the ‘‘pallium’’ or ‘‘cortex,’’ remains enigmatic.
To elucidate its origins, we set out to identify related
brain parts in phylogenetically distant animals, to
then unravel common aspects in cellular composi-
tion and molecular architecture. Here, we compare
vertebrate pallium development to that of the mush-
room bodies, sensory-associative brain centers, in
an annelid. Using a newly developed protocol for
cellular profiling by image registration (PrImR), we
obtain a high-resolution gene expression map for the
developing annelid brain. Comparison to the verte-
brate pallium reveals that the annelid mushroom
bodies develop from similar molecular coordinates
within a conserved overall molecular brain topology
and that their development involves conserved
patterning mechanisms and produces conserved
neuron types that existed already in the proto-
stome-deuterostome ancestors. These data indicate
deep homology of pallium and mushroom bodies
and date back the origin of higher brain centers to
prebilaterian times.
INTRODUCTION
In human and other mammals, the pallium represents the most
highly developed part of the forebrain, the site of learning and
memory (Kandel et al., 2000). It harbors huge densities of inter-
neurons arranged in cortical layers around a central neuropil
(hence, ‘‘cortex’’; Figure 1). The pallium is less elaborate in other
vertebrates and generally thought to function as a sensory-asso-
ciative center integrating primarily olfactory information (Nieu-
wenhuys, 2002). Insects and spiders, but also annelids, form
similar brain centers with densely packed neurons that send
out thousands, in some species hundreds of thousands, of
approximately parallel processes that form a central, lobed
neuropil (Strausfeld et al., 1998) (Figure 1). These likewise repre-
sent sensory-associative brain centers implicated in olfactory
800 Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc.
discrimination, as well as in olfactory learning and memory
(Heisenberg, 2003; Strausfeld et al., 2009). Anatomical, struc-
tural, and functional similarities of the evolutionarily more ancient
parts of the vertebrate pallium, the ‘‘paleopallium’’ and ‘‘archipal-
lium,’’ with, for example, mushroom bodies of neopteran insects
have long been noted and interpreted as convergent acquisi-
tions as a result of functional constraints in independent evolu-
tionary lineages (Farris, 2005, 2008; Strausfeld et al., 1998).
Also, the evolution of characteristic folds and fissures and the
subpartitioning into functional units have occurred indepen-
dently in the mammalian pallium and several times in insect
mushroom bodies (Farris, 2008). These multiple convergences,
however, do not rule out the possibility that vertebrate pallium
and invertebrate mushroom bodies ultimately trace back to
a common evolutionary precursor. Some less elaborate,
sensory-associative brain centers may have already existed in
the protostome ancestor (of, among others, insects and anne-
lids) or even in the last common ancestor of protostomes and
deuterostomes (to which vertebrates belong). These may have
given rise to vertebrate pallium and invertebrate mushroom
bodies alike.
We set out to decide between the two alternatives—indepen-
dent evolution versus common origin of pallium and mushroom
bodies from ancient sensory-associative brain centers—by the
detailed molecular comparison of their development and differ-
entiation. Previous studies revealed conserved expression of
various transcription factors in vertebrate and invertebrate fore-
brains, but a specific comparison of gene expression in associa-
tive brain centers has not been undertaken. For this, we investi-
gated mushroom body development in the marine annelid
Platynereis dumerilii (Fischer and Dorresteijn, 2004), to then
compare to available vertebrate data. Platynereis is an ‘‘errant’’
annelid with huge mushroom bodies (Müller, 1973) that actively
explores the environment and shows some sort of learning
(Evans, 1966). Fortunately, for the purpose of genetic compar-
ison, the Platynereis transcriptome has proven to be ‘‘slow
evolving,’’ exhibiting less evolutionary change than that of other
protostomes with mushroom bodies (Drosophila, Apis) (Raible
et al., 2005). Also, in line with a slower evolutionary pace of the
annelid lineage, we have shown that Platynereis neural
patterning exhibits ancient characteristics, many of which are
shared with the vertebrates but not with the faster-evolving fly

Figure 1. Comparison of Vertebrate Pallium and Annelid Mushroom
Body Development Based on Previous Work and on This Study
(A–E) Development of the mouse pallium from open neural plate early specifi-
cation (5 s; A and B) to E10.5 late specification stage (C and D) and adult (E).
(F–J) Development of the Platynereis mushroom bodies from 48 hpf specifica-
tion stage (F–I) to adult (J). Note that in mouse, elaborate morphogenesis
(neural tube and telencephalic vesicle formation) occurs between early and
late specification stages, which does not take place in the annelid.
(A) The mouse pallium (PAL) anlage with reference to the anterior-posterior
(otx, gbx, hox) and mediolateral (pax6, nkx2.1, nkx2.2) regional patterning
systems. (After Inoue et al. [2000] and Shimamura et al. [1995].)
(B) Expression of bf-1 (orange) outlining the telencephalon anlage including the
pallium. Wnt5/8 and Shh indicate spatially restricted activity of signaling
ligands implicated in telencephalon development.
(C) Regionalization of telencephalic vesicles into pallium (PAL) and lateral and
medial ganglionic eminences (LGE, MGE). Colored strip indicates level of
cross-section in (D). M, mesencephalon; HT, hypothalamus, Eye, position of
cutoff eye stalk.
(D) Gene expression along the telencephalic section in (C) indicated by black
bars and color code. For references, see the main text.
(E) Parasagittal section of the mouse cortex.
and nematode lineages (Denes et al., 2007; Tessmar-Raible
et al., 2007). Finally, Platynereis is easy to breed and thus acces-
sible to high-throughput approaches (Fischer and Dorresteijn,
2004), as a prerequisite for the analysis of multiple genes.
Previous comparisons of brain development between phylo-
genetically remote species have been restricted to gene-by-
gene comparisons that did not allow, or allowed only in a coarse
manner, relating expression data spatially for multiple genes. To
overcome this, we have established a new protocol for Profiling
by Image Registration (PrImR) that, for the first time, allows align-
ment of expression images with cellular resolution and thus
expression profiling for the whole brain at the single-cell level.
Our protocol builds on the highly stereotypical nature of Platyner-
eis neurodevelopment to align expression patterns for an unlim-
ited number of genes in an additive process.
Using PrImR, we first investigated whether the overall molec-
ular topography of the brain anlage is conserved between
annelid and vertebrate and whether the Platynereis mushroom
bodies develop from corresponding molecular regions when
compared to vertebrate pallium (as a test for homology) (Arendt,
2005; Woodger, 1945). We also tested whether the positioning of
both follows similar developmental patterning mechanisms.
Next, we used PrImR to determine the expression profile for
specific subregions, focusing on specifically expressed genes
making up the regulatory and differentiation signature of cells
developing from these regions and compared this ‘‘molecular
fingerprint’’ (Arendt, 2008) to that of vertebrate telencephalic
neuron types but also to that of the mushroom body neuroblasts
in Drosophila. We conclude that pallium and mushroom body
most likely evolved from the same (sensory associative) brain
center that was already present in the bilaterian ancestors.
RESULTS
In Silico Alignment of Gene Expression Patterns
To establish a high-resolution molecular topography of the
annelid brain and to investigate the molecular fingerprint of brain
subregions, we set out to obtain coexpression information for
a large number of specifically expressed genes. This is impos-
sible to achieve at large scale by two-color double whole-mount
in situ hybridization (Tessmar-Raible et al., 2005), because this
would require a nonlinearly increasing amount of technically
demanding experiments. To overcome this, we took advantage
of the observation that Platynereis development is highly stereo-
typic and synchronous within and across batches (Fischer and
Dorresteijn, 2004) and developed a computational protocol for
(F) The Platynereis mushroom body anlage with reference to the anterior-
posterior (otx, gbx, hox) and mediolateral (pax6, nkx2.1, nkx2.2) regional
patterning systems. MB, mushroom bodies.
(G) Expression of bf-1 (orange) in the annelid brain at 48 hpf. Wnt5/8 and Hh
indicate spatially restricted expression of conserved ligands.
(H) Brain molecular topography of the bf-1+ region at 48 hpf as revealed by this
study. Color code refers to (I).
(I) Gene expression in the Platynereis brain indicated by black bars and color
code referring to (H).
(J) The Platynereis adult mushroom bodies, parasagittal section.
See also Figure S2.
Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc. 801

(H) Violin plot illustrating the distribution of Pearson’s correlation coefficients of 171 gene expression images with their gene-specific average images. Magenta
shows the data density. Mean = 0.900, median = 0.904, standard deviation = 0.040.
Scale bars represent 50 mm. See also Figure S1 and Movies S1, S2, and S3.
accurately aligning the expression patterns of different genes
acquired from different stained individuals of identical develop-
mental stage (Profiling by Image Registration, PrImR, j,primərj).
In the first step, two-channel image stacks were acquired via
whole-mount reflection confocal laser-scanning microscopy
(Jékely and Arendt, 2007). One channel contained information
on the expression pattern of a given gene and the other on the
axonal scaffold of the Platynereis larval brain. More than three
biological replicas were acquired for every gene. Next, we
used the axonal scaffold channel to align these images to a
reference average axonal scaffold image (Figure 2A, Figure S1
available online), first via rigid alignment algorithms and then
by smooth nonrigid transformation (see the Experimental Proce-
dures) (Figures 2B–2D). For each gene, a normalized average
expression image from (in most cases) five individuals was
thus generated (Figure S1B) that was directly comparable to
those of an unlimited number of other genes. As a test for accu-
racy, we obtained almost perfect overlap of signal for single cells
stained for the same gene in different individuals (Figure 2E). We
also found that the coexpression images obtained by our PrImR
protocol fully reproduced those obtained by double-fluorescent
WMISH (Tessmar-Raible et al., 2005) (compare Figures 2F and
2G) but tended to be more ‘‘complete,’’ reflecting the higher
sensitivity of the NBT/BCIP staining and of the reflection micros-
copy technique (Jékely and Arendt, 2007). Finally, to challenge
the accuracy of the protocol, we systematically estimated the
extent of overlap between individually aligned scans by calcu-
lating the Pearson’s correlation coefficient for 171 individually
aligned scans with the gene-specific average image (see the
Experimental Procedures) and found that the probability is
highest to obtain a value above 0.9, which implies that the
average expression images used for this study should reliably
reproduce endogenous gene expression. Our protocol thus
allows the comparison of any newly added expression pattern
to all preexisting patterns at once, with high accuracy and in
cellular resolution.
802 Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc.
Figure 2. In Silico Alignment of Expression
Patterns
(A) Maximum Z projection of average reference
axonal scaffold image of 48 hpf larval brain.
(B–D) Step-wise alignment of an axonal scaffold
image to the average reference scaffold (cyan)
using rigid, affine, and smoothened freeform
nonrigid transformations. Images are 20 mm thick
optical sections. See Movies S1 and S2 for details.
(E) A positive control experiment to test the accu-
racy of the alignment algorithm. All images are
maximum Z projections of aligned expression
patterns of r-opsin, acquired from three different
larvae, and an RGB merge of them. See Movie
S3 for details.
(F and G) Experimental validation of an in silico co-
expression (F) using double fluorescence WMISH
double staining (G). White, colocalization of green
and red pixels. Cyan and blue arrows point to veri-
fied coexpression.
The Molecular Topography of the Platynereis Brain
We used our PrImR protocol to determine the molecular topog-
raphy of the Platynereis brain after 2 days of development, to test
whether any of the Platynereis forebrain regions would show
‘‘telencephalon-like’’ coordinates. At this stage, major subre-
gions of the developing Platynereis nervous system are already
established and larger populations of neurons have started
differentiation (Tessmar-Raible et al., 2007). For our purposes,
it was sufficient to focus on one stage only, because in the
absence of drastic morphogenetic changes (such as neurulation
and vesicle formation in vertebrates), gene expression patterns
in the developing Platynereis brain remain spatially constant
over time (as shown for example for pax6, six1/2 [Arendt et al.,
2002] or dach and bf-1 [Figures 5 and 6]), indicating colocaliza-
tion and temporal co-occurrence of differentiated neurons and
their precursors.
We had previously shown that both the annelid and vertebrate
brains exhibit a basic subdivision into medial nk2.1+ and lateral
pax6+ subregions (Tessmar-Raible et al., 2007) (Figure 1A)
(transformed into a ventral-dorsal arrangement during vertebrate
neurulation; Figures 1A–1C). It is within this conserved frame-
work that the vertebrate telencephalon anlage is established,
with the pallium developing from the anterior part of the pax6+
region (Figure 1A). The outline of the whole telencephalon anlage
is demarcated early on by expression of bf-1 (Hébert and Fishell,
2008) (orange in Figure 1B), a crucial regulator of telencephalic
development (Danesin et al., 2009). In the course of telence-
phalic morphogenesis, gsx+ and emx+ (Kimura et al., 2005)
domains complement the nk2.1+ and pax6+ subregions (Figures
1C and 1D) such that the gsx+, pax6+ overlap demarcates the
future pallial-subpallial boundary (Danesin et al., 2009; Yun
et al., 2001) and pax6+, emx+ coexpression specifies pallial
telencephalic precursors (Kimura et al., 2005). Besides bf-1,
lhx2 (expressed broadly in the telencephalon and acting highly
up in the hierarchy of cortical induction (Hébert and Fishell,
2008), COUP-TF1/seven-up (contributing to telencephalic

patterning (Hébert and Fishell, 2008) and soxB family members
(sox1[Ekonomou et al., 2005] and sox2 [Bani-Yaghoub et al.,
2006]) play important roles in early telencephalic development
(Figure 3E), although the spatial distribution and coexpression
of these factors along the anterior-posterior and mediolateral
brain axes is not fully resolved for the vertebrates.
We accordingly examined bf-1, lhx2, svp, soxB, pax6, emx,
gsx, and nk2.1 expression in Platynereis and indeed identified
a brain region with telencephalon-like molecular topography
(Figures 3A–3I; summarized in Figure 1). First, we found bf-1
expressed in a horseshoe-shaped domain in the Platynereis
brain (Figure 3A), where it is specifically coexpressed with lhx2,
svp, and soxB (Figures 3C–3E). (The more lateral part of bf-1+
domain represents the eye anlage; Figure 3B.) Second, from
lateral to medial, the Platynereis bf-1 brain domain is subdivided
by spatially restricted coexpression of emx, pax6, gsx, and nk2.1
(Figures 3F–3J) in a telencephalon-like fashion. Nowhere else
in the developing Platynereis larva was a similar sequence of
emx+/pax6+, pax6+, pax6+/gsx+, gsx+, and gsx+/nk2.1+ subre-
gions (Figure 3J) detected. The emx+/pax6+ and pax6+ regions
within the bf1+ domain thus represented candidate counterpart
regions to the vertebrate pallium anlage, a notion that we
decided to explore further.
Comparison of Patterning Mechanisms
In vertebrates, bf-1 coordinates the activity of two opposing
signaling centers that pattern the telencephalon anlage. bf-1
acts downstream of the ventral signal, Hh, to induce ventral
(subpallial) identities. At the same time, bf-1 inhibits dorsal
Wnt/b-catenin signaling through direct transcriptional repression
of Wnt8 (Danesin et al., 2009), which induces dorsal (pallial)
identities (Houart et al., 2002). Vertebrate telencephalic
Figure 3. Coexpression of ‘‘Telencephalic’’
and Other Marker Genes, Red, with bf-1,
Green, in the Platynereis 48 hpf Larval Brain
(A and B) Complete expression of bf-1 (green) (A).
Blue, reference axonal scaffold. Dotted circle
marks expression in the eye field, as indicated
by r-opsin coexpression in (B).
(C–E) genes covering the almost entire bf-1
domain.
(F–I) Lateral-to-medial sequence of regionalisa-
tion genes.
(J) Five-micron-thick optical section showing
lateral-to-medial coexpression of emx (green),
pax6(red), and gsx (cyan).
(K–O) Genes involved in mediolateral signaling.
(P–T) Genes showing mediolaterally restricted
expression.
All images are Maximum Z projections of slice
numbers 13 to 40, i.e., 28 mm thick (except K:
8–28, i.e., 21 mm thick) in the reference scaffold
image stack. White marks colocalizing pixels.
Scale bars represent 50 mm. See also Figure S5.
patterning can be perturbed by ectopic b-catenin activation
(mimicking the dorsal signal), which accordingly triggers the
upregulation of dorsal pallial and downregulation of ventral sub-
pallial markers (Backman et al., 2005). Investigating the expres-
sion of orthologous signals and factors in Platynereis, we indeed
detected Hh in the medial (Figure 3K) and Wnt8 in the lateral brain
(Figure 3L), matching the vertebrate situation. We also observed
lateral expression of Wnt5 (Figure 3M), another dorsal Wnt signal
implicated in corticospinal axon guidance (Keeble et al., 2006), of
WntA (not found in vertebrates; Figure 3N) and Gli (Figure 3O),
a zinc finger transcription factor essential for dorsal telencephalic
fates (Tole et al., 2000). None of the other Platynereis Wnt genes
was expressed in the brain (data not shown; note that Wnt3 does
not exist in annelids). For Platynereis, exposure to the chemical
compound 1-Azakenpaullone has been established as an effi-
cient means to ectopically induce b-catenin activation (Schneider
and Bowerman, 2007). We accordingly tested the effect of 1-Aza-
kenpaullone on Platynereis brain regionalization and observed
upregulation of lateral emx expression (Figures 4B and 4G),
downregulation of intermediate gsx (Figures 4D and 4I), and
medial nk2.1 expression (Figures 4E and 4J), while pax6 ap-
peared unaffected (Figures 4C and 4H). These results mirror
vertebrate telencephalic patterning (Figure 1) and are thus
strongly indicative of evolutionary conservation.
The Molecular Fingerprint of Mushroom Body Neurons
In vertebrates, distinct combinations of differentially expressed
transcription factors control the fate of the various telencephalic
subregions. We took advantage of PrImR to further determine
and compare molecular fingerprints. In mouse, dorsal telence-
phalic (pallial) pax6+ regions show differential activity of Dach
(Caubit et al., 1999) and ngn1/2, bHLH factors required for pallial
Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc. 803
Figure 4. Differential Gene Regulation in the Developing Platynereis Brain
Expression of bf-1, emx, pax6, gsx, and nk2.1 in DMSO control (A–E) and 1-Azakenpaullone (1 mM final concentration) treated 48 hpf larval brain (F–J). Changes in
expression patterns are highlighted by dashed circles. Blue, nuclear stain DAPI; green, immunostaining against acetylated tubulin. n for bf-1: treated > 200,
control > 200; emx: treated > 380, control > 300; pax6: treated > 350, control > 310; gsx: treated > 150, control > 160; nk2.1: treated > 320, control > 200.
Scale bars represent 50 mm.
neurogenesis (Nieto et al., 2001). In Platynereis, dach and ngn
are similarly restricted to emx+, pax6+ subregions (Figures 3P
and 3Q; compare Figures 1D and 1I). In mouse, ventral (subpal-
lial) gsx+ regions coexpress vax1, required for GABAergic inter-
neuron generation (Taglialatela et al., 2004), and coe/ebf1, an
HLH transcription factor affecting subpallial development (Garel
et al., 1999). We accordingly found both genes coexpressed in
Platynereis gsx+ regions (Figures 3R and 3S; compare Figures
1D and 1I). In mouse, telencephalic subregions express er81,
an ETS factor active in distinct subpallial domains (Yun et al.,
2001) and direct target of pax6 in the dorsal pallium (Tuoc
and Stoykova, 2008). Platynereis er81 is similarly expressed
in the pax6+/gsx+ and gsx+/nk2.1+ subregions but not in
between (Figure 3T; compare Figures 1D and 1I). Adding to
this, expression of Platynereis genes orthologous to ascl1/
mash1 (bHLH downstream effector of gsx1 [Wang et al., 2009];
expressed in ventral and dorsal telencephalon [Yun et al.,
2001]), to brn1/2/4 (POU domain transcription factors required
for cortical migration in dorsal and ventral telencephalon [Ekono-
mou et al., 2005; McEvilly et al., 2002]), and to many other genes
implicated in telencephalon development by function or by
expression is detailed for Platynereis in Figure S2 and has
been used to generate the molecular map in Figure 1H. The over-
all comparison suggests that the molecular fingerprints of
annelid and vertebrate emx+/pax6+, pax6+, pax6+/gsx+, gsx+,
and gsx+/nk2.1+ subregions are similar to large extent but may
also differ in the detail (for example, dlx is expressed more
broadly in annelid than in vertebrate; rx is expressed in the
emx+, pax6+, and gsx+ subregions in the annelid but excluded
in the vertebrate; Figure S2).
Positioning the Platynereis Mushroom Body Anlagen
Since our molecular comparison had identified one unique
candidate region in the Platynereis brain, which, by position,
similar responsiveness to Wnt signaling, and similar molecular
804 Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc.
fingerprint (bf1+, emx+, pax6+, dach+, svp+, tll+, ngn+, asc+)
qualified as possible evolutionary counterpart to the vertebrate
pallium anlage (encircled by stippled line in Figures 1C and 1D),
we set out to determine whether this paired region indeed
represented the Platynereis mushroom body anlagen by tracing
mushroom body development from adult (Figure 5, Figure S3,
Movie S4) and juvenile stages (10 days postfertilization [dpf],
when they are fully developed and can be easily identified
histologically) back to earlier larval stages. We combined
anatomical (the specific shape of mushroom bodies), histolog-
ical (the dense packing of neurons), and topological (the specific
spatial arrangement of mushroom bodies and palpal and
antennal nerves) evidence, as well as the expression of mush-
room body-specific marker genes (dach and pax6), to reidentify
and position the developing mushroom bodies at various stages
(6 weeks, 20, 10, 9, 8, 7, 6, 5, 4, 3, and 2 dpf; Figure 5 and data
not shown). This allowed unambiguous positioning of the mush-
room body anlagen to ventrolateral coordinates at 48 hr postfer-
tilization [hpf] that indeed matched the emx+, pax6+, dach+
candidate region.
In juvenile worms, the differentiating emx+, pax6+ mushroom
bodies and the more medial gsx+, nk2.1+ brain tissue (referred
to as ‘‘pars intercerebralis’’[Müller, 1973]) continued to express
bf-1 (Figure 6A) and started to specifically express arx (Fig-
ure 6B; expressed in mouse and fish telencephalon and ventral
thalamus [Miura et al., 1997]). We also detected specific expres-
sion of two conserved bilaterian microRNAs, miR-9 and mir-9*
(Christodoulou et al., 2010) (Figures 6C and 6D). In the verte-
brates, these microRNAs are expressed only in the telenceph-
alon, among all differentiated CNS tissues (Deo et al., 2006).
Regarding transmitter usage, large part of the Platynereis mush-
room bodies proved glutamatergic by expression of vglut,
encoding the vesicular glutamate transporter (Figure 6F). In
contrast, gad, a marker for GABAergic neurons, was restricted
to more medial brain tissue (Figure 6E), as was the dopaminergic

marker tyrosine hydroxylase (Figure 6H), again matching the
vertebrate situation (Marı́n and Rubenstein, 2001). However,
we also detected a difference in the spatial distribution of trans-
mitter type in that Platynereis mushroom bodies also harbored
cholinergic neurons (Figure 6G; not found in the vertebrate
pallium).
Molecular Fingerprint Comparison with Insect
Mushroom Bodies
Mushroom bodies of similar cellular composition also exist in
arthropods such as onychophorans (velvet worms), spiders,
and insects, yet ample molecular data exist only for the mush-
room bodies in insects. In Drosophila, mushroom body neuro-
blasts require pax6 and dach for their specification but do not
express sine oculis (six1/2) or eyes absent (eya) (Noveen et al.,
2000). We found that the only region of the Platynereis brain
coexpressing pax6 and dach but devoid of six1/2 and eya is
indeed the mushroom body anlage (Figures 7A and 7B and eya
Figure 5. Platynereis Mushroom Body
Development
Three-dimensional models were generated for
various stages according to confocal image
stacks (see Movie S4 for details). Depicted struc-
tures represent mushroom body pedunculi (red),
mushroom body perikarya (transparent red),
antennal nerve (dark blue), pars intercerebralis
(purple), eyes (yellow), palpae (green), and
a glomerulus-like structure (mediates the connec-
tion between palpae and mushroom bodies; light
green). Continuous expression of dach (top and
bottow row) in mushroom body neurons consis-
tent with anatomical backtracing. Blue arrows
point to the antennal nerve. A, P, D, V, L, and R
indicate body axes.
Scale bars represent 50 mm. See also Figure S3
and Movie S4.
data not shown), indicative of evolu-
tionary conservation. To challenge
homology of insect and annelid mush-
room bodies, we looked for a combination
of transcription factors that, by
coexpression, uniquely define Drosophila
mushroom body neuroblasts independent of pax6 and dach—to
then test whether a similar coexpression would locate to the
annelid mushroom body anlage in an equally specific manner.
Among the four Drosophila mushroom body neuroblasts, Pcd8
andPcd9 are uniquely identified by the specific coexpression
of asc/l’sc, tll, svp, slp/bf-1, and otx (Urbach and Technau,
2003). We determined by PrImR where in the Platynereis 48
hpf brain these five transcription factors would be coexpressed,
and indeed the only two bilateral spots of cells in the Platynereis
brain coexpressing them form part of the mushroom body
anlage (Figures 7C–7H). This is strong evidence in favor of an
evolutionary relatedness of insect and annelid mushroom
bodies. However, in contrast to annelid mushroom bodies
(Figure 7I) and to vertebrate pallium, the Drosophila dac+ mush-
room body precursors do not express emx, and the overall
molecular topography is conserved to lesser extent between
fly, annelid, and vertebrate (Urbach and Technau, 2003) than
between annelid and vertebrate. Our data suggest that this is
Figure 6. Gene Expression in the Brain of
Platynereis Juvenile Worms
Transcription factors (A and B), microRNAs (C and
D) and marker genes (E–H) for GABAergic (gad, E),
glutamatergic (vglut, F), cholinergic (vacht, G),
and dopaminergic (tyrosine hydroxylase, H) neu-
rons. Dashed circles outline mushroom bodies.
A, P, L, and R indicate body axes. Scale bars
represent 50 mm.
Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc. 805

due to secondary modifications of mushroom body develop-
ment in the insect lineage.
DISCUSSION
In this study, we have established a new protocol, PrImR, that
allows expression profiling by image registration with cellular
resolution for the entire organism. Expression profiling at
single-cell resolution has been reported previously (Liu et al.,
2009; Sugino et al., 2006), but our protocol is the first to allow
simultaneous profiling for the whole central nervous system,
where cells are densely packed. Our protocol thus sets the stage
for a new chapter of evolutionary developmental research that
has so far been restricted to gene-by-gene comparisons of
expression patterns with restricted coexpression information,
necessarily remaining fragmentary and sometimes inconclusive.
Instead, PrImR allows much broader comparisons, involving
hundreds of transcription factors and differentiation genes, and
a much higher level of integration. For the first time, molecular
fingerprinting as introduced here allows testing homology
hypotheses such as the comparison of mushroom bodies
between annelids and insects and of mushroom bodies with
vertebrate pallium (Strausfeld et al., 1998) that could not be
resolved on merely anatomical and physiological grounds.
Taking advantage of this new protocol, our data reveal that
mushroom bodies in annelid and vertebrate pallium develop
from the same, molecularly defined, subregion (emx+/pax6+) at
similar anterior-posterior and mediolateral coordinates within an
overall conserved molecular topography of the brain (Figures 1A
and 1F), which is established by similar patterning mechanisms
(Figures 1B and 1G). They express similar and unique combina-
806 Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc.
Figure 7. Coexpression of Insect Neuro-
blast Marker Genes, Red, with Dach, Green,
in the Platynereis 48 hpf Larval Brain
(A–J) Coexpression of dach (green) with pax6 (A),
six1/2 (B), ash2 (C), tll (D), svp (E), bf-1/slp (F), otx
(G), the intersection of ash2, tll, svp, bf-1/slp, and
otx (H), emx (I), and Wnt5 (J) (red). The intersection
of expression patterns was generated from
expression masks generated by smoothening
images with a Gaussian sigma of 2 and a signal
cutoff of 35. All images are Maximum Z projection
of slice numbers 13 to 40, i.e., 28 mm thick, in the
reference scaffold image stack. White marks
colocalizing pixels.
(K) Phylogenetic distribution of mushroom body
and pallium in Bilateria.
Scale bars represent 50 mm. See also Figure S4.
tions of additional transcription factors
(dach+, svp+, slp+, tll+, ngn+, asc+,
arx+) (Figures 1C, 1D, 1H, 1I, and 6).
They give rise to glutamatergic neurons,
while GABAergic and dopaminergic
neuron types emerge from more medial/
ventral brain regions. They are positive
for the same brain-specific microRNAs
(mir-9, -9*). We are confident that these
similarities are too complex to be plausibly explained by indepen-
dent (convergent) evolution. Instead, we propose that they were in
place already in the protostome-deuterostome ancestor, the last
common ancestor of annelids and vertebrates (Figure 7K).
To assess the statistical significance of our dataset, we con-
ducted permutation tests by randomly assigning gene expres-
sion to ‘‘pallium’’ or ‘‘mushroom body,’’ and we found the
observed similarity in molecular fingerprint extremely unlikely to
arise by chance (Figure S4; p value less than 2.0E-05). It should
be stressed, however, that this value might slightly overestimate
the significance of our comparison, because a subset of the
involved genes may be coregulated and should be treated as
a single observation.
Importantly, however, annelid mushroom body and vertebrate
pallium development and specification also differ to consider-
able extent, which is plausible given 600 million years of inde-
pendent evolution. First, vertebrate pallium development
involves neurulation and telencephalic vesicle formation (Figures
1B and 1C), which does not occur in the annelid. Second, impor-
tant differences are apparent in some expression domains. For
example, annelid nk2.1 expression exhibits lateral ‘‘wings’’
(Figure 1F), where it overlaps with the lateral pax6 domain (care-
ful 3D analysis, however, reveals that this overlap is outside of
the bf1+ mushroom body anlage). Also, the developing annelid
mushroom body is positive for the rx gene (Figure S2), which is
excluded from the vertebrate telencephalon anlage (Stigloher
et al., 2006). Finally, some genes are absent in the other species.
For example, Wnt3, expressed in the dorsomedial pallium
anlage, does not exist in annelids, and WntA, demarcating the
lateral mushroom body anlage (Figure 3N), is not found in the
vertebrate genomes.
 What is the significance of these findings for our under-
standing of mushroom body and pallium/telencephalon evolu-
tion? We can infer that the last common ancestor of protostomes
and deuterostomes (and thus, of annelids and vertebrates)
already possessed distinct brain centers that later gave rise to
the pallium in the vertebrate and to mushroom bodies in the
annelid (and arthropod) lineage (Figure 7K). Based on our data,
it is likely that these centers harbored subsets of glutamatergic
neurons that presumably functioned in some sort of integration
of sensory signals. Input was most likely chemosensory, as
deduced from the primarily chemosensory input of pallium and
mushroom bodies alike, in insects, spiders, and onychophorans,
as well as in annelids (Strausfeld et al., 2006). (The Platynereis
mushroom bodies receive connections from the chemosensory
palpae [Movie S4].) Notably, however, the sensory input of
arthropod mushroom bodies is received by distinct olfactory
appendages (Strausfeld et al., 2006) and can even be visual
(in spiders, honeybee and cockroach [Strausfeld and Barth,
1993]), indicating some plasticity in mushroom body connec-
tivity. Apparently, the sensory input into sensory associative
centers can change according to the needs of changing ecolog-
ical niches.
We can speculate that the first function of these chemosen-
sory integrative brain centers was to distinguish between food
and nonfood, to decide about directed locomotion toward
identified food sources, and, ultimately, to integrate previous
experiences into some sort of ‘‘learning’’ (Evans, 1966). Extrap-
olating from the similar motor output of today’s pallium and
mushroom bodies, we can assume that these integrative centers
already produced coordinated locomotion of some sort (Farris,
2005).
There is a long-standing notion that the vertebrate pallium
comprises distinct parts of different evolutionary age, paleo-,
archi-, and neopallium. Roughly, the paleopallium corresponds
to olfactory cortex and the archipallium to the hippocampus,
and the neopallium comprises the remainder of the cerebral
hemispheres (Swanson, 2000). Paleo- and archipallium are
found in all vertebrates, while the neopallium is restricted to
amniotes. In line with this concept, the cellular organization of
the more ancient archi- and paleopallium strikingly resembles
that of annelid and arthropod mushroom bodies, in that densely
packed, basophilic somata send out parallel processes inter-
sected by afferent and efferent networks (Strausfeld et al.,
1998). Intriguingly, when compared to neopallium, the hippo-
campus and dentate gyrus each consist of only one layer of
principle neurons (Förster et al., 2006). This contrasts with the
six-layered architecture of the neopallium but rather resembles
the simple architecture of the mushroom bodies. We propose
that such simple architecture, ancient for vertebrates, might
have been in place in the urbilaterian mushroom body/pallium
sensory-associative precursor structure.
As mentioned in the introduction, it is unlikely that a mushroom
body-like shape was already in place in the protostome-deutero-
stome ancestor. Rather, it seems plausible that the common
evolutionary precursor structure of mushroom bodies and
pallium was morphologically less complex and that higher
degrees of histological (and thus, functional) complexity were
acquired independently in divergent evolutionary lineages, as
repeatedly proposed (Farris, 2005, 2008; Strausfeld et al.,
1998; Strausfeld et al., 2009). To settle this issue, it will be
rewarding to identify brain regions with similar molecular coordi-
nates and fingerprint in species lacking overt mushroom body
morphology such as crustaceans or mollusks on the protostome
side or amphioxus on the deuterostome side (Figure 7K) and
to investigate possible sensory input, integrative function and
locomotor output of the corresponding brain parts. If, on one
hand, such less complex, ‘‘mushroom body-related’’ brain parts
indeed exist, they will help defining a minimum set of anatomical
and physiological characteristics associated with the conserved
molecular fingerprints described here. This way, we would gain
more insight into anatomy and function of an ancient sensory-
integrative center present in the protostome-deuterostome
ancestor. If, on the other, such regions do not exist, we can infer
secondary loss of brain complexity. Possible evolutionary
scenarios leading to reduced brain complexity include the tran-
sition to sessile or hemisessile lifestyle with filter feeding, to inter-
stitial burrowing with minute body size, or to parasitism.
Once a better and more complete picture of mushroom body-
related brain parts in bilaterians is obtained, this will set the stage
for a more comprehensive view on the evolution of insect and
annelid mushroom bodies and vertebrate pallium alike. What
do these groups have in common that is not found in bilaterians
in which mushroom body-related brain parts are present but less
elaborate? This will ultimately help to link mushroom body/
pallium evolution to changes in life style, locomotion, or food
capture.
EXPERIMENTAL PROCEDURES
Imaging Gene Expression Patterns
We used whole-mount reflection confocal microscopy (Jékely and Arendt,
2007) to acquire gene expression pattern images. Typically, an oil immersion
403 objective on the Leica TCS SPE Confocal Microscope was used.
Alignment of Expression Patterns
For achieving highly accurate alignments of axonal scaffolds, we developed
a multistep multiresolution image registration algorithm using Mutual Informa-
tion (Viola and Wells, 1995) as the image similarity metric. In the first step,
images were transformed rigidly with six degrees of freedom (three for trans-
lation and three for rotation) to roughly align the images. In the second step,
affine transformation with 12 degrees of freedom (three for rotation, three for
translation, three for scaling, and three for shear) was used to compensate
for the shrinking and shearing artifacts introduced by in situ hybridization
and imaging procedures. In the last step, we used a uniform grid of control
points to model freeform nonrigid local deformations and used third-order
B-splines to smoothly interpolate the transformations. For the affine and
nonrigid registration steps, we used a multiresolution approach in which 3D
images were smoothened and downsampled to various degrees before the
alignment.
Average Reference Axonal Scaffold
We generated an average Platynereis larval axonal scaffold using images
acquired from 36 different individual 48 hpf larvae (Figure S1), which was
used as reference for all the alignments.
Averaging Expression Patterns
For direct comparison of the spatial expression patterns of different genes, we
generated normalized average expression patterns from three or more indi-
vidual larvae (see Movie S3 and Figure S1B for examples). We used
Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc. 807
‘‘Colocalization Highlighter’’ plug-in in ImageJ 1.41e to highlight the colocaliz-
ing pixels as white pixels. Default settings were used for most of the cases.
Alignment Accuracy Estimates
We conducted a systematic analysis by testing images from 171 different
larvae. Pearson’s correlation coefficient was calculated for each aligned
expression image with the corresponding gene-specific average.
Modeling Platynereis Brain Anatomy
Amira 4.1.1 was used to segment various anatomical structures in the brain
(Movie S4) and to generate 3D surface renderings shown in Figure 5.
Ectopic Overexpression of Wnt Canonical Signaling
We used an established procedure (Schneider and Bowerman, 2007) of treat-
ment with 1-Azakenpaullone (cat. no. 191500, Calbiochem) to achieve ectopic
overexpression of canonical Wnt signaling in Platynereis dumerilii larvae.
A mixture of larvae from two to three different animal pairs was used for
treatment. The larvae were treated from 24 hpf to 48 hpf in 6 ml Natural Sea
Water at a final concentration of 1 mM 1-Azakenpaullone or DMSO and were
grown at 18 C. The larvae were fixed in 4% PFA/PTW, stored in 100% meth-
anol at 20 C, and proceeded for in situ hybridization. The staining reactions
were stopped simultaneously for control and treated larvae, much before the
saturation.
Cloning Genes
We used degenerate primers to clone fragments of brn1/2/4, dach, emx, svp,
tll, vglut, and gad and 50 ,30 -RACE extensions to extend the fragments (see the
Extended Experimental Procedures for primer sequences). We used cDNA
templates generated with SMART technology (SMART RACE cDNA amplifica-
tion kit, Clonetech) from total RNA extracted from larvae at several develop-
mental stages. Other genes investigated in this study were obtained either
as EST sequences or from previous publications. The GenBank IDs of arx,
ash2, bf-1, brn1/2/4, coe, dach, emx, er81, ets3, lhx1/5, svp, tll, vglut, Wnt5,
Wnt8, gad, and tyr hyd are GU169412–GU169428, respectively.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, five
figures, and four movies and can be found with this article online at doi:
10.1016/j.cell.2010.07.043.
ACKNOWLEDGMENTS
We thank F. Christodoulou for providing larvae stained with mir-9, 9*,
H. Snyman for expert technical assistance, and the Arendt lab for support.
The electron micrograph in Figure 5 was taken with C. Nielsen during a prac-
tical course. R.T. and K.T.-R. were supported by fellowships of the Marie Curie
RTN ZOONET (MRTN-CT-2004-005624), R.T. by DFG Collaborative Research
Network SFB-488, and A.D. by a Louis Jeantet Foundation fellowship. R.T.
conceived and developed PrImR, designed and performed the experiments,
analyzed and interpreted the data, and wrote the paper. A.S.D. contributed
clones of vglut and gad. K.T.-R. contributed clones of brn124 and emx. D.A.
proposed to study annelid mushroom bodies, provided ideas, interpreted
the data, and wrote the paper.
Received: March 12, 2010
Revised: May 22, 2010
Accepted: July 14, 2010
Published: September 2, 2010
REFERENCES
Arendt, D. (2005). Genes and homology in nervous system evolution:
comparing gene functions, expression patterns, and cell type molecular
fingerprints. Theory Biosci. 24, 185–197.
808 Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc.
Arendt, D. (2008). The evolution of cell types in animals: emerging principles
from molecular studies. Nat. Rev. Genet. 9, 868–882.
Arendt, D., Tessmar, K., de Campos-Baptista, M.I., Dorresteijn, A., and
Wittbrodt, J. (2002). Development of pigment-cup eyes in the polychaete
Platynereis dumerilii and evolutionary conservation of larval eyes in Bilateria.
Development 129, 1143–1154.
Backman, M., Machon, O., Mygland, L., van den Bout, C.J., Zhong, W.,
Taketo, M.M., and Krauss, S. (2005). Effects of canonical Wnt signaling
on dorso-ventral specification of the mouse telencephalon. Dev. Biol. 279,
155–168.
Bani-Yaghoub, M., Tremblay, R.G., Lei, J.X., Zhang, D., Zurakowski, B.,
Sandhu, J.K., Smith, B., Ribecco-Lutkiewicz, M., Kennedy, J., Walker, P.R.,
and Sikorska, M. (2006). Role of Sox2 in the development of the mouse
neocortex. Dev. Biol. 295, 52–66.
Caubit, X., Thangarajah, R., Theil, T., Wirth, J., Nothwang, H.G., Rüther, U.,
and Krauss, S. (1999). Mouse Dac, a novel nuclear factor with homology to
Drosophila dachshund shows a dynamic expression in the neural crest, the
eye, the neocortex, and the limb bud. Dev. Dyn. 214, 66–80.
Christodoulou, F., Raible, F., Tomer, R., Simakov, O., Trachana, K., Klaus, S.,
Snyman, H., Hannon, G.J., Bork, P., and Arendt, D. (2010). Ancient animal
microRNAs and the evolution of tissue identity. Nature 463, 1084–1088.
Danesin, C., Peres, J.N., Johansson, M., Snowden, V., Cording, A., Papalo-
pulu, N., and Houart, C. (2009). Integration of telencephalic Wnt and hedgehog
signaling center activities by Foxg1. Dev. Cell 16, 576–587.
Denes, A.S., Jékely, G., Steinmetz, P.R., Raible, F., Snyman, H., Prud’homme,
B., Ferrier, D.E., Balavoine, G., and Arendt, D. (2007). Molecular architecture of
annelid nerve cord supports common origin of nervous system centralization
in bilateria. Cell 129, 277–288.
Deo, M., Yu, J.Y., Chung, K.H., Tippens, M., and Turner, D.L. (2006). Detection
of mammalian microRNA expression by in situ hybridization with RNA oligonu-
cleotides. Dev. Dyn. 235, 2538–2548.
Ekonomou, A., Kazanis, I., Malas, S., Wood, H., Alifragis, P., Denaxa, M.,
Karagogeos, D., Constanti, A., Lovell-Badge, R., and Episkopou, V. (2005).
Neuronal migration and ventral subtype identity in the telencephalon depend
on SOX1. PLoS Biol. 3, e186.
Evans, S.M. (1966). Non-associative behavioural modifications in nereid poly-
chaetes. Nature 211, 945–948.
Farris, S.M. (2005). Evolution of insect mushroom bodies: old clues, new
insights. Arthropod Struct. Dev. 34, 211–234.
Farris, S.M. (2008). Evolutionary convergence of higher brain centers spanning
the protostome-deuterostome boundary. Brain Behav. Evol. 72, 106–122.
Fischer, A., and Dorresteijn, A.W.C. (2004). The polychaete Platynereis dumer-
ilii (Annelida): a laboratory animal with spiralian cleavage, lifelong segment
proliferation and a mixed benthic/pelagic life cycle. Bioessays 26, 314–325.
Förster, E., Zhao, S., and Frotscher, M. (2006). Laminating the hippocampus.
Nat. Rev. Neurosci. 7, 259–267.
Garel, S., Marı́n, F., Grosschedl, R., and Charnay, P. (1999). Ebf1 controls early
cell differentiation in the embryonic striatum. Development 126, 5285–5294.
Hébert, J.M., and Fishell, G. (2008). The genetics of early telencephalon
patterning: some assembly required. Nat. Rev. Neurosci. 9, 678–685.
Heisenberg, M. (2003). Mushroom body memoir: from maps to models. Nat.
Rev. Neurosci. 4, 266–275.
Houart, C., Caneparo, L., Heisenberg, C., Barth, K., Take-Uchi, M., and
Wilson, S. (2002). Establishment of the telencephalon during gastrulation by
local antagonism of Wnt signaling. Neuron 35, 255–265.
Inoue, T., Nakamura, S., and Osumi, N. (2000). Fate mapping of the mouse
prosencephalic neural plate. Dev. Biol. 219, 373–383.
Jékely, G., and Arendt, D. (2007). Cellular resolution expression profiling using
confocal detection of NBT/BCIP precipitate by reflection microscopy. Bio-
techniques 42, 751–755.
Kandel, E.R.S., James, H., and Jessell, T.M. (2000). Principles of Neural
Science, Fourth Edition (New York: McGraw-Hill Companies, Inc.).
Keeble, T.R., Halford, M.M., Seaman, C., Kee, N., Macheda, M., Anderson,
R.B., Stacker, S.A., and Cooper, H.M. (2006). The Wnt receptor Ryk is required
for Wnt5a-mediated axon guidance on the contralateral side of the corpus
callosum. J. Neurosci. 26, 5840–5848.
Kimura, J., Suda, Y., Kurokawa, D., Hossain, Z.M., Nakamura, M., Takahashi,
M., Hara, A., and Aizawa, S. (2005). Emx2 and Pax6 function in cooperation
with Otx2 and Otx1 to develop caudal forebrain primordium that includes
future archipallium. J. Neurosci. 25, 5097–5108.
Liu, X., Long, F., Peng, H., Aerni, S.J., Jiang, M., Sánchez-Blanco, A., Murray,
J.I., Preston, E., Mericle, B., Batzoglou, S., et al. (2009). Analysis of cell fate
from single-cell gene expression profiles in C. elegans. Cell 139, 623–633.
Marı́n, O., and Rubenstein, J.L. (2001). A long, remarkable journey: tangential
migration in the telencephalon. Nat. Rev. Neurosci. 2, 780–790.
McEvilly, R.J., de Diaz, M.O., Schonemann, M.D., Hooshmand, F., and
Rosenfeld, M.G. (2002). Transcriptional regulation of cortical neuron migration
by POU domain factors. Science 295, 1528–1532.
Miura, H., Yanazawa, M., Kato, K., and Kitamura, K. (1997). Expression of
a novel aristaless related homeobox gene ‘Arx’ in the vertebrate telenceph-
alon, diencephalon and floor plate. Mech. Dev. 65, 99–109.
Müller, W.A. (1973). [Autoradiographic studies on the synthetic activity of
neurosecretory cells in the brain of Platynereis dumerilii during sexual develop-
ment and regeneration]. Z. Zellforsch. Mikrosk. Anat. 139, 487–510.
Nieto, M., Schuurmans, C., Britz, O., and Guillemot, F. (2001). Neural bHLH
genes control the neuronal versus glial fate decision in cortical progenitors.
Neuron 29, 401–413.
Nieuwenhuys, R. (2002). Deuterostome brains: synopsis and commentary.
Brain Res. Bull. 57, 257–270.
Noveen, A., Daniel, A., and Hartenstein, V. (2000). Early development of the
Drosophila mushroom body: the roles of eyeless and dachshund. Develop-
ment 127, 3475–3488.
Raible, F., Tessmar-Raible, K., Osoegawa, K., Wincker, P., Jubin, C.,
Balavoine, G., Ferrier, D.E., Benes, V., de Jong, P., Weissenbach, J., et al.
(2005). Vertebrate-type intron-rich genes in the marine annelid Platynereis
dumerilii. Science 310, 1325–1326.
Schneider, S.Q., and Bowerman, B. (2007). beta-Catenin asymmetries after all
animal/vegetal- oriented cell divisions in Platynereis dumerilii embryos
mediate binary cell-fate specification. Dev. Cell 13, 73–86.
Shimamura, K., Hartigan, D.J., Martinez, S., Puelles, L., and Rubenstein, J.L.
(1995). Longitudinal organization of the anterior neural plate and neural tube.
Development 121, 3923–3933.
Stigloher, C., Ninkovic, J., Laplante, M., Geling, A., Tannhäuser, B., Topp, S.,
Kikuta, H., Becker, T.S., Houart, C., and Bally-Cuif, L. (2006). Segregation of
telencephalic and eye-field identities inside the zebrafish forebrain territory is
controlled by Rx3. Development 133, 2925–2935.
Strausfeld, N.J., and Barth, F.G. (1993). Two visual systems in one brain:
neuropils serving the secondary eyes of the spider Cupiennius salei.
J. Comp. Neurol. 328, 43–62.
Strausfeld, N.J., Hansen, L., Li, Y., Gomez, R.S., and Ito, K. (1998). Evolution,
discovery, and interpretations of arthropod mushroom bodies. Learn. Mem. 5,
11–37.
Strausfeld, N.J., Strausfeld, C.M., Loesel, R., Rowell, D., and Stowe, S. (2006).
Arthropod phylogeny: onychophoran brain organization suggests an archaic
relationship with a chelicerate stem lineage. Proc Biol Sci 273, 1857–1866.
Strausfeld, N.J., Sinakevitch, I., Brown, S.M., and Farris, S.M. (2009). Ground
plan of the insect mushroom body: functional and evolutionary implications.
J. Comp. Neurol. 513, 265–291.
Sugino, K., Hempel, C.M., Miller, M.N., Hattox, A.M., Shapiro, P., Wu, C.,
Huang, Z.J., and Nelson, S.B. (2006). Molecular taxonomy of major neuronal
classes in the adult mouse forebrain. Nat. Neurosci. 9, 99–107.
Swanson, L.W. (2000). Cerebral hemisphere regulation of motivated behavior.
Brain Res. 886, 113–164.
Taglialatela, P., Soria, J.M., Caironi, V., Moiana, A., and Bertuzzi, S. (2004).
Compromised generation of GABAergic interneurons in the brains of Vax1-/-
mice. Development 131, 4239–4249.
Tessmar-Raible, K., Steinmetz, P.R.H., Snyman, H., Hassel, M., and Arendt, D.
(2005). Fluorescent two-color whole mount in situ hybridization in Platynereis
dumerilii (Polychaeta, Annelida), an emerging marine molecular model for
evolution and development. Biotechniques 39, 460, 462, 464.
Tessmar-Raible, K., Raible, F., Christodoulou, F., Guy, K., Rembold, M.,
Hausen, H., and Arendt, D. (2007). Conserved sensory-neurosecretory cell
types in annelid and fish forebrain: insights into hypothalamus evolution. Cell
129, 1389–1400.
Tole, S., Ragsdale, C.W., and Grove, E.A. (2000). Dorsoventral patterning of
the telencephalon is disrupted in the mouse mutant extra-toes(J). Dev. Biol.
217, 254–265.
Tuoc, T.C., and Stoykova, A. (2008). Er81 is a downstream target of Pax6 in
cortical progenitors. BMC Dev. Biol. 8, 23.
Urbach, R., and Technau, G.M. (2003). Molecular markers for identified neuro-
blasts in the developing brain of Drosophila. Development 130, 3621–3637.
Viola, P.A., and Wells, W.M., III. (1995). Alignment by maximization of mutual
information. International Journal of Computer Vision 24, 137–154.
Wang, B., Waclaw, R.R., Allen, Z.J., 2nd, Guillemot, F., and Campbell, K.
(2009). Ascl1 is a required downstream effector of Gsx gene function in the
embryonic mouse telencephalon. Neural Dev 4, 5.
Woodger, J.H. (1945). On biological transformations. In Essays on Growth and
Form, presented to D’Arcy Wentworth Thompson, W.E. le Gros Clark and P.B.
Medawar, eds. (Oxford: Oxford University Press), pp. 95–120.
Yun, K., Potter, S., and Rubenstein, J.L. (2001). Gsh2 and Pax6 play comple-
mentary roles in dorsoventral patterning of the mammalian telencephalon.
Development 128, 193–205.
Cell 142, 800–809, September 3, 2010 ª2010 Elsevier Inc. 809
 Resource

The Protein Composition of Mitotic

Chromosomes Determined Using
Multiclassifier Combinatorial Proteomics
Shinya Ohta,1,6 Jimi-Carlo Bukowski-Wills,1,2,6 Luis Sanchez-Pulido,4 Flavia de Lima Alves,1 Laura Wood,1 Zhuo A. Chen,1
Melpi Platani,1 Lutz Fischer,1 Damien F. Hudson,1,5 Chris P. Ponting,4 Tatsuo Fukagawa,3 William C. Earnshaw,1,*
and Juri Rappsilber1,*
1Wellcome Trust Centre for Cell Biology
2Centre for Systems Biology at Edinburgh, School of Biological Sciences
University of Edinburgh, Edinburgh EH9 3JR, UK
3Department of Molecular Genetics, National Institute of Genetics and The Graduate University for Advanced Studies, Mishima,

Shizuoka 411-8540, Japan

4MRC Functional Genomics Unit, Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford OX1 3QX, UK
5Present address: Murdoch Children’s Research Institute, Royal Children’s Hospital, Melbourne, Victoria 3052, Australia
6These authors contributed equally to this work

*Correspondence: bill.earnshaw@ed.ac.uk (W.C.E.), juri.rappsilber@ed.ac.uk (J.R.)

DOI 10.1016/j.cell.2010.07.047

SUMMARY chores so that chromosomes can align and segregate on the

Despite many decades of study, mitotic chromo-
some structure and composition remain poorly char-
acterized. Here, we have integrated quantitative pro-
teomics with bioinformatic analysis to generate
a series of independent classifiers that describe the
4,000 proteins identified in isolated mitotic chro-
mosomes. Integrating these classifiers by machine
learning uncovers functional relationships between
protein complexes in the context of intact chromo-
somes and reveals which of the 560 uncharacter-
ized proteins identified here merits further study.
Indeed, of 34 GFP-tagged predicted chromosomal
proteins, 30 were chromosomal, including 13 with
centromere-association. Of 16 GFP-tagged pre-
dicted nonchromosomal proteins, 14 were confirmed
to be nonchromosomal. An unbiased analysis of the
whole chromosome proteome from genetic knock-
outs of kinetochore protein Ska3/Rama1 revealed
that the APC/C and RanBP2/RanGAP1 complexes
depend on the Ska complex for stable association
with chromosomes. Our integrated analysis predicts
that up to 97 new centromere-associated proteins
remain to be discovered in our data set.
INTRODUCTION
As cells enter mitosis, chromosomes undergo a remarkable
series of physiological and structural transformations known as
chromosome condensation. This process involves individualiza-
tion of the chromosomal territories to create the characteristic
mitotic chromosome morphology and maturation of the kineto-
mitotic spindle. Our understanding of the mechanisms under-
lying chromosome condensation is still fragmentary. These
processes can be fully understood only when all components
of mitotic chromosomes have been identified and functional
relationships between them determined. We have developed
a new approach that we term multiclassifier combinatorial pro-
teomics (MCCP) to do this.
The current list of mitotic chromatin proteins reported in pro-
teomic studies is surprisingly short. Early analyses described
62 and 79 proteins, respectively, in mitotic chromosome scaf-
folds (Morrison et al., 2002; Gassmann et al., 2005). A later study
identified > 250 proteins that bound to sperm chromatin in
Xenopus egg extracts in vitro, revealing the kinetochore protein
Bod1 (Porter et al., 2007). Other studies identified 240 proteins,
subsequently corrected to roughly 50 bona fide putative struc-
tural proteins (Uchiyama et al., 2005; Takata et al., 2007). In a tar-
geted study, 98 proteins were identified as shared in isolated
telomeres from wild-type and ALT cells (Dejardin and Kingston,
2009). Despite these efforts, currently available proteomics
reports miss a significant fraction of known mitotic chromosomal
proteins, particularly kinetochore components.
Biochemical analysis of important chromosome substruc-
tures such as kinetochores is extremely challenging. The
kinetochore is one of the most complex cellular substructures
(Cheeseman and Desai, 2008), with over 120 constituents
described by a range of approaches (Earnshaw and Rothfield,
1985), including targeted proteomic studies (Obuse et al.,
2004; Foltz et al., 2006; Okada et al., 2006; Hori et al., 2008;
Amano et al., 2009). Biochemical dissection of kinetochores is
complicated by the fact that it is not known to what extent
the constituent protein complexes can be recovered in soluble
form from chromosomes with their relevant intermolecular
associations intact. As described here, those problems can
be circumvented by purifying and analyzing whole mitotic
chromosomes.

810 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.

 Purifying large cellular structures or organelles free of con-
taminants is virtually impossible. Genuine components have
been distinguished from contaminants in such preparations by
subtractive (Schirmer et al., 2003) or quantitative (Foster et al.,
2003) proteomics by determining the difference between two
near-identical fractions, one enriched and the other depleted of
the target structure. In protein correlation profiling, a set of known
components was used to define a common intensity profile
across neighboring biochemical fractions from sucrose gradients
during purification of organelles and this was used to select other
proteins that show a similar profile (Andersen et al., 2003). These
methods do not recognize cellular background proteins that
adhere to the structure of interest due to nonspecific hydro-
phobic or electrostatic interactions. This type of contamination
is particularly relevant for vertebrate mitotic chromosomes.
In the present study, we identified approximately 4000 polypep-
tides in highly purified chromosomes. We developed a statistical
approach for analysis of proteomic data to confirm which known
and uncharacterized proteins from this long list are chromosomal.
An experimental test of our method led to the identification of 32
chromosomal proteins, including 13 kinetochore-associated
proteins. Key to our analysis is the innovative use of stable isotope
labeling with amino acids in cell culture (SILAC) (Ong et al., 2002),
plus development of a framework to integrate data from multiple
classifiers, including nonproteomic classifiers, to reveal proteins
of interest and determine functional relationships between protein
complexes in the context of whole chromosomes.
RESULTS
Identification of the Proteome of Isolated
Mitotic Chromosomes
We isolated mitotic chromosomes from chicken DT40 cells by
a refinement of the polyamine method (Lewis and Laemmli,
1982) (Figures 1A and 1B). These preparations are negative for
porin in immunoblots (Figure 1C), indicating that mitochondrial
contamination (common in chromosome preparations) is low.
Proteomic analysis (Cox and Mann, 2008; de Godoy et al.,
2008) of 250 mg of total chromosomal protein identified 4029
proteins in 28 functional categories (Figure 1D and Table S1
available online), including essentially all previously described
chromosomal proteins.
When vertebrate cells enter mitosis, the nuclear envelope
breaks down and chromosomes are newly exposed to cyto-
plasmic proteins, organelles and cellular membranes. Since
highly positively charged histones contribute 38% of the chro-
mosome mass and an equivalent amount is highly negatively
charged DNA, many charged nonchromosomal proteins exhibit
strong adventitious binding to chromosomes. These ‘‘hitch-
hikers’’ differ from conventional contaminants (e.g., mitochon-
dria), as they are physically associated with the chromosomes
before cell lysis and apparently cannot be separated by conven-
tional purification protocols. Thus, the 1331 cytoskeletal, cyto-
plasmic, mitochondrial, membrane and receptor proteins found
in our preparations, may be physically associated with chromo-
somes following nuclear envelope disassembly, but many not be
functionally relevant.
A Classifier Approach to Identify Genuine Mitotic
Chromosome Proteins
The presence of hitchhiker proteins complicates the definition
of what constitutes a ‘‘true’’ chromosomal protein, as well as
the design of biochemical control experiments. For example,
comparing mitotic to interphase chromatin is of limited use,
since the latter is shielded from the cytoplasm by the nuclear
envelope. In such a comparison, cytoplasmic hitchhiker proteins
would be scored as mitosis-specific chromosomal proteins.
Here, we describe an approach to study the complex chromo-
somal proteome that both identifies proteins that merit further
study and reveals functional relationships between all chromo-
somal proteins. We quantify the chromosomal association of
each protein in a series of quantitative proteomics experiments,
mostly using SILAC technology (Ong et al., 2002). Each experi-
ment provides an independent measure of a protein’s associa-
tion with mitotic chromosomes, which we term a ‘‘classifier.’’
Integration of the data obtained with all classifiers enables us
to detect patterns in the behavior of groups of proteins that
reveal shared membership in protein complexes as well as func-
tional dependency relationships.
The experimental protocols that define the five proteomic
classifiers are shown in Figure 1E and all classifiers are summa-
rized below.
Classifier I: Abundance estimation
To estimate the amounts of individual proteins in mitotic chromo-
somes, we used an established protocol (Rappsilber et al., 2002;
Ishihama et al., 2005, 2008) to calculate a scaled protein abun-
dance index based on the number of peptides observed and
the number of times that each peptide is observed (spectral
count) for each protein. This calculation and its validation are dis-
cussed in Extended Experimental Procedures.
In the conventional pie chart of Figure 1D, all proteins are
weighted equally, independent of their actual abundance in iso-
lated chromosomes. A more informative representation of chro-
mosome composition is obtained by normalizing each class by
its mass, obtained by multiplying the estimated abundance by
the predicted molecular mass of each protein (Figure 1F). As
expected, histones comprise the bulk of mitotic chromosomal
protein (48%). Overall 68% of the protein mass is annotated as
chromosomal.
Classifier II: Enrichment in Chromosomes
We expected core chromosomal components like histones or
structural proteins would be more abundant in isolated chromo-
somes than in cytoplasmic extracts. The reverse would be true of
background proteins. We therefore mixed isolated chromo-
somes from mitotic DT40 cells grown in light medium with an
equal mass of protein from post-chromosomal extracts of
parallel cultures grown with heavy SILAC medium (Figure 1E).
Classifier II was calculated as the ratio of light/heavy peaks for
each protein. Among the 20% most enriched proteins, chromo-
somal proteins outnumbered nonchromosomal proteins by 3 to
1. Conversely, among the 20% least enriched proteins, back-
ground proteins outnumbered chromosomal proteins 6 to 1.
Classifier III: In Vitro Exchange on Chromosomes
We ranked proteins based on their ability to stably bind to chro-
mosomes during an incubation in cytosol. A crude light chromo-
some fraction obtained by gentle centrifugation was mixed with
Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 811

an excess of heavy post-chromosomal extract and incubated
to allow proteins to exchange at 14 C for 30 min (Figure 1E).
The chromosomes were then subjected to rigorous purification.
All heavy proteins identified must have bound to the chromo-
somes during the incubation in vitro. Classifier III is the light/
heavy ratio for each protein identified in this experiment.
The most stable chromosomal-associated proteins were
histones (average classifier III = 63), topoisomerase IIa (classifier
812 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.
Figure 1. Proteomic Analysis of Mitotic
Chromosome Proteins
(A) Outline of chromosome isolation procedure.
(B) Silver-stained gel of isolated chromosomes.
(C) Immunoblot with markers for mitochondria
(porin) and chromosomes.
(D) The 28 classes of proteins found in chromo-
somes.
(E) Experimental designs to prepare samples
for determination of classifiers I (abundance), II
(enrichment in chromosomes versus post-chro-
mosomal extract), III (exchange onto chromo-
somes from post-chromosomal cytosol), IV
(dependency on SMC2/condensin) and V (depen-
dency on Ska3/RAMA1).
(F–J) (F) Estimated percentages of total chromo-
somal protein mass in the major classes of
proteins. Sample spectra used to calculate classi-
fier III (exchange) values for (G) Ribosomal protein
SA (a hitchhiker protein), (H) MAD2, (I) BUBR1
and (J) MAD1. Spectra are plotted as the peak
intensity (a measure of abundance) against the
mass divided by charge. Peaks of light peptides
from chromosomes are indicated by blue bars
and heavy peptides bound from cell extracts
during the incubation are indicated by red bars.
See also Figure S1.
III = 61) and condensin I (average classi-
fier III = 59). Interestingly, condensin II
was more exchangeable (average classi-
fier III = 14). In contrast, the ratios for
75% of ribosomal proteins were in the
range from 0.45 to 3.0 (Figure 1G), indi-
cating significant binding to chromo-
somes from cytosol during the incubation.
Although we sought to optimize purity
rather than preserve functionality of
chromosomes, the exchange experiment
revealed that at least one aspect of kinet-
ochore function was retained in purified
chromosomes. Similar to one recent
study (Kulukian et al., 2009), kinetochores
of the purified chromosomes can recruit
spindle checkpoint proteins Mad2, Bub3,
and BubR1 from cytosol (Figures 1H and
1I) but not Mad1 (Figure 1J).
Classifier IV: SMC2 Dependency
We used a conditional genetic knockout
of SMC2 in DT40 SMC2ON/OFF cells (Hud-
son et al., 2003) to compare the composi-
tion of mitotic chromosomes formed in the presence or absence
of condensin, which is required for structural integrity of mitotic
chromosomes. DT40 SMC2ON cells were cultured in SILACheavy
medium. To obtain chromosomes depleted of condensin, cells
grown in SILAClight medium were cultured with doxycycline for
30 hr to shut down SMC2 expression prior to the nocodazole
block (SMC2OFF). Equal numbers of mitotic cells from the two
different populations were mixed and mitotic chromosomes

isolated (Figure 1E). Classifier IV is the heavy/light ratio
(SMC2ON/SMC2OFF) for each of the proteins identified in this
experiment.
SMC2-depleted chromosomes contained 5.8% of the wild-
type level of SMC2. It is not possible to isolate chromosomes
from cultures completely lacking SMC2 as SMC2 is an essential
gene and dead cells do not accumulate in mitosis. These chro-
mosomes were similarly depleted of all condensin I and II
subunits.
Classifier V: Ska3/Rama1 Dependency
To demonstrate the targeted analysis possible with our
approach, we compared the association of kinetochore proteins
with chromosomes in cells with or without Ska3/Rama1/
C13orf3. Ska3/Rama1, which was identified in this analysis as
a chromosomal protein, was described in several recent publica-
tions (Daum et al., 2009; Gaitanos et al., 2009; Raaijmakers et al.,
2009; Theis et al., 2009; Welburn et al., 2009).
To obtain chromosomes depleted of Ska complex, we isolated
a genetic knockout of the Ska3/Rama1 gene (Figures S5A and
S5B). Ska3/Rama1 / cells (homozygous knockouts are viable)
were grown in SILAClight medium (Figure 1E). Equal numbers of
Ska3/Rama1 / and wild-type DT40 (cultured in SILACheavy
medium) mitotic cells from the two different populations were
mixed and mitotic chromosomes isolated. Classifier V is the
heavy/light ratio (wild-type/Ska3/Rama1 / ) for each protein
identified in this experiment.
Figure 2. Combining Classifiers Increases
Specificity
(A) The rank of all proteins observed for proteomic
classifiers I (abundance), II (enrichment), III (reten-
tion), IV (SMC2 dependency), V (Ska3/Rama1
dependency) and RF (combined random forest
ranking) is plotted as vertical color coded lines
(rug plots). Proteins are ranked from the highest
(left) to lowest values for each classifier. Centro-
mere proteins are indicated by longer bars
(indicated).
(B) Cumulative curves show that combining classi-
fiers by RF significantly increases the specificity
of identifying chromosomal proteins without com-
promising the completeness of the analysis.
(C) 2D scatter graph plotting classifiers I versus II
flanked by the relevant rug plots. Proteins most
enriched in chromosomes relative to cytoplasm
and proteins most abundant in chromosomes are
above and to the right, respectively. Each spot is
color-coded by category (Figure 1D).
(D) Zoom of the region indicated in (C) showing
members of the chromosomal passenger (CPC),
Mis12, Ndc80, Nup107-160 and Cohesin com-
plexes plus the constitutive centromere-associ-
ated network (CCAN) and anaphase-promoting
complex/cyclosome (APC/C).
(E) Enrichment for chromosomal proteins in the
region indicated in (C & D).
See also Figure S2.
Classifier VI: Domain Analysis
We added an additional nonproteomic classifier to our analysis
using the protein domains found in chromosomal and nonchro-
mosomal proteins (red/pink and green wedges in Figure 1D).
This made use of bioinformatic analysis in order to segregate
chromosomal from nonchromosomal proteins, but importantly
did not consider a protein’s relevance to mitosis. We counted
how often each domain was observed in chromosomal and
nonchromosomal proteins and assigned it this frequency as
a score (Table S2). Classifier VI was then determined for each
protein based on the sum of its domain scores.
Multiclassifier Combinatorial Proteomics
Traditional one-dimensional analysis (e.g., sorting the various
proteins according to their value for each classifier) was of
limited utility, as the data lacked a clear boundary between chro-
mosomal (red/pink in Figure 2A) and nonchromosomal proteins
(green in Figure 2A) for each classifier (Figure 2A and Figures
S2A–S2E).
By contrast, when classifiers were combined, our ability
to identify chromosomal proteins was vastly improved. For
example, an enrichment of centromeric or chromosomal pro-
teins relative to nonchromosomal proteins was obtained when
classifiers I (abundance) and II (enrichment) were plotted
(Figures 2C–2E). The clustering of protein complex subunits
in this plot (Figure 2D) reflects both their relative stoichiometry
Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 813
(x axis), and the similar degree to which subunits in a complex are
all present either on or off chromosomes (y axis). Members of the
APC/C, Ndc80 and Mis12 complexes form closely knit clusters.
It is important to note that this was achieved in the context of
entire chromosomes and without requiring solubilization of the
complexes.
We used random forest (RF) analysis, a machine learning
approach, to progress beyond two-dimensional analyses and
integrate the information present in all proteomics classifiers.
This analysis offered two powerful benefits. First, it enabled us
to work with data sets that contain missing values. This is a signif-
icant advantage in proteomics studies where not every protein is
observed in every experiment, as seen in Figure 2A and Fig-
ure S2F. Second, RF analysis allowed us to use any descriptor
of our proteins as a classifier and integrate it into our overall
analysis. Here, we also included a bioinformatic analysis of the
distribution of protein domains in our analysis distinguishing
chromosomal from nonchromosomal proteins (classifier VI).
In brief, RF is a decision tree analysis that separates data sets
into ‘‘true’’ and ‘‘false’’ groups. The decision trees are trained on
defined data sets and randomly built to optimize the separation
between them. Analysis of the experimental data set then occurs
by running each protein through all trees and adding up its overall
RF score (i.e., the fraction of trees that scored it as ‘‘true’’). RFs
perform much better on training data than application data, so
their performance is evaluated by ten-fold cross-validation.
The training data are split into random sections of 90% for
training and 10% for evaluation, so that successively the entire
set is used for evaluation. Here, the two training data sets chosen
were ‘‘nonchromosomal’’ (green wedges in Figure 1D) and
‘‘chromosomal’’ (red + pink wedges in Figure 1D), and the RF
score for a given protein is the fraction of trees that scored it
as ‘‘chromosomal.’’
RF analysis readily discriminated chromosomal from nonchro-
mosomal proteins. In the RF rug plot of Figure 2A, which repre-
sents the ranked list of proteins generated by RF analysis, the
left side is predominantly red, while the right side is predomi-
nantly green. To reach the 500th chromosomal protein on the
RF-ranked list only 229 nonchromosomal proteins are included
(Figure 2B). In contrast, 410-671 nonchromosomal proteins
would be included when considering ranked lists from individual
classifiers. Note that the RF-based sorting was done on the
complete data set, including proteins that failed to be observed
with some classifiers. Therefore, adding information from addi-
tional experiments did not decrease the number of proteins
covered.
The advantage of combining classifiers can be statistically
expressed by ROC curves (Figure S3A), with increased area
under the curve (AUC) for our combined analysis when com-
pared to each of the individual classifiers (AUCRF(cI-V) = 0.81,
AUCscI-V = 0.41-0.76). The combined classifiers assigned 88.8%
of our gold standard, the 125 centromere proteins, correctly as
chromosomal, at the cut-off that minimizes misassignment of
chromosomal versus nonchromosomal proteins. This specificity
was further improved when bioinformatic domain analysis (cVI)
was integrated with our proteomic classifiers (AUCRF(cI-VI) =
0.97; identification of the first 500 chromosomal proteins yields
17 nonchromosomal proteins, Suppl. Figures S3B and S3C,
814 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.
proteins lacking known domains are excluded from this boost,
Figure S3D). Now, 92.4% of the centromere proteins were
assigned as chromosomal. In summary, RF analysis provides
us with a tool for productively combining the outcomes of our
individual proteomics classifier experiments and further empow-
ering our analysis by including data from other sources.
If results of a random forest based on the five proteomic clas-
sifiers plus the bioinformatics-based classifier VI were plotted
against those from the initial random forest analysis (Fig-
ure 3A), a near-perfect separation of the training data was
achieved. Only a single chromosomal protein of the training set
and two nonchromosomal proteins were misassigned when
placing manually a separation line.
Using ten-fold cross-validation, we found that 118 centromere
proteins positioned right and only 7 left of the separation line
(Figure 3B). This compares to 14 and 9 centromere proteins
being missed when using the one-dimensional ranked lists by
classifiers I-V and I-VI, respectively. Accepting the line as a
threshold returns known centromere proteins with a yield of
94.4% and all other chromosomal proteins with 93.1% success.
In contrast, 83.1% of nonchromosomal proteins are rejected.
Thus, the classifier approach is sufficiently powerful to suggest
chromosomal proteins from among hitherto uncharacterized
proteins.
Identification of New Chromosomal Proteins
To test the predictive power of our RF analysis, we cloned and
tested the location in mitosis of 50 previously uncharacterized
proteins including 15 without known domains: 34 predicted to
be and 16 predicted not to be on chromosomes in mitosis.
Reasoning that important proteins would be conserved, we
expressed GFP-tagged human homologs of these chicken
proteins in U2OS cells. Remarkably, 30 of 34 cloned proteins
from the chromosomal region were confirmed as chromosomal,
contrasting with only 2 of the 16 predicted ab initio to be
nonchromosomal. This confirms the power of our analysis and
indicates a success rate of 88%, with 44 of 50 tagged proteins
localizing as predicted. Of 50 newly cloned proteins, 13 were
associated with kinetochores in mitosis, 12 had a more general
distribution on mitotic chromosomes and 7 others were peri-
chromosomal, a class whose new members we propose to
term chromosome periphery proteins (cPERPs A-G) (Figures
S4B–S4G and Table S3). The chromosome periphery (perichro-
mosomal layer) is enriched in ribosomal and nucleolar hitchhiker
proteins, and is of unknown function (Van Hooser et al., 2005).
The new centromere proteins all appeared to localize to the
outer kinetochore, relative to CENP-C and HEC1 as standards
(Figures 4B–4J, Figure S4A). In keeping with established nomen-
clature, we propose to name these proteins CENP-Y, CENP-Z
and CENP-27 through CENP-33. Beyond ‘Z’, the 26th letter of
the basic modern Latin alphabet, we propose to designate the
new proteins with numbers starting with CENP-27.
Functional Analysis of New Kinetochore-Associated
Proteins
We focused our initial functional analysis on kinetochore
proteins. Clustering analysis (Gentleman et al., 2004) allowed
us to combine data for proteins identified by all classifiers, and

Figure 3. Random Forest Analysis Predicts New Proteins of Interest
(A) Separation of the two training sets (color-coded groups from Figure 1D) by
RF analysis. The line provides optimal separation of the two training sets.
x axis: RF analysis of classifiers I–V; y axis: RF analysis of classifiers I–VI,
including the bioinformatic domain analysis.
(B) Positions of previously known centromere proteins from ten-fold cross-
validation.
(C) Position of cloned proteins and remaining uncloned proteins in the same 2d
analysis.
(D) Classification of newly cloned proteins from regions to the left and right of
the dividing line, respectively.
See also Figure S3.
look for informative groupings. This revealed a striking tendency
for functionally related proteins to form clusters, as exemplified
by members of the NDC80, CPC, Nup107-160 and APC/C
complexes (Figure 4). Interestingly, our clustering sorted
CENP-27 as a component of the APC/C. This was confirmed
and the protein named APC16 in three recent reports (Hutchins
et al., 2010; Kops et al., 2010; Hubner et al., 2010). To further
test the predictive value of cluster analysis for proteomic data
sets, we examined two kinetochore proteins in greater detail.
Ska3/Rama1 and Functional Analysis of Kinetochore
Subcomplexes
C13orf3 was located adjacent to Ska2 in the cluster analysis.
This protein now known as Ska3/Rama1 has been suggested
to be involved in microtubule attachment to kinetochores (Gaita-
nos et al., 2009; Raaijmakers et al., 2009; Welburn et al., 2009) or
coordination of the spindle checkpoint response (Daum et al.,
2009; Theis et al., 2009). We analyzed the kinetochore proteome
in the presence or absence of Ska3/Rama1 (defined as classifier
V) in order to determine the role of this protein in kinetochore
structure (Figure 5).
A map of the Ska3/Rama1 locus in DT40 is shown in Figures
S5A and S5B, together with a targeting strategy for inactivating
the gene. The Ska3/Rama1 gene is not essential for life in DT40
cells (Figure S5C). However, these cells struggle to achieve
a normal chromosome alignment, and show a 33 increase in
mitotic index (Figure S5D and S5E), a 33 increase in the
percentage of apoptotic cells and a 6x increase in the number
of bi-nucleated cells.
Proteomic analysis of isolated chromosomes revealed that
loss of Ska3/Rama1 was accompanied by the loss of Ska1 and
Ska2. Loss of the Ska complex caused no systematic changes
in the chromosomal association of proteins of the constitutive
centromere-associated network (CCAN), Knl-1/Mis12/Ndc80
(KMN), Mis18, Ndc80, CPC, and Nup107-160 kinetochore sub-
complexes. However, striking changes were seen in the levels
of the APC/C, RanBP2/RanGAP1, spindle checkpoint, Rod/
Zw10/Zwilch (RZZ), and dynein/dynactin complexes. We con-
firmed that the RanBP2/RanGAP1 complex is indeed depleted
from kinetochores when Ska3/Rama1 is deleted in HeLa cells
(Figure S5F and S5G). Attempts to confirm the specific kineto-
chore depletion of the APC/C were uninformative, as we were
unable to reproducibly obtain kinetochore staining for the
APC/C in HeLa cells using four independent antibodies.
We conclude that combining genetic and SILAC analysis
provides a powerful new method for analysis of multicomplex
protein superstructures.
A Protein Involved in Chromosome Alignment
and Spindle Organization
A second new kinetochore-associated protein, CENP-32/
C9orf114, was sorted on our kinetochore cluster diagram next
to CLASP1 and CLASP2, two paralogues known to be involved
in the regulation of microtubule dynamics. Like CLASPs,
CENP-32/C9orf114 mapped to the outer kinetochore (Figure 6A),
and its depletion caused a significant accumulation of cells in
later prometaphase (Figure S6A) with misaligned chromosomes
(Figure 6B). These cells frequently had bipolar spindles, however
60% of those spindles exhibited remarkable abnormalities
where centrosomes appeared to have detached from the poles
(Figures 6C and 6D and Figure S6B). In one remarkable case,
the centrosomes appeared at the midzone of a bipolar spindle
(Figure 6C9-12).
Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 815
Figure 4. Cluster Analysis and Imaging of New Kinetochore Proteins
(A) Heat map and cluster dendrogram for 101 centromere-associated proteins identified by all 5 proteomic classifiers. Known complexes are color-marked (see
bottom) and show a tendency to cluster in this analysis. Kinetochore proteins identified in this study are also marked (stars).
(B–J) Localization of GFP-tagged kinetochore-associated proteins (panels 2) relative to DNA detected with DAPI (panels 1) and CENP-C (panels 3 in [B]–[D], [F]–
[H], and [J] or ACA (panels 3 in [E] and [I]). Insets show merged images with blow-ups of representative kinetochores (GFP-novelCENP, green and CENP-C or
ACA, red).
See also Figure S4.

DISCUSSION complex proteome. The key approach of combining classifier

data can in principle be expanded indefinitely and can include
Multiclassifier Combinatorial Proteomics nonproteomics data sets such as the bioinformatic protein
The approach described here for analysis of the proteome of domain analysis used here. Other classifiers that could
vertebrate mitotic chromosomes can be used to study any be used in the future include microarray, protein interaction

816 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.


Figure 5. Ska3/Rama1 Dependency for Chromosomal Association
and Analysis of CENP-27/APC16
Kinetochore proteins with increased abundance on chromosomes in the
absence of Ska3/Rama1 are shown with blue bars and proteins with
decreased abundance are shown with red bars. Note covariance of protein
complexes. See also Figure S5.
(e.g., two hybrid screens or pull-down), protein phosphoryla-
tion, and localization data and, indeed, data from any experi-
mental approach in which the proteins of interest are sorted
systematically.
We first showed that plotting pairs of classifiers against one
another improved our ability to delineate potential chromosomal
proteins. As that approach could not be generalized when the
number of classifiers exceeded three, we adopted a random
forest (RF) analysis approach. This allowed us to integrate infor-
mation from all classifiers into decision trees on which known
and unknown proteins could be classified. Importantly, RF anal-
ysis handles missing values systematically. This is crucial when
not every protein is observed in every experiment. In contrast,
cluster analysis, which has been used both in this study and in
other recent work (Theis et al., 2009; Neumann et al., 2010),
can only integrate data for proteins that have a value for every
classifier.
Integrity of the Isolated Chromosomes
Our methods focused on optimizing the purity of the chromo-
somes. Thus, our list of proteins is likely to represent the minimal,
stably associated components of mitotic chromosomes. None-
theless kinetochores of isolated chromosomes retain some
function, as judged by their ability to recruit components of
the mitotic checkpoint complex from cytoplasm. This may be
because chromosomes were isolated from nocodazole-treated
cells, with kinetochores actively engaged in spindle checkpoint
signaling.
New Insights into Kinetochore Functional Organization
Remarkably, although no biochemical enrichment for centro-
meres was performed, our data set contained all known centro-
meric subcomplexes, with peptides from 125 reported cen-
tromere proteins (eight present as multiple isoforms) (Table S1).
We identified all members of the CCAN, KMN, Mis12 and Mis18
complexes and all members of the RZZ complex except Zwint
(which is not yet annotated in the chicken genome). Our success
in identifying centromere and telomere proteins may be
explained because 66 of the 78 chicken DT40 chromosomes
are microchromosomes whose purification provides a natural
enrichment for centromeres and telomeres since the chromo-
some arms are so short.
We combined genetics with whole proteome analysis in order
to identify complexes and structural dependencies in their
‘‘native environment’’ (e.g., kinetochore proteins in actual kinet-
ochores). Chromosomes lacking Ska3/Rama1 were depleted for
the entire Ska complex, confirming that these three proteins are
interdependent for chromosome binding. Similarly, depletion of
key condensin subunit SMC2 caused a loss of all seven subunits
of the condensin I and II complexes from chromosomes. Impor-
tantly, this analysis did not require tagging of any proteins or
attempts to solubilize functional complexes from large subcel-
lular structures.
In addition to these primary effects of depletion, loss of Ska3/
Rama1 also caused a significant secondary depletion of the
APC/C and RanBP2/RanGAP1 complexes from chromosomes
but had no consistent effect on most other kinetochore proteins.
Importantly, all members of the secondary-depleted complexes
Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 817
Figure 6. Initial Characterization of CENP-32
(A) Localization of GFP-CENP-32 (panel 2, green) in the kinetochore relative to DNA (panel 1, blue) and CENP-C (panel 3, red).
(B) CENP-32 RNAi causes problems with chromosome alignment. DNA, red; CENP-C, green; microtubules, white.
(C) CENP-32 RNAi causes detachment of centrosomes from bipolar spindles. DNA (panels 1,5,9,13), blue; pericentrin (panels 2,6) or g-tubulin (panels 10,14),
green; a-tubulin (panels 3,7,11,15), red.
(D) Quantitation of spindle phenotypes following CENP-32 RNAi.
(E) Schematic representation of domain architectures for new kinetochore proteins tagged in this study (drawn to approx. scale). The locations of domains are
according to Pfam and SMART family databases (Letunic et al., 2004; Finn et al., 2010), complemented by REP web server analysis (Andrade et al., 2000).
The localization of the PEHE domain has been assigned from (Marı́n, 2003). Phyletic distributions of proteins are indicated in blue, yellow, green and violet
for Metazoa, Fungi, Plantae and Archaea, respectively. CENP-35 has been truncated. Abbreviations: BCNT, Bucentaur or craniofacial development protein;
Fbox, cyclin-F motif; G-Patch, Glycine-rich nucleic binding domain; JmjC, jumonji C domain; LRR, Leucine-rich repeats; Nua4, Nucleosomal acetyltransferase
of H4; OB, Oligonucleotide/oligosaccharide Binding; PEHE, Pro-Glu-His-Glu conserved region; PHD, Plant HomeoDomain; RA, Ras associated (RalGDS/AF-6)
domain; SPOUT, spoU and trmD RNA methyltransferase; WD40, WD/beta-transducin repeats; Zf-AD, Zinc finger-associated domain; and Zf-C2H2, Classical
zinc finger Cys(2)His(2).
See also Figure S6.

818 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.

also behaved coordinately. Our results (e.g., the behavior of as specific for mitosis. Furthermore, their association might
CENP-27/APC16 as a component of the APC/C) confirm the depend on a fully assembled kinetochore and thus be lost
utility of single protein depletion analysis for the identification when attempting other than whole chromosome analysis.
of protein complexes and determination of their mutual interde-
pendencies for association with chromosomes. Characterization of a Kinetochore Protein
Our data suggest that the Ska complex may provide a docking CENP-32 is required both for chromosome alignment and for
site for the APC/C in the outer kinetochore. Alternatively, sumoy- association of the centrosomes with the poles of the bipolar
lation by RanBP2 may have a role in APC/C binding to chromo- spindle during metaphase. This latter phenotype is very similar
somes. The RanBP2-RanGAP1 complex is known to be involved to an unusual spindle morphology phenotype seen in Drosophila
in kinetochore-microtubule interactions and localization of cells following depletion of the CLASP homolog Mast/Orbit
several spindle checkpoint proteins (Joseph et al., 2004). We (Maiato et al., 2002). Indeed, in our analysis, CENP-32 clusters
note that among the recent spate of publications on Ska3/ with CLASP1 and CLASP2. A yeast homolog of CENP-32 inter-
Rama1, our observations appear to support a role in integration acts with CBF5, an enzyme involved in the posttranscriptional
and regulation of the spindle checkpoint response (Daum et al., modification of rRNA, that was shown to bind to budding yeast
2009; Theis et al., 2009). centromeres and microtubules (Jiang et al., 1993). Bioinformatics
Our whole-proteome analysis revealed that Cdc20 behaved analysis suggests that CENP-32 is a member of the SPOUT family
like a member of the APC/C and was distinct from other compo- of methyltransferases but is atypical in possessing a possible
nents of the spindle checkpoint pathway with respect to its Ska RNA-binding OB fold inserted into its catalytic domain (Tkaczuk
complex dependency. Spindle checkpoint components asso- et al., 2007). It is tempting to speculate that CENP-32 may func-
ciate with one another in cytoplasm as a mitotic checkpoint tion at kinetochores by interacting with an as-yet unknown RNA.
complex (MCC), containing BubR1, Bub3, Cdc20 and Mad2.
Our data suggest that once the MCC associates with chromo-
How Complex Is the Kinetochore?
somes, Cdc20 stably associates with the APC/C.
MCCP analysis allows us to predict how many more kinetochore
proteins remain to be identified in our data set. In the plot of
What Classes of Kinetochore Proteins Remain
Figure 3, where the chromosomal proteome is displayed in two
to Be Discovered?
dimensions, we found 35% of novel tagged proteins from region
We identified 13 kinetochore-associated proteins among previ-
R and 6% from region L to associate with the kinetochore during
ously uncharacterized proteins and, as discussed below, we
mitosis. These regions have 224 and 287 as-yet uncharacterized
predict that many more remain to be described. We therefore
novel proteins, respectively. Assuming no bias in the proteins we
asked whether there is any functional relationship between these
cloned, this suggests that approximately 97 more kinetochore
new proteins. That is, what sorts of kinetochore proteins had
proteins remain to be discovered. Taking into account the 13
been missed in the many previous genetic and biochemical
kinetochore-associated proteins confirmed in our work, this
screens? An interesting answer has emerged.
roughly doubles the currently known protein complexity of the
Since the new kinetochore proteins were identified solely
kinetochore during mitosis, confirming it as one of the most
based on their occurrence in chromosomes, they could poten-
complex cellular substructures.
tially represent a wide range of functions. Nevertheless, it is
striking that five of the new centromere proteins (namely,
Conclusions
CENP-28, 29, 31, 35, and 36) are subunits of complexes
Multiclassifier combinatorial proteomics and the data sets
that modify and/or bind histones (Figure 6E). Yeast orthologs
described here open the door to the identification of all functional
of two of these proteins (namely, CENP-28/C1orf149 and
components of mitotic chromosomes despite the adventitious
CENP-29/CFDP1) contribute to NuA4 histone acetyltransferase
binding of cellular background proteins during mitosis. Further-
(HAT) and SWR1 ATP-dependent chromatin remodelling com-
more, MCCP can be extended by adding additional classifiers
plexes, respectively. These complexes are known to share com-
to delineate protein complexes and define functional dependen-
ponents (Wu et al., 2005) and together stimulate the exchange of
cies between them in the context of intact mitotic chromosomes.
histone H2A for H2A.Z, following acetylation of H2A or H4 (Altaf
This will serve both as a starting point for systematic determina-
et al., 2010). A third centromere-associated protein, CENP-36/
tion of the full range of functions involved in mitotic chromosome
MSL1v1, is necessary for the activity of MOF, another HAT, on
segregation, and as a basis for the development of detailed
nucleosomal H4 (Li et al., 2009). Finally, CENP-31/PHD6 and
structural and functional interaction maps of key chromosomal
CENP-35/PHF2 each contain PHD (plant homeodomain) zinc
subdomains. MCCP should also prove useful for the analysis
fingers, which are usually associated with chromatin-mediated
of other cellular structures that lack defined boundaries, e.g.,
transcriptional regulation. The PHD of CENP-35, which also
membrane associated complexes like the post-synaptic density.
contains a JmjC (likely histone demethylase) domain, appears
to be required for demethylation of H3 at the promoters of ribo-
somal RNA genes (Wen et al., 2010). EXPERIMENTAL PROCEDURES

Why were these proteins not discovered earlier as kineto-

Preparation of Mitotic Chromosomes
chore-associated? One likely explanation is that they may have DT40 cells were incubated with Nocodazole for 13 hr, resulting in a mitotic
essential functions in other chromatin regions as well. Thus, index of 70%–90%. Mitotic chromosomes were isolated in the polyamine-
mutations might have pleiotropic phenotypes not recognized EDTA buffer system optimized for chicken DT40 cells (Lewis and Laemmli,

Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 819

1982). 19.3 OD260 units were obtained from pooling the material of four inde-
pendent preparations totaling 7.5 3 109 DT40 cells and solubilized in SDS-
polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer.
Preparation of Chromosome-Free Mitotic Cell Extracts
Nocodazole blocked DT40 cells were dounce-homogenized under hypotonic
conditions. Mitotic chromosomes were removed by centrifuging the superna-
tant twice at 10,000 x g and discarding the pellets to prepare a cell extract free
of chromosomes.
To measure the ratios between chromosomal and nonchromosomal
proteins, SILAC based mass spectrometry was performed with 150 mg of
labeled cell extract from 7.0 3 106 cells and 150 mg of nonlabeled proteins con-
tained in isolated chromosomes from 2.0 3 109 cells.
To measure the exchange ratio, we isolated mitotic chromosomes from
roughly 1.0 3 109 cells. Mitotic chromosomes were pelleted after centrifuging
at 3000 3 g and mixed into 10 ml cell extract that were made from 3.0 3 108
cells. This mixture was incubated at 14 C for 30 min. Finally, we re-isolated the
chromosomes as described above.
Mass Spectrometric Analysis
Proteins were separated into a high and a low molecular weight fraction by
SDS-PAGE, in-gel digested using trypsin (Shevchenko et al., 2006), and frac-
tionated into 30 fractions each using SCX. The individual SCX fractions were
desalted using StageTips (Rappsilber et al., 2003) and analyzed using
LC-MS on a LTQ-Orbitrap (Thermo Fisher Scientific) coupled to HPLC via
a nanoelectrospray ion source. The six most intense ions of a full MS acquired
in the orbitrap analyzer were fragmented and analyzed in the linear ion trap.
The MS data were analyzed using MaxQuant (Cox and Mann, 2008) and
proteins identified by searching MS and MS/MS data using the MASCOT
search engine (Matrix Science, UK). For more details, see the Extended Exper-
imental Procedures.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, six
figures, and five tables and can be found with this article online at doi:10.
1016/j.cell.2010.07.047.
ACKNOWLEDGMENTS
We dedicate this paper to Uli Laemmli on the occasion of his 70th birthday. We
thank Mayumi Takahashi for assistance with preparation of the Ska3/Rama1
knockout, Frauke Melchior for anti-RanBP2 and David Tollervey, Margarete
Heck, Robin Allshire, Iain Cheeseman, Kumiko Samejima, Susana Ribeiro
and Jan Bergmann for critical reading of the manuscript. This work was sup-
ported by a European Community Marie Curie Excellence Grant (JR), the
MRC (CPP), EMBO (LSP) and the Wellcome Trust (WCE, JR). WCE and
JR are Principal and Senior Research Fellows of The Wellcome Trust,
respectively.
Received: December 7, 2009
Revised: May 20, 2010
Accepted: July 14, 2010
Published: September 2, 2010
REFERENCES
Altaf, M., Auger, A., Monnet-Saksouk, J., Brodeur, J., Piquet, S., Cramet, M.,
Bouchard, N., Lacoste, N., Utley, R.T., Gaudreau, L., et al. (2010). NuA4-
dependent acetylation of nucleosomal histone H4 and H2A directly stimulates
incorporation of H2A.Z by the SWR1 complex. J. Biol. Chem., in press.
Amano, M., Suzuki, A., Hori, T., Backer, C., Okawa, K., Cheeseman, I.M., and
Fukagawa, T. (2009). The CENP-S complex is essential for the stable assembly
of outer kinetochore structure. J. Cell Biol. 186, 173–182.
820 Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc.
Andersen, J.S., Wilkinson, C.J., Mayor, T., Mortensen, P., Nigg, E.A., and
Mann, M. (2003). Proteomic characterization of the human centrosome by
protein correlation profiling. Nature 426, 570–574.
Andrade, M.A., Ponting, C.P., Gibson, T.J., and Bork, P. (2000). Homology-
based method for identification of protein repeats using statistical significance
estimates. J. Mol. Biol. 298, 521–537.
Cheeseman, I.M., and Desai, A. (2008). Molecular architecture of the kineto-
chore-microtubule interface. Nat. Rev. Mol. Cell Biol. 9, 33–46.
Cox, J., and Mann, M. (2008). MaxQuant enables high peptide identification
rates, individualized p.p.b.-range mass accuracies and proteome-wide
protein quantification. Nat. Biotechnol. 26, 1367–1372.
Daum, J.R., Wren, J.D., Daniel, J.J., Sivakumar, S., McAvoy, J.N., Potapova,
T.A., and Gorbsky, G.J. (2009). Ska3 is required for spindle checkpoint
silencing and the maintenance of chromosome cohesion in mitosis. Curr.
Biol. 19, 1467–1472.
de Godoy, L.M., Olsen, J.V., Cox, J., Nielsen, M.L., Hubner, N.C., Frohlich, F.,
Walther, T.C., and Mann, M. (2008). Comprehensive mass-spectrometry-
based proteome quantification of haploid versus diploid yeast. Nature 455,
1251–1254.
Dejardin, J., and Kingston, R.E. (2009). Purification of proteins associated with
specific genomic Loci. Cell 136, 175–186.
Earnshaw, W.C., and Rothfield, N. (1985). Identification of a family of human
centromere proteins using autoimmune sera from patients with scleroderma.
Chromosoma (Berl) 91, 313–321.
Finn, R.D., Mistry, J., Tate, J., Coggill, P., Heger, A., Pollington, J.E., Gavin,
O.L., Gunasekaran, P., Ceric, G., Forslund, K., et al. (2010). The Pfam protein
families database. Nucleic Acids Res. 38, D211–D222.
Foltz, D.R., Jansen, L.E., Black, B.E., Bailey, A.O., Yates, J.R., and Cleveland,
D.W. (2006). The human CENP-A centromeric nucleosome-associated
complex. Nat. Cell Biol. 8, 458–469.
Foster, L.J., De Hoog, C.L., and Mann, M. (2003). Unbiased quantitative pro-
teomics of lipid rafts reveals high specificity for signaling factors. Proc. Natl.
Acad. Sci. USA 100, 5813–5818.
Gaitanos, T.N., Santamaria, A., Jeyaprakash, A.A., Wang, B., Conti, E., and
Nigg, E.A. (2009). Stable kinetochore-microtubule interactions depend on
the Ska complex and its new component Ska3/C13Orf3. EMBO J. 28, 1442–
1452.
Gassmann, R., Henzing, A.J., and Earnshaw, W.C. (2005). Novel components
of human mitotic chromosomes identified by proteomic analysis of the chro-
mosome scaffold fraction. Chromosoma 113, 385–397.
Gentleman, R.C., Carey, V.J., Bates, D.M., Bolstad, B., Dettling, M., Dudoit, S.,
Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al. (2004). Bioconductor: open soft-
ware development for computational biology and bioinformatics. Genome
Biol. 5, R80.
Hori, T., Amano, M., Suzuki, A., Backer, C.B., Welburn, J.P., Dong, Y.,
McEwen, B.F., Shang, W.H., Suzuki, E., Okawa, K., et al. (2008). CCAN makes
multiple contacts with centromeric DNA to provide distinct pathways to the
outer kinetochore. Cell 135, 1039–1052.
Hubner, N.C., Bird, A.W., Cox, J., Splettstoesser, B., Bandilla, P., Poser, I.,
Hyman, A., and Mann, M. (2010). Quantitative proteomics combined with
BAC TransgeneOmics reveals in vivo protein interactions. J. Cell Biol. 189,
739–754.
Hudson, D.F., Vagnarelli, P., Gassmann, R., and Earnshaw, W.C. (2003).
Condensin is required for nonhistone protein assembly and structural integrity
of vertebrate mitotic chromosomes. Dev. Cell 5, 323–336.
Hutchins, J.R., Toyoda, Y., Hegemann, B., Poser, I., Heriche, J.K., Sykora,
M.M., Augsburg, M., Hudecz, O., Buschhorn, B.A., Bulkescher, J., et al.
(2010). Systematic analysis of human protein complexes identifies chromo-
some segregation proteins. Science 328, 593–599.
Ishihama, Y., Oda, Y., Tabata, T., Sato, T., Nagasu, T., Rappsilber, J., and
Mann, M. (2005). Exponentially modified protein abundance index (emPAI)
for estimation of absolute protein amount in proteomics by the number of
sequenced peptides per protein. Mol. Cell. Proteomics 4, 1265–1272.
Ishihama, Y., Schmidt, T., Rappsilber, J., Mann, M., Hartl, F.U., Kerner, M.J.,
and Frishman, D. (2008). Protein abundance profiling of the Escherichia coli
cytosol. BMC Genomics 9, 102.
Jiang, W., Middleton, K., Yoon, H.J., Fouquet, C., and Carbon, J. (1993). An
essential yeast protein, CBF5p, binds in vitro to centromeres and microtu-
bules. Mol. Cell. Biol. 13, 4884–4893.
Joseph, J., Liu, S.T., Jablonski, S.A., Yen, T.J., and Dasso, M. (2004). The Ran-
GAP1-RanBP2 complex is essential for microtubule-kinetochore interactions
in vivo. Curr. Biol. 14, 611–617.
Kops, G.J., van der Voet, M., Manak, M.S., van Osch, M.H., Naini, S.M., Brear,
A., McLeod, I.X., Hentschel, D.M., Yates, J.R., 3rd, van den Heuvel, S., and
Shah, J.V. (2010). APC16 is a conserved subunit of the anaphase-promoting
complex/cyclosome. J. Cell Sci. 123, 1623–1633.
Kulukian, A., Han, J.S., and Cleveland, D.W. (2009). Unattached kinetochores
catalyze production of an anaphase inhibitor that requires a Mad2 template to
prime Cdc20 for BubR1 binding. Dev. Cell 16, 105–117.
Letunic, I., Copley, R.R., Schmidt, S., Ciccarelli, F.D., Doerks, T., Schultz, J.,
Ponting, C.P., and Bork, P. (2004). SMART 4.0: towards genomic data integra-
tion. Nucleic Acids Res. 32, D142–D144.
Lewis, C.D., and Laemmli, U.K. (1982). Higher order metaphase chromosome
structure: evidence for metalloprotein interactions. Cell 29, 171–181.
Li, X., Wu, L., Corsa, C.A., Kunkel, S., and Dou, Y. (2009). Two mammalian
MOF complexes regulate transcription activation by distinct mechanisms.
Mol. Cell 36, 290–301.
Maiato, H., Sampaio, P., Lemos, C.L., Findlay, J., Carmena, M., Earnshaw,
W.C., and Sunkel, C.E. (2002). MAST/Orbit has a role in microtubule-kineto-
chore attachment and is essential for chromosome alignment and mainte-
nance of spindle bipolarity. J. Cell Biol. 157, 749–760.
Marı́n, I. (2003). Evolution of chromatin-remodeling complexes: comparative
genomics reveals the ancient origin of ‘‘novel’’ compensasome genes.
J. Mol. Evol. 56, 527–539.
Morrison, C., Henzing, A.J., Jensen, O.N., Osheroff, N., Dodson, H., Kandels-
Lewis, S.E., Adams, R.R., and Earnshaw, W.C. (2002). Proteomic analysis of
human metaphase chromosomes reveals topoisomerase II alpha as an Aurora
B substrate. Nucleic Acids Res. 30, 5318–5327.
Neumann, B., Walter, T., Heriche, J.K., Bulkescher, J., Erfle, H., Conrad, C.,
Rogers, P., Poser, I., Held, M., Liebel, U., et al. (2010). Phenotypic profiling
of the human genome by time-lapse microscopy reveals cell division genes.
Nature 464, 721–727.
Obuse, C., Iwasaki, O., Kiyomitsu, T., Goshima, G., Toyoda, Y., and Yanagida,
M. (2004). A conserved Mis12 centromere complex is linked to heterochro-
matic HP1 and outer kinetochore protein Zwint-1. Nat. Cell Biol. 6, 1135–1141.
Okada, M., Cheeseman, I.M., Hori, T., Okawa, K., McLeod, I.X., Yates, J.R.,
Desai, A., and Fukagawa, T. (2006). The CENP-H-I complex is required for
the efficient incorporation of newly synthesized CENP-A into centromeres.
Nat. Cell Biol. 8, 446–457.
Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey,
A., and Mann, M. (2002). Stable isotope labeling by amino acids in cell culture,
SILAC, as a simple and accurate approach to expression proteomics. Mol.
Cell. Proteomics 1, 376–386.
Porter, I.M., McClelland, S.E., Khoudoli, G.A., Hunter, C.J., Andersen, J.S.,
McAinsh, A.D., Blow, J.J., and Swedlow, J.R. (2007). Bod1, a novel kineto-
chore protein required for chromosome biorientation. J. Cell Biol. 179,
187–197.
Raaijmakers, J.A., Tanenbaum, M.E., Maia, A.F., and Medema, R.H. (2009).
RAMA1 is a novel kinetochore protein involved in kinetochore-microtubule
attachment. J. Cell Sci. 122, 2436–2445.
Rappsilber, J., Ishihama, Y., and Mann, M. (2003). Stop and go extraction tips
for matrix-assisted laser desorption/ionization, nanoelectrospray, and LC/MS
sample pretreatment in proteomics. Anal. Chem. 75, 663–670.
Rappsilber, J., Ryder, U., Lamond, A.I., and Mann, M. (2002). Large-scale pro-
teomic analysis of the human spliceosome. Genome Res. 12, 1231–1245.
Schirmer, E.C., Florens, L., Guan, T., Yates, J.R., 3rd, and Gerace, L. (2003).
Nuclear membrane proteins with potential disease links found by subtractive
proteomics. Science 301, 1380–1382.
Shevchenko, A., Tomas, H., Havlis, J., Olsen, J.V., and Mann, M. (2006). In-gel
digestion for mass spectrometric characterization of proteins and proteomes.
Nat. Protoc. 1, 2856–2860.
Takata, H., Uchiyama, S., Nakamura, N., Nakashima, S., Kobayashi, S., Sone,
T., Kimura, S., Lahmers, S., Granzier, H., Labeit, S., et al. (2007). A comparative
proteome analysis of human metaphase chromosomes isolated from two
different cell lines reveals a set of conserved chromosome-associated
proteins. Genes Cells 12, 269–284.
Theis, M., Slabicki, M., Junqueira, M., Paszkowski-Rogacz, M., Sontheimer,
J., Kittler, R., Heninger, A.K., Glatter, T., Kruusmaa, K., Poser, I., et al.
(2009). Comparative profiling identifies C13orf3 as a component of the Ska
complex required for mammalian cell division. EMBO J. 28, 1453–1465.
Tkaczuk, K.L., Dunin-Horkawicz, S., Purta, E., and Bujnicki, J.M. (2007). Struc-
tural and evolutionary bioinformatics of the SPOUT superfamily of methyltrans-
ferases. BMC Bioinformatics 8, 73.
Uchiyama, S., Kobayashi, S., Takata, H., Ishihara, T., Hori, N., Higashi, T.,
Hayashihara, K., Sone, T., Higo, D., Nirasawa, T., et al. (2005). Proteome anal-
ysis of human metaphase chromosomes. J. Biol. Chem. 280, 16994–17004.
Van Hooser, A.A., Yuh, P., and Heald, R. (2005). The perichromosomal layer.
Chromosoma 114, 377–388.
Welburn, J.P., Grishchuk, E.L., Backer, C.B., Wilson-Kubalek, E.M., Yates,
J.R., 3rd, and Cheeseman, I.M. (2009). The human kinetochore Ska1 complex
facilitates microtubule depolymerization-coupled motility. Dev. Cell 16,
374–385.
Wen, H., Li, J., Song, T., Lu, M., Kan, P.Y., Lee, M.G., Sha, B., and Shi, X.
(2010). Recognition of histone H3K4 trimethylation by the plant homeodomain
of PHF2 modulates histone demethylation. J. Biol. Chem. 285, 9322–9326.
Wu, W.H., Alami, S., Luk, E., Wu, C.H., Sen, S., Mizuguchi, G., Wei, D., and Wu,
C. (2005). Swc2 is a widely conserved H2AZ-binding module essential for
ATP-dependent histone exchange. Nat. Struct. Mol. Biol. 12, 1064–1071.
Cell 142, 810–821, September 3, 2010 ª2010 Elsevier Inc. 821
Scientific Editor Position Available
Cell is seeking a scientist to join its editorial team. The minimum qualification for this position is
a PhD in a relevant area of biomedical research, although additional post-doctoral or editorial
experience is preferred. This is a superb opportunity for a talented individual to play a critical role
in promoting science by helping researchers disseminate their findings to the wider community.

As an editor, you would be responsible for assessing submitted research papers and overseeing
the refereeing process, and you would commission and edit material for Cell’s Leading Edge.
You would also travel frequently to scientific conferences to follow developments in research and
to establish and maintain close ties with the scientific community. You would have opportunities
to pioneer and contribute to new trends in scientific publishing. The key qualities we look for
are breadth of scientific interest and the ability to think critically about a wide range of scientific
issues. The successful candidate will be highly motivated and creative and able to work inde-
pendently as well as in a team.

This is a full-time, in-house editorial position, based at the Cell Press office in Kendall Square
near MIT in Cambridge, Massachusetts. Cell Press offers an attractive salary and benefits pack-
age and a stimulating working environment that encourages innovation.

To apply, please send your CV and a letter describing your qualifications, general research inter-
ests, and motivation for pursuing a career in scientific publishing as soon as possible to our online
jobs site: http://reedelsevier.taleo.net/careersection/51/jobdetail.ftl?lang=en&job=CEL0000R.

No phone inquiries, please. Cell Press in an equal opportunity/affirmative action employer, m/f/d/v.
The American Journal of Human Genetics
Editor Position Available

The American Society of Human Genetics is seeking an Editor for The American Journal of Human
Genetics. The Editor leads one of the world’s oldest and most prestigious journals publishing pri-
mary human genetics research.

Among the Editor’s responsibilities are determining the scope and direction of the scientific con-
tent of The Journal, overseeing manuscripts submitted for review and their publication, selecting
and supervising a staff consisting of an Editorial Assistant and doctoral-level Deputy Editor, direct-
ing interactions with the publisher (currently Cell Press), reviewing quarterly reports provided by the
publisher, evaluating the performance of the publisher, and if required, supervising the process of
the selection a new publisher. The Editor serves as a member of the Board of Directors of the Ameri-
can Society of Human Genetics (ASHG), as well as the ASHG Finance Committee, and presents
semiannual reports to the Board. All Associate Editors of The Journal are appointed by the Editor,
who also determines their duties. At the ASHG annual meeting, the Editor presides over a meeting
of the Associate Editors and presents an annual report to the ASHG membership.

The term of the appointment is five years and includes a yearly stipend. The new Editor will be
selected by the end of 2010 and will begin receiving manuscripts approximately in September 2011;
there will be partial overlap with the Boston office. Applicants should be accomplished scientists
in the field of human genetics and should have a broad knowledge and appreciation of the field.
Nominations, as well as applications consisting of a letter of interest and curriculum vitae, should be
sent to:

AJHG Editorial Search Committee

American Society of Human Genetics
9650 Rockville Pike
Bethesda, MD 20814
Cell Host & Microbe
Scientific Editor Position Available

Cell Host & Microbe, the premier microbiology primary research journal, is seeking a full-time sci-
entific editor to join its editorial team. As a scientific editor, you would be responsible for assessing
submitted research papers, overseeing the refereeing process, and choosing and commissioning
review material. You would also travel frequently to scientific conferences to follow developments
in research and establish and maintain close ties with the scientific community. This is a superb
opportunity for a talented individual to play a critical role in the research community away from the
bench.

This is a full-time in-house editorial position, based at the Cell Press office in Cambridge, Mas-
sachusetts. Cell Press offers an attractive salary and benefits package and a stimulating working
environment that encourages innovation.

The minimum qualification for this position is a PhD in a relevant area of biomedical research,
although additional postdoctoral or editorial experience is preferred. We will consider qualified can-
didates with scientific expertise in any area that falls within the scope of the journal, including molec-
ular and cell biology of microbes, host immune response to microbes, immune evasion by microbes,
microbiology, therapeutics, and vaccine design. The key qualities we look for are breadth of scien-
tific interest and the ability to think critically about a wide range of scientific issues. The successful
candidate will also be highly motivated and creative and able to work independently as well as in a
team.

Applications will be accepted through August 31, 2010. To apply, please visit the Careers page at
http://www.elsevier.com and search on keywords “Cell Host & Microbe.” Please submit a resume
and cover letter describing your qualifications, general research interests, and motivation for pursu-
ing a career in scientific publishing. Applications will be held in the strictest confidence.
Cell Press Business Project Editor
Position Available

Cell Press is seeking a Business Project Editor to plan, develop, and implement projects that have
commercial or sponsorship potential. By drawing on existing content or developing new material, the
Editor will work with Cell Press’s commercial sales group to create collections of content in print or
online that will be attractive to readers and sponsors. The Editor will also be responsible for leverag-
ing new online opportunities for engaging the readers of Cell Press journals.

The successful candidate will have a PhD in the biological sciences, broad scientific interests, a
fascination with technology, good commercial instincts, and a true passion for both science and
science communication. They should be highly organized and dedicated, with excellent written and
oral communication skills, and should be willing to work to tight deadlines.

The position is full time and based in Cambridge, MA. Cell Press offers an attractive salary and
benefits package and a stimulating work environment. Applications will be considered on a rolling
basis. For consideration, please apply online and include a cover letter and resume. To apply, visit
the career page at http://www.elsevier.com and search on keywords “Business Project Editor.”
Positions Available
 Positions Available

TENURE TRACK FACULTY POSITION in NEUROSCIENCE

The Dept of Neurobiology & Anatomy at the University of Utah (http://www.neuro.utah.edu/) is seeking an outstanding scientist for a tenure track
faculty position at the Assistant Professor level. After a successful faculty search this past year, we continue with our expansion of the department in the
area of neuroscience.

We are interested in candidates who are using innovative combinations of molecular, genetic, and cellular approaches to pursue fundamental problems
in neuroscience. Areas of interest include but are not limited to in vivo imaging, genetic and epigenetic mechanisms underlying neural circuitry plasticity,
and behavior, as well as aging, regeneration and repair.

Individuals holding Ph.D. and/or M.D., or equivalent degrees, with two or more years of postdoctoral experience are encouraged to apply. Applicants
should demonstrate excellence in research and strong potential for securing and sustaining independent and collaborative extramural funding.
The University of Utah offers excellent resources to support new faculty, including competitive salary and start-up support, a highly collegial research
environment, core facilities and strong interdepartmental graduate training programs. A successful applicant will be expected to develop an innovative,
independent research program, and to share our commitment to excellence in graduate and medical education.

Only electronic applications will be accepted. Please submit a single PDF document including: 1) cover letter, 2) curriculum vitae, 3) research statement
4) one recent publication. Email the application to: facsearch@neuro.utah.edu Three letters of reference should be sent independently to: facsearch@
neuro.utah.edu

For full consideration, applications should be received by October 29, 2010.

The University of Utah is an Affirmative Action/Equal Opportunity employer and does not discriminate based upon race, national origin, color, religion, sex,
age, sexual orientation, gender identity/expression, disability, or status as a Protected Veteran. Upon request, reasonable accommodations in the application
process will be provided to individuals with disabilities. To inquire about the University’s nondiscrimination policy or to request disability accommodation,
please contact: Director, Office of Equal Opportunity and Affirmative Action, 201 S. Presidents Circle, Rm 135, (801) 581-8365.

The University of Utah values candidates who have experience working in settings with students from diverse backgrounds, and possess a demonstrated
commitment to improving access to higher education for historically underrepresented students.

ASSISTANT PROFESSOR
DEPARTMENT OF GENETICS
HARVARD MEDICAL SCHOOL

The Department of Genetics at Harvard Medical School invites applicants for a tenure-track faculty position at the
rank of Assistant Professor. The Department of Genetics consists of faculty working on diverse problems using a
variety of approaches and model organisms, unified in their focus on the genome as an organizing principle for
understanding biological phenomena. We are seeking outstanding applicants, with a demonstrated potential for
imaginative research and a clear vision for the future, who are working on exciting problems in any area of genet-
ics, broadly defined. The successful candidate is expected to direct innovative and independent research and
participate in the teaching of graduate and medical students. Significant scholarly and scientific resources are
readily available with this position. Our highly interactive Department provides the opportunity to interact and
collaborate with other dedicated researchers within the diverse Harvard research community. For further informa-
tion about our Department, please see our Web Page: http://genetics.med.harvard.edu.

Applicants should submit electronic copies of curriculum vitae, bibliography, a brief description of research
accomplishments and future research interests (limit to 500 words) by October 31, 2010, and ask three refer-
ences to provide letters of recommendation. These materials should be sent to the following email address:
faculty_search@genetics.med.harvard.edu.

We strongly encourage applications from women and minority candidates.

Harvard University is an Equal Opportunity/Affirmative Action Employer.

Positions Available

THE UNIVERSITY OF TEXAS

SOUTHWESTERN MEDICAL CENTER
AT DALLAS

The Department of Molecular Biology at the University of

Texas Southwestern Medical Center invites applications for
Assistant Professor tenure-track faculty positions. We are
seeking individuals with innovative research programs in
the areas of stem cell biology, cell signaling, development
and gene regulation. Successful applicants should be capa-
ble of establishing a vigorous independent research program
and teaching in one of several active graduate programs.
Attractive start-up packages through the Endowed Scholars
Program, state-of-the-art core facilities, and exceptional
laboratory space are available.
Applicants should submit a curriculum vitae containing a
summary of past research accomplishments, a statement of
future objectives, and names of three references to:
Eric N. Olson
Chairman
Department of Molecular Biology
University of Texas
Southwestern Medical Center at Dallas
5323 Harry Hines Boulevard
Dallas, Texas 75390-9148
UTSW is an Equal Opportunity Employer
 Positions Available

Department of Pharmacology
(with potential joint appointment in the Institute for Translational Neuroscience)
University of Minnesota Medical School
TENURE/TENURE TRACK POSITION
(Assistant Professor, Associate Professor, Professor)

The Department of Pharmacology at the University of Minnesota invites applications for a tenure/tenure track faculty position (Assistant,
Associate or Full Professor). Applicants using molecular, biochemical, cellular, and/or integrative translational approaches to study
problems relevant to pharmacological sciences are encouraged to apply. Positions are also available in this joint recruitment with the
Institute for Translational Neuroscience, whose interest is in how basic science discoveries may lead to new therapeutic principles of
certain neurological disorders (see their website for more details). Qualifications include a Ph.D. in biomedical science, or an M.D.
degree, and relevant postdoctoral research experience. Applicants must have a strong record of research accomplishments, as docu-
mented by publications in leading peer-reviewed journals. The successful Assistant Professor candidate will be expected to develop
an innovative, competitive research program supported by extramural funding and to participate in departmental teaching activities.
Applicants for Associate Professor and Professor positions must demonstrate distinction in published research, evidence of consistent
extramural funding, and a commitment to teaching. For additional information about the department and the Institute for Translational
Neuroscience, visit: www.pharmacology.med.umn.edu and http://www.itn.umn.edu.

Interested applicants should apply online at http://employment.umn.edu/applicants/Central?quickFind=88959 for the Assistant

Professor position or at http://employment.umn.edu/applicants/Central?quickFind=88961 for the Assoc. Prof. or Prof. position, and
attach a letter of interest, curriculum vitae, brief statement of research interests, and contact information for three references.

DEPARTMENT OF STRUCTURAL BIOLOGY

ASSISTANT PROFESSOR OF STRUCTURAL BIOLOGY

Applications are invited for a tenure-track junior faculty appointment in the Department of Structural Biology, Stanford University School of
Medicine. Candidates should have expertise and a commitment to future research in the broad area of structural biology and biophysics. The
predominant criterion for appointment in the University Tenure Line is a major commitment to research and teaching.
To be considered for this position, interested individuals should:
1. Fully complete the Structural Biology Faculty Recruitment Form available at: https://med.stanford.edu/survey/sbio_fclty_recruit.
2. Prepare the application materials as one .pdf file and e-mail the file to sbio_faculty_recruit@lists.stanford.edu. PDF file must contain
all of the following documents:
i) Cover letter
ii) Curriculum vitae
iii) Description of research interests (4 page limit)
iv) List of publications
v) Maximum of 5 representative reprints
vi) Names and contact information of three referees
3. Arrange for three reference letters to be e-mailed by each referee directly to: sbio_fclty_reference@lists.stanford.edu
or send by mail to:
Chair, Faculty Search Committee, Department of Structural Biology, Stanford University School of Medicine, 299 Campus Drive West, D100
Fairchild Bldg., Stanford, CA 94305-5126
Candidates must have a PhD and/ or MD degree and a minimum of two years of postdoctoral research experience. Application material and
reference letters must be received by November 1, 2010.
Stanford University is an equal opportunity employer and is committed to increasing the diversity of its faculty. It welcomes
nominations of and applications from women and members of minority groups, as well as others who would bring
additional dimensions to the university's research, teaching and clinical missions.
Positions Available

FACULTY POSITION IN
MOLECULAR, CELLULAR, AND
DEVELOPMENTAL BIOLOGY
University of Colorado at Boulder
The Department of MCD Biology invites applications for a
tenure-track Assistant Professor in the area of molecular, cel-
lular, or developmental biology with an emphasis on basic
molecular biological problems. Applicants must have a Ph.D.,
M.D., or equivalent; and postdoctoral research experience.
The candidate is expected to develop a vigorous and innova-
tive research program, and have enthusiasm for teaching at the
undergraduate and graduate levels.
Review of applications will begin on November 1, 2010 and
continue until the position is filled. Application materials are
accepted electronically at https://www.jobsatcu.com, posting
number 810875. Applicants should submit a curriculum vitae
and a concise statement of research and teaching interests, and
arrange to have three reference letters sent.
For questions or concerns on submitting your materials elec-
tronically, please contact mcdbsrch@Colorado.EDU.
The University of Colorado at Boulder is committed to diver-
sity and equality in education and employment.
See www.colorado.edu/ArtsSciences/Jobs/ for full job
description.
Positions Available
POSTDOCTORAL POSITIONS
AVAILABLE AT SLOAN-KETTERING
INSTITUTE
Seeking research scientists with postdoctoral experi-
ence to join a laboratory in order to study the molecular
pathogenesis of multiple myeloma and utilize mouse
models to develop new therapeutic approaches.
Applicants must have a thorough knowledge of plasma
cell biology and expertise using genetically modified
mice to study cancer pathogenesis. Send a description
of your research interests, curriculum vitae, and the
names and phone numbers of three references to
Dr. Stephen D. Nimer
Memorial Sloan-Kettering Cancer Center
1275 York Avenue, Box 575
New York, NY 10065
The Memorial Sloan-Kettering Cancer Center is an
Affirmative Action/Equal Opportunity Employer.
Positions Available

THE UNIVERSITY OF CALIFORNIA AT BERKELEY

Department of Molecular and Cell Biology,

Helen Wills Neuroscience
Institute and Center for Computational Biology

The Department of Molecular and Cell Biology (MCB), the Helen Wills Neuroscience Institute (HWNI)
and the Center for Computational Biology (CCB) are seeking applications for four faculty positions in
the areas listed below. We seek candidates with Ph. D. and/or M.D. degrees who have a strong inter-
est in undergraduate and graduate teaching and demonstrated excellence, originality and productivity
in research.

Biomedical Sciences (MCB)

We seek candidates interested in molecular and cellular mechanisms of disease processes. Areas of
research could include cancer, human genetics, aging, molecular physiology, animal models of
human disease, infectious and neurodegenerative diseases, or any other area of biomedical science.
This position is open at the Assistant Professor level (tenure-track).

Molecular, Cellular or Developmental Neuroscience (MCB / HWNI)

We seek candidates in any area of molecular, cellular, and developmental neurobiology, with a particular
emphasis on molecular genetic approaches to understanding neural circuit function, development, or
regeneration. Approaches could include (but are not limited to) human genetics, use of genetic model
organisms, genomics, behavioral analysis, optical imaging, and/or computational modeling. This posi-
tion is open at the Assistant Professor level (tenure-track).

Human Genome Variation (MCB / CCB)

We seek candidates whose research focuses on an understanding of genome and epigenome variation
of humans as it relates to phenotype, particularly disease predisposition, using both computational and
laboratory approaches. This position is open at the Assistant Professor level (tenure-track).

Stem Cell Biology (MCB)

We seek candidates whose research focuses on any aspect of stem cell biology, including (but not
limited to) the use of stem cells to develop models of human biology or disease, molecular mecha-
nisms of transcriptional regulation in pluripotent and differentiating stem cells, the regulation of stem
cell renewal and differentiation during development, and the biology of cancer stem cells. This posi-
tion is open at any level (tenured or tenure-track).

Applications and letters of reference should be submitted online through http://mcb.berkeley.edu.

Applications should include a curriculum vitae; a list of publications; copies of three significant
publications; a brief description of research accomplishments; and a statement of research objectives
and teaching interests. In addition, junior applicants applying for a non-tenured position should
arrange to have three letters of reference submitted online. Potential reviewers should be referred to
the Statement of Confidentiality found at: http://apo.chance.berkeley.edu/evalltr.html. The deadline
for applications is November 15, 2010.

We are interested in candidates who will contribute to diversity and equal

opportunity in higher education through their teaching, research, and service. We are
also committed to addressing the family needs of faculty. The University of California
is an Affirmative Action/Equal Opportunity Employer.
 Positions Available

The Cell Biology Department seeks exceptional

candidates at the rank of Assistant and Associate
Professor who wish to engage in independent research
programs in the field. Successful candidates will have
PhD, and/or M.D. degrees and a demonstrated record
of originality and productivity in research during gradu-
ate training and/or post-doctoral research, as well as
well-formulated plans for independent research.

60
Although the focus is on tenure-track junior faculty
appointments, applicants for senior faculty ranks may
also be considered. The Department is undergoing a
major expansion under the Chair, Dr. James E. Rothman.
Exceptional candidates in all areas of cell biology will
be considered.

Please email your CV with a list of publications, a years of leadership in

summary of doctoral research [1 page], a summary of
post-doctoral research [max. 2 pages], and a research human genetics research,
plan [max. 3 pages], along with the names and addresses education and service.
(including email) of three potential references by
October 25, 2010 to: cellbio.search@yale.edu
1948–2008
Applications from, or nominations of, women and
minority scientists are encouraged. Yale is an www.ashg.org
Affirmative Action/Equal Opportunity Employer.
 EuPA now has
http://www.eupa.org/
its own journal!

Editor in Chief:
Juan J. Calvete, Valencia, Spain Covered by
PubMed
Executive Editors:
Proteomics in Cell Biology
Jean-Jacques Diaz, Lyon, France

Proteomics in Microbiology
Concha Gil, Madrid, Spain

Proteomics in Plant Systems

Jesus V. Jorrín, Córdoba, Spain

Proteomics in Animal Models

Dario Neri, Zürich, Switzerland

Proteomics in Protein Science

Jasna Peter-Katalinic, Münster,
Germany

Biomedical Applications of
Proteomics and Congress
Proceedings
Jean-Charles Sanchez, Geneva,
Switzerland

Proteomics of Body Fluids and

Proteomic Technologies
Pier Giorgio Righetti, Milan, Italy Submitting Authors
Manuscripts can be submitted to the
Bioinformatics in Proteomics Journal of Proteomics at
Peter Højrup, Odense, Denmark http://ees.elsevier.com/jprot
For a complete listing of the
editorial board, visit the
journal’s homepage

To receive more information register at:

http://www.elsevier.com/locate/jprot
 For the past 40 years
FEBS Letters has
published award-winning
research papers crafted
by distinguished
experts in the field.

www.febsletters.com
 Just Published!

John H. Byrne and James L. Roberts, who treated us to a superb textbook in

their first edition of From Molecules to Networks, have now returned with another
outstanding effort in the second edition.
— Eric R. Kandel, Howard Hughes Medical Institute; Columbia
University College of Physicians and Surgeons, New York, USA

From Molecules to Networks

An Introduction to Cellular and Molecular
Neuroscience, Second Edition

Editors-in-Chief:
John H. Byrne, W.M. Keck Center for the
Neurobiology of Learning and Memory,
University of Texas Medical School at Houston;
James L. Roberts, Trinity University,
San Antonio, Texas

s Now includes an introduction to

mathematical modeling and simulation
approaches and a chapter on the
Neurobiology of Disease
s Full color, professional graphics throughout
s Two expansive, frequently updated
support websites:
- For students—study materials, updated
further reading references, and quizzes
- For instructors—all figures, solutions,
PowerPoint lecture slides, and more

Feb 2009, hardback 656 pp.

ISBN: 9780123741325
$94.95 USD / €69.95 EUR / £47.99 GBP

Secure online ordering at elsevierdirect.com/9780123741325

Register at our instructor site: http://textbooks.elsevier.com

This site allows registered/approved instructors to browse our list of cutting-edge books at
student-friendly prices, request print review copies, access ancillaries, view selected books
electronically, and contact your textbook representative.
 60
years of leadership in
human genetics research,
education and service.
1948–2008
www.ashg.org
 T H E N E W E D I T I O N O F F U N D A M E N TA L N E U R O S C I E N C E I S N O W AVA I L A B L E !

FUNDAMENTAL
NEUROSCIENCE
THIRD EDITION
EDITED BY:
Larry R. Squire, VA Medical Center and University
of California, San Diego, La Jolla, California, USA

February 2008 SENIOR EDITORS:

Darwin Berg, Floyd E. Bloom, Sascha du Lac,
Hardback Anirvan Ghosh, and Nicholas C. Spitzer
1,280 pp.
9780123740199 This comprehensive text reference presents the entire spectrum of modern neuroscience.
Print list price: Addressing feedback on the previous edition, Fundamental Neuroscience, Third Edition is
$99.95 now more concise and reader-friendly than ever before. Each chapter is heavily illustrated
and provides clinical boxes describing experiments, disorders, and methodological
approaches and concepts. A companion site contains an image bank of the figures for
use in poster presentations, slides, and handouts.
Capturing the promise and excitement of this fast-moving field, Fundamental Neuroscience,
Third Edition is the book that you will reference throughout your neuroscience career!

NEW TO THIS EDITION:

• 30% new material including new chapters on Dendritic Development and Spine
Morphogenesis; Chemical Senses; Cerebellum; Eye Movements; Circadian Timing;
Sleep and Dreaming; and Consciousness
• Companion Web site with all figures and Web links to additional material
• Multiple model system coverage beyond rats, mice, and monkeys
• Extensively expanded index for easier referencing

WHY BUY THIS EDITION?

• It presents the entire spectrum of modern neuroscience as a cohesive reference
rather than just a collection of review articles
• It is written at a level appropriate for graduate-level students and researchers with
various science backgrounds who need a solid foundation in neuroscience principles
• Clinical conditions are discussed in neuroscience to help readers establish the
connection between the basic biology and the neurological symptoms and syndromes
• It includes more than 650 illustrations, including color photographs, imaging, and
micrographs, also available on the companion Web site

SECURE ONLINE ORDERING AT ELSEVIERDIRECT.COM

 .%7

Find Your Ideal Job!

så3EARCHåJOBSåBYåKEYWORDå
LOCATIONååTYPEå

så0OSTåYOURåRESUMEåå
ANONYMOUSLYå

så#REATEåAå*OBå!LERTåå
ANDåLETåYOURåIDEALåå
JOBålNDåYOU

careers.cell.com
 ,OOKING¬FOR¬WHAT¬#ELL¬
HAS¬PUBLISHED¬IN¬¬
%VOLUTION¬LATELY

TION
LIANå%VOLU
O å6 IS IO N åI Nå-AMMA å   å PPån
å
D A P TS åT PR å 
Tå3PECIES
å!  å!
Rå#ELLS å#ELLå
Få 2 O D å0 H OTORECEPTO REMERå4å'UCKå*å*OFFEå" " E TW E E N å4WOå9EAS
åO å# RILITYå
RCHITECTURE å+ÙSEMå
3å0EICHLå, å(YBRIDå3TE
.UCLEARå! - å,ANCTØTå# N O M ESå#AUSES  åPPån R
YS IN Gå Lå ' E ESISå-OTO
3OLOVEIå)å+
RE HONDRIA ECåå EDå#YTOKIN
. U C LE A Rå ANDå-ITOC EU å* 9 å# ELLåå$
A å# O N S E RV
BILITYåOFå å9ANGå39 å,
LLSå$EPRIV
EDåOFå
)NCOMPATI
åå
å#HANGå.( DELå#7å,Iå2
å# HO Uå *9 å#HEONGå, O LU TI O N åO Få9EASTå#E G ( AM PTONå+å3EI
,EEå( 9 TIVEå% V !å3TAEH LIN
APIDå!DAP Nå+å0ERERAå
P LO ID Yå 5 NDERLIESå2 å.OLLå!å4RIMBLEå2å7ALTO Gå9EAST +2åå
!NE U å&LEHARTYå" O N åINå"UDDIN '2å&OSTERå
2ANCATIå'å0
AV ELKA å.
å  n  IK E å# O O P E RA TI
# å, AT GÎå*0å&INKå
 å PP IOlLM , Nå!å0REVO
ST å-
å.OVåå Tå$RIVESå" Så-$å*ANSE
#ELLå E A RD å'ENEåTHA å9ANå#å6INCE
N å" QUENCING
RE E GN INIå3
å6ARIABLEå' HETå.å"EAUVAISå!å'UA
DA
UGHPUTå3E å3IEBAUERå-å"URBANO
å
&,/åISåA RA å- å0 OC å  n  åB Yå ( IG H 4 H RO
FERå+
å# AL DA   å PP
ETERM IN E D EL å5å0Rß åå
3MUKALLAå3 å.OVå
å
EQUENCEå$ å4å3TENZ å0ÊÊBOå3
å+*åå#ELLå *-å-ARICIC å*å3LATKINå-
6ERSTREPEN H O N D RI A Lå 'ENOMEå3 HLERå#å-EYERå-å'OODå å, AA KK ON ENå,å+ELSO
RTALå-ITOC NSONå0,å5 | )å7IKSTRÙ
Må-
TEå.EANDE RAUSEå*å"RIGGSå!7å*OH OVICå | Nå:å'US
ICå
!å#OMPLE å+ RA JK | $å+UCA ON
CEå%VOLUTI
Så ! 3 å0 å"
ALASPINA å-å2UDAN
'REENå2%å-
OT HB ER Gå *-å%GHOLM O N å# O D IN G 3EQUEN
(!å2ONAN
å-å2 n NSTRAINTå
 åPPå MINANTå#O
#ELLå
å!UGåå
IS FO LD IN GåASåAå$O ROSOPHILA

N D U C E D å0 ROTEINå-  å PP ån IM O RP H IC å4RAITSåINå$
TION ) å*ULåå
 UALLYå$
-ISTRANSLA #/å#ELLå OLLINGå3EX ån
M ON Då $!å7ILKEå N E TI C å3 W ITCHå#ONTR å!UGåå åPP ,,å&AMILY

+./8"%
$RU M Få A å' E å# EL Lå
V O LU TI O N åO å# AR RO LLå 3 "
E å0 LA N Tå
å% ! NåOFåTH
LATIONåAND ERNERå4å'OMPELå.å+OP
På
4HEå2EGU Då%VOLUTIO
4- å3 EL EG UEå*%å7
TE RO D IM E RIZATIONåAN
7ILLIAMSå INå(E n
MEOPROTE  åPPå
E X U A Lå / RI GINSåOFå(O 5å#ELLåå-AYåå
%ARLYå3 OODENOUGHå
å(å*OOå3å'
,EEå*(å,IN

,OOKå!GAIN
$ISCOVERå-ORE
WWWCELLCOM
SnapShot: Nuclear Receptors I
Neil J. McKenna and Bert W. O’Malley
Baylor College of Medicine, Houston, TX 77030, USA

Receptor/ Disease
Family* Symbols Ligands Major Functions Associations Target Genes
Estrogen ERα/NR3A1; Endogenous: 17β-estradiol Regulation of cell growth and proliferation Cancer, cardiovascular, ↑ MYC, NGF, BCL2, CXCL2,
receptors* ERβ/NR3A2 in multiple tissues (e.g., female reproductive immune and IGF1, TYMS; ↓ CD36, NDRG1,
Clinical: Mixed agonists (e.g. tamoxifen, tissues, bone, and CNS) inflammatory, NCOR1, NCOA3
raloxifene, and toremifene in breast cancer) metabolic, neurological,
Xenobiotics: Bisphenol A, PCBs reproductive

Androgen AR/NR3C4 Endogenous: Testosterone, Key role in male reproductive organs in Cancer, cardiovascular, ↑ MYC, VEGF, BCL2, IGF1,
receptor dihydrotestosterone addition to other systems (e.g., CNS) immune, metabolic, MUC1, P66(Shc), CCND1;
neurological, ↓ TSHA, TSHB, PTEN, FAS,
Clinical: Flutamide and bicalutamide for reproductive CASP2, CTNND2, ESR2,
prostate cancer and alopecia TMPRSS2

Glucocorticoid GR/NR3C1 Endogenous: Cortisol (hydrocortisone) Diverse developmental and physiological Metabolic, ↑ SCNN1A, GADD45B, GILZ,
receptor roles (e.g., antagonism of inflammatory cardiovascular, immune TAT; ↓ BGLAP, POMC, INS
Clinical: Fluticasone and prednisolonein signaling pathways, mediation of the stress and inflammatory,
inflammatory disorders response, and gluconeogenesis) memory

Vitamin D VDR/NR1I1 Endogenous: Calcitriol (1′,25′ dihdroxy Maintenance of serum calcium and Bone, cancer, ↑ FGF23, CYP24A1, CALB1,
receptor vitamin D3) phosphate levels for skeletal integrity; cardiovascular, BGLAP, SPP1; ↓ IL2, PHEX
antiproliferative in many tissues metabolic, immune and
Clinical: Paracalcitol for 2o inflammatory, renal,
hyperparathyroidism in renal patients; neurological
Tacalcitol for psoriasis

Thyroid hormone TRα/NR1A1; Endogenous: Thyroxine Regulation of oxygen consumption; protein, Thyroid conditions, ↑ ADRB1, PCK2, GH1, UCP1; ↓
receptors* TRβ/NR1A2 (T4),Triiodothyronine (T3) carbohydrate, lipid, and vitamin metabolism cancer DIO2, EHHADH, PRL, EGFR

Clinical: Levo-thyroxine, triiodothyroacetic

acid (TRIAC) in resistance to thyroid
hormone

Progesterone PR/NR3C3 Endogenous: Progesterone Diverse reproductive functions (e.g., Cancer, metabolic, ↑ SERPINB14, FAS, MT2A,
receptor establishing and maintaining pregnancy, reproductive PGC, EGFR , IHH; ↓ ESR1, PGR,
Clinical: RU486 (Mifepristone) as an developing breast tissue, and stopping ANXA6
abortifacient proliferation in the uterus)

Mineralocorticoid MR/NR3C2 Endogenous: Aldosterone Regulating electrolyte and fluid balance in Metabolic ↑ SCNN1A, ATP1A1, ATP1B1,
receptor the kidney; specific roles in the CNS GILZ, SGK1, NDRG2
Clinical: Spironolactone in hypertensive
cardiovascular disease

Peroxisome- PPARγ/NR1C3 Endogenous: FAs and FA intermediates Regulation of adipocytes, insulin sensitivity Cardiovascular, ↑ FABP4, UCP1, AP2, PCK1,
proliferator- and lipogenesis, and broader integration of metabolic, cancer, LPL, ADIPOQ, CD36, AQP7
activated Dietary: FAs and PUFAs energy, lipid, and carbohydrate metabolism neurological
receptor-γ
Clinical: Thiazolidinediones (e.g.,
rosiglitazone) in type II diabetes

Peroxisome- PPARα/ Endogenous: FAs and FA intermediates Regulating energy expenditure; modulating Cardiovascular, ↑ ACBP, ACOX1, APOA1,
proliferator- NR1C1 fatty acid oxidation systems (mitochondria), metabolic, cancer, CPT1A, CYP1A1, CYP4A1,
activated Clinical: Fibrates (e.g., fenofibrate) in peroxisome β-oxidation, and microsomal neurological CYP7A1, SLC27A1, LCAS,
receptor-α hyperlipidemia ω-oxidation MLYCD, SCD, FADS2, RETN,
Dietary: FAs and PUFAs MYC, CCND1, IGFBP1, UCP1,
KRT23, IL6, TF, PEX11A
Xenobiotics: DEHP, DEHA

Peroxisome- PPARδ/ Endogenous: FAs and FA intermediates Regulating cell proliferation, differentiation, Cardiovascular, ↑ ACSL3, CPT1A, RGS3, RGS4,
proliferator- NR1C2 and migration in wound healing and metabolic, cancer, RGS5, ISG20, CXCL7, CCL21,
activated Dietary: FAs and PUFAs inflammatory processes neurological RETN, CPT1A
receptor-δ (β)

Retinoic acid RARα/NR1B1; Endogenous: All-trans and 9-cis retinoic Pleiotropic control of embryonic patterning Neurological and ↑ Numerous HOX genes, STRA6,
receptors* RARβ/NR1B2; acid and organogenesis, cell proliferation, psychiatric, cancer HNF3A, CRABP2, ACADM,
RARγ/NR1B3 differentiation, apoptosis and homeostatic MECOM; ↓ CYP1A1, HOXB9
Clinical: Tretinoin for treating acne and control
acute promyelocytic leukemia

Liver X receptors* LXRα/NR1H3; Endogenous: Oxysterols Cholesterol and steroid sensors with roles in Metabolic ↑ SREBP1C, CYP7A1, ABC8,
LXRβ/NR1H2 lipid and carbohydrate metabolism APOA1, APOE, LPL, PLTP

Retinoid X RXRα/NR2B1; Endogenous: 9-cis retinoic acid Embryonic cell patterning and Neurological and ↑↓ many genes as heterodimers
receptors* RXRβ/NR2B2; organogenesis, cell proliferation and psychiatric, immune with other receptors (e.g.,LXRs,
RXRγ/NR2B3 differentiation, other functions as PPARs, FXR, TRs, RARs; ↑ ABC1
heterodimers with other nuclear receptors (with LXR); ↓ CYP7A1 (with FXR)

Pregnane X PXR/NR1I2 Endogenous: Bile acids Metabolism and transport of pharmaceutical Immune ↑ Multiple CYP2 and CYP3
receptor drugs, xenobiotics, and toxic bile acids in gene family members, MDR1,
Xenobiotic: St. John’s Wort (hyperforin), the liver and GI tract MRP2, OATP2, UGT1A1, SULT,
Taxol, rifampicin, phenobarbital ↓ CYP7A1
Dietary: Vitamin E, sulforaphane, Gugulipid

Constitutive CAR/NR1I3 Endogenous: Androstanol, androstenol Metabolism of xenobiotics and endogenous Involved in ↑ CYP2B10, CYP311A, CYP3A4,
androstane lipids by regulating expression of hepatotoxicity of CYP1A2, CYP2B6 THRSP,
receptor Xenobiotics: Phenobarbital, DEHP, cytochrome P450 genes acetaminophen SLC21A6, MRP2, MDR1, OATP2
Meclizine

Farnesoid X FXR/NR1H4 Endogenous: Bile acids (e.g., A sensor for bile acid that helps regulate bile Metabolic ↑ SLC10A2, ABCB1, ABCB11,
receptor chenodeoxycholic acid) acid homeostasis NR0B2, HSD3B2, FETUB,
ABCB4, FGF19, NOS2; ↓
Dietary: Cafestol, guggulsterone CYP7A1, HNF1A, HNF4A,
SLCO1B1, SLC10A2

822  Cell 142, September 3, 2010 ©2010 Elsevier Inc.  DOI 10.1016/j.cell.2010.08.026 See online version for legend and references.
4HEå"IOPHYSICALå3OCIETYå
ANDå#ELLå0RESS
!åNEWåPARTNERSHIP

3EEåWHATSåNEWåAT
WWWBIOPHYSJORG
SAN TA CR UZ B I OT ECH N O LO GY, I N C . SANTA CRUZ
®

Leading (#1 worldwide) supplier BIOTECHNOLOGY

of specialty biochemicals for

The Power to Question

proteomics research.

Products include over 71,300 unique biochemicals

sorted alphabetically, by CAS number, by formula
and on the basis of functional activity.

See our web site for product details

and ordering information.

www.scbt.com

HEADQUARTERS
TOLL FREE: 800.457.3801 TEL: 831.457.3800
FAX: 831.457.3801 E-MAIL: scbt@scbt.com

EUROPEAN SUPPORT
TOLL FREE: +00800.4573.8000 PHONE: +49.6221.4503.0
FAX: +49.6221.4503.45 E-MAIL: europe@scbt.com

ASIAN SUPPORT
JAPAN TOLL FREE: (010) 800.40402026 S. KOREA TOLL FREE: 00798.1.1.002.0297
N. CHINA TOLL FREE: 10.800.711.0752 S. CHINA TOLL FREE: 10.800.110.0694
FAX: 831.457.3801 E-MAIL: asia@scbt.com

ChemCruz ™