JEM Immunity-and-Cancer Special Issue PDF

THE
JOURNAL OF
EXPERIMENTAL
MEDICINE
SELECTED ARTICLES FEBRUARY 2015 www.jem.org
IMMUNITY AND CANCER
RE LI
SE PI
av AR DO
a C
M
cl ntili H P IC
ic p
S
R
i
on ds.c OGR
LR om AM
P
AVANTIS NEW PROBES

PACFA
pacFA is a new technology that Avanti is making available to probe cellular
protein-lipid interactions in vivo. The pacFA lipid contains a photoactivable
diazirine ring with a clickable alkyne group and was developed by Dr. Per Haberkant at the European Molecular Biology Laboratory.
pacFA
Avanti Number 900401
pacFA Ceramide
pacFA Glucosyl Ceramide
pacFA Galactosyl Ceramide

16:0-pacFA PC
pacFA-18:1 PC
To screen for protein-lipid interactions cells are fed pacFA as a precursor for the biosynthesis of bifunctional lipids.
Proteins in contact with the bifunctional lipids are then cross-linked
by UV irradiation of the diazirine ring. Finally, click chemistry is
used to label the alkyne with a reporter molecule.
Labeling with a biotinylated azide allows for the affinity purification
and profiling of cross-linked proteins with mass spectrometry; labeling with a fluorescent azide allows visualization of cross-linked
proteins with microscopy.
Haberkant, P., R. Raijmakers, M. Wildwater, T. Sachsenheimer, B. Brugger, K. Maeda, M. Houweling, A.C. Gavin, C. Schultz, G. van
Meer, A.J. Heck, and J.C. Holthuis. (2013). In vivo profiling and visualization of cellular protein-lipid interactions using bifunctional
fatty acids. Angew Chem Int Ed Engl 52:4033-8.
AVANTI
IS THE
WORLDS
FIRST CHOICE FOR
PHOSPHOLIPIDS, SPHINGOLIPIDS, DETERGENTS
&
STEROLS
CGMP LIPIDS FOR PHARMACEUTICAL PRODUCTION

LIPID ANALYSIS
VISIT AVANTILIPIDS.COM
Welcome
I m mu n i t y a n d C a n c e r
The Journal of Experimental Medicine now prints topic-specific mini collections to showcase
a handful of our recent publications. In this installment, we highlight papers focusing on
cancer immunology and immunotherapy.
Our collection begins with an Insight from Christian Jobin discussing the article from
Bongers and colleagues on the influence of intestinal bacteria on tumorigenesis. The
study finds that the microbiota can modulate site-specific tumor development in a mouse
model of colorectal cancer. The transgenic mice (HBUS) express HB-EGF throughout the
intestine but develop serrated polyps (SPs) only in the cecum.The authors demonstrate that
the host microbiome is topologically associated with SPs and changes in the microbiota due
to antibiotic treatment or from embryo transfer rederivation can inhibit the development
of SPs in the cecum.
The search for appropriate combination immunotherapies in cancer treatment is a
fast-growing field, and one that harbors great potential. An article by Fan et al. explores how combination therapy of CTLA-4
blockade together with ICOS engagement can enhance anti-tumor immune responses.The authors show that irradiated B16/F10
melanoma cellsand other tumor modelstransduced with the ICOSL, and used as a vaccine in combination with CTLA-4
blockade in mice, results in 80% tumor rejection with great efficacy and increased function of effector CD8+ T cells.
Medullary thymic epithelial cells (mTECs) contribute to self-tolerance through thymic expression of tissue-specific antigens
(TSAs). Modulating central tolerance is an attractive therapeutic strategy for cancer treatment. Khan et al. demonstrate that in vivo
blockade of RANKL can inhibit the development and turnover of mTECs and alter central T cell tolerance. In vivo RANKL
blockade in mice transiently depletes Aire and TSA expression in the thymus inhibiting negative selection. RANKL blockade
can also rescue melanoma-specific T cells from thymic deletion, with tumor-specific effector CD4+ T cells facilitating host survival
in response to tumor challenge.
The mutational repertoire of cancers creates neoepitopes which make cancers immunogenic. An article by Duan et al.
provides novel tools for neoepitope prediction in anti-tumor T cell responses. Fibrosarcomas/sarcomas in mice are used to perform
transcriptome sequencing, identifying novel single nucleotide variants corresponding to neoantigens of MHC class I alleles. A
differential agretopic index (DAI) compares MHC-peptide binding between wild type parental and candidate neoantigens and
allows the identification of neoepitopes. Surprisingly, identified anti-tumor protective neoepitopes exhibit low MHC-peptide
binding affinity, concomitant with increased C-terminal conformational stability.
Adult T-cell leukemia/lymphoma (ATLL) is an aggressive malignancy caused by human T cell lymphotropic virus type1
(HTLV-1). Increased expression of CCR4 is a hallmark of ATLL, but it is not clear whether dysregulated CCR4 contributes to
disease pathogenesis. An article by Nagakawa et al. identifies recurring somatic mutations of CCR4 in ATLL patients, finding
that a CCR4 gain-of-function mutation impairs CCR4 internalization and increases cell migration and chemotaxis in vitro.
An accompanying Insight by Kevin Shannon discusses the implications of CCR4 mutations, PI-3K signaling downstream of
CCR4, and the translational potential of CCR4 blockade.
Collectively, the presented articles identify pathways, strategies, and tools which may contribute to therapeutic approaches
to combat oncogenic processes.We hope you enjoy this complimentary copy of our Immunity and Cancer collection.We invite
you to explore additional collections at www.jem.org and to follow JEM on Facebook, Google+, and Twitter.
Selected Articles February 2015

Do bugs define cancer geography?
Christian Jobin
Interplay of host microbiota, genetic perturbations, and inflammation promotes local development of
intestinal neoplasms in mice
Gerold Bongers, Michelle E. Pacer, Thais H. Geraldino, Lili Chen, Zhengxiang He, Daigo Hashimoto, Glaucia C. Furtado,
Jordi Ochando, Kevin A. Kelley, Jose C. Clemente, Miriam Merad, Harm van Bakel, and Sergio A. Lira
Engagement of the ICOS pathway markedly enhances efficacy of CTLA-4 blockade in

cancer immunotherapy
Xiaozhou Fan, Sergio A. Quezada, Manuel A. Sepulveda, Padmanee Sharma, and James P. Allison
Enhancement of an anti-tumor immune response by transient blockade of central T cell tolerance
Imran S. Khan, Maria L. Mouchess, Meng-Lei Zhu, Bridget Conley, Kayla J. Fasano, Yafei Hou, Lawrence Fong,
Maureen A. Su, and Mark S. Anderson
Genomic and bioinformatic profiling of mutational neoepitopes reveals new rules to predict
anticancer immunogenicity
Fei Duan, Jorge Duitama, Sahar Al Seesi, Cory M. Ayres, Steven A. Corcelli, Arpita P. Pawashe, Tatiana Blanchard,
David McMahon, John Sidney, Alessandro Sette, Brian M. Baker, Ion I. Mandoiu, and Pramod K. Srivastava
CCR4 drives ATLL jail break
Kevin M. Shannon
Gain-of-function CCR4 mutations in adult T cell leukemia/lymphoma
Masao Nakagawa, Roland Schmitz, Wenming Xiao, Carolyn K. Goldman, Weihong Xu, Yandan Yang, Xin Yu,
Thomas A. Waldmann, and Louis M. Staudt.
THE
JOURNAL OF
EXPERIMENTAL
MEDICINE
Executive Editor
Marlowe S.Tessmer
phone (212) 327-8575

email: jem@rockefeller.edu
Senior Editor
Heather L. Van Epps
Scientific Editors
Teodoro Pulvirenti
Catarina Sacristn
Editors
Jean-Laurent Casanova
Douglas T. Fearon
David Holtzman
Lewis L. Lanier
William A. Muller
Carl Nathan
Michel Nussenzweig
Anne OGarra
Alexander Rudensky
Alan Sher
Sasha Tarakhovsky
Andreas Trumpp
David Tuveson
Editor Emeritus
Alan N. Houghton
Manuscript Coordinator
Sylvia F. Cuadrado
phone (212) 327-8575
email: jem@rockefeller.edu
Preflight Editor
Rochelle Ritacco
Assistant Production Editor
Shauna OGarro
Production Editor
Brianna Caszatt
Production Manager
Camille Clowery
Advisory Editors
Shizuo Akira
Kari Alitalo
Frederick W. Alt
K. Frank Austen
Albert Bendelac
Michael J. Bevan
Christine A. Biron
Christian Bogdan
Hal E. Broxmeyer
Meinrad Busslinger
Arturo Casadevall
Ajay Chawla
Yongwon Choi
Robert L. Coffman
Daniel J. Cua
Myron I. Cybulsky
Riccardo Dalla Favera
Glenn Dranoff
Michael Dustin
Vincent A. Fischetti
Richard A. Flavell
Adolfo Garcia-Sastre
Patricia Gearhart
Ronald N. Germain
Christopher Goodnow
Siamon Gordon
Or Gozani
Sergio Grinstein
Philippe Gros
Kristian Helin
Chyi Hsieh
Christopher A. Hunter
Kayo Inaba
Gerard Karsenty
Jay Kolls
Paul Kubes
Vijay K. Kuchroo
Ralf Kuppers
Tomohiro Kurosaki
Bart N. Lambrecht
Klaus F. Ley
Yong-Jun Liu
Clare Lloyd
Tak Mak
Bernard Malissen
James S. Malter
Philippa Marrack
Diane Mathis
Ira Mellman
Matthias Merkenschlager
Sean J. Morrison
Muriel Moser
Christian Mnz
Cornelis Murre
Benjamin G. Neel
Michael Neuberger
Victor Nussenzweig
John J. OShea
Paul H. Patterson
Fiona Powrie
Lluis Quintana-Murci
Klaus Rajewsky
Gwendalyn J. Randolph
Jeffrey Ravetch
Sergio Romagnani
Nikolaus Romani
David L. Sacks
Shimon Sakaguchi
Matthew D. Scharff
Olaf Schneewind
Stephen P. Schoenberger
Hans Schreiber
Gerold Schuler
Robert A. Seder
Rafick-P. Skaly
Charles N. Serhan
Nilabh Shastri
Ethan M. Shevach
Roy L. Silverstein
Jonathan Sprent
Janet Stavnezer
Andreas Strasser
Stuart Tangye
Steven L.Teitelbaum
Thomas J.Templeton
Kevin J.Tracey
Giorgio Trinchieri
Shannon Turley
Marcel R.M. van den Brink
Ulrich von Andrian
Harald von Boehmer
Christopher M.Walker
Raymond M.Welsh
E. John Wherry
Linda S.Wicker
Ian Wilson
Thomas Wynn
Monitoring Editors
Marco Colonna
Jason Cyster
Stephen Hedrick
Kristin A. Hogquist
Andrew McMichael
Luigi Notarangelo
Anjana Rao
Federica Sallusto
Louis M. Staudt
Toshio Suda
Consulting
Biostatistics Editors
Glenn Heller
Madhu Mazumdar
Production Designer
Erinn A. Grady
Copyright to articles published in this journal is held by the authors. Articles are published by
The Rockefeller University Press under license from the authors. Conditions for reuse of the
articles by third parties are listed at http://www.rupress.org/terms
Print ISSN 0022-1007 Online ISSN 1540-9538
INSIGHTS
Do bugs define cancer geog raphy?
Although genetic factors are essential to the development of colorectal cancer (CRC), the intestinal microbiota were recently recognized as an important environmental contributor. In this issue, Gerold Bongers,
Sergio Lira, and colleagues demonstrate for the first time that the microbiota influence site-specific development
of tumors in a mouse model of CRC.
HBUS mice express two oncogenic transgenes throughout their intestinescytomegalovirus chemokine
receptor US28 and heparin-binding EGF-like growth factorbut they develop tumors (serrated polyps or SPs)
only in the cecum. When the authors treated HBUS mice with broad-spectrum antibiotics they noticed that
Insight from
development of cecal SPs was virtually ablated. Importantly, changing the HBUS microbiome by transferring Christian Jobin
HBUS embryos into female mice obtained from a different animal vendor also attenuated the development
of SPs. Microbiome analysis indicated that antibiotic-mediated attenuation of SPs
was correlated with reduced invasion of cecal tissue by bacteria belonging to the
Clostridiales/Lachnospiraceae family. Together, these findings link bacterial topology
and intestinal barrier defects with site-specific tumor development.
The work of Bongers et al. raises questions about the relationship between bacteria
and site-specific cancer development. First, longitudinal analysis of the microbiota in
HBUS mice would help define the geographical establishment of microbial niches
in relation to tumor development. Similarly, longitudinal assessment of intestinal
barrier integrity in HBUS mice would determine whether the barrier defect is the
consequence of neoplasia, as observed in a mouse model of colon adenoma, or due to
an intrinsic host malfunction that bacteria exploit to promote tumorigenesis.
Interestingly,enrichment of the mucin-degrading bacterium Akkermansia muciniphila
in SPs of HBUS mice may strip the mucosa of its protective layer, thereby creating
favorable conditions for Clostridiales/Lachnospiraceae invasion into the mucosal tissues
Development of serrated polyps (SPs, hatched
with ensuing inflammatory host responses. In the context of dysregulated EGFR
line) in HBUS mice is attenuated (right panel)
when embryos are rederived into foster females
signaling in HBUS mice, this situation favors tumor promotion. Experiments using
from a different animal vendor.
germ-free HBUS mice would help assess this possibility as well as establish the functional
interaction between various microbial communities and tumor locations.
In summary, this study adds additional layers of complexity to cancer etiology by highlighting the interplay between host
genetics, microbial location, and tumor geography. Whether microbial niche localization influences SP development in humans
requires further investigation.
Bongers, G., et al. 2014. J. Exp. Med. http://dx.doi.org/10.1084/jem.20131587.
Christian Jobin, Department of Medicine and Department of Infectious Diseases and Pathology, University of Florida: Christian.Jobin@medicine.ufl.edu
Article
Interplay of host microbiota, genetic

perturbations, and inflammation promotes local
development of intestinal neoplasms in mice
Gerold Bongers,1 Michelle E. Pacer,1 Thais H. Geraldino,1 Lili Chen,1
Zhengxiang He,1 Daigo Hashimoto,1 Glaucia C. Furtado,1 Jordi Ochando,1
Kevin A. Kelley,2 Jose C. Clemente,1,3,4 Miriam Merad,1 Harm van Bakel,3,4
and Sergio A. Lira1
1Immunology
Institute; 2Department of Developmental and Regenerative Biology, Mouse Genetics Shared Resource Facility;
of Genetics and Genomic Sciences; and 4Icahn Institute for Genomics and Multiscale Biology; Icahn School
of Medicine at Mount Sinai, New York, NY 10029
3Department
The preferential localization of some neoplasms, such as serrated polyps (SPs), in specific
areas of the intestine suggests that nongenetic factors may be important for their development. To test this hypothesis, we took advantage of transgenic mice that expressed HB-EGF
throughout the intestine but developed SPs only in the cecum. Here we show that a hostspecific microbiome was associated with SPs and that alterations of the microbiota induced
by antibiotic treatment or by embryo transfer rederivation markedly inhibited the formation of SPs in the cecum. Mechanistically, development of SPs was associated with a local
decrease in epithelial barrier function, bacterial invasion, production of antimicrobials, and
increased expression of several inflammatory factors such as IL-17, Cxcl2, Tnf-, and IL-1.
Increased numbers of neutrophils were found within the SPs, and their depletion significantly reduced polyp growth. Together these results indicate that nongenetic factors contribute to the development of SPs and suggest that the development of these intestinal
neoplasms in the cecum is driven by the interplay between genetic changes in the host, an
inflammatory response, and a host-specific microbiota.
CORRESPONDENCE
Sergio A. Lira:
sergio.lira@mssm.edu
Abbreviations used: FDR, false
discovery rate; GI, gastrointesti
nal; OTU, operational taxo
nomic unit; qPCR, quantitative
PCR; rDNA, ribosomal DNA;
SP, serrated polyp.
The preferential localization of some neoplasms in

specific areas of the intestine suggests that non
genetic factors may be important for their devel
opment. In contrast to adenomas, which occur
throughout the large intestine, serrated polyps
(SPs) occur in specific areas of the gut in a subtypespecific manner (Huang et al., 2004; Noffsinger,
2009). SPs encompass a heterogeneous set of le
sions and are associated with perturbations of the
MAPK pathway through, e.g., activating muta
tions in KRAS or BRAF (Noffsinger, 2009).
We have recently shown that EGFR activa
tion is associated with SPs in human biopsies
and that expression of the EGFR ligand HBEGF in transgenic mice (HBUS mice) pro
motes development of SPs that mostly resemble
human hyperplastic polyps (Bongers et al.,
2012). Strikingly, despite expression of the
HB-EGF transgene throughout the gut, SPs were
only observed in the cecum, suggesting that be
side genetic alterations, environmental factors
played a pivotal role in their development.
The Rockefeller University Press $30.00

J. Exp. Med. 2014 Vol. 211 No. 3 457-472
www.jem.org/cgi/doi/10.1084/jem.20131587
In the intestine, the microbiota is in close

proximity to the intestinal epithelial cells that
form a protective barrier separating commensal
bacteria from the host. Changes in the micro
biota have been associated with inflammatory
conditions and cancer (Plottel and Blaser, 2011;
Honda and Littman, 2012; Schwabe and Jobin,
2013). For example, activation of TLR4 by the
intestinal microbiota is important for the pro
motion of hepatocellular carcinoma induced by
the carcinogen diethylnitrosamine/CCl4 (Dapito
et al., 2012). In the gastrointestinal (GI) tract,
the presence of Helicobacter pylori is strongly as
sociated with an increased risk for the develop
ment of peptic ulcers, gastric mucosaassociated
lymphoid tissue tumors, and gastric adenocarci
nomas (McColl, 2010). The colonic microbiota
2014 Bongers et al. This article is distributed under the terms of an AttributionNoncommercialShare AlikeNo Mirror Sites license for the first six
months after the publication date (see http://www.rupress.org/terms). After six
months it is available under a Creative Commons License (AttributionNoncommercialShare Alike 3.0 Unported license, as described at http://creativecommons
.org/licenses/by-nc-sa/3.0/).
457
Figure 1. Expression changes in SPs of HBUS mice. (A) Expression of HBUS transgenes (n = 4) US28 (red) and HB-EGF (blue) in duodenum (Duo),
jejunum (Jej), Ileum (Ile), cecal pouch (Pou), cecalcolonic junction (CJ), and colon (Col). The experiment was performed once. R.E., relative expression.
(B) Number of SPs found in various gut segments of HBUS mice (n = 210). (C) HBUS SPs and surrounding (Surr.) cecal tissue were analyzed by RNASeq/edgeR using a paired design (n = 3/group). Plot of logFC (log fold change) versus logCPM (log counts per million) of all detected transcripts.
458
Microbiota and serrated polyps | Bongers et al.
Ar ticle
has also been suggested to play a role in the pathogenesis

of colorectal cancer (Cho and Blaser, 2012), either by provok
ing an inflammatory response or alterations in metabolic pro
cessing (Plottel and Blaser, 2011). Specifically, activation of
Th17 cells by the enterotoxigenic Bacteroides fragilis has been
shown to promote colonic tumorigenesis in the APCmin
mouse model (Wu et al., 2009). More recently, the polyketide
synthase genotoxic island expressed by Escherichia coli NC101
was implicated in the development of carcinomas in Il10/
mice treated with the colonic carcinogen azoxymethane
(Arthur et al., 2012).
In this study we explored the possibility that the topologi
cal distribution of SPs was dependent on the microbiota. We
show that alterations of the microbiota induced by antibiotic
treatment or by embryo transfer rederivation attenuated the
formation of cecal SPs in HBUS mice. SP development was
associated with bacterial invasion of the lamina propria, which
was associated with production of Il-17 by innate immune
cells, neutrophil recruitment, and production of antimicro
bials.Together these results indicate that the interplay between
genetic changes in the host, an inflammatory response, and a
host-specific microbiota accounts for the development of SPs
in HBUS mice.
RESULTS
Antimicrobial defense response genes are induced in SPs
Expression of the transgenes in HBUS mice (Bongers et al.,
2012) is highest in the small intestine (Fig. 1 A), but HBUS
mice only develop SPs at the cecalcolonic junction (Fig. 1 B).
To further elucidate potential pathways involved in the for
mation of SPs, we compared the transcriptome of SPs in
HBUS mice with adjacent histologically normal cecal tissue
(n = 3). Whole transcriptome sequencing (RNA-Seq) re
vealed 404 genes with increased and 272 genes with decreased
expression in SPs, respectively (Fig. 1 C and Table S1; false
discovery rate [FDR] Q < 0.05). Gene Ontology (GO) analy
sis of all 676 differentially regulated genes using ClueGO
(Bindea et al., 2009) found significant enrichment of 59 over
view GO terms such as inflammatory response and cellular
response to chemical stimulus that encompassed the MAPK
cascade GO term (Fig. 1 E and Table S2; Q < 0.05). Consis
tent with the genetic alterations in HBUS mice and our ear
lier observations (Bongers et al., 2012), we see significant
enrichment of the overview GO term response to growth
factor stimulus and induction of genes associated with HBEGF and EGFR signaling, including the Egfr and its ligands
Ereg, Areg, and Nrg1.
Interestingly, we found a significant enrichment of the

GO overview term associated with response to molecule
of bacterial origin (Fig. 1 E and Table S2). This encom
passed the GO term for response to bacterium and can
be triggered in response to the presence of a bacterium and
includes genes associated with bactericidal activity such as
Arg1, Ptgs2 (also known as COX-2), Serpine1, Reg3b, and
Reg3g (Fig. 1, D and E). REG-3 and REG-3 have previ
ously been shown to play an important role in bacterial
defense (Dessein et al., 2009; van Ampting et al., 2012) and
were among the genes with the largest increase in expres
sion in SPs compared with surrounding tissue (12- and
32-fold, respectively; Fig. 1 C). Immunostaining of cecal
sections localized REG-3 and REG-3 expression to
cells in the basal epithelium of the SPs (Fig. 1, F and G),
whereas they were mostly absent in unaffected surround
ing cecal tissue. This localization pattern suggested that the
epithelium of the SPs was actively responding to bacteria
present in the local environment.
Broad-spectrum antibiotic treatment
prevents SP development in HBUS mice
The elevated expression of antimicrobial genes in SPs
prompted us to investigate whether the intestinal microbiota
played a role in SP pathogenesis.To this end, we treated HBUS
mice for 9 or 29 wk with a broad-spectrum antibiotic cock
tail (metronidazole, ampicillin, neomycin, and vancomycin)
in the drinking water and assessed the presence of SPs after
15 or 35 wk (Fig. 2 A). Treatment with antibiotics led to a sig
nificant decrease in the relative abundance of bacterial 16S
ribosomal DNA (rDNA; Fig. 2 B) and diversity (species
richness, P < 0.001) within each sample of the cecal mucosa
associated microbiota of HBUS mice (Fig. 2 C), indicating
that most bacteria were greatly reduced or absent altogether.
Remarkably, none of the HBUS mice treated with antibiotics
for 9 (n = 14) or 29 (n = 6) wk developed gross SPs (Fig. 2 D)
or microscopic lesions (Fig. 2 E), suggesting that the micro
biota is essential for formation of SPs.
Next we explored whether reconstitution of the original
microbiota after antibiotic treatment would restore develop
ment of SPs. After treatment of HBUS mice with antibiotics
for 9 wk, we gavaged half of the mice with stool samples from
untreated HBUS mice in the colony, three times per week for
6 wk, while continuing antibiotic treatment for the other
HBUS mice (Fig. 2 F). At 35 wk of age, 20 wk after treatment
with antibiotics was stopped, HBUS mice were examined for
the presence of cecal SPs by gross inspection and histologically.
Points are colored according to expression status: nonsignificant genes (gray), significant genes (676 genes; Q < 0.05; black), and defense response
genes (red). The experiment was performed once. (D) Z-scored heat map of genes associated with the GO term response to bacterium; red indicates
increased and green decreased expression in SPs compared with surrounding tissue. (E) ClueGO analysis of significantly regulated genes shown in C.
Shown are the GO overview terms selected by %Genes/Term in color and relevant individual GO terms in black (Q < 0.05; terms > 25 genes; kappa 0.5).
(F and G) Increased transcript levels (left, n = 3/group) and immunoreactivity (right, n = 8/group) of REG-3 (F) and REG-3 (G) in SPs compared
with surrounding tissue (representative figure of three independent experiments). All histology sections were counterstained with DAPI (F and G) and
pan-Keratin (F) or E-cadherin (G). Bars: (F) 250 m; (G) 100 m.
JEM Vol. 211, No. 3
459
Figure 2. Alterations in the microbiota affect formation

of SPs in HBUS mice. (A) Overview of the antibiotics (Abx)
treatment plan. At 6 wk of age, HBUS mice started receiving
broad-spectrum antibiotics or water and were examined for the
presence of SPs at 15 and 35 wk of age. (B) Bacterial content in
antibiotics-treated HBUS mice at 15 (n = 7) and 35 (n = 5) wk of
age compared with water-treated HBUS mice (n = 13). Determined by pan-bacterial 16S qPCR amplification relative to host
ubiquitin (UBI). ***, P < 0.001 (pairwise Wilcoxon rank sum test
[FDR]; representative of two independent experiments).
(C) diversity, a measure of species richness, in antibioticstreated HBUS mice (n = 11) and HBUS mice on regular water
(n = 20). ***, P < 0.001 (Wilcoxon rank sum test; experiment was
performed once). (D) SP incidence in antibiotics-treated HBUS
mice after 9 (n = 14) and 29 (n = 6) wk compared with HBUS
mice on regular water (75% at 15 [n = 16] and 35 wk [n = 12]).
(right) Gross picture of the cecum of antibiotics and watertreated HBUS mice at 35 wk of age showing an SP (dashed
lines). *, P < 0.05; ***, P < 0.001 (Fishers exact test; combined
data of two independent experiments). (E) Representative histological section of the cecum of 35-wk-old HBUS mice treated
with water or antibiotics as described in D. (F) HBUS mice were
treated with antibiotics for 9 wk and subsequently gavaged with
stools from unmolested HBUS mice (Abx/Stool). Control HBUS
mice were maintained on antibiotics. Ceca of HBUS mice were
analyzed by gross inspection and histological sectioning. SP
incidence in HBUS mice treated with Abx/Stool (n = 8) and HBUS
mice maintained on antibiotics (n = 9). *, P < 0.05 (Wilcoxon
rank sum test; experiment was performed once). (G) At 35 wk of
age, the drinking water of HBUS mice was supplemented with
antibiotics, whereas control mice were maintained on regular
water. After 5 wk of antibiotic treatment, HBUS mice were
checked for the presence of SPs by gross and histological analysis. Graph represents SP size of HBUS mice treated with antibiotics for 5 wk starting at 35 wk of age (n = 10) compared with
age-matched HBUS controls on water (n = 17). *, P < 0.05
(Wilcoxon rank sum test; experiment was performed once).
Bars: (D, F, and G) 5 mm; (E) 250 m. Error bars indicate SEM.
As expected, none of the mice on continued antibiotic treat

ment developed SPs. In contrast, 67% of the mice that were
first treated with antibiotics and subsequently gavaged with
stools obtained from untreated HBUS mice developed SPs
(n = 8; Fig. 2 F).
Finally, we investigated whether continued exposure to
the gut microbiota is required for maintenance of SPs. We
460
treated a group of animals at 35 wk of age with antibiotics for

5 wk (Fig. 2 G). At 40 wk, the size of SPs in HBUS mice sub
jected to antibiotic treatment (10.7 4.6 mm2; n = 10) was
significantly smaller compared with untreated HBUS mice
(52 11 mm2; n = 17). Collectively, our data suggest that the
cecal microbiota is essential for both the development and
maintenance of SPs in HBUS mice.
Ar ticle
Figure 3. Biota modification through rederivation prevents SP formation in HBUS mice. Two HBUS colonies were
established; one was maintained by interbreeding with C57BL/6
mice obtained from the Jackson Laboratory and developed SPs
(dashed line); the second colony was generated through embryo
transfer using Swiss Webster mothers freshly obtained from
Taconic (rederived). (AC) HBUS mice were examined grossly (A)
and histologically (B and C) for the presence of SPs. Shown is the
SP incidence of regular HBUS mice (n = 16, n = 10, and n = 12
at 12, 17, and 25 wk of age, respectively) and rederived HBUS
mice (n = 7, n = 9, and n = 10 at 12, 17, and 25 wk of age,
respectively) as determined by gross and histological analysis.
*, P < 0.05; **, P < 0.01 (Fishers exact test; experiment was performed once). Inset shows a higher-magnification image of the
area in C indicated by the asterisk. (D) Weighted UniFrac analysis
of Taconic mothers (n = 8) compared with HBUS mice that develop SPs (n = 20) and rederived HBUS mice (n = 13; Adonis
test; experiment was performed once). (E) diversity of Taconic
mothers, rederived HBUS mice, and HBUS mice with SPs (pairwise Wilcoxon rank sum test; experiment was performed once).
Bars: (A) 5 mm; (C) 500 m. Error bars indicate SEM.
A specific cecal microbiota is necessary

for development of SPs in HBUS mice
Treatment with broad-spectrum antibiotics leads to a marked
reduction in microbial diversity and biomass, and long-term
treatment promotes mucosal imbalance (Willing et al., 2011),
which all could affect disease susceptibility. To exclude non
specific effects of antibiotic treatment on SP development
and to circumvent the massive changes associated with germfree rederivation, we transferred one-cell embryos into pseu
dopregnant foster mothers of a different strain to promote
adoption of a different microbiota. It is well established that
the newborn gut microbiota is shaped by the maternal micro
biota (Clemente et al., 2012).We reasoned that we could alter
the microbiota of HBUS mice by transferring HBUS em
bryos (generated from and interbred with C57BL/6 mice
obtained from the Jackson Laboratory) into Swiss Webster
females obtained from Taconic.The results from this rederiva
tion experiment were striking: HBUS mice rederived using
Taconic mothers and reared separately for 12 (n = 7) or 17 wk
(n = 8) did not develop SPs as determined by gross (Fig. 3 A) and
histological analysis (Fig. 3, B and C). Interestingly, by 25 wk
of age, we also observed SPs in Taconic-derived mice (n = 6),
JEM Vol. 211, No. 3
although at a significantly reduced frequency (17 vs. 63%;

Fig. 3 B) and size compared with Jackson-derived HBUS mice
(4.3 4 vs. 57 8 mm2).
Analysis of the cecal microbiota of the rederived mice by
16S sequencing and weighted UniFrac distances showed that
the microbiota of rederived HBUS mice was significantly dif
ferent from both the Jackson-derived HBUS mice (Adonis:
P < 0.001) and their Taconic mothers (Adonis: P < 0.001;
Fig. 3 D), suggesting that rederived HBUS mice had a biome
that contained, as expected, elements of the Taconic maternal
biota as well as the local environment. In contrast to antibiotic
treatment (Fig. 2 C), rederivation did not significantly affect
microbiome diversity (P > 0.05; Fig. 3 E). Collectively, our
results demonstrate that the presence of a host-specific micro
bial environment is key to the development of cecal SPs in
HBUS mice.
Identification of a subset of bacteria
enriched in cecal mucosa of SPs
Next, we sought to more narrowly define which bacteria
contributed to polyp formation. To do so, we compared the
cecal mucosaassociated microbiota of affected HBUS mice
461
Figure 4. Biome analysis of SPs. (A) Relative abundance of phyla present in HBUS mice with SPs (n = 20), WT littermates (n = 7), rederived HBUS
mice (Red; n = 13), and HBUS mice treated with antibiotics (Abx; n = 11). Data shown represent the most abundant phyla, whereas low abundant and
unclassified OTUs were grouped in Other. (B, top) Pearson hierarchical clustering of the abundance profiles of all 703 OTUs after filtering. Phyla
(P) are colored according to the legend in A. Mice were clustered (Spearman) by cage (color coded, top) and sample group (colored according to the
462
Ar ticle
with those of unaffected rederived HBUS mice, antibiotictreated HBUS mice, and cohoused WT control littermates.
We defined operational taxonomic units (OTUs) based on
sequence similarity of the 16S rRNA fragments with Green
genes reference sequences (McDonald et al., 2012). This re
sulted in the identification of 704 unique OTUs across all
samples after filtering as described in Materials and methods
(Table S3). Hierarchical clustering (Fig. 4 B) and weighted
UniFrac distances (Fig. 3 D) of OTU abundance profiles
separated the microbiota of the individual sample groups,
whereas WT mice clustered with cohoused HBUS mice
that developed SPs (Fig. 4 B). At the phylum level, we found
a distinct expansion of Verrucomicrobia and a decrease in Deferribacteres in SPs compared with rederived HBUS mice
(Fig. 4 A). Closer inspection revealed these differences to be
caused by the increased abundance of Akkermansia muciniphila (Greengenes ID: 4306262) and a decrease of Mucispirillum
schaedleri (Greengenes ID: 1136443). Although enriched in
SPs, A. muciniphila is one of the few species resistant to longterm treatment of HBUS mice with broad-spectrum antibi
otics (Fig. 4 D, arrowhead), which makes it unlikely to be
important for SP development.
To determine which OTUs were specially associated
with SPs, we compared the relative abundance of OTUs in
SPs with cohoused WT, rederived, and antibiotic-treated
HBUS mice by ANOVA (Q < 0.05) and determined differ
ences between groups by a Tukey post-hoc test (P < 0.05,
1.6 > fold > 1.6; Table S4). Cluster analysis led to the
identification of four distinct clusters (C1C4; Fig. 4 D).
OTUs present in clusters C1, C3, and C4 were also present
in rederived or antibiotic-treated HBUS mice, i.e., HBUS
mice that did not develop SPs. The second cluster (C2) rep
resented 44 OTUs that were specifically enriched in SPs
(Fig. 3 B). Of these, only 15 OTUs were found in >75% of
all SP samples (Fig. 3 C).
Bacterial infiltration SPs and decreased barrier function in SPs
Various bacteria have tissue-invasive properties (PizarroCerd and Cossart, 2006; Sartor, 2008). To examine whether
bacteria were present within SPs, we performed in situ hy
bridization with a eubacterial probe. As expected, probe sig
nal was observed throughout the cecal lumen. In the lamina
propria of SPs we observed the presence of bacteria, but not
in unaffected surrounding cecal tissue. The invasive bacteria
were found in close proximity to infiltrating neutrophils
(Fig. 5 A). The majority (7/15) of OTUs enriched in HBUS
SPs belonged to the order of Clostridiales. To test whether

the invasive bacteria corresponded to these OTUs, we per
formed in situ hybridization with a Clostridium coccoides
Eubacterium rectale probe (Clostridium cluster XIVa and XIVb,
pb-00963) that targets all the Clostridiales that were in
creased in HBUS SPs compared with the rederived HBUS
mice (Fig. 5 B). Analysis of HBUS SPs and surrounding tis
sue revealed positive signal for this probe within the lamina
propria of eight out eight tested HBUS polyps (Fig. 4 B). In
antibiotic-treated HBUS mice, as expected, no signal was
observed (Fig. 6). To examine whether Clostridiales would
be of relevance to the formation of SPs in HBUS mice, we
treated HBUS mice for 9 or 29 wk with 0.5 mg/ml vanco
mycin in the drinking water and assessed the presence of
SPs after 15 or 35 wk. Treatment with vancomycin targets
gram-positive bacteria and shifts the composition of the
microbiota, particularly the family Clostridiales/Lachnospi
raceae, without affecting the total number of cecal bacteria
(Sekirov et al., 2008; Ubeda et al., 2010; Willing et al.,
2011). Vancomycin-treated HBUS mice showed a signifi
cant decrease in incidence (11%, P = 0.002, n = 9) and size
of SPs (Fig. 5 C).
Next, we examined whether bacterial invasion was asso
ciated with a decreased barrier function. We observed an
apical loss of adherens junction proteins E-cadherin and
-catenin in SPs of HBUS mice compared with unaffected
surrounding cecal tissue (Fig. 4, C and D). Besides the 54
genes associated with the GO term biological adhesion,
such as Claudin-4, Claudin-23, and Cadherin-1 (Fig. 1), we
also found altered expression of tight junction protein Claudin-2, but not Claudin-1, in SPs compared with unaffected
surrounding cecal tissue (Fig. 4 E). To examine whether
changes in junction molecules had functional consequences,
we treated ceca obtained from HBUS mice with sulfoNHS-biotin. Sulfo-NHS-biotin can be used to examine
permeability of tight junction proteins ex vivo (Tamura et al.,
2008; Ding et al., 2011; Lei et al., 2012). Sulfo-NHS-biotin
(molecular weight 443.43) is a membrane- and tight junc
tionimpermeable molecule that efficiently biotinylates pri
mary amine-containing macromolecules such as proteins.
Ceca from HBUS mice were dissected and treated with
sulfo-NHS-biotin solution. Sulfo-NHS-biotin signal was
only observed in SPs of HBUS mice and not in adjacent un
affected cecal tissue (n = 5; Fig. 5 G). These results suggest
that bacterial invasion of SPs is facilitated by a decreased in
testinal epithelial barrier function.
legend). (bottom) Number of OTUs above background in each sample (OTU count) and fraction of total sample reads accounted for by the set of 703
filtered OTUs (read fraction). (D) Pearson hierarchical clustering identified four major clusters (C1C4) in the abundance profiles of 106 OTUs that
were significantly different between HBUS mice with SPs and rederived HBUS mice or WT controls (ANOVA Q < 0.05 [FDR]; Tukey P < 0.05; fold > 1.6).
OTUs significantly enriched in SPs compared with rederived (R), WT (W), or antibiotics-treated (A) mice are shaded blue (left). Phyla (P) are colored
according to the legend in A. OTU abundance is expressed as the log2-normalized read count in each sample. The OTU corresponding to A. muciniphila
is indicated (arrowhead). (E) 15 OTUs from C2 that were present in >75% of SPs. OTUs are ranked by abundance (vertical axis) and according to
abundance within each dataset (horizontal axis). OTU are annotated with Greengenes ID, color coded phyla (P) annotations (according to A), taxonomic family, and genus assignments. Each bar (A and C) or column (B, D, and E) represents a different mouse.
JEM Vol. 211, No. 3
463
Figure 5. Bacterial infiltration SPs

and decreased barrier function in SPs.
(A and B) In situ hybridization with a eubacterial
probe (A) or Clostridium cluster XIVa and XIVb
(pb-00963; B) on frozen sections obtained from
HBUS mice. Shown is a representative image
obtained from two independent experiments
of surrounding cecal tissue (Surr.) and an SP
(middle; higher magnification of on right;
n = 6). S100A9-positive cells are indicated by
asterisks. Arrowheads indicate bacteria that
invaded the lamina propria (A) or bacteria
recognized by pb-00963 that invaded the
lamina propria (B). (C) At 35 wk of age, the
drinking water of HBUS mice was supplemented with 1 mg/ml vancomycin, whereas
control mice were maintained on regular
water. After 4 wk of treatment, HBUS mice
were checked for the presence of SPs by gross
and histological analysis. Shown is SP size of
HBUS mice treated with vancomycin for 4 wk
starting at 35 wk of age (n = 9) compared
with age-matched HBUS controls on water
(n = 17). ***, P < 0.001 (Wilcoxon rank sum
test; experiment was performed once).
(D and E) Immunofluorescent analysis of the
adherens junction protein E-cadherin (D) and its
binding partner -catenin (E) in surrounding
cecal tissue (left) and an SP (right). Insets show
a higher-magnification image of . Arrowheads
indicate the apical adherens junction. Representative images of three independent experiments are shown (n = 5). (F) Expression of
Claudin-2 and Claudin-1 mRNA in SPs of HBUS
mice compared with surrounding tissue
(n = 5/group). **, P < 0.01 (Wilcoxon rank
sum test, experiment was performed once).
(G) Immunofluorescent analysis of proteins
labeled with sulfo-NHS-biotin in the ceca of
HBUS mice. Shown is a representative image
of two independent experiments of biotin
signal in cecal surrounding tissue and SPs of
HBUS mice. The arrowhead indicates biotin
signal in the lamina propria (inset shows
higher-magnification image of ). Immunofluorescent sections were counterstained with
DAPI and pan-Keratin (A, B, D, E, and G). Bar
graphs represent combined data of two independent experiments (n = 5). Bars: (A [left and
middle], B [left], D, and E) 50 m; (A, right)
2 m; (B, right) 10 m; (G) 100 m; (G, inset)
25 m. Error bars indicate SEM.
The inflammatory response in SPs

is consistent with bacterial invasion
The combination of bacterial invasion and the increased ex
pression of inflammatory and antimicrobial defense genes in
SPs prompted us to further characterize the immune response
in SPs of HBUS mice. Therefore, we enriched for CD45+
464
cells from SPs and surrounding unaffected cecal tissue and

analyzed gene expression profiles using the Illumina BeadAr
ray system. We found significant changes in expression of 153
genes in SPs compared with unaffected cecal tissue (Q <
0.05, 1.5 > fold > 1.5; Fig. 7 A and Table S5), consistent
with a signature of activated leukocytes, as well as response to
Ar ticle
bacterium (Fig. 7 B and Table S6). Analysis of cytokine and

chemokine expression by BeadArray and quantitative PCR
(qPCR) showed that the proinflammatory cytokines Il1-,
Il1-, and Tnf and the chemokines Ccl1, Ccl2, Ccl17, Cxcl2,
and Cxcl16 were significantly up-regulated in SPs compared
with unaffected cecal tissue (n = 5; Fig. 5 C).
Next we examined the cecal leukocyte composition of
WT and HBUS mice by FACS analysis. No differences in the
relative abundance of T cells, DCs, or macrophage subsets were
observed between SPs and unaffected surrounding tissue or
WT ceca. The most enriched GO overview term was associ
ated with T cell activation (Fig. 7 B). Detailed analysis of the
T cell subsets showed no change in the frequency of Th17
cells, but there was an increase in the frequency of IL-17
producing TCR- cells in SPs compared with unaffected
surrounding tissue (Fig. 5, D and E).
Among the most pronounced induced genes were S100a8,
S100a9, and Lgals2, which are expressed by neutrophils (KehlFie et al., 2011). To further evaluate this, we examined the
CD11b/c+/MHCII myeloid subsets that consist of neutro
phils, eosinophils, and monocytes. No significant differences
were observed in the Ly6G subset composed of eosinophils
and monocytes (Fig. 7, F and G); however, the number of
Ly6G and Ly6C double-positive cells was increased in SPs
(Fig. 8 A). Further analysis identified these cells as Ly6C/G+
and Gr-1High/Siglec-F neutrophils (Fig. 8 B), consistent with
the increased expression of S100a9 observed in the CD45
BeadArray analysis (Fig. 7 A).To further characterize the neu
trophil infiltrate, we stained cecal sections of HBUS mice
with an antibody against S100A9, which stains neutrophils.
Strong staining was observed close to the lumen in SPs, but
not in unaffected cecal surrounding tissue (Fig. 8 C). To de
termine whether neutrophils play a role in the pathology of
SPs, we subjected 35-wk-old HBUS mice to antiLy-6G
(1A8) treatment and examined the SP size after 30 d. Treat
ment with Ly-6Gspecific (1A8) antibodies can be used to
deplete neutrophils in mice (Daley et al., 2008). Treatment
with 1A8 antibody led to a significant decrease in the number
of S100A9+ neutrophils in SPs (10-fold) that was accompa
nied by a significant decrease in SP size (2.5-fold). These re
sults suggest that neutrophils are functionally relevant for the
growth of SPs in HBUS mice.
Inflammation can affect the expression of MMP-3, a ma
trix metalloproteinase shown to cleave HB-EGF and thereby
promote autocrine/paracrine EGFR signaling (Suzuki et al.,
1997).We therefore examined MMP-3 mRNA expression in
SPs and compared it with the nonaffected surrounding tissue.
We observed an increased mRNA expression of MMP-3 in
SPs. Immunostaining of SPs with an antiMMP-3 antibody
showed an increased number of MMP-3positive cells in the
lamina propria (>50 per 10 field) compared with unaffected
cecal tissue where MMP-3positive cells were only found in
the submucosa (<5 per 10 field; Fig. 8 E).
In summary, the presence of bacteria in the lamina pro
pria of SPs triggered an inflammatory response that included
expression of several cytokines, chemokines, and expression
JEM Vol. 211, No. 3
Figure 6. Clostridial probe signal in antibiotic-treated HBUS mice.

Representative image of an experiment performed once showing in situ
hybridization with a eubacterial probe (left) or Clostridium cluster XIVa
and XIVb (pb-00963) on frozen sections obtained from HBUS mice treated
with antibiotics (n = 3). Bars, 50 m.
of an HB-EGFprocessing metalloproteinase. These find

ings mechanistically link microbe-induced inflammation,
HB-EGF/EGFR signaling, and development of SPs in
HBUS mice.
DISCUSSION
Intestinal SPs occur at specific locations in the gut in a sub
type-specific manner (Huang et al., 2004; Noffsinger, 2009).
In this study, we show that the development of SPs in the
cecum of HBUS mice (Bongers et al., 2012) depends on the
presence of a host-specific microbiota. Alteration of the cecal
microbiota through antibiotic treatment or embryo transfer
mediated rederivation significantly attenuated the formation
of SPs in HBUS mice.To our knowledge, this study shows for
the first time that relatively minor changes in the microbiota
can dramatically affect the formation of neoplasms.
The GI tract is home to a complex set of microbial com
munities that differ between the various regions (Sartor,
2008). Longitudinal sampling of the GI tract has shown clear
differences in the bacterial composition of samples from the
mouth, stomach, duodenum, colon, and stool (Stearns et al.,
2011). Proximal and distal colon samples differ both in their
microbiota composition as well as in abundance of OTUs
(Wang et al., 2010). Overall, nutrient availability, pH, bile salts,
or other host factors can regulate the composition of the
microbiota in different regions of the GI tract (Merritt and
Donaldson, 2009; von Rosenvinge et al., 2013). Here, we
show that the exclusive penetrance of the polyp phenotype in
the cecum is related to bacteria living at this particular loca
tion. Antibiotic treatment completely prevents development
of SPs, and conversely, reconstitution of the flora using stools
from mice with SPs promoted SP development. Furthermore,
treatment of animals with well-developed SPs with antibiot
ics reduced polyp size, indicating that the presence of cecum
bacterial communities was important for SP maintenance and
growth. Importantly, our results indicate that there is a re
quirement for a host-specific cecum flora for development of
SPs. We show here that a relatively minor change in the cecal
microbiome markedly affected the incidence of SPs. Transfer
of HBUS embryos into Taconic foster mothers promoted
465
Figure 7. Marked inflammatory changes in SPs. (A) Leukocytes (CD45+ cells) isolated from HBUS SPs and surrounding (Surr.) tissue were analyzed by
Illumina BeadArray/limma. Quantile-normalized expression values were analyzed using a paired design (n = 3/group) and filtered for Q < 0.05 and 1.5 >
fold change > 1.5. Z scorenormalized data were subjected to hierarchical clustering (left): red indicates increased and green indicates decreased expression in SPs compared with surrounding tissue. Plot of logFC (log fold change) versus mean expression (right) of all detected transcripts (gray) and significant genes (153 genes; black). (B) ClueGO analysis of significantly regulated genes in an Illumina BeadArray analysis of SPs of HBUS mice compared with
unaffected surrounding cecal tissue. Shown are GO overview terms selected by %Genes/Term in color (Q < 0.05; terms > 15 genes; kappa 0.5). (C) Cytokine and chemokine mRNA expression in tissue obtained from SPs compared with unaffected surrounding proximal cecal tissue (n = 5/group). *, P < 0.05;
**, P < 0.01 (Wilcoxon rank sum test; experiment was performed twice). R.E., relative expression. (D) Il-17 production by TCR- cells in SPs (n = 6) from
HBUS mice compared with WT (n = 4) and unaffected surrounding HBUS cecal tissue (n = 6); cells were gated on CD45+/CD4/TCR-+. *, P < 0.05 (Wilcoxon rank sum test; experiment was performed once). Shown are representative FACS plots and a summary scatter dot plot. (EG) FACS analysis in SPs
from HBUS mice (P; n = 3) compared with unaffected surrounding HBUS cecal tissue (S; n = 4) and WT cecal tissue (W; n = 4). For production of IL-17
and IL-22 by CD4/TCR-, cells were gated on CD45+/CD4+/TCR-+ (E). For DC subsets, cells were gated on CD45+/CD11b+/MHC-II+/Ly6C (F). For
monocytes and eosinophils as defined by Ly6C and CD24 expression, cells were gated on CD45+/CD11b+/MHC-II/Ly6G (G). Shown are representative
summary scatter dot plots. *, P < 0.05 (Wilcoxon rank sum test; experiment was performed once). Error bars indicate SEM.
partial adoption of a Taconic cecal microbiome by the off

spring, as verified by weighted UniFrac and OTU analysis.
These effects were not dominant because after 25 wk of
466
age we observed SPs in rederived HBUS mice, likely reflect

ing a progressive drift from the maternal microbiota to the
local environment.
Ar ticle
Figure 8. Marked neutrophil infiltration in SPs. (A and B) Relative number of neutrophils determined by the number of Ly6G/Ly6C double-positive
cells (A; gated on CD45+/CD11b+/MHC-II) and Gr-1positive, Siglec-Fnegative cells (B; gated on CD45+/CD11b+/MHC-II) in SPs (n = 3) of HBUS mice
compared with unaffected surrounding cecal tissue (Surr.; n = 4) and WT littermate controls (n = 4). Shown are representative FACS plots and a summary
scatter plot. *, P < 0.05 (Wilcoxon rank sum test; experiment was performed once). (C) Representative image of three independent experiments showing
JEM Vol. 211, No. 3
467
Detailed analysis of the SP-associated microbiome indi

cated that 15 OTUs, from the four orders Clostridiales, Bac
teroidales, RF32, and Desulfovibrionales, were consistently
different between SPs and unaffected rederived and antibiotic
HBUS mice. While association does not prove that these
OTUs are causative of SP formation, members of the order of
Clostridiales have been known to predominantly colonize the
cecum and proximal colon and promote the production of
MMPs (Honda and Littman, 2012). Similarly, toxin B pro
duced by Clostridium difficile has been show to promote
EGFR-mediated activation of ERK1/2 in human colono
cytes in an MMP- and TGF-dependent manner (Na et al.,
2005). In the lamina propria of SPs in HBUS mice, we ob
served increased MMP-3 expression, which has been shown
to cleave HB-EGF (Suzuki et al., 1997). This raises the possi
bility that Lachnospiraceae promotes expression of MMP-3
in the SPs that facilitates the cleavage of HB-EGF and sub
sequent activation of the EGFRMAPK pathway. Previously,
we have shown that this pathway plays a key role in the devel
opment of SPs in HBUS mice (Bongers et al., 2012). Alterna
tively, MMP-3 production and release could also be increased
as part of a general inflammatory response (Louis et al., 2000).
Among the OTUs that were enriched in SPs was the
genus Bilophila whose members have been associated with
bile resistance and are often recovered from appendicitis spec
imens (Finegold and Jousimies-Somer, 1997). Bilophila wadsworthia has recently been shown to flourish on diet-induced
taurine-conjugated bile acid and promote colitis in IL10/
mice. The mechanism by which B. wadsworthia induces colitis
is unknown but was proposed to involve an immune response
to tissue damage induced by the bacterial byproducts, such
as H2S or secondary bile acids (Devkota et al., 2012). Interest
ingly, most bile acids secreted by the host are absorbed in
the terminal ileum, whereas secondary bile acids produced
by bacteria increase in more distal sections of the intestine
(Ridlon et al., 2006; Hamilton et al., 2007). Therefore, a spe
cific niche of cecal bile acidtransforming bacteria could ex
plain the specific cecal localization of SPs.
The bacterial invasion we observed in SPs could be caused
by properties encoded by the bacteria or specific interactions
with the genetic alterations in the host; e.g., EGFR signaling
can be used by infectious agents to increase their survival and
promote invasion (Galn et al., 1992; Fiske et al., 2009). HBEGF signaling has also been associated with changes in ex
pression of tight junction proteins such as Claudin-2 (Singh
et al., 2007), which we also found in this study. Currently, it is
unclear whether apical loss of the adherens junctionassociated
proteins -catenin and E-cadherin we observed in SPs are
functionally related to the changes in Claudin-2 expression, as

shown for Claudin-7 (Lioni et al., 2007), or brought about by
changes in inflammation-associated cytokine expression
(Bruewer et al., 2003).
Previous studies have shown that bacteria can regulate cy
tokine production by specific leukocyte subsets. For example,
segmented filamentous bacteria and Clostridium species have
been shown to induce IL-17 production by Th17 and regula
tory T cells, respectively (Ivanov et al., 2009; Nagano et al.,
2012). In SPs of HBUS mice, we identified TCR- T cells
as the main source of the increased IL-17 production. IL-17
production by resident TCR- cells is critical for the
recruitment of neutrophils and for the clearance of E. coli
(Shibata et al., 2007).
Development of SPs in HBUS mice was associated with a
distinct neutrophilic infiltrate, potentially as a result of in
creased levels of CXCL2, e.g., induced by the TNF or IL-17
produced by TCR- cells. Neutrophils are the final effectors
of an inflammatory response with an important role in the
clearance of pathogens. There is mounting evidence that
tumor-associated neutrophils (TANs) play an important role
in tumor initiation and progression (Mantovani et al., 2011).
Depletion experiments indicate that neutrophils play a role in
the growth of SPs in HBUS mice.
It is clear that the pathway to formation of SPs at specific
locations in the gut is complex and involves both bacterial
and host factors, given that prevention of EGFR signaling
(Bongers et al., 2012) or alterations of the microbiota are suf
ficient to decrease the incidence of SPs. Based on our obser
vations, we propose that SPs in HBUS mice are likely initiated
by increased basal HB-EGF/EGFR signaling, leading to a de
creased barrier function by altering tight/adherens-associated
proteins. This in turn would facilitate the invasion of specific
bacterial species that trigger a host antibacterial response
characterized by the production of antimicrobial molecules
and an innate immune response. Subsequently, bacterial or in
flammatory signals promote an expression of MMP-3 that in
turn further facilitates HB-EGF/EGFR signaling, resulting in a
positive feedback loop that ultimately promotes growth of SPs.
In summary, we have identified a requirement for inter
action between host genetic alterations and host-specific micro
biota for development of SPs.The discovery of this mechanism
may help to explain the preferential localization of SPs in spe
cific locations in the human intestine (Noffsinger, 2009).
MATERIALS AND METHODS
Mice. VS28, HBGF, and HBUS mice were described previously (Bongers
et al., 2010, 2012). No statistical method was used to determine sample size,
and when applicable, mice were assigned to a treatment group using a simple
S100A9-positive cells in SPs and unaffected surrounding cecal tissue (n = 6). (D and E) At 35 wk of age, HBUS mice were treated with 0.4 mg anti-1A8 i.p.
every other day for 30 d. Shown are representative immunofluorescent images (left) and quantification (right) of the number of S100A9-positive cells in
SPs (D) and the size of SPs (E) in control (n = 11) and 1A8-treated (n = 9) HBUS mice. *, P < 0.05 (Wilcoxon rank sum test; experiment was performed
once). (F) Expression of Mmp3 mRNA by qPCR analysis and immunofluorescent staining (MMP-3) in SPs and surrounding tissue (n = 4/group). Inset
shows higher-magnification image of the area indicated by . *, P < 0.05 (Wilcoxon rank sum test; experiment was performed twice). Sections were counterstained with DAPI and pan-Keratin. Bars: (C [left and middle] and F) 100 m; (C, right) 10 m; (D) 250 m. Error bars indicate SEM.
468
Ar ticle
randomization (coin flip). Histological analysis was performed blinded to the

treatment. All experiments involving mice were performed in accordance
with the guidelines of the Animal Care and Use Committee of Mount Sinai
School of Medicine.
Antibiotic treatment. Mice were treated for the indicated time with
1 g/liter ampicillin, 1 g/liter neomycin, 1 g/liter metronidazole (all SigmaAldrich), and 0.5 g/liter vancomycin (Western Medical Supply) ad libitum in
the drinking water for the indicated time.
In vivo cell depletion. For depletion of Ly6G+-expressing neutrophils, antiLy6G mAb clone 1A8 (Bio X Cell) was injected at 0.4 mg i.p. every other day
for 30 d, as previously described (Daley et al., 2008; Garcia et al., 2010).
Pan-bacterial PCR amplification. Total genomic DNA was isolated from
tissue using the DNeasy Blood and Tissue kit (QIAGEN). qPCR analysis and
primer sequences were described previously (Hartman et al., 2009). Relative
quantity was calculated by the Ct method and normalized by the presence
of mouse ubiquitin.
RNA extraction and qPCR. Total RNA was extracted using the RNeasy
Mini kit (QIAGEN) according to the manufacturers protocol. RT was per
formed using 3 g total RNA. RT-PCR was conducted in triplicates using
SYBR green (Fermentas). Relative expression levels were calculated by the
Ct method and normalized by the presence of mouse ubiquitin. Primers
were designed using Primer Express 2.0 software (Applied Biosystems) or
Primer3Plus (Untergasser et al., 2007).
Rederivation. Specific pathogenfree mice were produced by in vitro fer
tilization using standard procedures. In brief, sperm from VS28 males and
oocytes from HBGF females were implanted into naturally ovulating pseu
dopregnant Swiss Webster female mice freshly obtained from Taconic.
Isolation of leukocytes. Intestines were cut into 2-cm pieces and washed
in ice-cold PBS.To release the intestinal epithelial cells, the pieces were incu
bated in PBS containing 1.3 mM EDTA for 30 min at 4C. After vigorous
shaking, intestinal epithelial cells were collected in the supernatant and used
to prepare RNA extracts using the RNeasy Mini kit (QIAGEN). The re
maining tissue was digested in DMEM containing 1 mg/ml Dispase II (Roche)
for 20 min at 37C, incubated with CD45 microbeads (Miltenyi Biotec), and
purified using MACS columns (Miltenyi Biotec) according to manufacturers
protocol. RNA was extracted from the CD45-enriched purified cells using
the RNeasy Mini kit.
Histology and immunofluorescence. Organs were dissected, fixed in
10% formalin, and processed for paraffin sectioning. 4-m sections were
dewaxed by immersion in xylene and hydrated by serial immersion in etha
nol and PBS. Antigen retrieval was performed by incubating sections in a
pressure cooker for 15 min in Target Retrieval Solution (DAKO). Sections
were washed with PBS (twice for 10 min), and blocking buffer (TBS con
taining 10% BSA and 0.3% Triton X-100) was added for 1 h. Sections were
incubated with primary antibody in blocking buffer overnight at 4C and
then incubated with Alexa Fluor 488, Alexa Fluor 594, or Alexa Fluor
647labeled secondary antibodies for 1 h. Sections were mounted with
Fluoromount-G (Beckman Coulter). Antibodies were obtained from R&D
Systems (REG-3), Abcam (pan-keratin, S100a9, and MMP-3), Santa Cruz
Biotechnology, Inc. (REG-3), or Cell Signaling Technology (-catenin
and E-cadherin).
Fluorescent in situ hybridization. Whole ceca were dissected and di
rectly incubated overnight in fixing solution (PBS containing 1.6% parafor
maldehyde and 20% sucrose), embedded in O.C.T. (Tissue-Tek), and snap
frozen in 2-methylbutylene on dry-ice. 10-m sections were washed with
ice-cold PBS (twice for 10 min), and blocking buffer (TBS containing 10%
BSA) was added for 30 min. Sections were incubated with 0.45 pmol/l
JEM Vol. 211, No. 3
eubacterial oligonucleotide probe ([AminoC6 + Alexa Fluor 594] 5-GCT

GCCTCCCGTAGGAGT-3; Operon) in prechilled hybridization buffer
(Sigma-Aldrich) overnight at 4C (Bates et al., 2006). Sections were counter
stained with 30 nM DAPI (Invitrogen) in PBS for 10 min, washed for 10 min
in ice-cold PBS, and mounted with Fluoromount-G. Immunofluorescent
imaging was performed on the same day.
Intraepithelial cellular barrier assay. The barrier function assay based on
sulfo-NHS-biotin was performed as described previously (Tamura et al.,
2008). In brief, sulfo-NHS-biotin (1 mg/ml PBS; Sigma-Aldrich) was ap
plied to dissected ceca of HBUS mice. After 30 min of incubation in a hu
midified chamber at room temperature, the ceca were washed three times
with PBS for 5 min, fixed overnight with 4% paraformaldehyde/20% sucrose
in PBS, and frozen in OCT. Sections were stained with Alexa Fluor 594
labeled streptavidin (Invitrogen) for 30 min and counterstained with 30 nM
DAPI (Invitrogen) in PBS for 10 min, washed for 10 min in ice-cold PBS,
and mounted with Fluoromount-G.
Fluorescent imaging. Immunofluorescent imaging was performed using
Cy3 HYQ, FITC HYQ, and a fluorescence microscope (E600; Nikon) with
Plan Apochromat objective lenses. Images were acquired using a digital cam
era (DXM1200F; Nikon) and Act-1 software version 2.63 (Nikon). Images
were composed in Photoshop CS6 (Adobe).
FACS analysis on cecal tissue. The cecum was cut into small pieces (<1 mm),
digested with 2.5 mg/ml Collagenase D and 1 mg/ml Dispase II for 1 h, and
purified using a Percoll gradient (44 and 66%). For cytokine analysis, the cells
were cultured for 3 h with 10 ng/ml PMA and 1 g/ml ionomycin and for
1.5 h with 10 g/ml Monensin. Otherwise, the cells were immediately resus
pended in Fc-Block (BD) for 20-min RT and subsequently stained with
surface markers for 30 min at 4C. For intracellular staining, cells were per
meabilized in permeabilization buffer (BD) and stained with intercellular anti
bodies diluted in permeabilization buffer for 45 min at 4C. DC and
macrophage subsets in SPs were analyzed by FACS analysis using the markers
CD45, CD11c, CD11b, MHCII, Ly6C, and CD103. FACS antibodies were
obtained from BD or eBioscience.
BeadArray analysis. BeadArray analyses were performed with 500 ng total
RNA using the TotalPrep RNA Amplification kit (Illumina) and Mouse
Ref-8 BeadChip kit 430 2.0 Expression BeadChip arrays (Illumina) with
n = 3 per group. All arrays in this study were subjected to variance-stabilizing
transformation (Lin et al., 2008), quantile normalization, and quality control
using lumi version 2.14.0 Bioconductor/R (Du et al., 2008). Fold changes
and statistical significance were determined using a paired sample analysis in
limma version 3.18.7 Bioconductor/R (Smyth, 2004) and corrected for mul
tiple testing (FDR/Benjamini and Hochberg). Probe sets were selected based
on 1.5 > fold change > 1.5, Q < 0.05. The expression values were plotted
with heatmap.2 (gplots version 2.11.0, CRAN/R).
DNA extraction, 16S rDNA amplification, and multiplex sequencing.
All mice used for 16S rDNA sequencing were housed in specific pathogen
free conditions that were free of Helicobacter hepaticus and H. bilis. Rederived
(n = 13) and antibiotic-treated HBUS mice (n = 11) were house in separate
cages, whereas WT control littermates (n = 7) were cohoused with HBUS
mice that developed SPs (n = 20). Each sample was obtained from a different
mouse. HBUS mice never developed more than one SP and were only found
at the cecalcolonic junction. Before resection, surface stool was removed
and SPs from HBUS mice or 4-mm2 piece of the cecalcolonic junction was
processed directly using the DNeasy Blood and Tissue kit (QIAGEN). Bacte
rial 16S rRNA genes were amplified using the primers as described in
Caporaso et al. (2012). Each sample was amplified in quadruplicate, combined,
and cleaned using the Agencourt AMPure XP beads (Beckman Coulter).
PCR reactions contained 0.5 M for each primer, 100 ng genomic DNA,
and Phusion High-Fidelity PCR Master Mix (New England Biolabs, Inc.).
Reactions were held at 98C for 30 s, proceeding to 35 cycles at 98C for 10 s,
469
50C for 30 s, and 72C for 30 s and a final extension of 10 min at 72C.
Cleaned amplicons were quantified using Picogreen dsDNA, and a compos
ite sample for sequencing was created by combining equimolar ratios of am
plicons from the individual samples, agarose gel purifying (QIAquick Gel
Extraction; QIAGEN), and mixing with 50% PhiX (Illumina). The sample
concentration was verified by qPCR and loaded on Illumina MiSeq or
HiSeq sequencer (members of all groups were run on both machines) for se
quencing as previously described (Caporaso et al., 2012).
Analysis of 16S rDNA sequences. The raw reads obtained from 150-bp
paired-end MiSeq or HiSeq runs were preprocessed through the QIIME
version 1.7.0 pipeline (Caporaso et al., 2010). After demultiplexing and base
quality trimming (q20), a reference-based OTU picking protocol was applied
and 97% OTUs were picked against the May 2013 Greengenes database
(McDonald et al., 2012; prefiltered at 97% identity) using UCLUST (Edgar,
2010). Reads were assigned to OTUs based on their best match to a Green
genes sequence (89% of the reads), and reads that did not match a Green
genes sequence at 97% or greater sequence identity were discarded. The
Greengenes taxonomy associated with the best match in Greengenes was as
signed to each OTU, and the Greengenes tree was used for phylogenetic
diversity calculations. The resulting OTU table was normalized by downsampling to the sample with the lowest counts (35,000) and then quality fil
tered, discarding OTUs with less than five counts across all samples or present
in less than nine samples (Bokulich et al., 2013). Filtering did not significantly
affect the read fraction for each group (Fig. 4 C). Finally, filtered OTU counts
were used to calculate Jackknifed Weighted UniFrac -diversity indices
(Lozupone et al., 2011) and diversity (Faith and Baker, 2006) at an even
depth of 35,000 sequences/sample and generate taxonomic plots. To evaluate
differences between groups, one-way ANOVA (CRAN/R version 3.0.2)
was performed on log2-transformed count data (all zeroes were set to 0.5 of
the smallest nonzero entry) assuming equal variance, OTUs with Q < 0.05
(FDR/Benjamini and Hochberg) were subjected to Tukey HSD post hoc test
(CRAN/R version 3.0.2), and significant differences among groups were
defined as P < 0.05 and 1.6 > fold change > 1.6. Abundance profiles were
hierarchically clustered using Spearman correlation as the distance metric,
and heat maps were generated using R.
RNA extraction from tissues. Approximately 200 mg of tissue sample was
placed in 1.5 ml RNAlater buffer (Ambion) and snap-frozen in liquid nitro
gen. RNA extraction was performed as described previously (Giannoukos
et al., 2012). In brief, at time of extraction, samples were centrifuged for
10 min at 16,000 g at room temperature. Pellets were resuspended in 150 l
lysis buffer (Tris/HCl, pH 8, 1 mM EDTA, 15 mg/ml Lysozyme [SigmaAldrich], and 15 l of 20 mg/ml proteinase K) and incubated at room tem
perature for 10 min with brief mixing every 2 min. After addition of 1.2 ml
QIAGEN RLT buffer containing 1% vol/vol -mercaptoethanol, 1 ml of
0.1-mm glass beads (BioSpec) was added, and samples were homogenized in
a FastPrep at setting 5 (four pulses of 20 s). Samples were kept on ice for
1 min between pulses. Lysates were then homogenized with a QIAshredder
spin column, and RNA was isolated using the AllPrep mini kit (QIAGEN),
according to the manufacturers protocols, which included an on-column di
gestion with DNase I. 25 g RNA from each tissue sample was processed
with the MICROBEnrich kit (Ambion), and 5 g of processed RNA was
further depleted of rRNAs using the Meta-Bacteria RiboZero rRNA re
moval kit (Epicentre).The final samples consisted of a mix of host and micro
bial mRNAs in a 2:1 ratio.
cDNA library construction and sequencing. rRNA-depleted RNA
was prepared for Illumina paired-end sequencing using the Next mRNA Li
brary Prep Master Mix Set for Illumina (New England Biolabs, Inc.); manu
facturers protocols were followed with the following modifications. RNA
was fragmented for 10 min and then purified with RNeasy MinElute spin
columns (QIAGEN). RT was performed with SuperScript III (Invitrogen).
Library preparation reactions were cleaned up using Agencourt AMPure XP
beads. Size selection was performed before ligation-mediated PCR using
470
Invitrogen E-Gel 2% with SYBR Safe staining. Excised gel fragments were
purified with the QIAQuick Gel Extraction kit (QIAGEN). Adapters and
primers were synthesized by IDT according to published Illumina sequences.
Enrichment PCR was performed with Kapa HiFi HotStart ReadMix. Prim
ers were used at a final concentration of 500 nM; cycling parameters were
as follows: 94C for 5 min, 15 cycles of 94C for 1 min, 62C for 30 s, 72C
for 45 s, and then 72C for 10 min. Libraries were quantified using the Bio
Analyzer DNA 1000 chips (Agilent Technologies), diluted to 12 pM, and
sequenced for 100 cycles (paired-end) on the HiSeq 2000 (Illumina) using
standard methods.
Mouse transcriptome analysis. RNA-Seq data from HBUS tissue sam
ples were mapped to the mouse reference genome and transcriptome
(GRCm38 and the Jul 2012 ENSEMBL gene build, respectively) using
Tophat (Trapnell et al., 2009). Gene-level sequence counts were extracted for
all annotated protein-coding genes using htseq-count by taking the strict in
tersection between reads and the transcript models associated with each gene.
Raw count data were filtered to remove low expressed genes with less than
five counts in any sample. Remaining data (12,697 genes) were normalized
with trimmed mean of M (TMM) normalization and analyzed for differen
tially expressed genes using the Bioconductor EdgeR package version 2.11
Bioconductor/R (Robinson et al., 2010). To take into account the experi
mental design where paired unaffected cecal tissue and polyp tissue samples
were isolated for each mouse, we fitted an additive generalized linear model
that incorporated mouse + tissue effects to adjust for any baseline differences
between the mice. Statistically significant differentially expressed genes be
tween polyp and surrounding cecal (normal) tissues (Q < 0.05) were selected
in gene-wise log-likelihood ratio tests that were corrected for multiple test
ing by Benjamini and Hochberg FDR. GO analysis on significantly regulated
genes was performed using ClueGO (Bindea et al., 2009) for GO terms
containing at least 25 genes, redundancy was reduced using GO term fusion,
connections were based on kappa 0.5, the leading overview GO term was
selected based on %Genes/Term, and GO terms with Q < 0.05 (FDR) were
considered significantly enriched.
Accession numbers. Accession numbers for all primary array and se
quencing data are available from the NCBI under BioProject accession no.
PRJNA207540 and GEO accession no. GSE47736.
Statistical analysis. Statistical analysis for BeadArray, Microbiome, and
RNA-Seq was performed as described above. For all other experiments, dif
ferences among means were evaluated by a 2 2 contingency table using
Fishers exact test (Prism version 5; GraphPad Software), a two-tailed Wil
coxon rank sum test, or pairwise Wilcoxon rank sum test (CRAN/R version
3.0.2); P < 0.05 was considered significant. For multiple comparisons, p-values
were adjusted using Benjamini and Hochberg (FDR). No samples were ex
cluded from analysis. All results shown represent mean SEM.
Online supplemental material. Table S1, included as a separate PDF
file, shows differentially regulated genes by RNA-Seq analysis in tissue
isolated from HBUS polyps compared with unaffected proximal cecum,
as show in the heat map. Table S2, included as a separate PDF file, shows
ClueGO analysis of the RNA-Seq analysis in tissue isolated from HBUS
polyps compared with unaffected proximal cecum. Table S3, included as a
separate PDF file, shows a subsampled and filtered OTU table of Taconic
moms, rederived HBUS mice, and HBUS and WT mice obtained through
interbreeding with mice obtained from the Jackson Laboratory. Table S4,
included as a separate PDF file, shows statistical analysis of OTUs in Table S3.
Table S5, included as a separate PDF file, shows differentially regulated
genes by BeadArray in CD45+ cells isolated from HBUS polyps compared
with unaffected proximal cecum, as show in the heat map. Table S6, in
cluded as a separate PDF file, shows ClueGO analysis of the BeadArray
analysis in CD45+ cells isolated from HBUS polyps compared with unaf
fected proximal cecum. Online supplemental material is available at http://
www.jem.org/cgi/content/full/jem.20131587/DC1.
Ar ticle
We would like to thank the Genomics Core Facility of the Icahn Institute for
Genomics and Multiscale Biology for help with RNA-Seq, Jeremiah J. Faith for
helpful discussions, and Taciana Salviano and Alan Soto for experimental help.
We thank Jenny and Jon Steingart and Jenna and Paul Segal for a grant
supporting G. Bongers and the CAPES Foundation (Brazil) for a grant supporting
T.H. Geraldino. This work was supported by National Institutes of Health grants
1R01CA161373-01 and P01 DK072201 to S.A. Lira and in part by the Icahn School
of Medicine at Mount Sinai for providing scientific computing resources.
The authors declare no competing financial interests.
Author contributions: G. Bongers, H. van Bakel, and S.A. Lira conceived and designed
the experiments; G. Bongers, M.E. Pacer, L. Chen, Z. He, D. Hashimoto, T.H. Geraldino,
K.A. Kelley, and G.C. Furtado performed the experiments; G. Bongers, L. Chen, Z. He,
D. Hashimoto, J.C. Clemente, and H. van Bakel analyzed the data; J. Ochando, J.C.
Clemente, M. Merad, and H. van Bakel contributed reagents/materials/analysis tools;
and G. Bongers, J.C. Clemente, H. van Bakel, and S.A. Lira wrote the manuscript.
Submitted: 26 July 2013
Accepted: 29 January 2014
REFERENCES
Arthur, J.C., E. Perez-Chanona, M. Mhlbauer, S. Tomkovich, J.M. Uronis,

T.J. Fan, B.J. Campbell, T. Abujamel, B. Dogan, A.B. Rogers, et al. 2012.
Intestinal inflammation targets cancer-inducing activity of the microbiota.
Science. 338:120123. http://dx.doi.org/10.1126/science.1224820
Bates, J.M., E. Mittge, J. Kuhlman, K.N. Baden, S.E. Cheesman, and K.
Guillemin. 2006. Distinct signals from the microbiota promote different
aspects of zebrafish gut differentiation. Dev. Biol. 297:374386. http://
dx.doi.org/10.1016/j.ydbio.2006.05.006
Bindea, G., B. Mlecnik, H. Hackl, P. Charoentong, M. Tosolini, A. Kirilovsky,
W.H. Fridman, F. Pags, Z. Trajanoski, and J. Galon. 2009. ClueGO:
A Cytoscape plug-in to decipher functionally grouped gene ontology and
pathway annotation networks. Bioinformatics. 25:10911093. http://dx
.doi.org/10.1093/bioinformatics/btp101
Bokulich, N.A., S. Subramanian, J.J. Faith, D. Gevers, J.I. Gordon, R. Knight,
D.A. Mills, and J.G. Caporaso. 2013. Quality-filtering vastly improves
diversity estimates from Illumina amplicon sequencing. Nat. Methods.
10:5759. http://dx.doi.org/10.1038/nmeth.2276
Bongers, G., D. Maussang, L.R. Muniz, V.M. Noriega, A. Fraile-Ramos, N.
Barker, F. Marchesi, N.Thirunarayanan, H.F.Vischer, L. Qin, et al. 2010.The
cytomegalovirus-encoded chemokine receptor US28 promotes intesti
nal neoplasia in transgenic mice. J. Clin. Invest. 120:39693978. http://dx
.doi.org/10.1172/JCI42563
Bongers, G., L.R. Muniz, M.E. Pacer, A.C. Iuga, N. Thirunarayanan, E. Slinger,
M.J. Smit, E.P. Reddy, L. Mayer, G.C. Furtado, et al. 2012. A role for the
epidermal growth factor receptor signaling in development of intesti
nal serrated polyps in mice and humans. Gastroenterology. 143:730740.
http://dx.doi.org/10.1053/j.gastro.2012.05.034
Bruewer, M., A. Luegering, T. Kucharzik, C.A. Parkos, J.L. Madara, A.M.
Hopkins, and A. Nusrat. 2003. Proinflammatory cytokines disrupt epithe
lial barrier function by apoptosis-independent mechanisms. J. Immunol.
171:61646172.
Caporaso, J.G., J. Kuczynski, J. Stombaugh, K. Bittinger, F.D. Bushman, E.K.
Costello, N. Fierer, A.G. Pea, J.K. Goodrich, J.I. Gordon, et al. 2010.
QIIME allows analysis of high-throughput community sequencing data.
Nat. Methods. 7:335336. http://dx.doi.org/10.1038/nmeth.f.303
Caporaso, J.G., C.L. Lauber,W.A.Walters, D. Berg-Lyons, J. Huntley, N. Fierer,
S.M.Owens,J.Betley,L.Fraser,M.Bauer,et al.2012.Ultra-high-throughput
microbial community analysis on the Illumina HiSeq and MiSeq plat
forms. ISME J. 6:16211624. http://dx.doi.org/10.1038/ismej.2012.8
Cho, I., and M.J. Blaser. 2012. The human microbiome: At the interface of
health and disease. Nat. Rev. Genet. 13:260270.
Clemente, J.C., L.K. Ursell, L.W. Parfrey, and R. Knight. 2012. The impact of
the gut microbiota on human health: an integrative view. Cell. 148:1258
1270. http://dx.doi.org/10.1016/j.cell.2012.01.035
Daley, J.M., A.A.Thomay, M.D. Connolly, J.S. Reichner, and J.E. Albina. 2008.
Use of Ly6G-specific monoclonal antibody to deplete neutrophils in
mice. J. Leukoc. Biol. 83:6470. http://dx.doi.org/10.1189/jlb.0407247
JEM Vol. 211, No. 3
Dapito, D.H., A. Mencin, G.-Y. Gwak, J.-P. Pradere, M.-K. Jang, I. Mederacke,
J.M. Caviglia, H. Khiabanian, A. Adeyemi, R. Bataller, et al. 2012. Promotion
of hepatocellular carcinoma by the intestinal microbiota and TLR4.
Cancer Cell. 21:504516. http://dx.doi.org/10.1016/j.ccr.2012.02.007
Dessein, R., M. Gironella, C. Vignal, L. Peyrin-Biroulet, H. Sokol, T. Secher,
S. Lacas-Gervais, J.-J. Gratadoux, F. Lafont, J.-C. Dagorn, et al. 2009.
Toll-like receptor 2 is critical for induction of Reg3 beta expression and
intestinal clearance of Yersinia pseudotuberculosis. Gut. 58:771776. http://
dx.doi.org/10.1136/gut.2008.168443
Devkota, S., Y. Wang, M.W. Musch, V. Leone, H. Fehlner-Peach, A. Nadimpalli,
D.A. Antonopoulos, B. Jabri, and E.B. Chang. 2012. Dietary-fat-induced
taurocholic acid promotes pathobiont expansion and colitis in Il10/
mice. Nature. 487:104108.
Ding, L.,Y. Zhang, R. Tatum, and Y.H. Chen. 2011. Detection of tight junc
tion barrier function in vivo by biotin. Methods Mol. Biol. 762:91100.
http://dx.doi.org/10.1007/978-1-61779-185-7_7
Du, P., W.A. Kibbe, and S.M. Lin. 2008. lumi: a pipeline for processing
Illumina microarray. Bioinformatics. 24:15471548. http://dx.doi.org/10
.1093/bioinformatics/btn224
Edgar, R.C. 2010. Search and clustering orders of magnitude faster than
BLAST. Bioinformatics. 26:24602461. http://dx.doi.org/10.1093/
bioinformatics/btq461
Faith, D.P., and A.M. Baker. 2006. Phylogenetic diversity (PD) and biodi
versity conservation: Some bioinformatics challenges. Evol. Bioinform.
Online. 2:121128.
Finegold, S.M., and H. Jousimies-Somer. 1997. Recently described clinically
important anaerobic bacteria: Medical aspects. Clin. Infect. Dis. 25(s2,
Suppl 2):S88S93. http://dx.doi.org/10.1086/516237
Fiske, W.H., D. Threadgill, and R.J. Coffey. 2009. ERBBs in the gastrointesti
nal tract: Recent progress and new perspectives. Exp. Cell Res. 315:583
601. http://dx.doi.org/10.1016/j.yexcr.2008.10.043
Galn, J.E., J. Pace, and M.J. Hayman. 1992. Involvement of the epidermal
growth factor receptor in the invasion of cultured mammalian cells by
Salmonella typhimurium.Nature.357:588589.http://dx.doi.org/10.1038/
357588a0
Garcia, M.R., L. Ledgerwood, Y. Yang, J. Xu, G. Lal, B. Burrell, G. Ma, D.
Hashimoto, Y. Li, P. Boros, et al. 2010. Monocytic suppressive cells me
diate cardiovascular transplantation tolerance in mice. J. Clin. Invest.
120:24862496. http://dx.doi.org/10.1172/JCI41628
Giannoukos, G., D.M. Ciulla, K. Huang, B.J. Haas, J. Izard, J.Z. Levin, J.
Livny, A.M. Earl, D. Gevers, D.V. Ward, et al. 2012. Efficient and robust
RNA-seq process for cultured bacteria and complex community transcrip
tomes. Genome Biol. 13:r23. http://dx.doi.org/10.1186/gb-2012-13-3-r23
Hamilton, J.P., G. Xie, J.-P. Raufman, S. Hogan, T.L. Griffin, C.A. Packard,
D.A. Chatfield, L.R. Hagey, J.H. Steinbach, and A.F. Hofmann. 2007.
Human cecal bile acids: Concentration and spectrum. Am. J. Physiol.
Gastrointest. Liver Physiol. 293:G256G263. http://dx.doi.org/10.1152/
ajpgi.00027.2007
Hartman, A.L., D.M. Lough, D.K. Barupal, O. Fiehn, T. Fishbein, M. Zasloff,
and J.A. Eisen. 2009. Human gut microbiome adopts an alternative
state following small bowel transplantation. Proc. Natl. Acad. Sci. USA.
106:1718717192. http://dx.doi.org/10.1073/pnas.0904847106
Honda, K., and D.R. Littman. 2012.The microbiome in infectious disease and
inflammation. Annu. Rev. Immunol. 30:759795. http://dx.doi.org/10
.1146/annurev-immunol-020711-074937
Huang, C.S., M.J. Obrien, S. Yang, and F.A. Farraye. 2004. Hyperplastic
polyps, serrated adenomas, and the serrated polyp neoplasia pathway.
Am. J. Gastroenterol. 99:22422255. http://dx.doi.org/10.1111/j.15720241.2004.40131.x
Ivanov, I.I., K. Atarashi, N. Manel, E.L. Brodie, T. Shima, U. Karaoz, D. Wei,
K.C. Goldfarb, C.A. Santee, S.V. Lynch, et al. 2009. Induction of intes
tinal Th17 cells by segmented filamentous bacteria. Cell. 139:485498.
http://dx.doi.org/10.1016/j.cell.2009.09.033
Kehl-Fie, T.E., S. Chitayat, M.I. Hood, S. Damo, N. Restrepo, C. Garcia, K.A.
Munro, W.J. Chazin, and E.P. Skaar. 2011. Nutrient metal sequestration
by calprotectin inhibits bacterial superoxide defense, enhancing neutro
phil killing of Staphylococcus aureus. Cell Host Microbe. 10:158164. http://
dx.doi.org/10.1016/j.chom.2011.07.004
471
Lei, Z., T. Maeda, A. Tamura, T. Nakamura, Y. Yamazaki, H. Shiratori, K.

Yashiro, S. Tsukita, and H. Hamada. 2012. EpCAM contributes to for
mation of functional tight junction in the intestinal epithelium by re
cruiting claudin proteins. Dev. Biol. 371:136145. http://dx.doi.org/
10.1016/j.ydbio.2012.07.005
Lin, S.M., P. Du, W. Huber, and W.A. Kibbe. 2008. Model-based variancestabilizing transformation for Illumina microarray data. Nucleic Acids Res.
36:e11. http://dx.doi.org/10.1093/nar/gkm1075
Lioni, M., P. Brafford, C. Andl, A. Rustgi, W. El-Deiry, M. Herlyn, and
K.S.M. Smalley. 2007. Dysregulation of claudin-7 leads to loss of Ecadherin expression and the increased invasion of esophageal squamous
cell carcinoma cells. Am. J. Pathol. 170:709721. http://dx.doi.org/10
.2353/ajpath.2007.060343
Louis, E., C. Ribbens, A. Godon, D. Franchimont, D. De Groote, N. Hardy,
J. Boniver, J. Belaiche, and M. Malaise. 2000. Increased production of
matrix metalloproteinase-3 and tissue inhibitor of metalloproteinase-1
by inflamed mucosa in inflammatory bowel disease. Clin. Exp. Immunol.
120:241246. http://dx.doi.org/10.1046/j.1365-2249.2000.01227.x
Lozupone, C., M.E. Lladser, D. Knights, J. Stombaugh, and R. Knight. 2011.
UniFrac: An effective distance metric for microbial community compar
ison. ISME J. 5:169172. http://dx.doi.org/10.1038/ismej.2010.133
Mantovani, A., M.A. Cassatella, C. Costantini, and S. Jaillon. 2011. Neutrophils
in the activation and regulation of innate and adaptive immunity. Nat.
Rev. Immunol. 11:519531. http://dx.doi.org/10.1038/nri3024
McColl, K.E.L. 2010. Clinical practice. Helicobacter pylori infection. N. Engl. J.
Med. 362:15971604. http://dx.doi.org/10.1056/NEJMcp1001110
McDonald, D., M.N. Price, J. Goodrich, E.P. Nawrocki, T.Z. DeSantis, A.
Probst, G.L. Andersen, R. Knight, and P. Hugenholtz. 2012. An improved
Greengenes taxonomy with explicit ranks for ecological and evolutionary
analyses of bacteria and archaea. ISME J. 6:610618. http://dx.doi.org/10
.1038/ismej.2011.139
Merritt, M.E., and J.R. Donaldson. 2009. Effect of bile salts on the DNA and
membrane integrity of enteric bacteria. J. Med. Microbiol. 58:15331541.
http://dx.doi.org/10.1099/jmm.0.014092-0
Na, X., D. Zhao, H.W. Koon, H. Kim, J. Husmark, M.P. Moyer, C. Pothoulakis,
and J.T. LaMont. 2005. Clostridium difficile toxin B activates the EGF
receptor and the ERK/MAP kinase pathway in human colonocytes.
Gastroenterology. 128:10021011. http://dx.doi.org/10.1053/j.gastro.2005
.01.053
Nagano,Y., K. Itoh, and K. Honda. 2012. The induction of Treg cells by gutindigenous Clostridium. Curr. Opin. Immunol. 24:392397. http://dx.doi
.org/10.1016/j.coi.2012.05.007
Noffsinger, A.E. 2009. Serrated polyps and colorectal cancer: new pathway
to malignancy. Annu. Rev. Pathol. 4:343364. http://dx.doi.org/10.1146/
annurev.pathol.4.110807.092317
Pizarro-Cerd, J., and P. Cossart. 2006. Bacterial adhesion and entry into host
cells. Cell. 124:715727. http://dx.doi.org/10.1016/j.cell.2006.02.012
Plottel, C.S., and M.J. Blaser. 2011. Microbiome and malignancy. Cell Host
Microbe. 10:324335. http://dx.doi.org/10.1016/j.chom.2011.10.003
Ridlon, J.M., D.-J. Kang, and P.B. Hylemon. 2006. Bile salt biotransforma
tions by human intestinal bacteria. J. Lipid Res. 47:241259. http://
dx.doi.org/10.1194/jlr.R500013-JLR200
Robinson, M.D., D.J. McCarthy, and G.K. Smyth. 2010. edgeR: A
Bioconductor package for differential expression analysis of digital gene
expression data. Bioinformatics. 26:139140. http://dx.doi.org/10.1093/
bioinformatics/btp616
Sartor, R.B. 2008. Microbial influences in inflammatory bowel diseases.
Gastroenterology. 134:577594. http://dx.doi.org/10.1053/j.gastro.2007
.11.059
472
Schwabe, R.F., and C. Jobin. 2013. The microbiome and cancer. Nat. Rev.
Cancer. 13:800812. http://dx.doi.org/10.1038/nrc3610
Sekirov, I., N.M. Tam, M. Jogova, M.L. Robertson, Y. Li, C. Lupp, and B.B.
Finlay. 2008. Antibiotic-induced perturbations of the intestinal micro
biota alter host susceptibility to enteric infection. Infect. Immun. 76:4726
4736. http://dx.doi.org/10.1128/IAI.00319-08
Shibata, K., H.Yamada, H. Hara, K. Kishihara, and Y.Yoshikai. 2007. Resident
V1+ T cells control early infiltration of neutrophils after Escherichia
coli infection via IL-17 production. J. Immunol. 178:44664472.
Singh, A.B., K. Sugimoto, P. Dhawan, and R.C. Harris. 2007. Juxtacrine acti
vation of EGFR regulates claudin expression and increases transepithelial
resistance. Am. J. Physiol. Cell Physiol. 293:C1660C1668. http://dx.doi
.org/10.1152/ajpcell.00274.2007
Smyth, G.K. 2004. Linear models and empirical bayes methods for assessing
differential expression in microarray experiments. Stat. Appl. Genet. Mol.
Biol. 3:e3.
Stearns, J.C., M.D.J. Lynch, D.B. Senadheera, H.C. Tenenbaum, M.B.
Goldberg, D.G. Cvitkovitch, K. Croitoru, G. Moreno-Hagelsieb, and J.D.
Neufeld. 2011. Bacterial biogeography of the human digestive tract. Sci
Rep. 1:170. http://dx.doi.org/10.1038/srep00170
Suzuki, M., G. Raab, M.A. Moses, C.A. Fernandez, and M. Klagsbrun. 1997.
Matrix metalloproteinase-3 releases active heparin-binding EGF-like
growth factor by cleavage at a specific juxtamembrane site. J. Biol. Chem.
272:3173031737. http://dx.doi.org/10.1074/jbc.272.50.31730
Tamura, A.,Y. Kitano, M. Hata,T. Katsuno, K. Moriwaki, H. Sasaki, H. Hayashi,
Y. Suzuki, T. Noda, M. Furuse, et al. 2008. Megaintestine in claudin15-deficient mice. Gastroenterology. 134:523534: e3. http://dx.doi.org/
10.1053/j.gastro.2007.11.040
Trapnell, C., L. Pachter, and S.L. Salzberg. 2009. TopHat: Discovering splice
junctions with RNA-Seq. Bioinformatics. 25:11051111. http://dx.doi
.org/10.1093/bioinformatics/btp120
Ubeda, C.,Y. Taur, R.R. Jenq, M.J. Equinda, T. Son, M. Samstein, A.Viale, N.D.
Socci, M.R. van den Brink, M. Kamboj, and E.G. Pamer. 2010.Vancomycinresistant Enterococcus domination of intestinal microbiota is enabled by an
tibiotic treatment in mice and precedes bloodstream invasion in humans.
J. Clin. Invest. 120:43324341. http://dx.doi.org/10.1172/JCI43918
Untergasser,A., H. Nijveen, X. Rao,T. Bisseling, R. Geurts, and J.A.M. Leunissen.
2007.Primer3Plus,an enhanced web interface to Primer3.NucleicAcids Res.
35(Web Server):W71W74. http://dx.doi.org/10.1093/nar/gkm306
van Ampting, M.T.J., L.M.P. Loonen, A.J. Schonewille, I. Konings, C. Vink,
J. Iovanna, M. Chamaillard, J. Dekker, R. van der Meer, J.M. Wells, and
I.M.J. Bovee-Oudenhoven. 2012. Intestinally secreted C-type lectin
Reg3b attenuates salmonellosis but not listeriosis in mice. Infect. Immun.
80:11151120. http://dx.doi.org/10.1128/IAI.06165-11
von Rosenvinge, E.C.,Y. Song, J.R. White, C. Maddox, T. Blanchard, and W.F.
Fricke. 2013. Immune status, antibiotic medication and pH are associated
with changes in the stomach fluid microbiota. ISME J. 7:13541366.
http://dx.doi.org/10.1038/ismej.2013.33
Wang, Y., S. Devkota, M.W. Musch, B. Jabri, C. Nagler, D.A. Antonopoulos, A.
Chervonsky, and E.B. Chang. 2010. Regional mucosa-associated microbiota
determine physiological expression of TLR2 and TLR4 in murine colon.
PLoS ONE. 5:e13607. http://dx.doi.org/10.1371/journal.pone.0013607
Willing, B.P., S.L. Russell, and B.B. Finlay. 2011. Shifting the balance:
Antibiotic effects on host-microbiota mutualism. Nat. Rev. Microbiol.
9:233243. http://dx.doi.org/10.1038/nrmicro2536
Wu, S., K.J. Rhee, E. Albesiano, S. Rabizadeh, X.Wu, H.R.Yen, D.L. Huso, F.L.
Brancati, E.Wick, F. McAllister, et al. 2009. A human colonic commensal
promotes colon tumorigenesis via activation of T helper type 17 T cell re
sponses. Nat. Med. 15:10161022. http://dx.doi.org/10.1038/nm.2015
Article
Engagement of the ICOS pathway markedly

enhances efficacy of CTLA-4 blockade
in cancer immunotherapy
Xiaozhou Fan,1 Sergio A. Quezada,3 Manuel A. Sepulveda,5,6
Padmanee Sharma,1,2,4 and James P. Allison1,4,5,6
1Department
of Immunology and 2Department of Genitourinary Medical Oncology, The University of Texas

MD Anderson Cancer Center, Houston, TX 77030
3Cancer Immunology Unit, Research Department of Haematology, University College London Cancer Institute,
London WC1E 6DD, England, UK
4Ludwig Center for Cancer Immunotherapy, 5Howard Hughes Medical Institute, and 6Immunology Program,
Memorial Sloan-Kettering Cancer Center, New York, NY 10065
Cytotoxic T lymphocyte antigen-4 (CTLA-4) blockade with a monoclonal antibody yields

durable responses in a subset of cancer patients and has been approved by the FDA as a
standard therapy for late-stage melanoma. We recently identified inducible co-stimulator
(ICOS) as a crucial player in the antitumor effects of CTLA-4 blockade. We now show that
concomitant CTLA-4 blockade and ICOS engagement by tumor cell vaccines engineered to
express ICOS ligand enhanced antitumor immune responses in both quantity and quality
and significantly improved rejection of established melanoma and prostate cancer in mice.
This study provides strong support for the development of combinatorial therapies incorporating antiCTLA-4 and ICOS engagement.
CORRESPONDENCE
Padmanee Sharma:
padsharma@mdanderson.org
OR
James P. Allison:
jallison@mdanderson.org
Abbreviations used: ICOS,
inducible co-stimulator; ICOSL,
ICOS ligand; i.d., intradermal(ly).
Harnessing T cell responses to eradicate tumors

has been difficult in part because of the complexity of regulation of T cell responses. Early
T cell activation requires an antigen-specific
signal mediated by the TCR plus additional costimulatory signals generated by engagement of
molecules such as CD28 with their ligands
(Harding et al., 1992). CD28 co-stimulation is
subject to down-regulation by inhibitory molecules such as cytotoxic T lymphocyte antigen-4
(CTLA-4; Walunas et al., 1994; Krummel and
Allison, 1995). Beginning in 1996, we showed
that inhibitory signals mediated by CTLA-4 were
responsible for limiting antitumor responses in a
series of mouse models because administration of
antibodies blocking the interaction of CTLA-4
with its ligands could result in tumor rejection
and long-lived immunity (Leach et al., 1996).
These preclinical studies led to the generation of antibodies to human CTLA-4, ipilimumab and tremelimumab (Sharma et al., 2011).
To date, over 20,000 patients have been treated
with these antibodies, the majority receiving ipi
limumab. Objective responses have been observed
P. Sharma and J.P. Allison contributed equally to this paper.
M.A. Sepulvedas present address is Janssen Research & Development, Spring House, PA 19002.

J. Exp. Med. 2014 Vol. 211 No. 4 715-725
in patients with melanoma, ovarian, prostate, renal

cell, and lung cancers. A randomized phase III
clinical trial with ipilimumab was reported in
2010, showing a significant increase in survival for
patients with advanced melanoma who received
ipilimumab therapy (Hodi et al., 2010).Treatment
with ipilimumab improved median overall survival by 3.7 mo and 23% of treated patients
were alive with durable clinic benefit for the
4.5 yr of follow up. Ipilimumab was the first therapy of any kind to show a survival benefit in phase
III trials (Hodi et al., 2010; Robert et al., 2011) for
patients with advanced melanoma and was approved in March 2011 by the Food and Drug
Administration (FDA) as both first and second
line therapy for the treatment of patients with
advanced melanoma. A recent retrospective study
of 177 metastatic melanoma patients from the
earliest clinical trials of ipilimumab showed an
88-mo median duration of objective responses
(Prieto et al., 2012). And a recent trial of ipilimumab in combination with an antibody to PD-1
2014 Fan et al. This article is distributed under the terms of an Attribution
NoncommercialShare AlikeNo Mirror Sites license for the first six months
after the publication date (see http://www.rupress.org/terms). After six months
it is available under a Creative Commons License (AttributionNoncommercial
Share Alike 3.0 Unported license, as described at http://creativecommons.org/
licenses/by-nc-sa/3.0/).
715
Figure 1. Treatment of B16/F10 tumors with antiCTLA-4 led to increased frequency of ICOS expression on tumor-infiltrating CD8 and CD4 Teff
cells. (A) Frequency of ICOS expression on CD8, CD4 Foxp3, and CD4 Foxp3+ T cells in the tumor. Horizontal bars represent means. (B) Breakdown of total intratumoral ICOS+ T cells in terms of CD8, CD4 Foxp3, and CD4 Foxp3+ subsets. Data are pooled from two independent experiments (n = 3 mice per group). Error
bars represent means SEM. Data were analyzed with one-way ANOVA and Bonferronis multiple comparisons test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
(nivolumab) in metastatic melanoma showed an objective response rate of 50% (Wolchok et al., 2013).
Together these data demonstrate that blockade of inhibitory signals mediated by CTLA-4 can be quite effective
against large bulky tumors and metastatic disease. However,
there is clearly a need to extend the therapeutic benefit of this
treatment to more patients.We have uncovered a novel immunebased strategy that can significantly enhance the efficacy of
CTLA-4 blockade.
In a presurgical clinical trial in which patients with localized bladder cancer were treated with ipilimumab, the frequency of T cells expressing inducible co-stimulator (ICOS)
was significantly increased both in tumor tissues and peripheral
blood of patients (Liakou et al., 2008). ICOS is a T cellspecific
molecule that belongs to the CD28/CTLA-4 family (Hutloff
et al., 1999; Sharpe and Freeman, 2002). ICOS expression is
up-regulated upon T cell activation, which is enhanced in the
setting of CTLA-4 blockade, thereby leading to a higher frequency of ICOS+ T cells detected in cancer patients receiving
antiCTLA-4 therapy, with the ICOS+ population containing
the bulk of tumor-specific, IFN-producing CD4 T cells
(Liakou et al., 2008; Carthon et al., 2010; Vonderheide et al.,
2010). In a retrospective study of advanced melanoma patients,
we also found a significant correlation between sustained elevation of ICOS+ CD4 T cells in the peripheral blood after ipilimumab treatment and increased survival (Carthon et al., 2010).
These clinical studies suggested that ICOS might play an important role in the therapeutic effect of antiCTLA-4. Our finding
that mice deficient in ICOS or ICOS ligand (ICOSL) had impaired antitumor responses after treatment with antiCTLA-4,
as compared with wild-type mice, further supported the notion
that the ICOS/ICOSL pathway is critical for the therapeutic
effect of antiCTLA-4 (Fu et al., 2011). These data prompted
us to investigate the potential benefit of providing additional
signal to the ICOS pathway in the setting of CTLA-4 blockade
as a strategy to further improve antitumor responses.
716
RESULTS
ICOS is selectively up-regulated on intratumoral CD8
and CD4 effector T cells (Teff cells)
Similar to what we previously observed in cancer patients but
even more dramatically, ICOS was up-regulated on CD8 and
CD4 Foxp3 Teff cells in mouse B16/F10 melanoma after
treatment with CTLA-4 blockade. We used irradiated parental B16 tumor cells as a control vaccination approach, which
did not affect ICOS expression on any T cell compartment
(unpublished data). In this situation, a very small fraction of
CD8 T cells in the tumor expressed ICOS, whereas about half
of CD4 Teff cells and the majority of CD4 Foxp3+ regulatory
T cells (Treg cells) were ICOS positive (Fig. 1 A). Blockade of
CTLA-4 in addition to the vaccination released the limit on
T cell activation and generally increased ICOS expression in
all of the T cell compartments, but the most significant change
was observed on CD8 T cells, with a six- to eightfold increase
in frequency.This trend led to a much greater presence of CD8
T cells, but much less presence of Treg cells in the total ICOSpositive pool inside the tumor (Fig. 1 B). These data further
support the rationale of activating the ICOS pathway as immunotherapy, as it would be more likely to benefit the anti
tumor CD8 T cells rather than immunosuppressive Treg cells.
Synergistic tumor protection by CTLA-4 blockade
and ICOS engagement
In light of the significant up-regulation of ICOS on intra
tumoral CD8 T cells, we developed a strategy to activate the
ICOS pathway by transducing tumor cells with the cognate
ligand, ICOSL (Yoshinaga et al., 1999), and using the irradiated ICOSL-positive tumor cells as a vaccine (IVAX) to treat
tumor-bearing mice. B16/F10 melanoma cells were engineered to express ICOSL on the cell surface and tested for
stable expression throughout the treatment process both
in vitro and ex vivo (Fig. 2 A). ELISA of tissue culture super
natant from these cells failed to show the presence of soluble/
ICOS engagement synergizes with CTLA-4 blockade | Fan et al.
Ar ticle
Figure 2. Cellular vaccine with ICOSL-expressing tumor cells (IVAX) synergized with CTLA-4 blockade to provide protection against B16/F10
tumors. (A) Treatment schedule of vaccination and CTLA-4 blockade and the verification of ICOSL expression on IVAX. Expression of ICOS on B16 and
IVAX were followed in vivo up to day 14 after tumor challenge. (B) Individual tumor growth curves after B16/F10 challenge. Numbers on the top right
side represent tumor-free mice. Data are representative of three independent experiments (n = 10 mice per group). (C) Tumor growth curves depict
average tumor volume in each group. Error bars represent means SEM. Data are representative of three independent experiments (n = 10 mice per
group). (D) Cumulative survival curves from two independent experiments (n = 10 mice per group). Survival curves were analyzed with Log-rank test.
****, P < 0.0001. (E) Cumulative survival curves of ICOS-deficient hosts from two independent experiments (n = 5 mice per group).
shed ICOS (detection range >0.1 ng/ml). Mice were given

an intradermal (i.d.) tumor challenge with parental (ICOSL
negative) B16/F10 cells and subsequently treated with anti
CTLA-4 plus a vaccine comprised of irradiated ICOSL-negative
JEM Vol. 211, No. 4
B16/F10 cells or ICOSL-positive B16/F10 cells (IVAX).

Neither IVAX nor the control vaccine of irradiated untransduced B16/F10 cells had any protective effect against tumor
growth in the absence of antiCTLA-4, probably because of
717
Figure 3. Combination of IVAX and CTLA-4 blockade led to rejection of higher doses of injected tumor cells and increased eradication of established tumors. (AF) Mice were challenged with 100K B16/F10 cells and treated from day 3 (AC) or challenged with 50K B16/F10 but treated from day 6
(DF). (A) Individual tumor growth curves after challenge with 100K B16/F10 cells. (B) Tumor growth curves depict average tumor volume in each group.
(C) Overall survival curves representative of two independent experiments (n = 10 mice per group). (D) Individual tumor growth curves after challenge with
50K B16/F10 cells. (A and D) Numbers on the top right side represent tumor-free mice. Data are representative of two independent experiments (n = 10 mice
per group). (E) Tumor growth curves depict average tumor volume in each group. (B and E) Error bars represent means SEM. Data are representative of two
independent experiments (n = 10 mice per group). (F) Overall survival curves representative of two independent experiments (n = 10 mice per group).
poor activation status and low frequency of ICOS expression

on CD8 and CD4 Teff cells (Fig. 1 A). In this model, anti
CTLA-4 treatment alone also failed to elicit tumor rejection.
The combination of antiCTLA-4 with control vaccine
resulted in tumor rejection in a minority of mice, whereas
antiCTLA-4 in combination with IVAX led to tumor rejection in >80% of mice, a fourfold increase in efficacy (Fig. 2,
BD). The increase in efficacy was specific to ICOS engagement, as the effect was completely lost in ICOS-deficient
hosts (Fig. 2 E). To further test the potency of this combination therapy in a more stringent and clinically relevant situation, we doubled the initial dose of tumor challenge or delayed
the onset of therapy. The combination of IVAX and CTLA-4
blockade still generated robust protection against a high dose
718
of B16/F10 challenge (Fig. 3, AC) or a more established

tumor (Fig. 3, DF).
The combination of IVAX and CTLA-4 blockade also
improved memory response against secondary challenge in
the tumor survivors (Fig. 4). We pooled the mice surviving
from either the combination of control vaccine and CTLA-4
blockade or the IVAX and antiCTLA-4 combination. At
least 4 mo after their initial tumor rejection, they were challenged with a very high dose (four times higher than regular)
of the same B16/F10 tumor without any further intervention. About half the mice from the control combination group
obtained enough memory response against the secondary
challenge, but all of the mice that had undergone IVAX and
antiCTLA-4 treatment rejected the second tumor without
Ar ticle
Teff cells, a cellular vaccine expressing ICOSL can trigger the

ICOS pathway to synergize with CTLA-4 blockade to provide potent tumor protection.
Figure 4. Stimulation of the ICOS pathway also improved memory

response against B16/F10 rechallenge. Mice that had been treated
with the indicated combination therapies and survived the primary
B16/F10 tumor were rechallenged with 200K B16/F10 cells but with no
further treatment. Data are representative of two independent experiments.
Survival curves were analyzed with Log-rank test. **, P < 0.01.
need of additional therapy. This is especially promising because one of the major advantages of tumor immunotherapy
is immune memory.
These results suggest that the elevated expression of ICOS
on T cells in antiCTLA-4treated tumor-bearing hosts is
not just a marker for T cell activation, but ICOS can actively
participate in further enhancing immune responses against
tumors. Thus, in the context of CTLA-4 blockade, which
leads to significant up-regulation of ICOS on CD8 and CD4
Changed balance of antitumor Teff cells

and immunosuppressive Treg cells
We next sought to dissect the basis for the enhanced efficacy
of antiCTLA-4/IVAX combination. We have previously
shown that antiCTLA-4, when combined with GM-CSF
secreting cellular vaccine (GVAX), increases the ratio of CD8
T cells to Treg cells in the tumor (Quezada et al., 2006). To
evaluate the impact of IVAX and CTLA-4 blockade on the
intratumoral cell composition, we counted the absolute numbers of CD8, CD4 Foxp3, and CD4 Foxp3+ T cells in B16/
F10 tumors on day 14 and normalized these numbers by the
tumor weight. IVAX alone did not change the composition
of T cells in the tumor but, when administered together with
antiCTLA-4, dramatically increased the density of CD8 (approximately fivefold) and CD4 Teff cells (approximately threefold) compared with the combination of control vaccine and
antiCTLA-4 (Fig. 5 A, left and middle). It is notable that the
density of CD4 Foxp3+ Treg cells was unaffected (Fig. 5 A,
right). The enrichment of CD4 and CD8 Teff cells was primarily observed at the tumor site but not in draining lymph
nodes or spleen (unpublished data). Because the density of Treg
cells remained unchanged and CD8 or CD4 Teff cells increased
several fold, the combination of IVAX and antiCTLA-4
blockade raised the intratumoral CD8/Treg cell ratio almost
Figure 5. Combination of IVAX and CTLA-4

blockade enriched CD8 and CD4 Foxp3 T cells
in the tumor and raised the intratumoral
CD8/Treg and CD4 Teff/Treg cell ratios. (A) Density
of CD8, CD4 Foxp3, and CD4 Foxp3+ T cells depicted
as absolute number of cells per milligram of tumor on
day 14 after tumor challenge. Numbers of T cells in
tumors were calculated as described in Materials and
methods. Data are pooled from two independent
experiments (n = 3 mice per group). (B) Cumulative
quantification of CD8/Treg and CD4 Teff/Treg cell ratios
in day 14 B16/F10 tumors from two independent
experiments (n = 3 mice per group). Horizontal bars
represent means. Data were analyzed with one-way
ANOVA and Bonferronis multiple comparisons test.
*, P < 0.05; **, P < 0.01.
JEM Vol. 211, No. 4
719
Figure 6. Combination of IVAX and CTLA-4

blockade enhanced proinflammatory cytokine production by CD4 Foxp3 T cells and
cytotoxicity of CD8 T cells. (A) Dot plots of
IFN- and TNF staining in tumor-infiltrating CD4
Foxp3 T cells. Numbers in the quadrants are
relative frequency. Data are representative of
three independent experiments (n = 3 mice per
group). (B) Cumulative quantification of the
frequency of IFN- and TNF production in tumorinfiltrating CD4 Foxp3 T cells from three
independent experiments (n = 3 mice per group).
(C) Dot plots of granzyme B and CD107a staining
in tumor-infiltrating CD8 T cells. Numbers in the
quadrants are relative frequency. Data are representative of two independent experiments (n = 3
mice per group). (D) Cumulative quantification of
the frequency of granzyme B+ CD107a+ in tumorinfiltrating CD8 T cells from two independent
experiments (n = 3 mice per group). (E) Density of
IFN-+ TNF+ CD4 Foxp3 T cells (left) and granzyme B+ CD107a+ CD8 T cells (right) depicted as
absolute numbers of these cells per milligram of
tumor. Numbers of T cells in tumors were calculated as described in Materials and methods. Data
are pooled from two or three independent experiments (n = 3 mice per group). Horizontal bars
represent means. Data were analyzed with oneway ANOVA and Bonferronis multiple comparisons test. **, P < 0.01; ***, P < 0.001.
sixfold and doubled the CD4 Teff/Treg cell ratio as compared with values in mice treated with control vaccine and
antiCTLA-4 (Fig. 5 B). The enhanced ratio of effector to
720
regulatory T cells marked the shift from an immunosuppressive to immunostimulatory tumor microenvironment and
provides one possible explanation for the potent antitumor
Ar ticle
efficacy observed with the combination therapy of anti

CTLA-4 and IVAX.
IVAX combination therapy greatly enhances
the function of antitumor Teff cells
The higher intratumoral CD8/Treg and CD4 Teff/Treg cell
ratios could quantitatively contribute to the tumor protection effect of IVAX and CTLA-4 blockade. Furthermore,
there were also profound qualitative changes in the immune
functions of these enriched CD8 and CD4 Teff cells. We isolated tumor-infiltrating lymphocytes from day 14 tumors,
briefly restimulated them ex vivo with DCs loaded with
B16 lysate and assayed their cytokine production and cytotoxic activity. It has been reported that ICOS signaling
in human T cells induced Th17 polarization (Paulos et al.,
2010). However, in our models, we were unable to detect
IL-17A expression in CD4 Teff cells isolated from the tumors
or the vaccination sites (unpublished data). Although mice
treated with cellular vaccine alone (either control vaccine or
IVAX) or a combination of control vaccine and antiCTLA-4
had quite few CD4 Teff cells producing IFN-, there was a
distinct population of polyfunctional CD4 Teff cells secreting
both IFN- and TNF, with an increase of 25-fold in the
frequency of IFN-+ TNF+ cells after treatment with anti
CTLA-4 plus IVAX (Fig. 6 A). This cytokine profile indicated that these cells were potent tumor antigenspecific
Th1 cells. We also observed that the frequency of either
IFN-+ or TNF+ CD4 Teff cells was significantly higher in
tumors treated with IVAX and antiCTLA-4 blockade
as compared with the combination of control vaccine plus
antiCTLA-4 (Fig. 6 B). These data suggested that the synergy between IVAX and antiCTLA-4 was capable of inducing Th1 polarization in vivo and thus providing strong
help to the antitumor cytotoxic CD8 T cells. The CD8
T cells in the treated tumors did not produce as much proinflammatory cytokines (IFN- or TNF) as CD4 Teff cells, although the combination of IVAX and CTLA-4 blockade still
induced higher IFN- and TNF production than the combination of control vaccine and antiCTLA-4 (unpublished
data). However, the cytolytic activity of tumor-infiltrating
CD8 T cells was considerably enhanced by IVAX and anti
CTLA-4, as measured by coexpression of granzyme B and
CD107a (LAMP-1; Fig. 6, C and D), which is one of the
most striking features of this novel combinatorial approach.
Thus the combination of IVAX and antiCTLA-4 not only
enriched CD8 and CD4 Teff cells in the tumors, but also enhanced their antitumor functions, including secretion of proinflammatory cytokines and cytolytic activity at the tumor
site. Calculating the effect of these changes altogether, the
numbers of CD4 Teff cells producing IFN- and TNF and
degranulating cytotoxic CD8 T cells were increased by 70and 8-fold, respectively (Fig. 6 E). Overall, the combination
of antiCTLA-4 and IVAX could increase the density of
tumor-reactive helper and killer T cells in the tumor to significantly higher levels, which might be the leading cause
for tumor rejection.
JEM Vol. 211, No. 4
Figure 7. CD8, CD4 T cells, and IFN- were indispensable for

therapeutic efficacy. B16/F10 tumor-bearing mice were treated with
IVAX and CTLA-4 blockade as described in Fig. 2 A and were depleted of
CD8 cells with anti-CD8 (clone 2.43; n = 20). The tumor protection rate
was also measured in MHC class II KO (n = 13) or IFN-R KO (n = 20)
hosts. Data were pooled from two independent experiments. Survival
curves were analyzed with Log-rank test. ****, P < 0.0001.
CD8, CD4 Teff cells, and IFN- are required

for the tumor protection
The expanded pool of CD8 and CD4 Teff cells were critical in
the IVAX-driven tumor rejection. We repeated our tumor
protection experiments with additional treatment with depleting antibody or with genetically deficient hosts. Mice receiving
CD8 depleting antibody or lacking MHC class II molecules
suffered greatly diminished therapeutic efficacy (Fig. 7). Of
note, although both the CD8 and CD4 population were required for the maximum protection, other cell populations also
play a role in the absence of either one, suggesting that the tumor
rejection caused by this treatment regimen involves multiple
helper and effector cell populations. However, the Th1 cytokine
IFN- was indispensable for any tumor protection as IFN-R
KO mice completely lost the survival benefit (Fig. 7).
IVAX combination therapy has robust protection
efficacy in multiple tumor models
To show that the protection efficacy of IVAX combination
therapy is not tumor model specific, we also used the same
protocol to generate ICOSL-expressing tumor cell vaccines
from another mouse melanoma cell line, B16/BL6, as well as
the mouse prostate cancer cell line TRAMP C2. As we observed with the B16/F10 model, the combination of IVAX
and antiCTLA-4 can also cure TRAMP prostate cancer. All
of the mice rejected TRAMP C2 tumors after treatment with
IVAX plus antiCTLA-4, as compared with the 50% protection rate in the group of control vaccine plus antiCTLA-4
(Fig. 8 A). The control TRAMP C2 vaccine did not synergize
with antiCTLA-4, as the tumor growth kinetics is very similar between the group of antiCTLA-4 alone and the one of
control vaccine plus antiCTLA-4 (Fig. 8 B). In contrast,
there was statistically significant survival benefit with the
combination of IVAX plus antiCTLA-4 (Fig. 8 C). These
data were also reproducible in a third tumor model, B16/BL6
(unpublished data). Thus, the synergy between IVAX and
CTLA-4 blockade was observed across different tumor models, and in each model, the combination therapy significantly
improved tumor rejection and overall survival. These results
721
Figure 8. Combination therapy of IVAX and CTLA-4 blockade was also therapeutic against mouse prostate tumors. (A) Individual tumor
growth curves after challenge with TRAMP C2 cells. Numbers on the top right side represent tumor-free mice. Data are representative of two independent
experiments (n = 10 mice per group). (B) Tumor growth curves depict average tumor volume in each group. Error bars represent means SEM. Data are
representative of two independent experiments (n = 10 mice per group). (C) Cumulative survival curves from two independent experiments (n = 10 mice
per group). Survival curves were analyzed with Log-rank test. ***, P < 0.001.
suggest that the underlying principle of this combination

therapy could potentially be applied to multiple types of cancer in clinical application.
Tumor protection by IVAX requires presentation in cis
One aspect that could potentially expand the clinical application of IVAX therapy is whether the ICOS signal can be delivered independently of the cognate tumor antigen. This, if
true, would lead to easier development of off-the-shelf ICOS
agonist that can enhance T cell immunity against any target
antigen. We tested this hypothesis by treating mice with B16
tumors with TRAMP-based IVAX or a 1:1 mixture of irradiated wild-type ICOSL-negative B16 and TRAMP-based
IVAX. In this setting, the primary TCR signal (B16 tumor
antigen) and the secondary ICOS signal were presented in
trans. This strategy was unable to generate the same degree of
tumor protection as with B16-based IVAX where the two
722
signals were presented in cis (Fig. 9). The cognate TCR signal is required in order for ICOS signal to take effect, as
TRAMP-based IVAX alone was no more effective than irradiated wild-type B16. The ICOS signal in trans did provide
some additional survival benefit when compared with mice
treated with irradiated B16 control vaccine alone, but the difference was not significant. This result suggests that cognate
TCR signal and ICOS stimulation should be incorporated
on the same vehicle for optimal therapeutic effect.
DISCUSSION
With the FDA approval of PROVENGE and more recently
ipilimumab, the effectiveness of immunotherapy in the treatment of cancer is firmly established. Ipilimumab has quickly
become a standard-of-care agent for the treatment of latestage melanoma, and its application will possibly expand as
results are reported from ongoing phase III trials in prostate
Ar ticle
Figure 9. Tumor protection by IVAX requires presentation in cis. B16/F10 tumor-bearing mice were treated with TRAMP-based IVAX or a 1:1 mixture of irradiated wild-type ICOSL-negative B16 and TRAMP-based IVAX. Irradiated wild-type B16 and B16-based IVAX were included as control. CTLA-4
blockade was given in all the treatment groups. (A) Individual tumor growth curves after B16/F10 challenge. Numbers on the top right side represent
tumor-free mice. Data are representative of two independent experiments (n = 10 mice per group). (B) Tumor growth curves depict average tumor volume in each group. Error bars represent means SEM. Data are representative of two independent experiments (n = 10 mice per group). (C) Cumulative
survival curves from two independent experiments (n = 10 mice per group). Survival curves were analyzed with Log-rank test. **, P < 0.01; ***, P < 0.001.
and other tumor types. As with previous standard-of-care therapies, it will be necessary to develop combination strategies to
improve clinical benefit. Here, we demonstrate that the efficacy
of antiCTLA-4 therapy is greatly enhanced by targeting the
ICOS/ICOSL pathway with a cellular vaccine (IVAX).
The synergy of IVAX with antiCTLA-4 results in a dramatic enhancement of tumor rejection. However, in the absence of ICOS up-regulation in CD8 and CD4 Teff cells as a
result of CTLA-4 blockade, IVAX monotherapy has minimal
effects.This finding is consistent with previous reports of minimal efficacy when ICOS was targeted as monotherapy. For
example, it has been shown that ectopic expression of ICOSL
could elicit tumor-specific T cell response, but antitumor responses could only be generated in a prophylactic and not a
therapeutic setting (Liu et al., 2001;Wallin et al., 2001; Zuberek
et al., 2003). Similarly, engagement of the ICOS pathway with
JEM Vol. 211, No. 4
an ICOSL-Ig fusion protein alone failed to induce rejection

of poorly immunogenic tumors such as B16 melanoma (Ara
et al., 2003; Zuberek et al., 2003).This was probably caused by
the fact that in the absence of CTLA-4 blockade, ICOS was
expressed by only a few CD8 and/or CD4 Teff cells but expressed highly by the majority of Treg cells (Fig. 1), so that
engagement of the ICOS pathway primarily stimulated the
regulatory rather than the effector population.This hypothesis
was supported by the fact that pretreatment of mice with cyclophosphamide, which preferentially depleted Treg cells, could
help improve the efficacy of ICOSL-Ig against poorly immunogenic tumors (Ara et al., 2003). A recent study showed that
human melanoma cells expressing high levels of ICOSL facilitated expansion and IL-10 production in the Treg cell population in the setting of high-dose IL-2 treatment (Martin-Orozco
et al., 2010). In our study, however, there was no expansion of
723
Treg cells in mice treated with IVAX alone or with antiCTLA-4,

but the balance between effector and regulatory T cells in the
tumor was enhanced considerably by the combination of
IVAX and CTLA-4 blockade as the result of selective accumulation of effector CD4 and CD8 cells.
CTLA-4 blockade has been reported to selectively reduce
the frequency and number of Treg cells in B16 tumors but
not in the peripheral lymphoid organs (Quezada et al., 2006;
Simpson et al., 2013). The depletion is strongest in the combination of GVAX and antiCTLA-4 on adoptively transferred TCR-transgenic Trp1 CD4 cells when compared with
GVAX monotherapy. This effect is potentially driven by
heavy infiltration of macrophages in the tumor (Simpson et al.,
2013), which is a hallmark of GVAX therapy but not observed in the IVAX model.We did verify that CTLA-4 blockade reduced the frequency of Treg cells in the tumor, but that
was primarily because of relatively greater expansion of CD8
and CD4 Teff cells rather than a decrease in the number of
Treg cells per milligram of tumor. In contrast, this also opens
up new avenues of combination therapy by incorporating
Treg cell depletion regimen into IVAX therapy to further improve the efficacy.
The underlying mechanism of IVAX therapy is likely to
be distinct from GVAX and Flt3L-secreting vaccine (FVAX;
Curran and Allison, 2009). An optimal cancer immunotherapy entails ample tumor antigen presentation, co-stimulation,
and/or removal of co-inhibition on T cells. Both GVAX and
FVAX enhance the differentiation, maturation, and chemo
attraction of DCs, which synergizes very well with CTLA-4
blockade. Although IVAX and antiCTLA-4 combination
strategy focuses more on the two facets of T cell activation,
adding positive signals and blocking negative ones, using irradiated tumor cell vaccine as the vehicle to carry the positive
ICOS signal still provides some help to tumor antigen presentation, but we reason that it would provide further benefit to
combine the strength of IVAX and other strategies like GVAX
or FVAX.
Our working hypothesis is that CTLA-4 blockade leads
to enhanced activation of tumor-reactive T cells with concomitant up-regulation of ICOS, thereby enabling their responses to be enhanced by ICOS engagement. Thus, in the
context of CTLA-4 blockade, IVAX triggers the ICOS pathway to enhance the proliferation, survival, and/or migration
of effector cells into the tumor, which led to a higher density
of Teff cells inside the tumor, as indicated by an increase in the
Teff/Treg cell ratio and marked increases in both Th1 CD4 Teff
and cytolytic CD8 T cells. These results clearly demonstrate
a proof-of-concept that antitumor responses enhanced by
CTLA-4 blockade can be greatly improved by targeting the
ICOS/ICOSL pathway with ICOSL-expressing tumor cell
vaccines.We are currently exploring other strategies that might
be more suitable for clinical application for providing agonistic
signals through ICOS locally and systemically in combination
with CTLA-4 blockade. We anticipate that these combination
strategies will translate to the clinic to increase the number of
patients who derive benefit from antiCTLA-4 therapy.
724

Mice. 6-wk-old C57BL/6 and ICOS/ mice were purchased from the
Jackson Laboratory. Mice were housed in specific pathogenfree conditions
in accordance with institutional guidelines. All animal experiments were
approved by the Memorial Sloan-Kettering Cancer Center Institutional
Animal Care and Use Committee.
Cell lines. The poorly immunogenic mouse melanoma cell lines B16/F10 and
B16/BL6 were obtained from I. Fidler (The University of Texas MD Anderson
Cancer Center, Houston,TX) and described previously (van Elsas et al., 1999).
The prostate cancer cell line TRAMP C2 was maintained as described previously (Foster et al., 1997). B16/F10, B16/BL6, and TRAMP C2 were transduced
with retrovirus to express full-length mouse ICOSL on the cell membrane
and tested for stable expression throughout the treatment process.
Development of IVAX. HEK 293T cells were transfected with vectors
encoding full-length mouse ICOSL (provided by W. Sha, University of California, Berkeley, Berkeley, CA), envelope glycoprotein from the vesicular stomatitis virus (VSV-G), and Gag-Pol using FuGENE HD (Roche). Supernatant
containing packaged virus was collected 48 and 72 h later, filtered with 0.45-m
microfilters, and applied to cultured B16/F10, B16/BL6, or TRAMP C2 cells.
5 g/ml polybrene was also added to the virus solution. Target cells were
spun at 2,600 rpm and 32C for 2 h before being transferred into 37C incubators.Tumor cells stably expressing ICOSL were selected by surface staining
with anti-ICOSL.
Antibodies. AntiCTLA-4 (9H10) was purchased from Bio X Cell and administered i.p. Antibodies for flow cytometry were purchased from eBioscience and BD.
Tumor challenge and treatments. Mice were challenged i.d. on the right
flank with 5 104 B16/F10, 2 104 B16/BL6, or 7.5 105 TRAMP C2
tumor cells on day 0. In experiments in which mice would be sacrificed on
day 14, initial B16/F10 challenge was 2 105. Mice were then treated with
i.p. injection of 100 g antiCTLA-4 (clone 9H10) and i.d. vaccination on
the left flank with 106 irradiated (150 Gy) ICOSL-expressing tumor vaccine
(IVAX) on days 3, 6, 9, and 12. The dose of antiCTLA-4 was doubled on
day 3. The mice were then followed for tumor growth or sacrificed on day
14 for dissection of lymphoid organs and tumors.
Phenotypic and functional analyses of tumor-infiltrating lymphocytes. Mice used for functional experiments were sacrificed on day 14 after
tumor challenge, and tumor draining lymph nodes, vaccine draining lymph
nodes, and tumors were isolated. Tumors were digested in Liberase TL
(Roche) and DNase I (Roche) at 37C for 30 min, filtered, and centrifuged
over Histopaque-1119 (Sigma-Aldrich). Tumor-infiltrating T cells were restimulated for 4 h at 37C with 5 104 DCs loaded with B16 lysate, in the
presence of Golgi-Plug (BD). When cytolytic activity was measured, tumor
infiltrates was incubated with anti-CD107a at 37C for 2 h before staining
with other antibodies, in the presence of monensin (BD).
Flow cytometry and quantification. Samples were stained with antiCD4APCeFluor 780 (L3T4), antiCD8-PerCP-Cy5.5 (53-6.7), and antiICOS-PE
(17G9), fixed and permeabilized (eBioscience) according to the manufac
turers instructions, and stained with anti-Foxp3Alexa Fluor 700 (FJK-16s),
antiIFN-Alexa Fluor 488 (XMG1.2), anti-TNFPacific Blue (MP6XT22), antigranzyme Ballophycocyanin (GB11), and antiCD107a-PE
(1D4B). Flow cytometry reference beads (PeakFlow blue; Invitrogen) were
added to the samples before analysis for quantification of T cells in each tumor.
The absolute number of a subset of T cells per milligram of tumor was calculated as the following example shows: density of CD8 cells = (number of beads
added to each sample count of CD8 cells/count of beads)/tumor weight.
Statistical analyses. Data were analyzed with Prism 5.0 (GraphPad Software). Experiments were repeated two to three times. Statistical significance
Ar ticle
was determined by one-way ANOVA and Bonferronis multiple comparisons

test. Tumor survival data were analyzed with the Kaplan-Meier method. The
log-rank test was used to compare survival curves for different groups on
univariate analyses. P < 0.05 was considered statistically significant.
We would like to thank Dr. W. Sha for providing the retroviral vector encoding mouse
ICOSL. We would also like to thank J. Geddes, J. Lu, W. Montalvo, and A. Trumble for
technical assistance and M. Curran and R. Waitz for helpful discussion of the project.
This work was supported in part by a Department of Defense Idea
Development Award (W81XWH-10-1-0073 01), a Cancer Prevention Research
Institute Individual Investigator Award (RP120108), and a National Institutes of
Health NCI R01 grant (R01CA163793-01), all to P. Sharma; a Prostate Cancer
Foundation Challenge Grant in Immunology to P. Sharma and J.P. Allison; and a
Cancer Research Institute Tumor Immunology Predoctoral Fellowship to X. Fan. S.A.
Quezada is funded by a Cancer Research UK Career Development Fellowship and a
Cancer Research Institute Investigator Award.
J.P. Allison is an inventor of intellectual property licensed by the University of
California to Medarex and is a consultant for Medarex and Bristol Meyers Squibb.
The authors declare no further competing financial interests.
Submitted: 21 March 2013
Accepted: 5 March 2014
REFERENCES
Ara, G., A. Baher, N. Storm, T. Horan, C. Baikalov, E. Brisan, R. Camacho,

A. Moore, H. Goldman, T. Kohno, et al. 2003. Potent activity of soluble B7RP-1-Fc in therapy of murine tumors in syngeneic hosts. Int. J.
Cancer. 103:501507. http://dx.doi.org/10.1002/ijc.10831
Carthon, B.C., J.D. Wolchok, J.Yuan, A. Kamat, D.S. Ng Tang, J. Sun, G. Ku, P.
Troncoso, C.J. Logothetis, J.P. Allison, and P. Sharma. 2010. Preoperative
CTLA-4 blockade:Tolerability and immune monitoring in the setting of a
presurgical clinical trial. Clin. Cancer Res. 16:28612871. http://dx.doi.org/
10.1158/1078-0432.CCR-10-0569
Curran, M.A., and J.P. Allison. 2009. Tumor vaccines expressing flt3 ligand
synergize with ctla-4 blockade to reject preimplanted tumors. Cancer Res.
69:77477755. http://dx.doi.org/10.1158/0008-5472.CAN-08-3289
Foster, B.A., J.R. Gingrich, E.D. Kwon, C. Madias, and N.M. Greenberg. 1997.
Characterization of prostatic epithelial cell lines derived from transgenic
adenocarcinoma of the mouse prostate (TRAMP) model. Cancer Res.
57:33253330.
Fu, T., Q. He, and P. Sharma. 2011. The ICOS/ICOSL pathway is required
for optimal antitumor responses mediated by antiCTLA-4 therapy.
Cancer Res. 71:54455454. http://dx.doi.org/10.1158/0008-5472.CAN-111138
Harding, F.A., J.G. McArthur, J.A. Gross, D.H. Raulet, and J.P. Allison. 1992.
CD28-mediated signalling co-stimulates murine T cells and prevents induction of anergy in T-cell clones. Nature. 356:607609. http://dx.doi.org/10
.1038/356607a0
Hodi, F.S., S.J. ODay, D.F. McDermott, R.W. Weber, J.A. Sosman, J.B.
Haanen, R. Gonzalez, C. Robert, D. Schadendorf, J.C. Hassel, et al. 2010.
Improved survival with ipilimumab in patients with metastatic melanoma.
N. Engl. J. Med. 363:711723. http://dx.doi.org/10.1056/NEJMoa1003466
Hutloff, A., A.M. Dittrich, K.C. Beier, B. Eljaschewitsch, R. Kraft, I.
Anagnostopoulos, and R.A. Kroczek. 1999. ICOS is an inducible T-cell
co-stimulator structurally and functionally related to CD28. Nature.
397:263266. http://dx.doi.org/10.1038/16717
Krummel, M.F., and J.P. Allison. 1995. CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation. J. Exp. Med. 182:459465.
http://dx.doi.org/10.1084/jem.182.2.459
Leach, D.R., M.F. Krummel, and J.P. Allison. 1996. Enhancement of antitumor
immunity by CTLA-4 blockade. Science. 271:17341736. http://dx.doi
.org/10.1126/science.271.5256.1734
Liakou, C.I., A. Kamat, D.N.Tang, H. Chen, J. Sun, P.Troncoso, C. Logothetis,
and P. Sharma. 2008. CTLA-4 blockade increases IFN-producing
CD4+ICOShi cells to shift the ratio of effector to regulatory T cells in cancer patients. Proc. Natl. Acad. Sci. USA. 105:1498714992. http://dx.doi
.org/10.1073/pnas.0806075105
JEM Vol. 211, No. 4
Liu, X., X.F. Bai, J. Wen, J.X. Gao, J. Liu, P. Lu, Y. Wang, P. Zheng, and Y. Liu.
2001. B7H costimulates clonal expansion of, and cognate destruction
of tumor cells by, CD8+ T lymphocytes in vivo. J. Exp. Med. 194:1339
1348. http://dx.doi.org/10.1084/jem.194.9.1339
Martin-Orozco, N., Y. Li, Y. Wang, S. Liu, P. Hwu, Y.J. Liu, C. Dong, and L.
Radvanyi. 2010. Melanoma cells express ICOS ligand to promote the
activation and expansion of T-regulatory cells. Cancer Res. 70:95819590.
http://dx.doi.org/10.1158/0008-5472.CAN-10-1379
Paulos, C.M., C. Carpenito, G. Plesa, M.M. Suhoski, A.Varela-Rohena, T.N.
Golovina, R.G. Carroll, J.L. Riley, and C.H. June. 2010. The inducible
costimulator (ICOS) is critical for the development of human TH17
cells. Sci.Transl. Med. 2:55ra78. http://dx.doi.org/10.1126/scitranslmed
.3000448
Prieto, P.A., J.C.Yang, R.M. Sherry, M.S. Hughes, U.S. Kammula, D.E.White,
C.L. Levy, S.A. Rosenberg, and G.Q. Phan. 2012. CTLA-4 blockade
with ipilimumab: Long-term follow-up of 177 patients with metastatic melanoma. Clin. Cancer Res. 18:20392047. http://dx.doi.org/10
.1158/1078-0432.CCR-11-1823
Quezada, S.A., K.S. Peggs, M.A. Curran, and J.P. Allison. 2006. CTLA4 blockade and GM-CSF combination immunotherapy alters the intratumor
balance of effector and regulatory T cells. J. Clin. Invest. 116:19351945.
http://dx.doi.org/10.1172/JCI27745
Robert, C., L. Thomas, I. Bondarenko, S. ODay, J.W. M D, C. Garbe, C.
Lebbe, J.F. Baurain, A. Testori, J.J. Grob, et al. 2011. Ipilimumab plus dacarbazine for previously untreated metastatic melanoma. N. Engl. J. Med.
364:25172526. http://dx.doi.org/10.1056/NEJMoa1104621
Sharma, P., K. Wagner, J.D. Wolchok, and J.P. Allison. 2011. Novel cancer immunotherapy agents with survival benefit: Recent successes and next
steps. Nat. Rev. Cancer. 11:805812. http://dx.doi.org/10.1038/nrc3153
Sharpe, A.H., and G.J. Freeman. 2002. The B7-CD28 superfamily. Nat. Rev.
Immunol. 2:116126. http://dx.doi.org/10.1038/nri727
Simpson, T.R., F. Li, W. Montalvo-Ortiz, M.A. Sepulveda, K. Bergerhoff, F.
Arce, C. Roddie, J.Y. Henry, H. Yagita, J.D. Wolchok, et al. 2013. Fcdependent depletion of tumor-infiltrating regulatory T cells co-defines
the efficacy of antiCTLA-4 therapy against melanoma. J. Exp. Med.
210:16951710. http://dx.doi.org/10.1084/jem.20130579
van Elsas, A., A.A. Hurwitz, and J.P. Allison. 1999. Combination immunotherapy of B16 melanoma using anticytotoxic T lymphocyteassociated
antigen 4 (CTLA-4) and granulocyte/macrophage colony-stimulating
factor (GM-CSF)-producing vaccines induces rejection of subcutaneous
and metastatic tumors accompanied by autoimmune depigmentation.
J. Exp. Med. 190:355366. http://dx.doi.org/10.1084/jem.190.3.355
Vonderheide, R.H., P.M. LoRusso, M. Khalil, E.M. Gartner, D. Khaira, D.
Soulieres, P. Dorazio, J.A. Trosko, J. Rter, G.L. Mariani, et al. 2010.
Tremelimumab in combination with exemestane in patients with advanced breast cancer and treatment-associated modulation of inducible
costimulator expression on patient T cells. Clin. Cancer Res. 16:3485
3494. http://dx.doi.org/10.1158/1078-0432.CCR-10-0505
Wallin, J.J., L. Liang, A. Bakardjiev, and W.C. Sha. 2001. Enhancement of CD8+
T cell responses by ICOS/B7h costimulation. J. Immunol. 167:132139.
Walunas, T.L., D.J. Lenschow, C.Y. Bakker, P.S. Linsley, G.J. Freeman, J.M.
Green, C.B. Thompson, and J.A. Bluestone. 1994. CTLA-4 can function
as a negative regulator of T cell activation. Immunity. 1:405413. http://
dx.doi.org/10.1016/1074-7613(94)90071-X
Wolchok, J.D., H. Kluger, M.K. Callahan, M.A. Postow, N.A. Rizvi, A.M.
Lesokhin, N.H. Segal, C.E. Ariyan, R.A. Gordon, K. Reed, et al. 2013.
Nivolumab plus ipilimumab in advanced melanoma. N. Engl. J. Med.
369:122133. http://dx.doi.org/10.1056/NEJMoa1302369
Yoshinaga, S.K., J.S. Whoriskey, S.D. Khare, U. Sarmiento, J. Guo, T. Horan, G.
Shih, M. Zhang, M.A. Coccia,T. Kohno, et al. 1999.T-cell co-stimulation
through B7RP-1 and ICOS. Nature. 402:827832. http://dx.doi.org/
10.1038/45582
Zuberek, K., V. Ling, P. Wu, H.L. Ma, J.P. Leonard, M. Collins, and K. DunussiJoannopoulos. 2003. Comparable in vivo efficacy of CD28/B7, ICOS/
GL50, and ICOS/GL50B costimulatory pathways in murine tumor models: IFN-dependent enhancement of CTL priming, effector functions,
and tumor specific memory CTL. Cell. Immunol. 225:5363. http://dx
.doi.org/10.1016/j.cellimm.2003.09.002
725
Brief Definitive Report
Enhancement of an anti-tumor immune

response by transient blockade
of central T cell tolerance
Imran S. Khan,1 Maria L. Mouchess,1 Meng-Lei Zhu,3,4 Bridget Conley,3,4
Kayla J. Fasano,1 Yafei Hou,2 Lawrence Fong,2 Maureen A. Su,3,4
and Mark S. Anderson1
1Diabetes
Center and 2Division of Hematology/Oncology, Department of Medicine, University of California, San Francisco,
San Francisco, CA 94143
3Department of Pediatrics and 4Department of Microbiology/Immunology, School of Medicine, and Lineberger Comprehensive
Cancer Center, University of North Carolina, Chapel Hill, NC 27599
Thymic central tolerance is a critical process that prevents autoimmunity but also presents
a challenge to the generation of anti-tumor immune responses. Medullary thymic epithelial
cells (mTECs) eliminate self-reactive T cells by displaying a diverse repertoire of tissuespecific antigens (TSAs) that are also shared by tumors. Therefore, while protecting against
autoimmunity, mTECs simultaneously limit the generation of tumor-specific effector T cells
by expressing tumor self-antigens. This ectopic expression of TSAs largely depends on
autoimmune regulator (Aire), which is expressed in mature mTECs. Thus, therapies to
deplete Aire-expressing mTECs represent an attractive strategy to increase the pool of
tumor-specific effector T cells. Recent work has implicated the TNF family members RANK
and RANK-Ligand (RANKL) in the development of Aire-expressing mTECs. We show that in
vivo RANKL blockade selectively and transiently depletes Aire and TSA expression in the
thymus to create a window of defective negative selection. Furthermore, we demonstrate
that RANKL blockade can rescue melanoma-specific T cells from thymic deletion and that
persistence of these tumor-specific effector T cells promoted increased host survival in
response to tumor challenge. These results indicate that modulating central tolerance
through RANKL can alter thymic output and potentially provide therapeutic benefit by
enhancing anti-tumor immunity.
CORRESPONDENCE
Mark S. Anderson:
manderson@diabetes.ucsf.edu
Abbreviations used: Aire,
autoimmune regulator; cTEC,
cortical thymic epithelial cell;
IRBP, interphotoreceptor
retinoid-binding protein; mTEC,
medullary thymic epithelial cell;
OPG, Osteoprotegerin; TSA,
tissue-specific antigen.
Medullary thymic epithelial cells (mTECs) contribute to self-tolerance through the ectopic expression of tissue-specific antigens (TSAs) in the
thymus (Derbinski et al., 2001; Anderson et al.,
2002; Metzger and Anderson, 2011). This TSA
expression in mTECs is largely dependent on
autoimmune regulator (Aire), which is expressed
in mature mTECs (Gbler et al., 2007; Gray et al.,
2007; Metzger and Anderson, 2011). Through
the recognition of TSAs, developing autoreactive
T cells are either negatively selected from the pool
of developing thymocytes or recruited into the
regulatory T (T reg) cell lineage (Liston et al., 2003;
Anderson et al., 2005; DeVoss et al., 2006; Shum
et al., 2009; Taniguchi et al., 2012; Malchow et al.,
2013). The overall importance of this process is
underscored by the development of a multi-organ
I.S. Khan and M.L. Mouchess contributed equally to
this paper.

J. Exp. Med. 2014 Vol. 211 No. 5 761-768
autoimmune syndrome in patients or mice with

defective AIRE expression (Consortium, 1997;
Nagamine et al., 1997; Anderson et al., 2002).
Although central tolerance provides protection against autoimmunity, this process also
represents a challenge for anti-tumor immunity
(Kyewski and Klein, 2006; Malchow et al., 2013).
Because many of the TSAs expressed in the thymus are also expressed in tumors, high-affinity
effector T cells capable of recognizing tumor
self-antigens may normally be deleted in the thymus (Bos et al., 2005; Cloosen et al., 2007; Trger
et al., 2012; Zhu et al., 2013). Transiently suppressing central tolerance by depleting mTECs
or modulating Aire expression may provide a
2014 Khan et al. This article is distributed under the terms of an Attribution
761
Figure 1. Selective depletion of mTECs with antiRANKL blockade. (A) Wild-type mice were treated for 2 wk
with either isotype (gray) or anti-RANKL (RANKL, blue)
antibody, and absolute numbers of mTECs and cTECs were
enumerated by flow cytometry. Bar graphs of total cell numbers are depicted by mean SEM. (B) Representative flow
cytometry plots of mTECs in A showing relative composition
of mTEC subsets. Values represent mean SEM. Bar graphs
(right) of total cell numbers of each mTEC subset depicted by
mean SEM. (C) Top panels show immunostaining for keratin-8
(green) and keratin-5 (red) on thymic sections, and bottom
panels show immunostaining for keratin-5 (red) and
Aire (green). Bars: (top) 500 m; (bottom) 50 m. (D) Gene
expression analysis on mTECs sorted from wild-type mice
treated with either isotype (gray) or anti-RANKL (blue) antibody. Results standardized to Cyclophilin A and normalized
to isotype-treated mice with bars depicting mean SD.
(E) Representative flow cytometry plots of thymocytes from
antibody-treated wild-type mice. Values depict mean SD.
Total thymocyte numbers (right) are indicated with each
circle depicting an individual animal and bars showing the
mean. (F) Flow cytometric analysis of Foxp3 staining in CD4
SP thymocytes from E. Plots (right) show percentage and
number of thymic CD4+ Foxp3+ cells, with bars showing the
mean. Data shown in Fig. 1 is representative of two to four
independent experiments containing three or more individual mice within each group. *, P 0.05; ***, P 0.001,
Students t test.
therapeutic window for the generation of T cells capable of

recognizing tumor self-antigens. Many current cancer immune
therapies rely on activating relatively weak tumor-specific T cell
responses through modulating peripheral tolerance (Swann and
Smyth, 2007; Chen and Mellman, 2013). In contrast, manipu
lation of central tolerance has the potential to increase the
pool and affinity of effector T cells that can recognize and contribute to effective anti-tumor responses. Furthermore, such
high-affinity, self-reactive T cells may be more resistant to peripheral tolerance mechanisms that typically restrain an antitumor response (Swann and Smyth, 2007).Thus, the development
of methods that selectively and transiently deplete Aire-expressing
mTECs may be an attractive method to enhance tumor-specific
immune responses.
Previous work has identified agents that can inhibit the
growth and development of TECs such as corticosteroids, cyclosporine, and some inflammatory cytokines (Anz et al., 2009;
Fletcher et al., 2009). Despite their clear inhibitory effects on
TECs, however, these agents do not appear to have selectivity
for blocking mTEC development. Interestingly, recent studies
have demonstrated a role for TNF family member pairs
RANKRANKL and CD40-CD40L in the embryological
development of Aire+ mTECs (Rossi et al., 2007; Akiyama et al.,
2008; Hikosaka et al., 2008; Roberts et al., 2012). Recent work
762
has also demonstrated that mTECs in particular have a relatively fast turnover in adult mice with an estimated half-life of
2 wk (Gbler et al., 2007; Gray et al., 2007). Given these findings, we speculated that in vivo blockade of RANKL in adult
hosts could both selectively and transiently inhibit the development and turnover of mTECs with potential to alter central
T cell tolerance. To this end, we performed in vivo RANKL
blockade in adult mice and investigated its effects on both TECs
and developing thymocytes.We show that anti-RANKL treatment not only depleted mTECs but could also be used therapeutically to break central tolerance and, as a result, increase
the generation of tumor-specific T cells.
RESULTS AND DISCUSSION
Depletion of mTECs with RANKL blockade
The RANKRANKL signaling pathway is important for
mTEC development, but it remains unclear what impact perturbation of this pathway might have on the adult thymus.
Previous work has linked the RANKRANKL pathway to
the development of Aire-expressing mTECs (Rossi et al.,
2007; Akiyama et al., 2008; Hikosaka et al., 2008; Roberts
et al., 2012), and we hypothesized that treating mice with
blocking anti-RANKL antibody would decrease Aire+ mTECs.
RANKL blockade enhances anti-tumor immunity | Khan et al.
Br ief Definitive Repor t
We treated wild-type mice with either anti-RANKL or isotype

control antibody for 2 wk and harvested their thymi for analy
sis. Interestingly, although mice treated with anti-RANKL
showed only a modest loss of cortical thymic epithelial cell
(cTEC) cellularity, they exhibited a severe depletion of >80%
of mTECs (Fig. 1 A). Using MHC II and Aire as markers of
mTEC maturation (Gbler et al., 2007; Gray et al., 2007), we
further analyzed the immature mTEClo (MHC IIlo Aire),
intermediate mTEChi (MHC IIhi Aire), and mature Aire+
mTEC subsets (MHC IIhi Aire+).The relative mTEC composition in anti-RANKLtreated mice revealed a substantial loss
of Aire+ and mTEChi cells along with enrichment of the remaining mTEClo cells (Fig. 1 B). Although absolute numbers
of mTECs were decreased across all mTEC subsets, mTEC
depletion was mostly due to the loss of >90% of all Aire+ and
mTEChi cells (Fig. 1 B). Immunostaining revealed a slight decrease in the density of keratin-5+ (K5) cells in the medulla of
anti-RANKLtreated mice, whereas Aire+ cells were nearly
undetectable (Fig. 1 C). Importantly, staining for keratin-8
(K8) and K5 showed that the overall corticomedullary thymic
architecture was preserved despite the loss of mature mTECs
(Fig. 1 C). Furthermore, consistent with the loss of the Aire+
and mTEChi subsets, Aire-dependent TSA gene expression in
sorted mTECs was decreased in anti-RANKLtreated mice
(Fig. 1 D).
Next, we characterized the impact of anti-RANKLmediated mTEC depletion on thymocyte selection. In the polyclonal T cell repertoire of wild-type mice treated with antiRANKL, we observed a modest increase in frequencies of
both CD4 single-positive (SP) and CD8 SP thymocytes, consistent with a lack of negative selection (Fig. 1 E). Importantly,
anti-RANKLtreated mice showed only a slight reduction in
the frequency of double-positive thymocytes while absolute
numbers were maintained, confirming normal positive selection. In addition, total thymocyte numbers were also maintained
in mice treated with anti-RANKL (Fig. 1 E). Furthermore,
anti-RANKLtreated mice showed a 50% reduction of Foxp3+
T reg cells within the CD4 SP subset (Fig. 1 F).
Given the dramatic impact of RANKL blockade on
mTECs, we sought to determine whether this effect could be
reversed in the context of increased RANK signaling. Osteoprotegerin (OPG; Tnfrsf11b) is a soluble decoy receptor for
RANKL and its role as a negative regulator of RANK signaling has been well described in bone physiology (Kearns et al.,
2008). Interestingly, in sorted wild-type mTECs, OPG expression was up-regulated in MHC IIhi mTECs when compared with MHC IIlo mTECs (Fig. 2 A). To test whether
OPG negatively regulates RANKRANKL signaling in
mTECs, we analyzed thymi from OPG/ mice.While cTEC
cellularity was increased threefold in OPG/ mice, mTEC
cellularity was increased by nearly 10-fold when compared
with OPG+/+ mice (Fig. 2 B). Although absolute numbers of
all mTEC subsets were increased in OPG/ mice, we observed an enrichment of Aire+ mTECs and a proportional
loss of the mTEClo subset (Fig. 2 C). Immunostaining of
K8 and K5 showed that OPG/ thymi maintained their
JEM Vol. 211, No. 5
Figure 2. Increased Aire+ mTECs in OPG/ mice. (A) Quantitative

PCR analysis of Aire and OPG gene expression in MHC IIlo and MHC IIhi
mTECs sorted from wild-type B6 mice. Results standardized to Cyclophilin
A and normalized to MHC IIlo, with error bars depicting mean SD.
(B) Absolute numbers of mTECs and cTECs in OPG+/+ (black) and OPG/
(red) mice were enumerated by flow cytometry. Bar graphs of total cell
numbers are depicted by mean SEM. (C) Representative flow cytometry
plots of mTECs in B showing relative composition of indicated mTEC subsets. Values depict mean SEM. Bar graphs (right) depict total cell numbers of each mTEC subset and indicate mean SEM. (D) Immunostaining
for keratin-8 (green) and keratin-5 (red) on frozen thymic sections from
OPG+/+ and OPG/ mice. Data are representative of at least three independent experiments with at least three mice per group. *, P 0.05;
**, P 0.01; ***, P 0.001, Students t test. Bars, 500 m.
corticomedullary thymic architecture despite these changes

in TEC cellularity and mTEC composition (Fig. 2 D).
Collectively, these data suggest that RANKRANKL
OPG interactions regulate both TEC cellularity and mTEC
maturation, and are particularly important for the induction of
Aire+ mTECs. Furthermore, anti-RANKL treatment selectively targeted mature mTECs and altered negative selection
without affecting the development of new thymocytes.
Regeneration of mTECs after withdrawal of anti-RANKL
We next examined the kinetics of mTEC recovery after withdrawal from anti-RANKL blockade. After a 2-wk treatment
window, we harvested mice at 2-wk intervals and observed
gradual recovery of the mTEChi and Aire+ subsets (Fig. 3,
A and B). By 10 wk, we observed the complete recovery of
all mTEC subsets and a normal level of Aire expression after
anti-RANKL treatment (Fig. 3 B). Interestingly, the pattern of
mTEC recovery appeared consistent with published reports
of mTEClo cells giving rise to mTEChi cells before the induction
of Aire+ mTECs (Fig. 3 B; Gbler et al., 2007; Gray et al., 2007;
Rossi et al., 2007). Notably, thymi from anti-RANKL and
763
Figure 3. Regeneration of mTECs after anti-RANKL withdrawal. (A) Experimental layout for analysis of mTEC recovery
after anti-RANKL withdrawal. Wild-type mice were treated for
1 wk with isotype or anti-RANKL (black arrows) and harvested at
indicated time points (red arrows). Graph (bottom) depicts relative composition of mTEClo mTEChi and Aire+ mTECs at each time
point in anti-RANKLtreated mice. Values represent mean
SEM. (B) Representative flow cytometry plot of mTECs from
each of the indicated time points from A. Values represent the
frequency of Aire+ mTECs. (C) Immunostaining for keratin-8
(green) and keratin-5 (red) on thymic sections from isotype or
anti-RANKLtreated mice harvested at 10 wk. Bars, 500 m.
Data shown is representative of at least two independent experiments with three to five mice per group.
isotype-treated mice at 10 wk were indistinguishable by immunostaining (Fig. 3 C). Thus, anti-RANKLmediated mTEC
depletion is a transient phenomenon with return of normal
thymic composition after withdrawal of antibody treatment.
Manipulation of thymic negative selection
with RANKL blockade
To further characterize the impact of anti-RANKL treatment
on negative selection, we first used the OT-II CD4+ TCR
x RIP-mOVA double transgenic mouse model. RIP-mOVA
transgenic mice express a membrane form of OVA under the

control of the rat insulin promoter which results in its expression in both the pancreatic islets and in mTECs (Anderson et al.,
2005). When the OT-II CD4+ TCR transgenic line is crossed
to the RIP-mOVA transgenic line, OVA-specific OT-II T cells
are deleted in the thymus (Anderson et al., 2005). We treated
both OT-II and OT-II x RIP-mOVA mice with either antiRANKL or isotype control antibody and performed thymocyte
analysis. Consistent with previous reports, OT-II x RIP-mOVA
mice treated with control antibody showed a significant
Figure 4. Anti-RANKLmediated mTEC
ablation leads to altered negative selection.
(A) OT-II and OT-II x RIP-mOVA mice were harvested after 2 wk of indicated antibody treatment. Percentages of CD4 SP (top) and OT-II
clonotype-specific (bottom) thymocytes are
shown. Values represent mean SD. (B) Quantification of OT-II clonotype-specific CD4 SP cells
from A. Graph shows mean SEM. **, P 0.01,
Students t test. (C) Representative histograms
for Foxp3 staining in CD4 SP events from A.
(D) Wild-type mice were treated with isotype or
anti-RANKL antibody for 3 wk, immunized with
P2 peptide, and then harvested 10 d later for
tetramer analysis. Representative flow cytometry plots of CD4+ T cells are shown for each
condition. Values indicate absolute numbers of
CD44+ P2-I-Abspecific T cells. IRBP/ mice
were included as positive controls. Each dot on
the graph represents an individual mouse, bars
show mean, and the dotted line represents the
limit of detection. Data shown is representative
of two to three independent experiments.
*, P 0.05, Mann-Whitney test.
764
Figure 5. RANKL blockade increases

anti-melanoma immune response.
(A and B) RAG1/ x TRP-1 TCR Tg mice were
treated with isotype or anti-RANKL antibody
for 2 wk and thymus and spleen were harvested for analysis. Flow cytometry plots in A
show percentage of CD4+ (top) and CD4-gated
V14+ (bottom) T cells. Bar graphs in B show
quantification of data in A. Values represent
mean SEM. *, P 0.05; **, P 0.01;
***, P 0.001, Students t test. Data shown is
representative of at least three independent
experiments with n = 47 mice per group.
(C) Ear skin from mice treated with isotype (gray)
or anti-RANKL (blue) antibody was analyzed
by qPCR for Tyrosinase and Tyrp1 expression.
Results standardized to 2m and normalized
to untreated B6 wild-type mice (black) with
bars depicting mean SD. (D) Survival curves
of RAG1/ x TRP-1 TCR Tg mice inoculated
with B16 melanoma after treatment with isotype (gray) or anti-RANKL (blue) antibody,
n = 78 mice for each group. P 0.05, Log-rank
test. Data shown is representative of four independent experiments. (E) Tumor sections from
isotype or anti-RANKLtreated mice in D were
immunostained for CD3 (pink) with DAPI counterstain (purple). Bars, 50 m. Mean densities
of CD3+ cells per tumor area are quantified on
the right. Four sections were evaluated for
each treatment group and 6 imaging fields
were randomly scored from each section.
Graphs depict mean SEM. *, P 0.05, Students t test. (F) RAG1/ x TRP-1 TCR Tg mice
were inoculated with B16 melanoma after
treatment with isotype (gray) or anti-RANKL
(blue) antibody. Mice were harvested at 21 d
after tumor inoculation and tumor-infiltrating
lymphocytes (TILs) were analyzed by flow cytometry. Graph depicts percentage of Foxp3+
CD25+ T reg cells among CD4+ T cells. Values
represent mean SEM. ***, P 0.001, Students t test. Data shown is n = 79 mice per group pooled from two independent experiments. (G) Survival curves
of RAG1/ recipients inoculated with B16 melanoma after adoptive transfer of splenocytes from either anti-RANKL (blue) or isotype (gray)treated
RAG1/ x TRP-1 TCR Tg mice, n = 10 mice for each group. P 0.001, Log-rank test. Data shown is representative of two independent experiments.
reduction in the proportions of CD4 SP thymocytes and also

had decreased frequencies and numbers of OT-II+ T cells (Fig. 4,
A and B; Anderson et al., 2005). In contrast, thymocyte profiles of anti-RANKLtreated OT-II x RIP-mOVA mice were
indistinguishable from that of OT-II mice, demonstrating that
RANKL blockade prevented thymic deletion of OT-II T cells
yet allowed their positive selection (Fig. 4, A and B). We also
observed a loss of thymic T reg cell development in these
mice, which suggested a loss of cognate antigen (OVA) expression from mTECs within the thymus (Fig. 4 C).
To expand these observations to the polyclonal T cell repertoire, we analyzed the development of Aire-dependent autoreactive T cells using a tetramer enrichment protocol. Previously,
we had shown that T cells specific for the self-antigen interphotoreceptor retinoid-binding protein (IRBP) are deleted in
JEM Vol. 211, No. 5
thymus of Aire+/+ mice, whereas these cells escape deletion

in Aire/ mice and cause autoimmune uveitis (DeVoss et al.,
2006; Taniguchi et al., 2012). Through the use of an IRBP peptide class II tetramer, P2-I-Ab, such autoreactive CD4+ T cells
can be detected in Aire/ mice. Given both the severe depletion of Aire+ mTECs and the loss of thymic IRBP expression
in anti-RANKLtreated mice, we hypothesized that P2-I-Ab
specific T cells could be detected in treated mice. After treating
wild-type mice with anti-RANKL antibody, we immunized
mice with a MHC IIbinding IRBP peptide epitope (P2) to
expand T cells for detection. 10 d after immunization, lymph
nodes and spleen were pooled for the enumeration of CD4+
P2-I-Abspecific T cells by flow cytometry. Consistent with the
loss of Aire+ mTECs, tetramer analysis of anti-RANKLtreated
mice revealed an expansion of CD4+ P2-I-Abspecific T cells
765
(Fig. 4 D). Overall, these data show a clear defect in negative

selection associated with the loss of Aire+ mTECs in antiRANKLtreated mice.
Treatment with anti-RANKL enhances anti-tumor immunity
Given its ability to selectively block mTECs and central tolerance, we next sought to determine whether anti-RANKL
treatment could be used to therapeutically enhance anti-tumor
immunity. Previous work has shown that many tumor selfantigens are expressed by mTECs as TSAs, and that high-affinity
T cells capable of recognizing these tumor self-antigens are efficiently deleted in the thymus (Bos et al., 2005; Trger et al.,
2012; Zhu et al., 2013). To test whether anti-RANKL treatment could rescue tumor self-antigenspecific T cells from
thymic deletion, we used the TRP-1 CD4+ TCR transgenic
mouse model which generates T cells specific for the melanoma antigen TRP-1 (Tyrp1; Muranski et al., 2008). TRP-1 is
expressed in B16 melanoma cells, and TRP-1 T cells are efficacious against established B16 melanoma tumors (Muranski et al.,
2008). Importantly, mTECs endogenously express TRP-1 in an
Aire-dependent manner, such that TRP-1specific T cells are
deleted in the thymi of Aire-sufficient, TRP-1sufficient hosts
(Muranski et al., 2008; Zhu et al., 2013). Given the depletion of
Aire+ mTECs and loss of thymic TRP-1 expression in antiRANKLtreated mice (Fig. 1 D), we hypothesized that antiRANKL treatment could rescue TRP-1 T cells from thymic
deletion. We treated RAG1/ x TRP-1 TCR transgenic mice
with either anti-RANKL or control antibody and harvested
their thymi and spleens for analysis. Consistent with previous
reports, isotype-treated mice showed efficient deletion of CD4
SP T cells in the thymus (Fig. 5 A; Muranski et al., 2008; Zhu
et al., 2013). In contrast, anti-RANKL treatment prevented
deletion of CD4 SP cells and also resulted in a much higher
percentage of V14+ T cells (Fig. 5, A and B). Furthermore,
TRP-1 T cells were detectable in the spleens of anti-RANKL
treated mice and also expressed higher levels of the V14 TCR
(Fig. 5, A and B). In addition, qPCR analysis revealed a loss of
Tyrosinase and Tyrp1 expression in the ear skin of anti-RANKL
treated mice which appeared consistent with the destruction of
melanocytes by TRP-1 T cells (Fig. 5 C).
Next, we challenged anti-RANKLtreated RAG1/ x
TRP-1 mice with B16 melanoma to determine whether a limited break in central tolerance could improve overall survival.
We observed a statistically significant increase in the survival of
anti-RANKLtreated mice compared with the isotype-treated
cohort (Fig. 5 D).We also found evidence of an enhanced T cell
response in anti-RANKLtreated mice by measuring CD3+
T cell infiltrates within the tumors of these mice (Fig. 5 E). Of
note, given the overall increase in T reg cells in the spleen and
tumors of anti-RANKLtreated mice, it appears unlikely that
the survival benefit observed in TRP-1 mice is due to differences in T reg cell numbers (Fig. 5, B and F). Furthermore, to
exclude potential effects of anti-RANKL antibody on the
tumor microenvironment, we performed adoptive transfer
studies in which splenocytes from antibody-treated RAG1/
x TRP-1 mice were transferred into RAG1/ mice. When
766
challenged with B16 melanoma, recipients of splenocytes from

anti-RANKLtreated donors again showed a statistically significant increase in survival (Fig. 5 G). Collectively, these data
demonstrate that short-term, reversible RANKL blockade of
mTEC development can be used to create a therapeutic window that allows tumor self-antigenspecific T cells to escape
thymic deletion. Furthermore, the generation of these highaffinity tumor-specific T cells confers a survival benefit in a
melanoma tumor model.
In conclusion, our findings provide strong evidence that
mTECs can be selectively and therapeutically targeted by
RANKL blockade in adult mice and that anti-RANKL can
be used as an approach to enhance anti-tumor responses. To
date, previous efforts on the therapeutic manipulation of
TECs have shown global effects that involve both cTECs and
mTECs and disrupt thymocyte development (Fletcher et al.,
2009). The selectivity for mTECs with anti-RANKL thus
provides evidence that central tolerance can be transiently suppressed in adult hosts while maintaining T cell generation. Although autoimmunity is a dangerous potential consequence
of this approach, the treatment may also be an attractive new
method to help break tolerance for cancer immunotherapy.
Interestingly, although we could detect an expansion of
retina-specific T cells in anti-RANKLtreated mice, we did
not observe development of spontaneous uveitis (unpublished
data), suggesting that this brief window of central tolerance
suppression was countered by peripheral tolerance mechanisms such as T reg cells which existed before treatment. Importantly, we find that the medullary epithelial compartment
of the thymus recovers after withdrawal of anti-RANKL antibody and, thus, only a transient window of central tolerance
suppression occurs. This window may be an attractive feature
of this approach, especially if coupled with methods that preferentially expand tumor-specific T cells over pathogenic autoreactive T cells. Currently, there has been intense interest and
progress in manipulating peripheral tolerance for immuno
therapy (Chen and Mellman, 2013). Given our results, it will
also be of interest to determine if a combination of methods
that target both central and peripheral tolerance could further enhance anti-tumor immune responses. Finally, it is important to note that our results also have implications for patients
in the clinical setting receiving denosumab, a humanized monoclonal antibody that blocks RANKL which is widely used in
the treatment of osteoporosis in adults (Cummings et al., 2009).
Further study will be needed in such patients regarding their
susceptibility to autoimmunity and for other potential defects
in central tolerance.
Mice. C57BL/6, B6.RAG1/ Tyrp1B-w TRP-1 TCR transgenic, and B6.
OPG/ mice were purchased from The Jackson Laboratory. Mice were
treated intraperitoneally with 100 g anti-RANKL (IK22/5) or isotype control (2A3; Bio X Cell) antibody three times per week as stated in the text.
B6.OT-II, B6.RIP-mOVA, and B6.IRBP/ mice were described previously
(Anderson et al., 2005; DeVoss et al., 2006). Mice were treated at 35 wk of
age and harvested at time points indicated in the text. All mice were housed
and bred in specific pathogen-free conditions in animal facilities at UCSF or
UNC, Chapel Hill. Animal experiments were approved by the Institutional

Animal Care and Use Committee (IACUC) at UCSF or UNC, Chapel Hill.
Submitted: 6 September 2013

Accepted: 3 April 2014
Quantitative PCR. RNA was extracted from sorted mTECs using the
RNeasy Micro Plus kit (QIAGEN) and cDNA was synthesized with the Invitrogen Superscript III kit. TaqMan gene expression assays (Applied Biosystems) were used for all targets.
REFERENCES
Histology and immunofluorescence. Thymi were harvested and embedded in O.C.T. media (Tissue-Tek). 8-m frozen thymic sections were fixed in
100% acetone, blocked, and stained for keratin-5, keratin-8 (Abcam), or Aire
(eBioscience). Secondary antibodies were purchased from Invitrogen. Thymic
sections were visualized using a widefield microscope (Apotome; Carl Zeiss).
For tumor immunostaining, tumors were harvested and fixed in 10% formalin
as previously described (Zhu et al., 2013). Fixed tumors were embedded in
paraffin and sectioned for staining with anti-CD3 antibody and counterstained
with DAPI. Tumor sections were visualized using a fluorescent microscope
(BX60; Olympus) and analyzed using ImageJ software (National Institutes
of Health).
Flow cytometry. TECs were isolated as previously described (Gardner
et al., 2008). In brief, thymi were minced and digested with DNase I and
Liberase TM (Roche) before gradient centrifugation with Percoll PLUS (GE
Healthcare). Enriched stromal cells were stained with the indicated surface
marker antibodies (BioLegend). mTECs were defined as CD45, EpCAM+,
MHC II+, Ly51 events and cTECs were defined as CD45, EpCAM+,
MHC II+, Ly51+ events. For lymphocyte staining, all surface marker antibodies were obtained from BioLegend. For intracellular staining, cells were
stained using the Foxp3 Staining Buffer Set and stained with anti-Foxp3
or anti-Aire (eBioscience). All data were collected using a flow cytometer
(LSR II; BD) and analyzed with either FlowJo software (Tree Star) or FACS
Diva (BD). Cell sorting was performed using a FACS Aria III cell sorter (BD).
IRBP P2 peptide immunization and tetramer analysis. Mice were
immunized with 100 g P2 peptide (IRBP; aa 271290) emulsified in complete Freunds adjuvant as described previously (Taniguchi et al., 2012).
Tetramer analysis was performed on pooled lymph nodes and spleen from
treated mice 10 d after immunization. P2-I-Ab tetramer was generated by the
National Institutes of Health Tetramer Core Facility, and tetramer staining
was performed as described previously (Taniguchi et al., 2012). After tetramer
enrichment, cells were stained with antibodies for flow cytometry, and counting beads (Invitrogen) were used to enumerate tetramer+ cells.
B16 melanoma tumor challenge. B6.RAG1/TRP-1 TCR transgenic
mice were injected subcutaneously in the left flank with 1.0 105 B16 melanoma cells. For adoptive transfer studies, spleens from donor mice were
pooled and CD25-depleted before transfer into B6.RAG1/ hosts. Recipient mice were inoculated with 7.5 104 B16 melanoma cells 7 d after transfer. Tumor growth was monitored by taking measurements of length (L) and
width (W), and tumor volume was calculated as LW2/2. For generation of
survival curves, death was defined as tumor size >1,000 mm3.
Statistical analysis. Statistical analysis was performed using Prism 6.0
(GraphPad Software). Mann-Whitney Rank sum testing was performed on
tetramer analysis. Students t test was performed for TEC and lymphocyte
analyses. Log-rank test was performed for Kaplan-Meier survival curves.
We thank T. Metzger, T. LaFlam, M. Cheng, and W. Purtha for critical reading of the
manuscript. We thank the National Institutes of Health Tetramer Core Facility for
providing tetramer reagent.
This work was supported by the US National Institutes of Health Grants
AI097457 (M.S. Anderson) and K12-GM081266 (M.L. Mouchess), and the UCSF
Medical Scientist Training Program (I.S. Khan). Flow cytometry data were generated
in the UCSF Parnassus Flow Cytometry Core, which is supported by the Diabetes
and Endocrinology Research Center (DERC) grant NIH P30 DK063720.
JEM Vol. 211, No. 5
Akiyama, T.,Y. Shimo, H.Yanai, J. Qin, D. Ohshima,Y. Maruyama, Y. Asaumi,

J. Kitazawa, H. Takayanagi, J.M. Penninger, et al. 2008. The tumor necrosis factor family receptors RANK and CD40 cooperatively establish
the thymic medullary microenvironment and self-tolerance. Immunity.
29:423437. http://dx.doi.org/10.1016/j.immuni.2008.06.015
Anderson, M.S., E.S.Venanzi, L. Klein, Z. Chen, S.P. Berzins, S.J. Turley, H. von
Boehmer, R. Bronson, A. Dierich, C. Benoist, and D. Mathis. 2002. Projec
tion of an immunological self shadow within the thymus by the aire protein. Science. 298:13951401. http://dx.doi.org/10.1126/science.1075958
Anderson, M.S., E.S.Venanzi, Z. Chen, S.P. Berzins, C. Benoist, and D. Mathis.
2005.The cellular mechanism of Aire control of T cell tolerance. Immunity.
23:227239. http://dx.doi.org/10.1016/j.immuni.2005.07.005
Anz, D., R.Thaler, N. Stephan, Z.Waibler, M.J.Trauscheid, C. Scholz, U. Kalinke,
W. Barchet, S. Endres, and C. Bourquin. 2009. Activation of melanoma
differentiation-associated gene 5 causes rapid involution of the thymus. J.
Immunol. 182:60446050. http://dx.doi.org/10.4049/jimmunol.0803809
Bos, R., S. van Duikeren,T. van Hall, P. Kaaijk, R.Taubert, B. Kyewski, L. Klein,
C.J. Melief, and R. Offringa. 2005. Expression of a natural tumor antigen by thymic epithelial cells impairs the tumor-protective CD4+ T-cell
repertoire. Cancer Res. 65:64436449. http://dx.doi.org/10.1158/00085472.CAN-05-0666
Chen, D.S., and I. Mellman. 2013. Oncology meets immunology: the cancerimmunity cycle. Immunity. 39:110. http://dx.doi.org/10.1016/j.immuni
.2013.07.012
Cloosen, S., J. Arnold, M. Thio, G.M. Bos, B. Kyewski, and W.T. Germeraad.
2007. Expression of tumor-associated differentiation antigens, MUC1
glycoforms and CEA, in human thymic epithelial cells: implications for
self-tolerance and tumor therapy. Cancer Res. 67:39193926. http://
dx.doi.org/10.1158/0008-5472.CAN-06-2112
Consortium, F.G.A.; Finnish-German APECED Consortium. 1997. An auto
immune disease, APECED, caused by mutations in a novel gene featuring
two PHD-type zinc-finger domains. Nat. Genet. 17:399403. http://
dx.doi.org/10.1038/ng1297-399
Cummings, S.R., J. San Martin, M.R. McClung, E.S. Siris, R. Eastell, I.R.
Reid, P. Delmas, H.B. Zoog, M. Austin, A. Wang, et al. FREEDOM
Trial. 2009. Denosumab for prevention of fractures in postmenopausal
women with osteoporosis. N. Engl. J. Med. 361:756765. http://dx.doi
.org/10.1056/NEJMoa0809493
Derbinski, J., A. Schulte, B. Kyewski, and L. Klein. 2001. Promiscuous gene
expression in medullary thymic epithelial cells mirrors the peripheral
self. Nat. Immunol. 2:10321039. http://dx.doi.org/10.1038/ni723
DeVoss, J., Y. Hou, K. Johannes, W. Lu, G.I. Liou, J. Rinn, H. Chang,
R.R. Caspi, L. Fong, and M.S. Anderson. 2006. Spontaneous autoimmunity prevented by thymic expression of a single self-antigen. J. Exp.
Med. 203:27272735. http://dx.doi.org/10.1084/jem.20061864
Fletcher, A.L., T.E. Lowen, S. Sakkal, J.J. Reiseger, M.V. Hammett, N.
Seach, H.S. Scott, R.L. Boyd, and A.P. Chidgey. 2009. Ablation and
regeneration of tolerance-inducing medullary thymic epithelial cells
after cyclosporine, cyclophosphamide, and dexamethasone treatment.
J. Immunol. 183:823831. http://dx.doi.org/10.4049/jimmunol.0900225
Gbler, J., J. Arnold, and B. Kyewski. 2007. Promiscuous gene expression and
the developmental dynamics of medullary thymic epithelial cells. Eur.
J. Immunol. 37:33633372. http://dx.doi.org/10.1002/eji.200737131
Gardner, J.M., J.J. Devoss, R.S. Friedman, D.J. Wong,Y.X. Tan, X. Zhou, K.P.
Johannes, M.A. Su, H.Y. Chang, M.F. Krummel, and M.S. Anderson.
2008. Deletional tolerance mediated by extrathymic Aire-expressing
cells. Science. 321:843847. http://dx.doi.org/10.1126/science.1159407
Gray, D., J. Abramson, C. Benoist, and D. Mathis. 2007. Proliferative arrest
and rapid turnover of thymic epithelial cells expressing Aire. J. Exp. Med.
204:25212528. http://dx.doi.org/10.1084/jem.20070795
Hikosaka, Y., T. Nitta, I. Ohigashi, K. Yano, N. Ishimaru, Y. Hayashi, M.
Matsumoto, K. Matsuo, J.M. Penninger, H.Takayanagi, et al. 2008.The
cytokine RANKL produced by positively selected thymocytes fosters
medullary thymic epithelial cells that express autoimmune regulator.
Immunity. 29:438450. http://dx.doi.org/10.1016/j.immuni.2008.06.018
767
Kearns, A.E., S. Khosla, and P.J. Kostenuik. 2008. Receptor activator of nuclear
factor kappaB ligand and osteoprotegerin regulation of bone remodeling
in health and disease. Endocr. Rev. 29:155192. http://dx.doi.org/10
.1210/er.2007-0014
Kyewski, B., and L. Klein. 2006. A central role for central tolerance. Annu.
Rev. Immunol. 24:571606. http://dx.doi.org/10.1146/annurev.immunol
.23.021704.115601
Liston, A., S. Lesage, J. Wilson, L. Peltonen, and C.C. Goodnow. 2003.
Aire regulates negative selection of organ-specific T cells. Nat. Immunol.
4:350354. http://dx.doi.org/10.1038/ni906
Malchow, S., D.S. Leventhal, S. Nishi, B.I. Fischer, L. Shen, G.P. Paner, A.S.
Amit, C. Kang, J.E. Geddes, J.P. Allison, et al. 2013. Aire-dependent thymic development of tumor-associated regulatory T cells. Science. 339:1219
1224. http://dx.doi.org/10.1126/science.1233913
Metzger, T.C., and M.S. Anderson. 2011. Control of central and peripheral tolerance by Aire. Immunol. Rev. 241:89103. http://dx.doi.org/10
.1111/j.1600-065X.2011.01008.x
Muranski, P., A. Boni, P.A. Antony, L. Cassard, K.R. Irvine, A. Kaiser, C.M.
Paulos, D.C. Palmer, C.E. Touloukian, K. Ptak, et al. 2008. Tumorspecific Th17-polarized cells eradicate large established melanoma. Blood.
112:362373. http://dx.doi.org/10.1182/blood-2007-11-120998
Nagamine, K., P. Peterson, H.S. Scott, J. Kudoh, S. Minoshima, M. Heino, K.J.
Krohn, M.D. Lalioti, P.E. Mullis, S.E. Antonarakis, et al. 1997. Positional
cloning of the APECED gene. Nat. Genet. 17:393398. http://dx.doi
.org/10.1038/ng1297-393
Roberts, N.A., A.J. White, W.E. Jenkinson, G. Turchinovich, K. Nakamura,
D.R. Withers, F.M. McConnell, G.E. Desanti, C. Benezech, S.M. Parnell,
768
et al. 2012. Rank signaling links the development of invariant T cell

progenitors and Aire(+) medullary epithelium. Immunity. 36:427437.
http://dx.doi.org/10.1016/j.immuni.2012.01.016
Rossi, S.W., M.Y. Kim, A. Leibbrandt, S.M. Parnell, W.E. Jenkinson, S.H.
Glanville, F.M. McConnell, H.S. Scott, J.M. Penninger, E.J. Jenkinson,
et al. 2007. RANK signals from CD4+3 inducer cells regulate development of Aire-expressing epithelial cells in the thymic medulla. J. Exp.
Med. 204:12671272. http://dx.doi.org/10.1084/jem.20062497
Shum, A.K., J. DeVoss, C.L. Tan, Y. Hou, K. Johannes, C.S. OGorman,
K.D. Jones, E.B. Sochett, L. Fong, and M.S. Anderson. 2009.
Identification of an autoantigen demonstrates a link between interstitial
lung disease and a defect in central tolerance. Sci. Transl. Med. 1:9ra20.
http://dx.doi.org/10.1126/scitranslmed.3000284
Swann, J.B., and M.J. Smyth. 2007. Immune surveillance of tumors. J. Clin.
Invest. 117:11371146. http://dx.doi.org/10.1172/JCI31405
Taniguchi, R.T., J.J. DeVoss, J.J. Moon, J. Sidney, A. Sette, M.K. Jenkins, and
M.S. Anderson. 2012. Detection of an autoreactive T-cell population
within the polyclonal repertoire that undergoes distinct autoimmune
regulator (Aire)-mediated selection. Proc. Natl. Acad. Sci. USA. 109:7847
7852. http://dx.doi.org/10.1073/pnas.1120607109
Trger, U., S. Sierro, G. Djordjevic, B. Bouzo, S. Khandwala, A. Meloni, M.
Mortensen, and A.K. Simon. 2012. The immune response to melanoma
is limited by thymic selection of self-antigens. PLoS ONE. 7:e35005.
http://dx.doi.org/10.1371/journal.pone.0035005
Zhu, M.L., A. Nagavalli, and M.A. Su. 2013. Aire deficiency promotes
TRP-1-specific immune rejection of melanoma. Cancer Res. 73:2104
2116. http://dx.doi.org/10.1158/0008-5472.CAN-12-3781
Article
Genomic and bioinformatic profiling

of mutational neoepitopes reveals new rules
to predict anticancer immunogenicity
Fei Duan,1 Jorge Duitama,2 Sahar Al Seesi,2 Cory M. Ayres,3
Steven A. Corcelli,3 Arpita P. Pawashe,1 Tatiana Blanchard,1
David McMahon,1 John Sidney,4 Alessandro Sette,4 Brian M. Baker,3
Ion I. Mandoiu,2 and Pramod K. Srivastava1
1Department
of Immunology and Carole and Ray Neag Comprehensive Cancer Center, University of Connecticut School
of Medicine, Farmington, CT 06030
2Department of Computer Science and Engineering, University of Connecticut, Storrs, CT 06269
3Department of Chemistry and Biochemistry and Harper Cancer Research Institute, University of Notre Dame,
Notre Dame, IN 46556
4LaJolla Institute of Allergy and Immunology, La Jolla, CA 92037
The mutational repertoire of cancers creates the neoepitopes that make cancers immunogenic. Here, we introduce two novel tools that identify, with relatively high accuracy, the
small proportion of neoepitopes (among the hundreds of potential neoepitopes) that protect the host through an antitumor T cell response. The two tools consist of (a) the numerical difference in NetMHC scores between the mutated sequences and their unmutated
counterparts, termed the differential agretopic index, and (b) the conformational stability
of the MHC Ipeptide interaction. Mechanistically, these tools identify neoepitopes that
are mutated to create new anchor residues for MHC binding, and render the overall
peptide more rigid. Surprisingly, the protective neoepitopes identified here elicit CD8dependent immunity, even though their affinity for Kd is orders of magnitude lower
than the 500-nM threshold considered reasonable for such interactions. These results
greatly expand the universe of target cancer antigens and identify new tools for human
cancer immunotherapy.
CORRESPONDENCE
Pramod Srivastava:
Srivastava@uchc.edu
OR
Ion I. Mandoiu:
ion@engr.uconn.edu
Abbreviations used: DAI,
differential agretopicity index;
RMSD, root mean square
deviation; RMSF, root mean
square fluctuation; SNV,
single-nucleotide variant.
The idea that neoepitopes created by random

mutations in tumor cells, termed as individually
specific tumor antigens or unique antigens, are
responsible for the immunogenicity of tumors
has been around for over 20 yr (Srivastava, 1993).
There has been strong experimental evidence
for their existence and activity in murine and
human tumors (Duan et al., 2009; van der
Bruggen et al., 2013), and mathematic modeling
has predicted the existence of tens to hundreds
of neoepitopes in individual human tumors
(Srivastava and Srivastava 2009). The recent
revolution in high-throughput DNA sequencing
and accompanying bioinformatics approaches
has finally made it possible to actually identify the
individually specific neoepitopes in individual
cancers. Using this methodology, Segal et al.

(2008) showed that human breast and colon
cancers harbor tens of putative mutational neoepitopes; Rajasagi et al. (2014) have also published similar data for chronic myelogenous
leukemia and other human cancers. A genomic/
bioinformatic approach to identify such epitopes
in a mouse melanoma also led to identification
of hundreds of epitopes (Castle et al., 2012).
A similar approach led to identification of a
neoepitope in a methylcholanthrene-induced
sarcoma in an immunocompromised mouse;
transplantation of this tumor into an immunecompetent animal led to epitope-dependent
tumor regression (Matsushita et al., 2012). In
human studies, association of favorable clinical
F. Duan, J. Duitama, and S. Al Seesi contributed equally to

this paper.
J. Duitamas present address is Agrobiodiversity Research
Area, International Center for Tropical Agriculture (CIAT),
6713 Cali, Colombia.
2014 Duan et al. This article is distributed under the terms of an Attribution

J. Exp. Med. 2014 Vol. 211 No. 11 2231-2248
2231
course of disease with a dominant immune response to mutated neoepitopes has been demonstrated (Lennerz et al.,
2005; Lu et al., 2013; van Rooij et al., 2013). These growing
numbers of studies strongly suggest that the host immune response to mutant neoepitopes plays the dominant role in protection of the host from tumor growth.
The opportunity to identify a vast number of putative
neoepitopes from individual human tumors creates a corresponding problem: how does one differentiate and identify
actual tumor protective neoepitopes from among the large
number of putative neoepitopes identified in silico? The problem is daunting in scale because an examination of the tumor
transcriptomes and their comparison with normal exomes in
the TCGA database shows that many tumors harbor hundreds
of putative neoepitopes. Presumably, only a small fraction of
these virtual neoepitopes are immunoprotective against cancer.
This question has been addressed before in viral systems.
Assarsson et al. (2007) performed a systematic analysis of the
putative and real epitopes of the vaccinia virus. This study revealed the magnitude of the problem: starting from all possible
910 amino acid peptides encoded by the vaccinia genome,
only 2.5% are high-affinity binders to a given HLA allele. Of
the high-affinity binders, half elicit a CD8 response. Of these,
only 15% are naturally processed and presented. Finally, they
observed little correlation between the dominance of an
epitope with HLApeptide affinity, HLApeptide stability,
TCR avidity, or the quantity of processed epitope. Thus,
without the benefit of information from T cell responses, one
would be unable to start from the vaccinia genome and identify useful epitopes.
The problem is orders of magnitude more complex for
identifying useful epitopes from cancer genomes because the
mammalian genome is considerably larger than that of vaccinia. Moreover, viral genomes are entirely nonself, whereas
the cancer genomes are mutated-self; hence the neoepitopes
may be cross-reactive with self, and tolerance or suppression
mechanisms are highly likely to come into play.
Here, we make the first systematic analysis of the transcriptomes and CD8 immunomes of tumors, and seek to
understand the rules that govern the immunogenicity and
tumor-protective ability of mutation-generated neoepitopes.
This effort has led to surprising observations regarding MHC I
peptide interactions that distinguish the recognition of neoepitopes from that of viral epitopes, and a recognition that the
proportion of putative mutational neoepitopes that is translatable in vivo is far smaller than the corresponding proportion
for viral systems.
RESULTS
From transcriptome to immunome
A methylcholanthrene-induced fibrosarcoma, CMS5 (BALB/c,
d haplotype) was used as the primary workhorse, and the results were cross-tested to varying degrees with another
chemically induced (Meth A) mouse tumor, as well as several
human cancers. The CMS5 sarcoma is well characterized,
as is the primary host in which this tumor first arose. It is a
2232
progressively growing tumor, which is lethal in a syngeneic

host. CTLs against a CMS5 line have led to the identification
of a single immunogenic and tumor-protective neoepitope
(Ikeda et al., 1997; Hanson et al., 2000).
Transcriptome sequencing was chosen over genome or
exome sequencing to identify mutations specifically in the
genes expressed in CMS5 and Meth A. Broadly speaking, the
cDNA sequences obtained were compared with the normal
mouse sequences, and single-nucleotide variants (SNVs) were
identified (Table 1). This pipeline is named Epi-Seq. The
SNVs were analyzed for their potential to generate MHC I
restricted epitopes of the murine H-2 K, D, or L alleles using
the NetMHC algorithm. The complete list of epitopes was
filtered based on default NetMHC 3.0 PWM peptide binding score thresholds for weak binders of 8.72, 8.08, and 8.19
for Kd, Dd, and Ld, respectively. Using these thresholds, CMS5
and Meth A were observed to harbor 112 and 823 potential
epitopes, respectively (see Table S1 for the total output of
the pipeline for both tumors); the difference in the number
of epitopes identified between these two lines is a reflection
of the depth to which their transcriptomes were sequenced
(Table 1). The putative neoepitopes are randomly distributed
over the entire genome.
We made an attempt to determine if the numbers of
MHC Irestricted neoepitopes in these mouse tumors are
within the range expected in actual primary human tumors.
An analysis of exome sequences of several human melanomas
and their comparison with corresponding normal sequences
(Wei et al., 2011) through our bioinformatic pipeline reveals
Table 1. Single nucleotide variants and predicted epitopes
of tumor lines as deduced from transcriptome sequencing
and bioinformatic analyses
Mouse strain
BALB/c
Tumor type
RNA-Seq reads (million)
Genome mapped
Transcriptome mapped
HardMerge mapped
After PCR amplification filter
HardMerge and filtered mapped bases (Gb)
High-quality heterozygous SNVs in CCDS
exonsa
Tumor specific
Non-synonymous
Missense
Nonsense
No-stop
NetMHC predicted epitopesb
H2 Kd-restricted
H2 Dd-restricted
H2 literd-restricted
Meth A
CMS5
105.8
75%
83%
65%
18%
1.15
1,528
23.4
54%
59%
48%
22%
0.24
208
1,504
77.1%
1,096
63
1
823
203
328
292
191
78.5%
146
4
112
15
58
39
aThe
number of mutations identified depends on the sequencing depth.

on default NetMHC 3.0 PWM peptide binding score thresholds for weak
binders, of 8.72, 8.08, and 8.19 for Kd, Dd, and Ld alleles, respectively.
bBased
Rules for immunogenicity of cancer neoepitopes | Duan et al.
Ar ticle
Table 2. Tumor-specific polymorphisms and epitopes for human melanomas

Sample
Somatic mutationsa
Synonymous/nonsynonymous
in UCSC annotation
Subset of nonsynonymous mutations

within CCDS coding regions
Synonymous Nonsynonymous
01T
05T
09T
12T
18T
22T
24T
35T
43T
51T
60T
91T
93T
96T
160
56
39
427
91
69
163
13
68
51
67
99
54
68
NetMHC predicted epitopesb
Total
304
115
83
741
190
126
397
34
94
136
129
215
130
118
292
106
81
706
179
116
381
32
91
126
121
209
125
112
Total
A0101
A0201
A0301
473
175
131
1193
286
181
625
68
175
229
209
329
184
192
26
12
5
50
15
20
26
1
4
7
14
18
6
10
262
116
77
657
144
94
263
44
90
148
91
176
105
101
185
47
49
486
127
67
336
23
81
74
104
135
73
81
Nonsynonymous Missense Non-sense

291
106
80
705
179
115
379
32
91
126
120
209
124
112
272
100
76
656
168
108
358
31
86
117
112
196
116
103
12
6
4
49
11
7
21
1
5
9
8
13
8
9
aCalculated
bMost
from mutation data published by Wei et al. (2011), and using the epitope prediction step of the Epi-Seq pipeline to call the epitopes based on the mutation data.
common HLA alleles were chosen according to the published frequencies of such alleles (Cao et al., 2001).
hundreds of putative neoepitopes per melanoma (Table 2).

Similarly, the list of mutations derived from transcriptome sequencing of 14 human prostate cancers and normal tissues
(Ren et al., 2012), led to identification of a median of 14 putative epitopes (range, 282) for the common HLA alleles
(Table 3). The smaller number of neoepitopes in prostate cancers is related to the fact that they harbor relatively smaller
numbers of mutations as compared with melanomas (Vogelstein

et al., 2013).The published data for the B16 melanoma line of
spontaneous origin also reveal the presence of >100 MHC
Irestricted neoepitopes. These measurements indicate that
regardless of their murine or human origin, and regardless
of etiology, tumors harbor a significant number of candidate
MHC Irestricted neoepitopes.
Table 3. Tumor-specific polymorphisms and epitopes for human prostate cancers

Tumor
sample
Somatic mutationsa
Synonymous Nonsynonymous
1T
2T
3T
4T
5T
6T
7T
8T
9T
10T
11T
12T
13T
14T
aCalculated
NetMHC predicted epitopes
Synonymous/nonsynonymous Subset of nonsynonymous mutations within

in UCSC annotation
CCDS coding regions
7
4
7
5
4
3
2
20
13
20
6
3
2
19
10
3
4
3
7
5
3
44
22
32
13
11
5
32
Total
10
2
3
3
7
5
2
40
18
29
9
10
5
29
Total
A1101
A2402
A0201
9
4
10
2
18
3
2
82
30
36
17
28
9
53
4
4
5
2
12
1
1
42
13
20
8
13
2
25
0
0
1
0
2
0
1
7
0
3
2
5
3
8
5
0
4
0
4
2
0
33
17
13
7
10
4
20
Heterozygous Missense Non-sense

10
2
3
3
7
5
2
40
18
29
9
10
5
29
9
2
3
3
7
4
2
38
16
28
9
10
5
24
1
0
0
0
0
1
0
2
2
1
0
0
0
5
from mutation data published by Ren et al. (2012), and using the epitope prediction step of the Epi-Seq pipeline to call the epitopes based on the mutation data.
JEM Vol. 211, No. 11
2233
Heterogeneity of neoepitopes
Meth A cells were cloned and 30 distinct clones were tested
for four SNVs picked at random (Tnpo3, NFkb1, Prp31, and
Psg17). To our surprise, all 30 clones harbored three of the
four SNVs and fourth SNV was detected in 29/30 clones.We
attribute this apparent lack of antigenic heterogeneity to the
relatively shallow depth of sequencing. It is also possible that
cancer cell lines show less antigenic heterogeneity than primary tumors. Most importantly, these results suggest it is possible to use a relatively shallow sequencing as a methodology
to identify the neoepitopes that are the most broadly distributed among cancer cells.
Immunogenicity of neoepitopes identified in silico
To reduce the complexity of analyses, attention was directed
toward the 218 Kd-restricted epitopes (for Meth A and CMS5
combined) from a total list of 935 for all three alleles (Table 1).
All the mutations used for immunological analyses were confirmed individually by Sanger sequencing.
The neoepitopes were ranked in descending order of their
NetMHC scores for Kd binding. The top 7 neoepitopes from
CMS5 and the top 11 from Meth A are shown in Table 4,
which also shows the NetMHC score of the WT peptide
corresponding to the neoepitope. Peptides corresponding to
these 18 putative neoepitopes and their WT counterparts were
synthesized, and the affinity of all peptides for Kd (IC50 values)
was determined experimentally as described in Materials and
methods. (NetMHC scores and experimentally determined

IC50 values were significantly correlated; r = 0.44, twotailed Students t test; P < 0.001). 10 of 18 top-ranked neoepitopes (56%) bound Kd with an affinity of 500 nM or
better, and 7/18 (39%) bound with an affinity of 100 nM
or better.
All 18 peptides were used to immunize BALB/c mice.
The draining LNs of immunized mice were harvested 1 wk
after the single immunization, and the cells were stimulated in
vitro for 16 h without any added peptide, the mutant peptide
used for immunization, or the corresponding WT peptide.
The CD8+ cells were analyzed for activation (CD44+) and
effector function (intracellular IFN-+; see Materials and
methods for details of FACS analysis.) All possible patterns of
immunoreactivity were observed (Fig. 1 A): no immune response (12/18), a mutant peptide-specific, i.e., tumor-specific
immune response (5/18), and a cross-reactive response between the mutant and corresponding WT peptides (1/18).
Altogether, 6/18 or 33% of the neoepitope candidates identified
in silico actually elicited functional effector T cells in vivo.
When analyzed by an IFN- ELISpot assay, four additional
neoepitopes showed immunogenicity, bringing the total to
10/18 or 56%.
Of the 10 peptides with a Kd-binding affinity of 500 nM
or better, 7 or 70% were determined to be immunogenic experimentally. Of the 7 peptides with a Kd-binding affinity
of 100 nM or better, 4 or 57% were immunogenic. Three
Table 4. CMS5 and Meth A epitopes with highest NetMHC PWM scores.
Gene
Mut/WT sequence
CMS5 Epitopes with highest NetMHC scores

Ssx2ip
CYAK(v/L)KEQL
Mapk1.1
(q/K)YIHSANVL
Farsb
HY(v/L)HIIESKPL
Ncoa3
(h/Q)YLQYKQEDL
Mapk1.2
(q/K)YIHSANV
Mapk1.3
LYQILRGL(q/K)YI
Serinc1
NYLLSLVAV(m/V)L
Meth A Epitopes with the highest NetMHC scores
Usp12
SY(l/R)VVFPL
Tfdp1
QYSGS(w/R)VETPV
Ufsp2
HYINM(i/S)LPI
Apc
AYCETCWE(l/W)
Hspg2
SY(l/Q)LGSGEARL
Ccdc85c
TYIRP(f/L)ETKV
Pacs2
HYLS(s/A)ILRL
Alms1.1
(l/S)YLDSKSDTTV
Alms1.2
YYVPLLKRVP(l/S)
Ckap5
K(y/D)MSMLEERI
Abr
GYFVSKAKT(s/R)V
Mut/WT Score
Measured
IC50 for Kd (Mut/WT)
ICS
ELISpot
Tumor rejection
14.5/14.1
13.2/12.4
13/13.6
11.5/11.7
11.4/10.7
11.3/11
11/10.2
26/3.2
57/0.2
423/52
2162/54074
2135/295
110/333
2679/20861
+
-
+++++
+++++
14.2/12.3
14.2/15.3
14.2/14.5
14/8.1
14/14.4
13.5/13.1
13.4/12.7
13.3/15.2
13.3/7.3
13.2/1.7
13.1/12.8
6835/1155
-a/603
0.23/+++b
23/60
2623/79
6155/118
41/1269
79/16
421/1485
17/7686
958/570
+
+
+
+
+
+++++
+++++
+++++
++
-
aIC
50 > 70,000 nM
b+++, IC < 0.1 nM
50
Note for ELISpot results: 19 spots/106 CD8 = +; 1020 spots/106 CD8 = ++; 2150 spots/106 CD8 = +++; 51100 spots/106 CD8 = ++++; >100 spots/106 CD8 = +++++.
2234
Ar ticle
Figure 1. Immunogenicity of epitopes

generated by point mutations. Mice were
immunized in the footpad with indicated
mutant peptides. 1 wk later, draining LNs
were harvested and their cells were stimulated overnight in vitro without peptides
(No pep) or with mutated (Mut) peptides or
their unmutated (WT) counterparts. Surface
CD44 and intracellular IFN-+ cells were
counted on 20,000-gated CD8+ cells. (A) Representative examples of mutant peptides that
elicited no response, tumor-specific (i.e., mutant peptide specific) response, or tumor/self
cross-reactive response. The right pie chart
shows the total percentage of each type of
T cell response elicited by mutated peptides
from Meth A (n = 39) and CMS5 (n = 27).
(B) Representative examples of unmutated
counterparts of selected mutant peptides
that elicited no response, unmutated peptidespecific response, or cross-reactive functional
CD8 response (as in A). The right pie chart
shows the percentage of type of T cell response elicited by unmutated peptides from
both tumors. (C) Mice were immunized 100 g
of the indicated peptide once (Prime), or twice
(Boost) with a 29-d interval. 7 d after the last
immunization, draining popliteal LNs were
harvested for intracellular cytokine assay.
Lymphocytes were stimulated, or without
stimulation, with 10 g/ml cognate peptides
for 16 h and stained for CD4, CD8 and CD44,
followed by permeabilization and staining for
IFN-. Shown are the percent CD44+ IFN-+
cells of total CD8+ cells stimulated with peptides and without peptide stimulation (No pep). Error bars represent SEM. *, P < 0.05; ***, P < 0.001. Immunogenicity of each peptide was tested in two to four mice each and the experiments were performed between four and six times. See Fig. S2 for FACS gating
strategy and representative primary data.
peptides with a Kd-binding affinity of 500 nM or worse

were immunogenic.
Lack of immunogenicity of WT peptides
While testing for immunogenicity of neoepitopes, their WT
counterparts were similarly tested (Fig. 1 B). Surprisingly, as
with the mutant neoepitopes (Fig. 1 A), all possible patterns
of immunoreactivity were observed (Fig. 1 B): no immune
response (11/18), a WT peptide-specific immune response
(5/18), and a cross-reactive response between the mutant and
corresponding WT peptides (2/18). Altogether, 7/18 or 39%
of the WT counterparts of neoepitope candidates identified
in silico elicited functional effector T cells in vivo (Alms1.1,
Alms1.2, Abr, Ccdc85c, Farsb, Mapk1.2, and Ufsp2; Table 4).
This was still a surprisingly large proportion of self-reactive
peptides in view of the strong role of negative selection in
sculpting of the T cell repertoire.We tested the possibility that
the WT peptides that are immunogenic are functionally tolerized, even though they show immunogenicity after a 16-h
stimulation in vitro. We reasoned that if a small proportion of
low affinity autoreactive T cells escaped negative selection, they
may still be clonally expanded to a degree upon immunization
with a self-peptide, but that such an expansion would prove

self-limiting. Hence, naive mice were immunized with the
peptides, followed by a second immunization, and the responses in naive mice, once-immunized mice, and twiceimmunized mice were compared. The ovalbumin-derived
Kb-binding epitope SIINFEKL was used as a positive control
in C57BL/6 mice, and indeed the magnitude of the antiSIINFEKL CD8 response was amplified by a second immunization (Fig. 1 C). This same phenomenon was observed
with the mutant neoepitope Tnpo3 in BALB/c mice. However, second immunization with 4/4 WT peptides tested did
not elicit an amplification of the response. Indeed responses
detected after the second immunization were significantly diminished as compared with the response after the first immunization. By this stringent criterion, not a single WT peptide
was observed to be immunogenic.
Lack of immunoprotective activity
of the strongest Kd-binding neoepitopes
All 18 neoepitopes (Table 4) were tested for their ability to
elicit tumor rejection of CMS5 or Meth A. BALB/c mice
were immunized with the individual peptides and were
2235
challenged with the appropriate tumor 1 wk after the last immunization. None of the peptides elicited tumor rejection
(Fig. 2 A). One of the CMS5 neoepitopes FarsB shows
significant protection from tumor growth in this panel;
however, this result could not be reproduced unambiguously,
leading us to assign this as a negative neoepitope with respect
to protection from tumor growth. Interestingly, one of the
neoepitopes identified by us Mapk1 (listed as Mapk1.1, 1.2,
and 1.3; Table 4), was also identified by Ikeda et al. (1997) in
the CMS5 sarcoma as a tumor rejection antigen. Immunization with it does not elicit tumor rejection in our hands, just
as it did not in the original paper. The authors of the original
paper noted that IL-12 treatment was essential to show
antitumor immunity in this system, because mice vaccinated
with 9m-pulsed spleen cells in the absence of exogenous
IL-12 showed no resistance to CMS5 challenge. This is an
unintended validation of our pipeline and also highlights the
fact that we have used a very stringent tumor rejection assay
in our analyses.
Differential agretopicity
From the aforementioned results, it is evident that the NetMHC score is not a valuable predictor of immunogenicity or
tumor rejection. A close examination of the data in Table 4
suggests an underlying possibility: for each neoepitope, both
the neoepitope and its WT counterpart have similar NetMHC scores characteristic of high-affinity peptide binding.
Moreover, examining the experimental IC50 values, 4/7 WT
peptides of CMS5 and 7/11 WT peptides of Meth A have
stronger affinity for Kd than the mutant peptides. Thus, unless
the mutations alter TCR contacts or the structural properties
of the peptides in the Kd binding groove, T cells potentially
reactive to the neoepitopes may have been centrally deleted
or peripherally tolerized.
We therefore created a new algorithm wherein the NetMHC scores of the unmutated counterparts of the predicted
mutated epitopes were taken into consideration by subtracting them from the corresponding NetMHC scores of the
mutated epitopes. We refer to this new property of an epi
tope as the differential agretopicity index (DAI; agretopicity is
the ability of a peptide to act as an agretope), and we expect
it to reflect the degree to which the peptide-binding determinants of the neoepitopes differ from those of their WT
counterparts. In the search for the rules for immunogenicity
of viral or other clearly nonself-epitopes, such a parameter is
not necessary, nor possible, and has therefore never been
sought. The putative epitopes were ranked on basis of the
DAI (Table 5). A review of this DAI-ranked list for both tumors shows curiously that all the neoepitopes in this new
ranking are mutated at one of the two primary anchor residues at position 2 or the C terminus (we did not identify
any neoepitope with changes at both anchor residues.) Consistent with the Kd preferences gleaned from structural analyses (Mitaksov and Fremont, 2006), all of these mutations
involve aspartic acid to tyrosine at position 2 or proline/arginine
2236
to leucine at the C terminus. Although the DAI was not

crafted with this intent, it is a perfectly reasonable outcome
in hindsight.
The top DAI-ranking 20 epitopes of CMS5 and 28 epi
topes of Meth A were tested in tumor rejection assays. The
results (Fig. 2 B) show that 6/20 or 30% CMS5 epitopes and
4/28 or 14% Meth A epitopes showed statistically significant
tumor protective immunogenicity. The six CMS5 neoepi
topes are particularly impressive in that they elicited near
complete or complete protection from a lethal tumor challenge. Fig. 2 C shows representative tumor rejection curves of
the two protective and one unprotective CMS5 epitope from
Table 5. Corresponding WT peptides did not mediate any
tumor rejection. Detailed data on tumor rejection elicited by
a Meth A neoepitope Tnpo3 are shown in Fig. 5. Statistical
comparison of the NetMHC alone versus the DAI algorithms
in predicting antitumor protective immunity, using one-tailed
Fishers exact test, shows the DAI to be far superior (0/18 vs.
10/48, for NetMHC and DAI, respectively, one-sided Fishers
exact test, P = 0.031).
Although the DAI algorithm yielded a far richer harvest
of tumor-protective epitopes than the reliance on the highest
NetMHC or MHC-binding scores, most epitopes identified
by DAI still fail to elicit tumor protection. Further, the DAIranked neoepitopes (Table 5) that elicit protection from
tumor growth are not necessarily the highest ranking in DAI.
Clearly, other properties of the putative epitopes (see Discussion) contribute to the tumor rejection activity of individual
neoepitopes. Regardless of its imperfection, the DAI is a statistically significant and novel predictor of tumor-protective
immunogenicity, and more importantly, permits a dissection
of the other potential criteria for antitumor immunogenicity
in vivo.
The DAI algorithm also uncovers a new paradox of
fundamental significance. All the six neoepitopes that elicit
protection against CMS5 have NetMHC scores between
6.8 and 1.2, which are well below 8.72, the PWM peptide
binding score threshold for weak binders for Kd (NetMHC3.0). Consistent with that observation, their measured
IC50 values for binding to Kd are >70,000 nM (for all except
Slit3, for which is IC50 is 50,000 nM). This observation is
surprising because epitopes are typically considered good
MHC Ibinders if they have an IC50 value of <100 nM, or
at least <500 nM. The IC50 values for the CMS5 neoepi
topes are so high (i.e., their binding to Kd is so poor) that
these neoepitopes would never be considered suitable candidates for being epitopes based simply on their Kd-binding
characteristics. For Meth A as well, all the four neoepitopes
that elicit tumor immunity have a NetMHC below the 8.72
threshold for weak binders for Kd; the measured Kd-binding
affinities of only two of the four Meth A neoepitopes are
<100 nM. To explore the possibility that these neoepitopes
may be binding to another allele, Dd or Ld, the six tumorprotection eliciting neoepitopes for CMS5 were tested for
binding to these two alleles by direct peptide-binding studies;
none of the peptides showed significant binding (unpublished
Ar ticle
Figure 2. Landscape of protective tumor immunity elicited by tumor-specific peptides. (A) Tumor-protective activity of the mutated epitopes
with top NetMHC or DAI scores for CMS5 and Meth A. (A and B), Mice were immunized with peptides that had the highest NetMHC scores (A) or top DAI
scores (B) and challenged with live tumor cells, and tumor growth monitored. Area under the curve (AUC; Duan et al., 2012) for each individual tumor
growth curve was calculated and normalized by setting the naive group to a value of 100, shown by a horizontal red line. Bars corresponding to peptides
that show statistically significant tumor-protective immunogenicity are filled in red, and indicated by an asterisk (P = 0.0150.03). The peptides are
arranged in order of increasing antitumor activity. The pie charts show the percentage of neoepitopes tested that did not (black) and did (red) elicit protection from tumor challenge. (C) Examples of tumor growth curves in untreated mice (naive) and mice immunized with indicated mutant peptides from
CMS5 or their WT counterparts. Stau1.2 is a representative neoepitope that does not elicit protective immunity, whereas Atxn10.1 and Alkbh6.2 are representative neoepitopes that do. Each line shows the kinetics of tumor growth in a single mouse. The experiments were performed three times.
data). Because the requirement for MHC I binding to be

<500 nM is so well established (Assarsson et al., 2007), the
possibility that the peptides identified as being potent in eliciting protection from tumor challenge (Fig. 2, B and C) may
do so through nonimmunological means was investigated.
CD8 dependence of tumor protection

Mice immunized with five of the six active neoepitopes of
CMS5 (Alkbh6.2, Slit3, Atxn10.1, Atxn10.2, and Ccdc136)
were tested for immunogenicity; as shown in Table 5, Alkbh6.2,
Slit3, and Atxn10.1 elicited modest CD8 T cell response that
2237
Table 5. CMS5 and Meth A epitopes with highest DAI scores

Gene
CMS5 Epitopes
Dhx8.1
Alkbh6.1
Dhx8.2
Dhx8.3
Alkbh6.2
Dhx8.4
Alkbh6.3
Alkbh6.4
Rangap1
Stau1.1
Stau1.2
Stau1.3
Stau1.4
9430016
H08Rik
Slit3
Atxn10.1
Sipa1l3
Atxn10.2
Ccdc136
Mast2
Meth A Epitopes
Tnpo3.1
Tnpo3.2
Tnpo3.3
Tnpo3.4
Trim26.1
Nus1
Tpst2.1
Fiz1
Kdm4b
Dis3l2.1
Ube4a.1
Ncdn
Gapdh
Ckap5
Prrc2a
Tmx3
Nfkb1
Dis3l2.2
Ugdh
Mll2
Galnt1
Tpst2.2
Cpsf2
Zfp236.1
Mut/WT sequence
Mut/WT Score
DAI
Measured IC50 for ICS

Kd (Mut/WT)
P(y/D)LTQYAIIML
D(y/D)VPMEQP
P(y/D)LTQYAI
P(y/D)LTQYAII
D(y/D)VPMEQPR
P(y/D)LTQYAIIM
D(y/D)VPMEQPRP
D(y/D)VPMEQPRPP
SEDKIKAI(l/P)
LKSEEKT(l/P)
KPALKSEEKT(l/P)
PALKSEEKT(l/P)
ALKSEEKT(l/P)
SWSSRRSLLG(l/R)
9.3/-2.3
4.3/-7.3
9.9/-1.6
9.3/-2.2
6.8/-4.7
6.6/-4.9
4.8/-6.7
4.7/-6.8
1.4/-5.4
0.6/-6.2
0.1/-6.7
1.3/-5.4
1.2/-5.5
5.9/-0.6
11.6
11.6
11.5
11.5
11.5
11.5
11.5
11.5
6.8
6.8
6.8
6.7
6.7
6.5
2192/1653
60858/
a/
244/1418
/51229
6571/6256
/23570
47053/2957
/5108
/
/
69546/
/
/
GFHGCIHEV(l/R)
QVFPGLME(l/R)
TTTPGGRPPY(l/R)
VFPGLME(l/R)
ELQGLLEDE(l/R)
KLQRQYRSPR(l/R)
4.7/-1.8
3.4/-3.1
2.7/-3.8
2.5/-4
2.4/-4.1
2.2/-4.3
6.5
6.5
6.5
6.5
6.5
6.5
51640/
/7054
/
/1107
/4537
10107/8511
(sy/LD)MLQALCI
(sy/LD)MLQALCIPT
(sy/LD)MLQALCIP
(sy/LD)MLQALC
A(y/D)ILAALTKL
P(y/D)LVLKFGPV
L(y/D)EAGVTDEV
H(y/D)LQGSNA
L(y/D)HTRPTAL
I(y/D)GVVARNRAL
A(y/D)AKQFAAI
S(y/D)CEPALNQA
V(y/D)LTCRLEKPA
K(y/D)MSMLEER
P(y/D)KRLKAEPA
D(y/D)IIEFAHRV
G(y/D)SVLHLAI
I(y/D)GVVARNRA
L(y/D)YERIHKKML
S(y/D)RLPSSRKK
L(y/D)VSKLNGP
L(y/D)EAGVTDE
L(y/D)DVDAAF
E(y/D)LDLQTQ
8.2/-5.2
7.1/-6.3
3/-10.4
5.9/-7.4
12.8/1.2
10.5/-1.1
10.3/-1.3
10.3/-1.3
10/-1.6
9.3/-2.3
9.3/-2.3
8.9/-2.7
8.9/-2.7
8.1/-3.5
7.9/-3.7
7.3/-4.3
6.9/-4.6
6.9/-4.7
6.4/-5.2
5.9/-5.7
5.6/-6
5.5/-6.1
5.3/-6.3
5.3/-6.3
13.4
13.4
13.4
13.3
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.6
11.5
11.6
11.6
11.6
11.6
11.6
11.6
11.6
82/146
9964/85
67/111
14927/89
622/1.1
2359/1.9
60/
2473/
264/
143/
12/
664/
1150/
58/
1450/261
7941/351
0.26/1615
1342/262
600/12
6673/17
27086/
/
1410/
1641/
ELISpot
Tumor
rejection
+
-
+++++
+++++
+++++
++
+++
+++++
+++++
+
+
-
+
+
-
+++
+++
++++
-
+
+
+
+
-
+
+
+
+
+
+
+
-
+++++
+++
+++++
+++++
+++
+
+
+++
+
+++
+++
+++
+
+
+
+
aIC
50 > 70,000 nM
2238
Ar ticle
Table 5. (Continued)
Gene
Mut/WT sequence
Mut/WT Score
DAI
Trim26.2
Zfp236.2
Ube4a.2
Dcaf6
A(y/D)ILAALTKLQ
E(y/D)LDLQTQG
A(y/D)AKQFAA
A(y/D)RLEGDRS
4.9/-6.7
4.9/-6.7
4.7/-6.9
3.7/-7.9
11.6
11.6
11.6
11.6
Measured IC50 for ICS

Kd (Mut/WT)
17599/
10028/
28583/
/
ELISpot
Tumor
rejection
++++
-
aIC
50 > 70,000 nM
was undetectable by the intracellular cytokine assay, and

detectable only by the ELISpot assay (Fig. 3 A), whereas
Atxn10.2 and Ccdc136 did not elicit a detectable response at
all. Immunized mice were depleted of CD8 or CD4 cells in
the effector phase only (i.e., after immunization but before
tumor challenge) and were challenged with CMS5 cells as in
Fig. 2. The results (Fig. 3 B) show that, compared with naive
mice, each mutant peptide elicited potent tumor rejection.
Depletion of CD8 cells completely abrogated the tumorrejecting ability of each peptide. Depletion of CD4 cells had
no such effect. These data show that the CMS5 neoepitopes
that elicit tumor rejection do so through elicitation of specific
CD8+ T cells, and not through nonimmunological means.
However, the data do not imply that CD4+ T cells do not
have a role in tumor rejection; an examination of the kinetics
of tumor rejection shows that in the absence of CD4+ cells,
Figure 3. Protective tumor immunity elicited by tumor-specific peptides is CD8-dependent. (A) Mice were immunized with mutant peptides
Alkbh6.2, Slit3, and Atxn10.1, Atxn10.2, and Ccdc136. Draining LNs harvested 7 d after immunization were not stimulated or stimulated in vitro with the
cognate peptides for 20 h, and were analyzed by ELISpot. Data for mice immunized with Alkbh6.2, Slit3, and Atxn10.1 peptides are shown. *, P < 0.05;
**, P < 0.01. (B) Naive mice or mice immunized twice with indicated peptides (Alkbh6.2, Slit3, Atxn10.1, Atxn10.2, and Ccdc136) were challenged intradermally with 300,000 live CMS5 tumor cells, and tumor growth was monitored. The numbers in the top left corner of each set of tumor growth curves denote the number of mice with growing tumors/total number of mice. Immunized mice were not depleted (NT), depleted of CD8 or CD4 cells, as indicated.
Each line shows the kinetics of tumor growth in a single mouse. The experiments were performed three times.
2239
tumors actually do begin to grow in almost all the immunized

mice, but begin to regress by days 710.
Conformational stability as an indicator of immunogenicity
The CD8-dependence of neoepitopes that have low predicted and measured affinity for their restricting allele led us
to seek other determinants of immunogenicity. Although the
DAI algorithm selects for mutations at primary anchor positions that improve peptideMHC binding affinity, unless the
mutations alter the structural properties of the peptide in the
binding groove as discussed above, potentially responding
T cells may be centrally or peripherally tolerized. Studies
have shown that anchor modification can have a range of
effects on MHC-bound peptides, in some cases having no
apparent influence (Borbulevych et al., 2005, 2007), and in
others leading to alterations in structural and motional properties (Sharma et al., 2001; Borbulevych et al., 2007; Insaidoo
et al., 2011).
To examine the consequences of anchor modification, we
used computational modeling to examine the structural properties of pairs of mutant and WT peptides bound to Kd. We
adapted an approach recently used to model the structures of
peptides bound to HLA-A*0201 (Park et al., 2013), using the
crystallographic structure of a viral peptideH-2Kd complex
as a template (Mitaksov and Fremont, 2006). As described
in Materials and methods, we used a workflow that included
homology modeling, simulated annealing, and molecular dynamics simulations to predict structural properties. Because
the conformations of peptide backbones in MHC I proteins
vary considerably with peptide length (Borbulevych et al.,
2007), we restricted the modeling to pairs of nonamers in
Table 5, matching the length of the peptide in the template
structure. This necessary restriction leads to noninclusion in
our modeling analyses of some of the more immunogenic
and protective neoepitopes in Table 5, simply because they are
8- or 10-mers, and not 9-mers. To ascertain how well the
modeling procedure was transferable from HLA-A*0201 to
H-2Kd, we applied the modeling procedure to the complex
of the immunodominant and highly immunogenic HBV core
peptide with Kd, a complex for which the crystallographic
structure is known (Zhou et al., 2004).
We began by looking for the presence of structural differences between the mutant and WT peptides. In each instance,
there were differences between the predicted conformations
of the mutant and WT peptides bound to H-2Kd. In all cases,
these differences propagated away from the site of the mutations (Fig. 4 A). These observations were not surprising, as
the mutations were all quite drastic, and even conservative
mutations at class I MHC-presented peptide anchor residues
can influence downstream conformation (Sharma et al., 2001;
Borbulevych et al., 2007).
However, although T cell receptors can be exquisitely
sensitive to changes in peptide conformation (Ding et al.,
1999), peptide immunogenicity (defined as having positive
ELISpot or ICS results) did not correlate with the magnitude of structural differences, expressed as root mean square
2240
deviations (RMSD) in ngstroms when all common atoms of

each peptide pair were superimposed (Fig. 4 B). Examining
the modeling data more closely, however, revealed a correlation, albeit imperfect, with conformational stability. In 10 out
of 14 cases, the mutant peptides were more rigid, sampling
fewer conformations during the molecular dynamics phase of
the modeling, expressed as the average root mean square fluctuations (RMSF) of all nine carbons of each peptide (Fig. 4,
CD; and Fig. S1). Of the 10 instances in which the mutant
peptides were more rigid, eight of these had positive ICS or
ELISpot results. Of the four instances in which the WT peptides were more rigid, one did not yield any positive immuno
logical outcomes. Altogether, greater structural stability was
a predictor of immunogenicity in 9 out of 14, or 65% of the
nonameric, high DAI-ranking neoepitopes.
In some instances though, the differences in the fluctuations between the mutant and WT peptides were very small
(for example, Ckap5, Dhx8.3, and Zfp236.2 have differences
0.03 ). This led us to question the fidelity of the mean
RMSF as a predictive tool. Examining the data for all peptides more closely, we observed a tendency for high structural
instability in the peptide C terminus, particularly in the WT
peptides (Fig. 4 C and Fig. S1). Previous studies have shown
that weak binding peptides can detach and dissociate from
MHC I proteins at the C terminus (Winkler et al., 2007;
Narzi et al., 2012), suggesting that in these cases the high
structural instability reflects at least partial peptide dissociation. However, some of the mutant peptides also had high instability at the C-terminus. Most notably, all three mutant
peptides which failed to elicit an immunological response had
high C-terminal instability. To quantitate this, we calculated
the mean C-terminal RMSF for all mutant peptides. The
C-terminal RMSF of the nonimmunogenic peptides were all
above this mean value (Fig. 4 E). Only three of the 14 immuno
genic peptides had C-terminal RMSF values above the mean
(Stau1.4, Dhx8.3, and Tpst2.2). Thus, the presence or absence
of C-terminal instability was a predictor of immunological
outcome in 11 out of 14, or 79% of the nonameric high DAI
ranking neoepitopes. (We note that the HBV control peptide
was relatively stable in the Kd binding groove, with a mean
Ca RMSD from the starting coordinates of 1.7 and an
average C-terminal RMSF of 0.57 , as would be predicted
from an immunogenic viral peptide).
Natural presentation of a neoepitope
The tumor-protective immunogenicity of the epitope
syMLQALCI (WT LDMLQALCI), the mutated Transportin 3
(Tnpo3)-derived epitope, the highest ranking (by DAI) epi
tope of Meth A (Table 5), was investigated in more detail.
Tnpo3 is a nuclear import receptor and is not a driver protein
for any tumor type reported thus far (Brass et al., 2008). The
mutant Tnpo3 epitope was shown to elicit strictly tumorspecific CD8+ immune response, as seen by the ability of mutant Tnpo3-immunized mice to show strong tumor-specific
CD8+, CD44+, IFN-+ response to the mutant but not the
WT peptide ex vivo or after stimulation in vitro (Fig. 5 A).
Ar ticle
Figure 4. Structural stability as a correlate with immunogenicity. (A) Mutations within neoepitopes lead to structural alterations across the peptide backbone, as illustrated with structural snapshots from the simulations of the mutant and WT Tnpo3.1 epitope bound to H-2Kd. (B) Summary of
structural differences for highly DAI ranked nonamers. Differences were quantified by superimposing average peptide conformations from the molecular
dynamics simulations and computing RMSDs for all common atoms. Green and red bars indicate epitopes that led to either positive or negative immunological responses, respectively. The yellow bar shows the results for control calculations for an immunogenic HBV core epitope. (alue for the HBV epitope
is the average Ca RMSD relative to the starting coordinates. (C) The numbers in the legend give the mean RMSF for the all amino acids of each peptide;
those at the right give the value for only the C-terminal carbon. Mutated amino acids are indicated by lower case in the x-axis. Results for all nonamers
are in Fig. S1. (D) Effects of mutations on the conformational stability of all nonamers, calculated as the difference between the mean RMSF of the mutant and the WT peptide. (E) Fluctuations at the peptide C-terminal ends and immunogenicity. The dashed vertical line shows the mean value for all
mutant nonamers. The yellow bar shows the C-terminal stability of the HBV core epitope control. Error bars represent SEM.
A Tnpo3-specific immune response was also detectable ex vivo

upon staining of cells with a Tnpo3-specific tetramer (Fig. 5 B).
Conversely, CD8+ CD44+ IFN-+ cells isolated from mice immunized with Meth A cells recognize mutant Tnpo3-pulsed
cells but not cells pulsed with an irrelevant Kd-biding peptide
Prpf31 ex vivo, as well as after stimulation in vitro (Fig. 5 C).

Although <0.1% Tnpo3-specific CD8+ T cells seen ex vivo is
a truly small response, we consider it real because ex vivo responses are bound to be weak, the response is statistically significant, and the response amplifies very significantly (>1.2%),
2241
Figure 5. Antigen presentation of neoepitope Tnpo3 and immune response and tumor protection elicited by it. (A) Mice were immunized with
mutant Tnpo3 peptide. Draining LNs, harvested 1 wk after immunization, were briefly stimulated ex vivo without (No pep) or with WT or mutant Tnpo3
peptides (left), or with a weekly in vitro stimulation with 1 M mutant Tnpo3 peptide (right). After 5 d, cells were tested for the responsiveness to mutant
Tnpo3-pulsed cells (Tnpo3) or Meth A cells (Meth A) by ELISpot. IFN-+ CD44+ CD8+ T cells were counted. (B) Mice were immunized twice with ovalbumin
peptide (SIINFEKL) or Tnpo3 mutant peptide. 6 d after the second immunization, splenocytes from both groups were stained with Kd/SYMLQALCI tetramer.
Tetramer positive cells were counted in CD8+ gate. (C) Mice were immunized with irradiated Meth A cells. (left) 6 d later, inguinal LN cells were stimulated
overnight without peptide, irrelevant Prpf31 peptide or Tnpo3 peptide. % activated effector CD8+ cells is shown, as assessed by flow cytometry. (right)
Splenocytes were stimulated in vitro in multiple rounds with 1 M of indicated peptides for a total of 19 d. Irrelevant peptide from Prpf31 was used as a
control. 5 d after stimulation, cells were tested for the responsiveness to indicated peptides by flow cytometry. Typically, for each sample, 150,000 lymphocytes, or at least 19,000 CD8+CD4 cells, were acquired. See in Fig. S3 for FACS gating strategy and representative primary data. (D) Mice were injected with 200,000 Meth A cells on the right flank. 21 d later, tumor-draining LNs and contralateral LNs were harvested and stained with anti-CD8
antibody and Tnpo3 and Nfkb1 tetramers Kd/SYMLQALCI and Kd/GYSVLHLAI, respectively (left and middle). Splenocytes were used to purify CD8+ cells to
assess the responsiveness to mutant Tnpo3-pulsed cells (Tnpo3) or Meth A cells (Meth A) by ELISpot assay with no peptide (No pep) stimulation as negative control (right). (E) Naive mice or Tnpo3 mutant peptide-immunized mice were challenged with Meth A cells and treated with anti-CD25 antibody or
anti-CTLA-4 antibody as indicated. AUC for each group is plotted (Duan et al., 2012), and complete tumor growth curves for all the mice in all groups are
shown. Between four and six mice per group were used in each experiment, and each experiment was repeated between three and five times. *, P < 0.05;
**, P < 0.01; ***, P < 0.001.
while maintaining specificity, after weekly in vitro stimulation

with Tnpo3 peptide for 19 d (Fig. 5 C, right). Interestingly,
Meth A tumor-bearing mice (day 21 after inoculation) harbor a low frequency of T cells recognizing two Kd-binding
peptides (Tnpo3 and Nfkb1, measured by tetramer staining)
in the tumor-draining LNs (Fig. 5 D). These observations
2242
confirm that the mutant Tnpo3 peptide is naturally presented by Meth A cells and also that immune response to it
is elicited upon immunization by whole tumor cells, as well
as in tumor-bearing mice. Attempts to identify this mutant
peptide by mass spectroscopic analysis of MHC Ieluted peptides from Meth A were unsuccessful, presumably because of
Ar ticle
the higher sensitivity of the T cell assays as compared with

mass spectroscopy. The structural modeling predicts that
the mutant Tnpo3 peptide is substantially more stable
the WT peptide across the center of the peptide (Fig. 4,
A and C).
Enhancement of tumor-protectivity
of neoepitopes by immune modulators
Combination of immunization with mutant neoepitopes was
tested using the Meth A neoepitope Tnpo3. This neoepitope
is only modestly tumor protective in monotherapy, thus allowing more dynamic range for testing of an enhanced effect
by combination therapy. Combination of immunization with
mutant Tnpo3 with antagonistic antibodies to CD25, which
has been shown to target regulatory T cells, showed synergy;
the anti-CD25 alone showed complete regression in all mice
(P = 0.008) and Tnpo3 alone also elicited significant protection (P = 0.03). The combination showed more significant
protection than either agent alone (P = 0.016; Fig. 5 E, left):
although tumors eventually regressed in all mice in both
groups, the kinetics of tumor regression was significantly
steeper in the combination group. A similar result was obtained with anti-CTLA4 antibody, which releases T cells from
checkpoint blockade (Egen et al., 2002; Callahan et al., 2010).
Each agent alone elicited statistically significant protection
and the combination was significantly more effective than
Tnpo3 alone (P = 0.04) or anti-CTLA4 antibody alone (P =
0.04; Fig. 5 E, right). Only a single tumor regressed in the
anti-CTLA4 antibody group, and no tumors regressed in the
Tnpo3 alone group (although the tumor growth was very
significantly retarded); the combination group showed complete regression of two tumors and a sustained trend toward
regression in two additional tumors.
DISCUSSION
Our results uncover a trove of truly tumor-specific antigenic
epitopes. Using novel tools reported in this study, we are able
to identify with high accuracy the small number of neoepi
topes (among the vast numbers of potential neoepitopes) that
truly elicit immunological protection against tumor growth.
The application of our method is described here for two independent tumors. In actuality, the pipeline, including the
DAI algorithm, was first derived empirically on the data from
the Meth A tumor, and was then tested on CMS5. Clearly,
the antitumor activity predicted from the DAI algorithm is
significantly stronger in CMS5 than in Meth A; this variation
is most likely a reflection of the immuno-suppressive mechanisms unique to the Meth A tumor (Levey et al., 2001), and
thus unrelated to the merits of the DAI algorithm per se. The
DAI algorithm has since been tested in yet another mouse
tumor, the B16 melanoma, and data on T cell responses in
this line as well, are consistent with significant superiority of
DAI over NetMHC alone (unpublished data). Although the
present study is focused on identification of MHC Irestricted
epitopes of CD8 T cells, the analysis can also be extended to
MHC II-restricted epitopes of CD4 T cells.
Although T cells play an unambiguously central role in

cancer immunity, they have been poor probes for identification of immunoprotective epitopes thus far. Extensive and laborious analyses of T celldefined tumor-specific antigens of
Meth A and CMS5 sarcomas over the years managed to yield
a total of five epitopes (Buckwalter and Srivastava, 2008),
none of which elicit particularly robust tumor rejection; in
contrast, this single study has uncovered nearly a dozen potent
tumor-protective epitopes of these two tumors. It is instructive to ponder the reason for this discrepancy. The use of
T cells as probes inherently requires generation of T cell lines
or clones, which is itself a highly selective process. It would
appear that the diversity of effector T cells in vivo is not readily
captured by the T cell lines or clones generated in vitro, leading to a distorted, and sparse, view of the T cell immunomes
of tumors. The genomics-driven analysis of the immunome
described in this study cuts through the bias in selection of
T cells, and thus illuminates the entire field of neoepitopes.
We introduce here the DAI score (the numerical difference between the NetMHC scores of the mutated epitope
and its unmutated counterpart), which allows significant enrichment for the extremely small number of truly immunoprotective neoepitopes from among the hundreds of putative
neoepitopes identified by the NetMHC algorithm.The demonstrated utility of the DAI score underscores the validity of its
premise: a tumor-protective immune response requires neoepitopes that differ from their WT counterparts, and the DAI
score is a means to quantify and rank such differences. Understandably, because our existing ideas about immunogenicity
are derived entirely from the study of viral and model antigens, which have no self-counterparts, there was no necessity
to devise a DAI for their studies. As follows from the design of
the NetMHC algorithm, amino acid substitutions at primary
anchor residues make for the biggest contributions to the DAI.
Indeed, every neoepitope with a high DAI ranking replaces
aspartic acid with tyrosine at position 2 or proline/arginine at
the C terminus with leucine. From structural considerations,
these substitutions would be expected to significantly impact
peptide binding, as tyrosine at P2 and leucine at the peptide C terminus are the most optimal Kd anchor residues (indeed, aspartic acid at P2 or arginine at the C terminus would
be expected to be considerably unfavorable due to substantial charge repulsion; Mitaksov and Fremont, 2006). Peptide
conformational stability, expressed as the fluctuations observed
during molecular dynamics simulations, is another tool that
suggests a novel correlate with immunogenicity.The majority
of the neoepitopes with high DAI rankings are predicted to
interact with Kd in a more stable fashion than their wild-type
counterparts; in these cases, alteration of the anchor residues
yields a more rigidly bound peptide. For mutations that replace charges at the P2 or C-terminal positions, greater conformational stability may follow from elimination of charge
repulsion as noted above. Effect of anchor modification on
peptide conformational stability has been noted previously,
and notably, increased peptide flexibility correlates with a loss of
immunogenicity. This may occur by reducing the opportunities
2243
for productive interactions with T cell receptors (Sharma et al.,

2001; Borbulevych et al., 2007; Insaidoo et al., 2011) or increasing the lifetime of the MHC-bound peptide.
In a recent study, Fritsch et al. (2014) have performed a
retrospective of 40 neoepitopes of human cancers, that induced immune responses, most of which were associated
with regression or long-term disease stability. Interestingly, in
all of these neoepitopes, the mutations were located in the
TCR-facing residues of the neoepitopes rather than the anchor residues. At first look, these data appear to contradict our
finding that the vast majority of bonafide neoepitopes arise
due to changes in anchor residues. However, a careful look at
the data shows that the comparison cannot really be made:
the criterion associated with aggression or long-term disease stability as determined subjectively by a dozen different
groups of clinical investigators, is a very different criterion
from actually testing tumor regression, (as we have done experimentally with mice). The true test of this question in a
human study would require a randomized trial where patients
would be immunized based upon one set of criteria or the
other. Fritsch et al. (2014) have just begun a study using their
criteria in human melanoma, and we are starting a clinical
study in epithelial ovarian cancer based on our criteria. These
studies will allow us to compare the two approaches.
A most surprising observation that emerges from our study
is that 10/10 neoepitopes that elicit protective immunity are
classified as nonbinders of Kd by NetMHC (cut-off value of
8.72). Correspondingly, the affinity of 8/10 neoepitopes for Kd
is well over 500 nM, the traditional threshold for fruitful interaction of viral epitopes with MHC I molecules. In five of five
instances tested, these presumed non-binders elicit classical
CD8 T celldependent tumor immunity.This observation challenges some of our basic assumptions about MHC Ipeptide
T cell receptor interactions, and exposes a far wider universe of
potential neoantigens than assumed thus far.
The observed dissociation between detectable CD8 responses and immunoprotection (Table 5) from tumor growth
merits comment. The neoepitopes that elicit immunoprotection and a CD8 response are straightforward and require
no comment. The neoepitopes that elicit immunoprotection
but not a detectable CD8 response (Fig. 3) may also be under
stood with the explanation that the CD8 response elicited
is too weak to be detected by the ELISpot assay, thus highlighting the need for developing more sensitive assays for
CD8 cells and their activities. It is, however, the epitopes that
elicit potent CD8 responses but not immunoprotection that
are difficult to understand. However, data in Table 5 may provide guidance in thinking about this dissociation. Note the
neoepitopes Tnpo3.1, 3.2, 3.3, and 3.4 for Meth A in Table 5.
They share the same N-terminal mutations (sy/LD), but differ in the extent of their extension on the C termini.Whereas
the first three elicit strong CD8 responses, Tnpo3.4 does not.
More interestingly, of the three neoepitopes that are immunogenic, only one,Tnpo3.1, elicits protection from tumor growth.
It is conceivable that Tnpo3.1 is the only neoepitope that is
naturally presented, while the others are not. This entirely
2244
testable hypothesis provides a framework for testing the dissociation between T cell responses and immunoprotection.
We note with satisfaction, but also with some surprise that
not a single WT epitope among the more than 100 tested (66
epitopes listed in Tables 4 and 5, and >35 additional epitopes)
elicited a measurable, amplifiable CD8 immune response.The
immune responses, when detected after a first immunization,
were abrogated rather than enhanced after a second immunization, consistent with them being peripherally tolerized responses.This study represents perhaps the largest in which the
immune responses to such a large number of self-epitopes
have been systematically tested, and testifies strongly to the
powerful scope of mechanisms of negative selection and peripheral tolerance.
These results present clear opportunities for clinical translation in human cancer immunotherapy. With the advent of
high-throughput and inexpensive DNA sequencing, it is now
possible to routinely sequence the exomes of cancers and
normal tissues of each cancer patient, and compare the two to
identify cancerspecific mis-sense mutations. The NetMHC
or other such commonly available algorithms can then be
used to identify the potential neoepitopes generated by the
mis-sense mutations, for each of the three to six HLA I alleles
of each patient. Peptides corresponding to the neoepitopes
can then be chemically synthesized and used to immunize
patients. However, the numbers of potential neoepitopes can
be vast, and it is impractical to immunize patients with such
vast numbers of peptides. The pioneering study of van Rooij
et al. (2013) is an excellent case in point. Using exome comparison, indexed to the tumor transcriptome, these authors
identified 448 potential neoepitopes in a melanoma patient;
of these, only two were truly immunogenic.The combination
of the NetMHC algorithm with the DAI and the C-terminal
stability algorithms, as identified here, now makes it possible
to reduce the large numbers of potential neoepitopes to a
much smaller number of truly immunogenic epitopes, which
can now be used to immunize patients in a realistic manner.
Our observations also present new opportunities to address some long-standing questions in immunology. Antigenic
heterogeneity of cancers has been the subject of much discussion, but due to the lack of bona fide tumor-specific antigens in significant numbers, the debate has been largely
theoretical. Uncovering of a large repertoire of true tumorspecific neoepitopes now allows the questions regarding antigenic heterogeneity (and antigen escape) to be asked in an
unprecedentedly robust manner. Inherently linked to the issue
of antigenic heterogeneity is the role of immunoediting of
growing cancers. Since the classical studies on immune surveillance against tumors (Burnet, 1970), Dunn et al. (2004)
have suggested that tumors go through elimination, equilibrium, and escape phases of immunoediting and have demonstrated evidence supporting the idea (Matsushita et al., 2012).
The availability of a large repertoire of tumor-specific neoepi
topes of any given tumor allows immuno-editing to be addressed in far more granularity. Epitope spreading is one of
the more exciting and underexplored ideas in cancer immunity
Ar ticle
(Markiewicz et al., 2001; Corbire et al., 2011). The major

reason that it is underexplored is the paucity of a suitable
number of antigenic neoepitopes. Identification of a large
number of neoepitopes for practically any mouse or human
tumor now permits a vigorous exploration of epitope spreading, along with its clinical utilization.
Mice and tumors. The BALB/cJ mice (68-wk-old female) were purchased from The Jackson Laboratory. Mice were maintained in the virusfree mouse facilities at the University of Connecticut Health Center and
their use was approved and monitored by the Institutional Animal Care and
Use Committee.
Sample preparation. Samples were prepared using the Illumina protocol
outlined in Preparing Samples for Sequencing of mRNA (Part# 1004898
Rev. A September 2008). The protocol consists of two parts: cDNA synthesis and paired-end library preparation. First, mRNA was purified from total
RNA using magnetic oligo(dT) beads, and then fragmented using divalent
cations under elevated temperature. cDNA was synthesized from the fragmented mRNA using Superscript II (Invitrogen), followed by second strand
synthesis. cDNA fragment ends were repaired and phosphorylated using
Klenow, T4 DNA Polymerase, and T4 Polynucleotide Kinase. Next, an A
base was added to the 3 end of the blunted fragments, followed by ligation
of Illumina Paired-End adapters via TA-mediated ligation. The ligated
products were size selected by gel purification and then PCR amplified
using Illumina Paired-End primers. The library size and concentration were
determined using an Agilent Bioanalyzer.
GAII run conditions. The RNA-seq library was seeded onto the flowcell
at 8 pM, yielding 282384 K clusters per tile. The library was sequenced
using 61 cycles of chemistry and imaging.
Analysis of sequencing data. Initial data processing and base calling, including extraction of cluster intensities, was done using RTA (SCS version
2.6 and SCS version 2.61). Sequence quality filtering script was executed in
the Illumina CASAVA software (ver 1.6.0, Illumina).
Epi-Seq bioinformatics pipeline. The pipeline starts by mapping RNASeq reads against the strain-specific genome sequences downloaded from the
Sanger Mouse Genomes Project (Keane et al., 2011) and a strain-specific
haploid transcript library derived from CCDS annotations (Li et al., 2009).
We used BALB/c genome/transcriptome sequences for CMS5 and Meth A
cell lines. DatabaseSNP polymorphisms were removed. Instead of comparing
to the mm9 reference genome, and then excluding SNPs in dbSNP, we created a strain-specific genome by applying strain SNVs to the mm9 reference
genome.The SNVs were downloaded from the mouse genomes.The created
genome was used to map the reads and call the mutations. Reads were
mapped using Bowtie (Langmead et al., 2009) with the default seed length of
28, maximum of 2 mismatches in the seed, and maximum sum of phred quality scores at mismatch positions of 125. After an initial round of mapping, we
calculated mismatch statistics for each read position and each sample. Based
on this analysis, 2 bases from the 5 end and 10 bases from the 3 end were
clipped from all aligned reads. The resulting read alignments were merged
using the HardMerge algorithm (Duitama et al., 2012). HardMerge discards
reads that align to multiple locations in the genome and/or transcriptome, as
well as reads that align uniquely to both, but at discordant locations. To reduce the effect of bias introduced by PCR amplification during library preparation (Aird et al., 2011), we replaced multiple reads with alignments starting
at the same genomic location with their consensus. The SNVQ algorithm
(Aird et al., 2011) was then used to call single nucleotide variants (SNVs)
from the filtered set of aligned reads. SNVQ uses Bayes rule to call the genotype with the highest probability while taking base quality scores into account. High confidence SNVs were selected by requiring a minimum phred
quality score of 50 for each called genotype, a minimum of 3 reads supporting the alternative allele, with at least one read mapping on each strand.
Haplotype inference over called SNV genotypes was performed using the
RefHap Single Individual Haplotyping algorithm in (Duitama et al., 2012)
that uses read evidence to phase blocks of proximal SNVs. Because residual
heterozygosity in the inbred mice used in our experiments is predicted to be
low (Bailey 1978 [specifically, Table 1 therein]), unique heterozygous SNVs
were considered to be novel somatic mutations. Homozygous SNVs as well
as heterozygous SNVs shared by more than one tumor with the same genome background were assumed to be germ-line mutations and were not
used for epitope prediction unless located near a unique heterozygous SNV.
For each unique heterozygous SNV, reference and alternative peptide sequences were generated based on the two inferred haplotypes for each
CCDS transcript. Generated amino acid sequences were then run through
the NetMHC 3.0 epitope prediction program (Lundegaard et al., 2008) and
scored using the Profile Weight Matrix (PWM) algorithm with default detection thresholds.
Binding assays. Binding of peptides to H-2 Kd was determined using
quantitative assays based on the inhibition of binding of a radiolabeled standard peptide to purified MHC molecules essentially as described previously
(van der Most et al., 1996; Sidney et al., 2013). Peptides were typically tested
at six different concentrations covering a 100,000-fold dose range, and in
three or more independent assays. Under the conditions used, where [label] <
[MHC] and IC50 [MHC], the measured IC50 values are reasonable approximations of the dissociation constant values (Gulukota et al., 1997).
Intracellular IFN- assay by FACS and ELISpot. Lymphocytes were
incubated either with or without 110 g/ml peptide. GolgiPlug (BD) was
added 1 h later. After incubation of 12 to 16 h, cells were stained for CD44
(clone IM7), CD4 (clone GK1.5) and CD8 (clone 536.7; BD), fixed and
permeabilized using the Cytofix/Cytoperm kit (BD), and stained for intracellular IFN- using Phycoerythrin-conjugated anti-mouse IFN- (clone
XMG1.2, BD). Cells were stained with 1 l antibody/million cells in 50 l
staining buffer (PBS with 1% bovine serum albumin) and incubated for
20 min at 4C in the dark, or according to the manufacturers instructions.
Cells without peptide stimulation were used as a negative control, and the
values for these controls was very close to the values seen with the negative
control peptide. Typically, 95,000129,000 lymphocytes (14,50017,000
CD8+CD4 cells) were acquired. Our background is consistently very low
(10% of the signal).
For the ELISpot assays, our negative controls are CD8+ cells from immunized mice without peptide stimulation. We consider the peptides to be
positive or immunogenic when spots from peptide-stimulated wells are significant higher by Mann-Whitney test, compared with wells without cognate peptide stimulation. We rate the magnitude of responses by mean spot
numbers per million CD8+ cells: 510(+); 1120 (++); 2150 (+++); 51
100(++++); and >100(+++++).
Tumor challenge and representation of tumor growth. AUC as a tool
to measure tumor growth has been described previously (Duan et al., 2012).
In brief, AUC was calculated by selecting Curves & Regression and then
Area under curve from the analyze tool, using the Prism 5.0 (GraphPad).
Grubbs test was used to remove up to one outlier from each group.
Depletion of T cell subsets. Immunized mice were depleted of CD8 cells
using anti-CD8 rat IgG2b monoclonal antibody 2.43, or depleted of CD4
cells using anti-CD4 rat IgG2b monoclonal antibody GK1.5. Depleting antibodies were given in PBS i.p. 2 d before tumor challenge and every 7 d for
the duration of the experiment. The first 3 injections of depleting antibodies
were 250 g per mouse; later injections were 500 g per mouse. For treatment with antagonistic antibodies, mice were treated with anti-CD25 antibody (clone PC61, 250 g, 2 d before tumor challenge) or antiCTLA-4
antibody (clone 9D9; 100 g; 7 d before and every 3 d after tumor challenge).
The appropriate T cell subsets were depleted by >95%.
2245
Modeling of peptide/H-2Kd complexes. Models of peptide/H-2Kd

complexes were built by adapting a previous method by Park et al. (2013)
used to identify immunogenic epitopes. Although developed on HLAA*0201, the approach is generally applicable to class I MHC proteins in general, taking advantage of common class I MHC structural features.
An initial model of each complex was generated by MODELLER and
the build mutants functionality implemented in Accelrys Discovery Studio
(Feyfant et al., 2007; Webb and Sali 2014). The protocol uses a heavy-atom
representation of the protein and includes homology-derived restraints combined with energy minimization and molecular dynamics/simulated annealing. The structure of the Flu peptide bound to H-2Kd was used as a template
(Mitaksov and Fremont 2006). Atoms within 4.5 of each altered residue
were allowed to repack during the modeling. For each pMHC, one hundred
initial models were generated, and the lowest energy model from this first set
was subjected to a more exhaustive, second phase of fully atomistic simulated
annealing and molecular dynamics. For the second phase, after adding hydrogens, the structure was heated to 1500 K over 200 ps of simulation followed
by cooling to 300 K over 800 ps. The annealed structure was then subjected
to five independent 10 ns molecular dynamics runs at 300 K, each time beginning with the structure that resulted from the second phase annealing step.
All second phase dynamics calculations were performed using AMBER
12 running on Nvidia GPU accelerators (Case et al., 2005).The ff99SB force
field was used with a 2 fs time step. Solvent was treated implicitly using the
generalized Born model to accelerate sampling (Gtz et al., 2012). The
SHAKE algorithm was used to constrain all bonds to hydrogens. A 20-kcal/mol
harmonic restraint was applied to the 1 and 2 helices (residues 5685 and
138175). Two additional 20 kcal/mol distance restraints were applied to hydrogen bonds at the N- and C-terminal ends of the peptide. The first was
between the P1 backbone oxygen and the hydroxyl of Tyr 159 of H-2Kd.The
second was between the P8 backbone oxygen and the ring nitrogen of Trp
147 of H-2Kd. As a positive control, we performed the simulated annealing
and molecular dynamics steps of the procedure on the structure of an HBV
peptide presented by Kd (Zhou et al., 2004). As shown in Figs. 4 B and 4D,
the viral peptide was predicted to be relatively rigid in the Kd binding groove.
For the data in Fig. 4 B, average peptide structures were calculated from
the 50 ns of simulation data for each pMHC, and all common atoms of pairs
of mutant and WT peptides superimposed to generate the RMSD. For the
data in Figs. 4 (CE) and Fig. S1, RMSFs were computed for the carbons
of the peptides from the 50 ns of simulations.
Statistical analysis. P-values for group comparisons were calculated using a
two-tailed nonparametric MannWhitney test, using GraphPad Prism 5.0
(GraphPad). For tumor rejection assays, Grubb test was used to remove up to
one outlier from each group. Fishers exact test was used to test association
between pairs of categorical parameters. Statistical significance of a Pearson
correlation coefficient was computed using two-sided Students t test as described in (Cohen et al., 2003).
Online supplemental material. Table S1 contains the entire output of the
EpiSeq pipeline for Meth A and CMS5 tumors. Fig. S1, which accompanies
Fig. 4, shows the complete data for root mean square fluctuations for the
carbons of all top DAI ranked nonamers from the structural modeling.
Figs. S2 and S3 show the FACS gating strategies and representative primary
data for Fig. 1 and Fig. 5 C, respectively. Online supplemental material is
available at http://www.jem.org/cgi/content/full/jem.20141308/DC1.
We acknowledge Drs. Anthony T. Vella and Adam Adler of UConn School of
Medicine for helpful discussions.
This research was funded in part by an award from the Cancer Research
Institute, New York (P.K. Srivastava), the Northeastern Utilities Chair in Experimental
Oncology (P.K. Srivastava), the Personalized Immunotherapy Core Interest Group of
the Connecticut Institute for Clinical and Translational Science (P.K. Srivastava), a
SPARK award (P.K. Srivastava, I.I. Mandoiu, F. Duan, S. Al Seesi), award IIS-0916948
(I.I. Mandoiu) from National Science Foundation, awards GM067079 and GM103773
from National Institutes of Health (B.M. Baker), Collaborative Research Grant
AG110891 on Software for Robust Transcript Discovery and Quantification from
2246
Life Technologies (I.I. Mandoiu), and the Agriculture and Food Research Initiative
Competitive Grant no. 2011-67016-30331 from the USDA National Institute of
Food and Agriculture (to I.I. Mandoiu).
This paper is dedicated to the cherished memory of Lloyd J. Old for his
mentorship and friendship to P.K. Srivastava.
P.K. Srivastava has a significant equity interest in Accuragen Inc., which has
obtained an option to license the intellectual property relating to the results
described in this manuscript. The authors have no additional financial interests.
Submitted: 11 July 2014
Accepted: 5 September 2014
REFERENCES
Aird, D., M.G. Ross, W.S. Chen, M. Danielsson, T. Fennell, C. Russ, D.B.
Jaffe, C. Nusbaum, and A. Gnirke. 2011. Analyzing and minimizing
PCR amplification bias in Illumina sequencing libraries. Genome Biol.
12:R18. http://dx.doi.org/10.1186/gb-2011-12-2-r18
Assarsson, E., J. Sidney, C. Oseroff, V. Pasquetto, H.H. Bui, N. Frahm, C.
Brander, B. Peters, H. Grey, and A. Sette. 2007. A quantitative analysis of
the variables affecting the repertoire of T cell specificities recognized after
vaccinia virus infection. J. Immunol. 178:78907901. http://dx.doi.org/
10.4049/jimmunol.178.12.7890
Bailey, D.W. 1978. Sources of subline divergences and their relative importance for sublines of six major inbred strains of mice. In Origins
of inbred mice. H.C. Morse III, editor. Elsevier http://dx.doi.org/10
.1016/B978-0-12-507850-4.50020-2
Borbulevych, O.Y., T.K. Baxter, Z. Yu, N.P. Restifo, and B.M. Baker.
2005. Increased immunogenicity of an anchor-modified tumor-associated
antigen is due to the enhanced stability of the peptide/MHC complex:
implications for vaccine design. J. Immunol. 174:48124820. http://dx
.doi.org/10.4049/jimmunol.174.8.4812
Borbulevych, O.Y., F.K. Insaidoo, T.K. Baxter, D.J. Powell Jr., L.A.
Johnson, N.P. Restifo, and B.M. Baker. 2007. Structures of MART126/27-35 Peptide/HLA-A2 complexes reveal a remarkable disconnect
between antigen structural homology and T cell recognition. J. Mol.
Biol. 372:11231136. http://dx.doi.org/10.1016/j.jmb.2007.07.025
Brass, A.L., D.M. Dykxhoorn, Y. Benita, N. Yan, A. Engelman, R.J. Xavier,
J. Lieberman, and S.J. Elledge. 2008. Identification of host proteins required for HIV infection through a functional genomic screen. Science.
319:921926. http://dx.doi.org/10.1126/science.1152725
Buckwalter, M.R., and P.K. Srivastava. 2008. It is the antigen(s), stupid and
other lessons from over a decade of vaccitherapy of human cancer. Semin.
Immunol. 20:296300. http://dx.doi.org/10.1016/j.smim.2008.07.003
Burnet, F.M. 1970. The concept of immunological surveillance. Prog. Exp.
Tumor Res. 13:127.
Callahan, M.K., J.D. Wolchok, and J.P. Allison. 2010. Anti-CTLA-4 anti
body therapy: immune monitoring during clinical development of a
novel immunotherapy. Semin. Oncol. 37:473484. http://dx.doi.org/
10.1053/j.seminoncol.2010.09.001
Cao, K., J. Hollenbach, X. Shi, W. Shi, M. Chopek, and M.A. FernndezVia. 2001. Analysis of the frequencies of HLA-A, B, and C alleles and
haplotypes in the five major ethnic groups of the United States reveals
high levels of diversity in these loci and contrasting distribution patterns
in these populations. Hum. Immunol. 62:10091030. http://dx.doi.org/
10.1016/S0198-8859(01)00298-1
Case, D.A., T.E. Cheatham III, T. Darden, H. Gohlke, R. Luo, K.M. Merz
Jr., A. Onufriev, C. Simmerling, B. Wang, and R.J. Woods. 2005. The
Amber biomolecular simulation programs. J. Comput. Chem. 26:1668
1688. http://dx.doi.org/10.1002/jcc.20290
Castle, J.C., S. Kreiter, J. Diekmann, M. Lwer, N. van de Roemer, J. de Graaf,
A. Selmi, M. Diken, S. Boegel, C. Paret, et al. 2012. Exploiting the mutanome for tumor vaccination. Cancer Res. 72:10811091. http://dx.doi
.org/10.1158/0008-5472.CAN-11-3722
Cohen, J., P. Cohen, S.G. West, and L.S. Aiken. 2003. Applied Multiple
Regression/Correlation Analysis for the Behavioral Sciences. Third edition. Lawrence Earlbaum Associates, Mahwah, NJ.
Corbire, V., J. Chapiro, V. Stroobant, W. Ma, C. Lurquin, B. Leth,
N. van Baren, B.J. Van den Eynde, T. Boon, and P.G. Coulie. 2011.
Antigen spreading contributes to MAGE vaccination-induced regression
Ar ticle
of melanoma metastases. Cancer Res. 71:12531262. http://dx.doi

.org/10.1158/0008-5472.CAN-10-2693
Ding, Y.H., B.M. Baker, D.N. Garboczi, W.E. Biddison, and D.C. Wiley.
1999. Four A6-TCR/peptide/HLA-A2 structures that generate very
different T cell signals are nearly identical. Immunity. 11:4556. http://
dx.doi.org/10.1016/S1074-7613(00)80080-1
Duan, F., Y. Lin, C. Liu, M.E. Engelhorn, A.D. Cohen, M. Curran, S.
Sakaguchi, T. Merghoub, S. Terzulli, J.D. Wolchok, and A.N.
Houghton. 2009. Immune rejection of mouse tumors expressing mutated self. Cancer Res. 69:35453553. http://dx.doi.org/10.1158/00085472.CAN-08-2779
Duan, F., S. Simeone, R. Wu, J. Grady, I. Mandoiu, and P.K. Srivastava.
2012. Area under the curve as a tool to measure kinetics of tumor
growth in experimental animals. J. Immunol. Methods. 382:224228.
http://dx.doi.org/10.1016/j.jim.2012.06.005
Duitama, J., P.K. Srivastava, and I.I. Mndoiu. 2012. Towards accurate detection and genotyping of expressed variants from whole transcriptome
sequencing data. BMC Genomics. 13:S6. http://dx.doi.org/10.1186/
1471-2164-13-S2-S6
Dunn, G.P., L.J. Old, and R.D. Schreiber. 2004. The immunobiology of
cancer immunosurveillance and immunoediting. Immunity. 21:137148.
http://dx.doi.org/10.1016/j.immuni.2004.07.017
Egen, J.G., M.S. Kuhns, and J.P. Allison. 2002. CTLA-4: new insights into
its biological function and use in tumor immunotherapy. Nat. Immunol.
3:611618. http://dx.doi.org/10.1038/ni0702-611
Feyfant, E., A. Sali, and A. Fiser. 2007. Modeling mutations in protein structures.
Protein Sci. 16:20302041. http://dx.doi.org/10.1110/ps.072855507
Fritsch, E.F., M. Rajasagi, P.A. Ott, V. Brusic, N. Hacohen, and C.J. Wu.
2014. HLA-binding properties of tumor neoepitopes in humans. Cancer
Immunol Res. 2:522529. http://dx.doi.org/10.1158/2326-6066.
CIR-13-0227
Gtz, A.W., M.J. Williamson, D. Xu, D. Poole, S. Le Grand, and R.C.
Walker. 2012. Routine Microsecond Molecular Dynamics Simulations
with AMBER on GPUs. 1. Generalized Born. J. Chem. Theory Comput.
8:15421555. http://dx.doi.org/10.1021/ct200909j
Gulukota, K., J. Sidney, A. Sette, and C. DeLisi. 1997. Two complementary methods for predicting peptides binding major histocompatibility complex molecules. J. Mol. Biol. 267:12581267. http://dx.doi
.org/10.1006/jmbi.1997.0937
Hanson, H.L., D.L. Donermeyer, H. Ikeda, J.M. White, V. Shankaran, L.J.
Old, H. Shiku, R.D. Schreiber, and P.M. Allen. 2000. Eradication of
established tumors by CD8+ T cell adoptive immunotherapy. Immunity.
13:265276. http://dx.doi.org/10.1016/S1074-7613(00)00026-1
Ikeda, H., N. Ohta, K. Furukawa, H. Miyazaki, L. Wang, K. Kuribayashi,
L.J. Old, and H. Shiku. 1997. Mutated mitogen-activated protein kinase: a tumor rejection antigen of mouse sarcoma. Proc. Natl. Acad. Sci.
USA. 94:63756379. http://dx.doi.org/10.1073/pnas.94.12.6375
Insaidoo, F.K., O.Y. Borbulevych, M. Hossain, S.M. Santhanagopolan,
T.K. Baxter, and B.M. Baker. 2011. Loss of T cell antigen recognition arising from changes in peptide and major histocompatibility complex protein flexibility: implications for vaccine design. J. Biol. Chem.
286:4016340173. http://dx.doi.org/10.1074/jbc.M111.283564
Keane, T.M., L. Goodstadt, P. Danecek, M.A. White, K. Wong, B. Yalcin, A.
Heger, A. Agam, G. Slater, M. Goodson, et al. 2011. Mouse genomic
variation and its effect on phenotypes and gene regulation. Nature.
477:289294. http://dx.doi.org/10.1038/nature10413
Langmead, B., C.Trapnell, M. Pop, and S.L. Salzberg. 2009. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome.
Genome Biol. 10:R25. http://dx.doi.org/10.1186/gb-2009-10-3-r25
Lennerz,V., M. Fatho, C. Gentilini, R.A. Frye, A. Lifke, D. Ferel, C. Wlfel, C.
Huber, and T. Wlfel. 2005. The response of autologous T cells to a human
melanoma is dominated by mutated neoantigens. Proc. Natl. Acad. Sci.
USA. 102:1601316018. http://dx.doi.org/10.1073/pnas.0500090102
Levey, D.L., H. Udono, M. Heike, and P.K. Srivastava. 2001. Identification
of a tumor-associated contact-dependent activity which reversibly
downregulates cytolytic function of CD8+ T cells. Cancer Immun. 1:5.
Li, H., B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G.
Marth, G. Abecasis, and R. Durbin. 1000 Genome Project Data
Processing Subgroup. 2009. The Sequence Alignment/Map format
and SAMtools. Bioinformatics. 25:20782079. http://dx.doi.org/10

.1093/bioinformatics/btp352
Lu, Y.C., X. Yao, Y.F. Li, M. El-Gamil, M.E. Dudley, J.C. Yang, J.R. Almeida,
D.C. Douek,Y. Samuels, S.A. Rosenberg, and P.F. Robbins. 2013. Mutated
PPP1R3B is recognized by T cells used to treat a melanoma patient who
experienced a durable complete tumor regression. J. Immunol. 190:6034
6042. http://dx.doi.org/10.4049/jimmunol.1202830
Lundegaard, C., O. Lund, and M. Nielsen. 2008. Accurate approximation
method for prediction of class I MHC affinities for peptides of length
8, 10 and 11 using prediction tools trained on 9mers. Bioinformatics.
24:13971398. http://dx.doi.org/10.1093/bioinformatics/btn128
Markiewicz, M.A., F. Fallarino, A. Ashikari, and T.F. Gajewski. 2001.
Epitope spreading upon P815 tumor rejection triggered by vaccination with the single class I MHC-restricted peptide P1A. Int. Immunol.
13:625632. http://dx.doi.org/10.1093/intimm/13.5.625
Matsushita, H., M.D. Vesely, D.C. Koboldt, C.G. Rickert, R. Uppaluri, V.J.
Magrini, C.D. Arthur, J.M.White,Y.S. Chen, L.K. Shea, et al. 2012. Cancer
exome analysis reveals a T-cell-dependent mechanism of cancer immunoediting. Nature. 482:400404. http://dx.doi.org/10.1038/nature10755
Mitaksov, V., and D.H. Fremont. 2006. Structural definition of the H-2Kd
peptide-binding motif. J. Biol. Chem. 281:1061810625. http://dx.doi
.org/10.1074/jbc.M510511200
Narzi, D., C.M. Becker, M.T. Fiorillo, B. Uchanska-Ziegler, A. Ziegler,
and R.A. Bckmann. 2012. Dynamical characterization of two differentially disease associated MHC class I proteins in complex with
viral and self-peptides. J. Mol. Biol. 415:429442. http://dx.doi.org/10
.1016/j.jmb.2011.11.021
Park, M.S., S.Y. Park, K.R. Miller, E.J. Collins, and H.Y. Lee. 2013.
Accurate structure prediction of peptide-MHC complexes for identifying highly immunogenic antigens. Mol. Immunol. 56:8190. http://
dx.doi.org/10.1016/j.molimm.2013.04.011
Rajasagi, M., S.A. Shukla, E.F. Fritsch, D.B. Keskin, D. DeLuca, E. Carmona,
W. Zhang, C. Sougnez, K. Cibulskis, J. Sidney, et al. 2014. Systematic
identification of personal tumor-specific neoantigens in chronic lymphocytic leukemia. Blood. 124:453462. http://dx.doi.org/10.1182/
blood-2014-04-567933
Ren, S., Z. Peng, J.H. Mao, Y. Yu, C. Yin, X. Gao, Z. Cui, J. Zhang,
K. Yi, W. Xu, et al. 2012. RNA-seq analysis of prostate cancer in the
Chinese population identifies recurrent gene fusions, cancer-associated
long noncoding RNAs and aberrant alternative splicings. Cell Res. 22:
806821. http://dx.doi.org/10.1038/cr.2012.30
Segal, N.H., D.W. Parsons, K.S. Peggs,V.Velculescu, K.W. Kinzler, B.Vogelstein, and
J.P. Allison. 2008. Epitope landscape in breast and colorectal cancer. Cancer
Res. 68:889892. http://dx.doi.org/10.1158/0008-5472.CAN-07-3095
Sharma, A.K., J.J. Kuhns, S. Yan, R.H. Friedline, B. Long, R. Tisch, and
E.J. Collins. 2001. Class I major histocompatibility complex anchor
substitutions alter the conformation of T cell receptor contacts. J. Biol.
Chem. 276:2144321449. http://dx.doi.org/10.1074/jbc.M010791200
Sidney, J., S. Southwood, C. Moore, C. Oseroff, C. Pinilla, H.M. Grey,
and A. Sette. Measurement of MHC/Peptide Interactions by Gel
Filtration or Monoclonal Antibody Capture. Curr Protoc Immunol.
2013; Chapter 18:Unit18.3.
Srivastava, P.K. 1993. Peptide-binding heat shock proteins in the endoplasmic reticulum: role in immune response to cancer and in antigen presentation. Adv. Cancer Res. 62:153177. http://dx.doi.org/10
.1016/S0065-230X(08)60318-8
Srivastava, N., and P.K. Srivastava. 2009. Modeling the repertoire of
true tumor-specific MHC I epitopes in a human tumor. PLoS ONE.
4:e6094. http://dx.doi.org/10.1371/journal.pone.0006094
van der Bruggen, P., V. Stroobant, N. Vigneron, and B. Van den Eynde.
Peptide database: T cell-defined tumor antigens. Cancer Immun 2013
http://www.cancerimmunity.org/peptide/
van der Most, R.G., A. Sette, C. Oseroff, J. Alexander, K. Murali-Krishna,
L.L. Lau, S. Southwood, J. Sidney, R.W. Chesnut, M. Matloubian, and
R. Ahmed. 1996. Analysis of cytotoxic T cell responses to dominant and
subdominant epitopes during acute and chronic lymphocytic choriomeningitis virus infection. J. Immunol. 157:55435554.
van Rooij, N., M.M. van Buuren, D. Philips, A. Velds, M. Toebes, B.
Heemskerk, L.J. van Dijk, S. Behjati, H. Hilkmann, D. El Atmioui,
2247
et al. 2013. Tumor exome analysis reveals neoantigen-specific T-cell reactivity in an ipilimumab-responsive melanoma. J. Clin. Oncol. 31:e439
e442. http://dx.doi.org/10.1200/JCO.2012.47.7521
Vogelstein, B., N. Papadopoulos, V.E. Velculescu, S. Zhou, L.A. Diaz Jr.,
K.W. Kinzler, L.J. van Dijk, B. Vogelstein, N. Papadopoulos, V.E.
Velculescu, et al. 2013. Cancer genome landscapes. Science. 339:1546
1558. http://dx.doi.org/10.1126/science.1235122
Webb, B., and A. Sali. 2014. Protein structure modeling with MODELLER.
Methods Mol. Biol. 1137:115. http://dx.doi.org/10.1007/978-14939-0366-5_1
Wei, X., V. Walia, J.C. Lin, J.K. Teer, T.D. Prickett, J. Gartner, S. Davis, K.
Stemke-Hale, M.A. Davies, J.E. Gershenwald, et al. NISC Comparative
2248
Sequencing Program. 2011. Exome sequencing identifies GRIN2A

as frequently mutated in melanoma. Nat. Genet. 43:442446. http://
dx.doi.org/10.1038/ng.810
Winkler, K., A. Winter, C. Rueckert, B. Uchanska-Ziegler, and U. Alexiev.
2007. Natural MHC class I polymorphism controls the pathway of peptide dissociation from HLA-B27 complexes. Biophys. J. 93:27432755.
http://dx.doi.org/10.1529/biophysj.106.096602
Zhou, M., Y. Xu, Z. Lou, D.K. Cole, X. Li, Y. Liu, P. Tien, Z. Rao, and
G.F. Gao. 2004. Complex assembly, crystallization and preliminary
X-ray crystallographic studies of MHC H-2Kd complexed with an
HBV-core nonapeptide. Acta Crystallogr. D Biol. Crystallogr. 60:1473
1475. http://dx.doi.org/10.1107/S0907444904013587
INSIGHTS
CCR4 dr ives ATLL jail break
Adult T cell leukemia/lymphoma (ATLL), caused by human T cell lymphotropic virus 1 (HTLV-1), is
an aggressive cancer that is refractory to current therapies. The long latency and low overall penetrance of
ATLL in HTLV-1infected individuals infers the need for cooperating events, which include somatic
JAK3, NOTCH1, and FAS mutations. While overexpression of CCR4 is a hallmark of ATLL, it is not clear
whether dysregulation of CCR4 function contributes to disease pathogenesis. In this issue, Nakagawa et
al. report recurrent somatic mutations in the CCR4 chemokine receptor in ~25% of ATLL cases, implicating these mutations in ATLL pathogenesis.
The authors performed RNA transcriptome analysis of two ATLL cases and targeted sequencing of addi- Insight from
tional ATLL patient samples and cell lines. Remarkably, the CCR4 mutations found in primary ATLL speci- Kevin Shannon
mens were heterozygous and introduced missense or truncating mutations in a conserved carboxy-terminal
domain of CCR4 involved in negative regulation. Together, these findings suggested a dominant gain-of-function mechansim of
action. Indeed, elegant functional studies revealed defective internalization of these mutant CCR4 proteins as well as enhanced
migration and chemotaxis in response to chemokines. The authors also demonstrated hyperactive PI3 kinase/Akt signaling in
ATLL cells expressing CCR4 mutant proteins.
So, how do CCR4 mutations promote malignant growth? The authors provide two logical and nonexclusive explanations. First, these mutations might enable ATLL cells to migrate to and colonize niches in tissues such as skin and lymph nodes
that are favorable for cancer cell survival and proliferation. If this
idea is correct, it is possible that patients with and without CCR4
mutations will exhibit specific patterns of tissue involvement and
disease evolution. Second, dysregulated PI3K signaling downstream of mutant CCR4 might be the key biochemical driver contributing to clonal selection of ATLL cells. Given this underlying
biology, it is reasonable to speculate that seed and soil both
contribute to the aberrant growth of ATLL tumors with somatic
CCR4 mutations. For example, because CCR4 mutations render
PI3K signaling hypersensitive to chemokine stimulation, specific
tissue microenvironments likely favor ATLL growth through paracrine mechanisms.
An exciting aspect of these new mechanistic insights is their
potential for clinical translation. A first generation anti-CCR4
antibody called KW-0761 is showing promise in early phase clinical
trials. It will be interesting to determine whether mutant CCR4 is
Schematic of CCR4 mutant isoforms in ATLL
a predictive biomarker of sensitivity to this and other anti-CCR4
agents. Deep genomic analysis of tumors with CCR4 mutations
that relapse after an initial response will provide additional insights. PI3 kinase inhibitorsboth alone and in combination with
other agentsare another potential therapeutic strategy for improving outcomes in this relentless cancer.
Nakagawa, M., et al. 2014. J. Exp. Med. http://dx.doi.org/10.1084/jem.20140987.
Kevin M. Shannon, University of California, San Francisco: shannonk@peds.ucsf.edu
Brief Definitive Report
Gain-of-function CCR4 mutations in adult

T cell leukemia/lymphoma
Masao Nakagawa,* Roland Schmitz,* Wenming Xiao, Carolyn K. Goldman,
Weihong Xu, Yandan Yang, Xin Yu, Thomas A. Waldmann,**
and Louis M. Staudt**
Lymphoid Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health,
Bethesda, MD 20892
Adult T cell leukemia/lymphoma (ATLL) is an aggressive malignancy caused by human T cell

lymphotropic virus type-I (HTLV-I) without curative treatment at present. To illuminate the
pathogenesis of ATLL we performed whole transcriptome sequencing of purified ATLL
patient samples and discovered recurrent somatic mutations in CCR4, encoding CC chemokine receptor 4. CCR4 mutations were detected in 14/53 ATLL samples (26%) and consisted exclusively of nonsense or frameshift mutations that truncated the coding region at
C329, Q330, or Y331 in the carboxy terminus. Functionally, the CCR4-Q330 nonsense
isoform was gain-of-function because it increased cell migration toward the CCR4 ligands
CCL17 and CCL22, in part by impairing receptor internalization. This mutant enhanced
PI(3) kinase/AKT activation after receptor engagement by CCL22 in ATLL cells and conferred a growth advantage in long-term in vitro cultures. These findings implicate somatic
gain-of-function CCR4 mutations in the pathogenesis of ATLL and suggest that inhibition
of CCR4 signaling might have therapeutic potential in this refractory malignancy.
CORRESPONDENCE
Louis M. Staudt:
lstaudt@mail.nih.gov
Abbreviations used: ATLL,
adult T cell leukemia/
lymphoma; CCR4, CC
chemokine receptor 4; GPCR,
G proteincoupled receptor;
huKO, human KusabiraOrange; PI3K, PI(3) kinase;
PTX, pertussis toxin; SNV,
single nucleotide variant.
Adult T cell leukemia/lymphoma (ATLL) is

one of the most aggressive forms of peripheral
T cell lymphoma, with a median survival of
<1 yr with current therapy, which consists primarily of cytotoxic chemotherapy (Campo et al.,
2011). Molecular analyses of ATLL cells revealed that high expression of CC chemokine
receptor 4 (CCR4) is a hallmark of this disease
(Yoshie et al., 2002; Ishida et al., 2003; Iqbal
et al., 2010). Clinical trials in ATLL of a therapeutic monoclonal antibody directed against
CCR4 (KW-0761) are ongoing, and promising
early results have been reported (Yamamoto et al.,
2010; Ishida et al., 2012).
CCR4 is a chemokine receptor that has a
critical role in immune cell trafficking.T-helper
type 2 cells (Th2), regulatory T cells (Treg),
interluekin-17producing T-helper cells (Th17),
and skin-homing memory T cells express CCR4
on their surface and migrate toward the chemokines CCL17 and CCL22 (Imai et al., 1997,
1998; Yoshie, 2005). The leukemic cells in 90%
of ATLL cases express CCR4 on their surface
(Ishida et al., 2003). Interestingly, the most

frequent sites of ATLL involvement are lymph
nodes and skin (Campo et al., 2011), where dendritic cells, M2-phenotype macrophages, Langerhans cells, and cutaneous venules can produce
CCL17 and/or CCL22 (Campbell et al., 1999;
Vissers et al., 2001; Vulcano et al., 2001; Chong
et al., 2004). These observations suggest that
CCR4 could have a role in ATLL biology, but it
is still unclear whether dysregulation of CCR4
function contributes to ATLL pathogenesis.
Human T cell lymphotropic virus type-I
(HTLV-I) is believed to be the causative agent
for ATLL (Matsuoka and Jeang, 2007; Campo
et al., 2011). However, only a small proportion
of HTLV-I carriers (27%) develop ATLL with
a long latency (4050 yr; Arisawa et al., 2000;
Campo et al., 2011). Thus, acquisition of somatic mutations in cellular genes is likely to be
crucial for the development of ATLL. Identifying such somatic mutations is essential not
only for understanding ATLL pathogenesis but
also for defining molecular targets for therapy.
*M. Nakagawa and R. Schmitz contributed equally to this paper.

**T.A. Waldmann and L.M. Staudt contributed equally to
this paper.
This article is distributed under the terms of an AttributionNoncommercial

Share AlikeNo Mirror Sites license for the first six months after the publication
date (see http://www.rupress.org/terms). After six months it is available under a
Creative Commons License (AttributionNoncommercialShare Alike 3.0 Unported
license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/).

J. Exp. Med. 2014 Vol. 211 No. 13 24972505
2497
Somatic mutations in p53, NOTCH1, JAK3, and FAS have

been reported in ATLL (Elliott et al., 2011; Yamagishi and
Watanabe, 2012), but our knowledge of genetic aberrations in
this malignancy is nonetheless incomplete.
In the present study, we used whole transcriptome analysis (RNA-seq) to discover activating mutations in CCR4,
which we found to be a frequent genetic event in this malignancy. Functional analysis clarified the gain-of function nature
of these mutations, suggesting that dysregulation of CCR4
function is key to the pathogenesis of ATLL.
that these four lines were established from the same patient
at different clinical time points.
In cancer genome studies cataloged in the COSMIC data
base (Catalogue of somatic mutations in cancer) or conducted by The Cancer Genome Atlas (TCGA) initiative, we
found only three primary human cancer samples with CCR4
nonsense mutations, and these mutations did not target the
carboxy-terminal cytoplasmic domain. Thus, truncation of
the CCR4 cytoplasmic domain by somatic mutation appears
to be a specific and frequent genetic event to ATLL.
RESULTS AND DISCUSSION

Frequent mutation of CCR4 in ATLL
We performed RNA-seq of peripheral blood leukemia samples from two ATLL patients, TW36R and TW51R, which
allowed us to identify 85 and 127 genes with potential coding
region mutations in these two samples, respectively. These
candidates included two genes that were mutated in both
samples: CCR4 and MICALL1. High expression of CCR4 is
a well-known hallmark of ATLL, whereas MICALL1 has not
been implicated in this disease. Importantly, both ATLL samples had the same nonsense CCR4 mutation affecting the
Y331 codon (Y331*), suggesting that CCR4 mutations might
play a critical role in ATLL pathogenesis. The percentage of
mutant CCR4 sequencing reads was 39% in TW36R and
55% in TW51R. Because both blood samples had a high proportion of malignant cells (TW36R: 90%, TW51R: 84%), the
CCR4 mutations are likely to be heterozygous and potentially might exert a dominant functional effect.
By Sanger sequencing of genomic DNA, we confirmed
the heterozygous nature of the CCR4 nonsense mutations in
TW36R and TW51R. We extended this analysis to an additional cohort of ATLL primary patient samples (n = 41)
and ATLL cell lines (n = 12). CCR4 mutations were detected in 26.4% (14/53) of ATLL samples (Fig. 1, A and B;
and Table S1). In five cases for which paired normal DNA
was available, three different CCR4 mutations were detected
only in the ATLL cells (Q330*, Q330 frameshift, and Y331*),
demonstrating that they were acquired somatically during
malignant transformation or progression (Fig. 1 C). CCR4
mutations were identified in both primary ATLL patient samples
(24%, 10/41) and in ATLL cell lines (33.3%, 4/12). Mutations were heterozygous in all samples except one cell line
(ATL42T+). We also confirmed that mutant CCR4 mRNAs
were heterozygously expressed in six primary ATLL samples (Table S1). These findings indicated that CCR4 mutations in ATLL might be either gain-of-function or potentially
dominant negative.
All CCR4 mutations were either nonsense or frameshift
in nature and affected the codons for four nearby amino acids
(F326, C329, Q330, and Y331) that are located in an evolutionarily conserved region in the carboxy terminus of the
protein (Fig. 1, A and B). The most frequent nonsense mutation affected codon Y331 (Y331*; 7/53, 13%). Four cell lines
(ED40515(+), ED40515(), ED41214(+), and ED41214 C())
had Q330* nonsense mutations, which is understandable given
Mutant CCR4 enhances chemotaxis toward CCR4 ligands

Chemokine receptors, including CCR4, belong to the seventransmembrane G proteincoupled receptor (GPCR) family.
All CCR4 mutations in ATLL encode truncated receptors that
lack most of the carboxy-terminal cytoplasmic domain, which
typically serves a regulatory role in GPCRs (Fig. 1, A and B;
Luttrell and Lefkowitz, 2002). The loss of this region of CCR4
might deregulate its activity, potentially altering the migration
of ATLL cells in response to CCR4 ligands. The CCR4 mutations in ATLL are reminiscent of mutations affecting the
chemokine receptor CXCR4 in WHIM syndrome, a human
immunodeficiency disease (Diaz, 2005). Most CXCR4 mutations in WHIM syndrome are nonsense or frameshift mutations that truncate the carboxy-terminal cytoplasmic domain
of the protein, conferring a gain-of-function phenotype
with respect to chemotaxis toward the CXCR4 ligand SDF1
(Balabanian et al., 2005; Diaz, 2005; Kawai et al., 2005).
We therefore hypothesized that mutant CCR4 isoforms
might enhance chemotaxis of the affected cells to CCR4 ligands. To study this, we infected a mouse myeloid cell line,
32D, with retroviral vectors expressing WT CCR4 (CCR4WT) or CCR4-Q330* coding regions and tested the migration of the transduced cells toward CCL22, the most potent
CCR4 ligand (Fig. 2 A; DAmbrosio et al., 2002). Surface CCR4
levels were similar between CCR4-WT and CCR4-Q330*
transduced 32D cells, whereas mock vectortransduced 32D
cells did not have detectable CCR4 expression (Fig. 2 B). Mock
vectorinfected 32D cells did not migrate toward CCL22
(Fig. 2, A and B). CCR4-WTtransduced 32D cells migrated with a typical bell-shaped doseresponse curve (Fig. 2,
A and B). Compared with CCR4-WT, CCR4-Q330*
transduced 32D cells migrated to a significantly greater extent (Fig. 2, A and B). These results support the view that the
CCR4 mutants in ATLL are gain-of-function with respect to
ligand-directed chemotaxis.
Next we evaluated the ability of CCR4 isoforms to mediate chemotaxis in ATLL cell lines (Fig. 2 C). Most ATLL
cell lines express CCR4 on their surface (not depicted), reflecting the high frequency (90%) of CCR4 expression in
primary ATLL cases (Ishida et al., 2003). To evaluate CCR4WT and CCR4-Q330* in ATLL cells, we first knocked
down the endogenous expression of CCR4 in ED40515(+)
cells using an shRNA targeting the 3 untranslated region
(UTR) of the CCR4 mRNA. This resulted in a >90% reduction of surface CCR4 expression (Fig. 2 D) and substantially
2498
Gain-of-function CCR4 mutations in ATLL | Nakagawa et al.
Figure 1. CCR4 mutations in ATLL. (A) Amino acid residues in the carboxy-terminal region of CCR4. (B) Schematic of CCR4 mutant isoforms in ATLL.
(C) DNA sequence of the CCR4-Y331* mutation in the TW14R ATLL biopsy sample by Sanger sequencing (bottom). The WT sequence of normal DNA obtained
from the same patient is shown in the top.
decreased chemotaxis of the cells (Fig. 2 C). Cells were then

transduced with retroviruses expressing CCR4-WT or
CCR4-Q330* coding regions, resulting in cell populations
with equivalent expression of CCR4 on the cell surface
(Fig. 2 D). In chemotaxis assays with CCL17, CCR4-WT
reconstituted cells exhibited greater dose-dependent migration than mock vectorreconstituted cells, but migration of
CCR4-Q330*reconstituted cells was significantly greater
than either of the other populations (Fig. 2 C, left). In response to CCL22, the most potent CCR4 ligand (DAmbrosio
et al., 2002), CCR4-Q330*reconstituted ED40515(+) cells
again displayed significantly greater chemotaxis than CCR4WTreconstituted cells and responded better to lower concentrations of the ligand (Fig. 2 C, right). Because CCR4
mutations in ATLL samples were heterozygous, we also
analyzed whether these mutations could enhance chemotaxis
in the presence of WT CCR4. To this end, we transduced
the shCCR4-ED40515(+) line with one retroviral vector
that coexpresses CCR4-WT and the human Kusabira-Orange
(huKO) fluorescent protein and another vector that co
expresses CCR4-Q330 and GFP. This strategy allowed us to
compare the phenotype of GFP/huKO double-positive cells
that expressed CCR4-WT and Q330* with GFP or huKO
single-positive cells that only ectopically expressed one CCR4
isoform (Fig. 2, E and F). Cells expressing both CCR4-WT
and CCR4-Q330* showed greater CCL22-directed migration than cells expressing CCR4-WT alone, which was
similar to the phenotype of cells expressing CCR4-Q330*
alone (Fig. 2 F). These results suggest that ATLL cells may
acquire CCR4 mutations to migrate more effectively toward
their ligands.
Mutant CCR4 impairs receptor internalization

after CCL22 binding
The carboxy-terminal region of CCR4 that is truncated by
mutations in ATLL contains a serine- and threonine-rich
motif that is shared by many GPCRs (Fig. 1 A, amino acid
positions 342351). These serine and threonine residues
become rapidly phosphorylated in response to ligand, which
results in receptor internalization and contributes to desensitization of the cells to ligand (Luttrell and Lefkowitz,
2002). Carboxy-terminal truncation of CXCR4 in WHIM
syndrome impairs receptor internalization, contributing to
the enhanced migration of these cells in response to ligand
(Balabanian et al., 2005; Kawai et al., 2005).
We therefore studied the change in surface CCR4 levels
after CCL22 exposure in CCR4-WT and CCR4-Q330*
reconstituted ED40515(+) cells. In CCR4-WTreconstituted
cells, surface CCR4 levels declined rapidly, with a 58% reduction at 5 min after CCL22 exposure and a maximum reduction of 75% at 20 min (Fig. 2 G). In comparison, CCR4
internalization in CCR4-Q330*reconstituted cells was
significantly impaired, with a 27% of reduction at 5 min and
reaching only a 54% reduction at 20 min (Fig. 2 G). These
results suggest that the ATLL CCR4 mutants impair desensitization by ligand, which likely contributes to the enhanced
chemotaxis of cells bearing these mutants.
Mutant CCR4 enhances PI(3) kinase (PI3K)/AKT signaling
in response to ligand
We explored the influence of the ATLL CCR4 mutants on
PI3Kdependent activation of AKT because it has been
2499
Figure 2. CCR4 mutant isoforms enhance chemotaxis and impair receptor internalization. (A) CCL22-mediated chemotaxis of mouse 32D cells
ectopically expressing CCR4-WT or CCR4-Q330*. In these Transwell assays, the lower chamber contained the indicated amount of CCL22. After a 2-h incubation, the number of cells migrating from the upper to lower chamber was determined and plotted as a percentage of the input cell number. (B) Surface CCR4
expression levels of 32D cells ectopically expressing CCR4-WT or CCR4-Q330* analyzed by FACS. (C) Chemotactic ability of CCR4-WT or CCR4-Q330*
reconstituted ED40515(+) ATLL cells toward CCL17 and CCL22. (D) Surface CCR4 expression levels in CCR4-WT or CCR4-Q330*reconstituted ED40515(+)
ATLL cells analyzed by FACS. (E) Surface CCR4 expression levels analyzed by FACS in ED40515(+) ATLL cells ectopically expressing CCR4-WT and/or CCR4-Q330*.
(B, D, and E) Isotype control IgG staining is indicated in gray. (F) CCL22-induced chemotaxis of ED40515(+) ATLL cells ectopically expressing CCR4-WT and/or
CCR4-Q330*. (G) Time course of surface CCR4 levels after CCL22 exposure in CCR4-WT or CCR4-Q330*reconstituted ED40515(+) ATLL cells. Surface CCR4
levels were analyzed by FACS and normalized to the levels at time 0. Data in all panels are presented as mean SEM of technical duplicates representative of
at least two biological replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001 for a comparison between CCR4-WT and CCR4-Q330*.
reported that binding of CCL22 to CCR4 activates AKT in

CEM leukemic T cells and in human Th2 cells (Cronshaw
et al., 2004). We first studied two ATLL cell lines: ED40515(+),
which bears a CCR4-Q330* mutant allele, and KOB, in
which CCR4 is WT. Immunoblot analysis revealed that
baseline levels of phospho-AKT (P-AKT), a measure of AKT
activation, were much lower in ED40515(+) than in KOB
(Fig. 3 A, lane 2 vs. lane 6). However, ED40515(+) showed
stronger induction of AKT phosphorylation at 10 min after
CCL22 exposure than KOB (Fig. 3 A, lane 3 vs. lane 7). The
activation of AKT was transient in both cell lines, decreasing
by 30 min after stimulation (Fig. 3 A, lanes 4 and 8). These
findings indicated that AKT activation is one of the signaling pathways downstream of CCR4 in ATLL cells and suggested that CCR4 mutations might affect the magnitude
of AKT activation. To accurately evaluate the relative ability
of CCR4 isoforms to activate AKT, we studied ED40515(+)
cells that were transduced with a CCR4 shRNA to knock
down endogenous CCR4 expression and were reconstituted
with CCR4-WT or CCR4-Q330*. Cells were treated with
CCL22, and P-AKT levels were measured by FACS at various time points after exposure (Fig. 3 B). Knockdown of
2500
endogenous CCR4 reduced AKT activation by CCL22 compared with mock-infected cells, confirming that AKT activation
was dependent on CCR4. Ectopic provision of CCR4-WT
did not restore AKT activation despite expression of CCR4
at a higher level than in mock-transduced ED40515(+)
cells (Fig. 2 D), presumably because the endogenous CCR4
locus in ED40515(+) encodes the CCR4-Q330* isoform.
In contrast, cells reconstituted with CCR4-Q330* restored
robust AKT activation in response to CCL22 (Fig. 3 B). In
KOB cells, reconstitution with CCR4-WT modestly increased P-AKT activation in response to CCL22, but cells
reconstituted with CCR4-Q330* again responded with a
significantly greater rise in P-AKT levels (Fig. 3, C and D).
The effect of heterozygous mutation of CCR4 on AKT
activation was analyzed using the dual fluorescence strategy described above for experiments depicted in Fig. 2
(E and F). After CCL22 treatment, ED40515(+) cells expressing both CCR4-WT and CCR4-Q330* or CCR4Q330* alone had elevated p-AKT levels compared with
cells expressing only CCR4-WT (Fig. 3 E). Together, these
data demonstrate that CCR4 mutations in ATLL enhance
AKT activation in response to ligand engagement.
Figure 3. Enhanced PI3K/AKT activation by mutated

CCR4. (A) Immunoblot analysis for P-AKT levels in
ED40515(+) and KOB ATLL cells after CCL22 exposure
(50 ng/ml). Blots using P-AKT (S473) antibody are shown
with short and long exposure times. Relative scanning densitometry estimates of P-AKT levels are depicted under the
panel showing the short exposure. The pan-PI3K inhibitor
BKM120 was used at 1 M. (B and C) Time course experiment of AKT activation after CCL22 exposure (50 ng/ml)
using CCR4-reconstituted ED40515(+) cells (B) and KOB
cells (C), transduced as indicated with shRNA and cDNA
expression vectors. P-AKT (S473) levels were analyzed by
FACS at the indicated times after CCL22 exposure. MFI,
mean fluorescence intensity. (D) Surface CCR4 expression
levels in CCR4-WT or CCR4-Q330*reconstituted KOB cells.
Isotype control IgG staining is indicated in gray. (E) Time
course experiment of AKT activation after CCL22 exposure (50 ng/ml) in ED40515(+) ATLL cells ectopically expressing CCR4-WT and/or CCR4-Q330*. (F) P-AKT (S473)
levels of ED40515(+) treated with inhibitors. Cells were
pretreated with 100 ng/ml PTX for 16 h or the indicated
amount of the pan-PI3K inhibitor BKM120 for 1 h and then
incubated with 50 ng/ml CCL22 for 5 min. P-AKT (S473) was
analyzed by FACS. (G) Chemotaxis of ED40515(+) cells
treated with inhibitors. Cells were pretreated with PTX or
BKM120 as in F and then used in a CCL22-mediated chemotaxis assay. Data in all panels are presented as mean SEM
of technical duplicates representative of at least two biological replicates. *, P < 0.05; **, P < 0.01 for a comparison
between CCR4-WT and CCR4-Q330*.
We next investigated whether AKT activation is involved

in CCR4-mediated chemotaxis in ATLL. The pan-PI3K inhibitor BKM120 abrogated CCR4-mediated AKT activation
in ED40515(+) cells in a dose-dependent manner, indicating
that PI3K signaling contributes to AKT activation (Fig. 3 F).
However, BKM120 treatment only partially inhibited CCR4mediated chemotaxis by ED40515(+) cells (Fig. 3 G). In con
trast, the Gi inhibitor pertussis toxin (PTX) abrogated AKT
activation and chemotaxis in these cells (Fig. 3, F and G).
Thus, CCR4-mediated chemotaxis of ATLL cells is not primarily caused by PI3K-dependent AKT signaling, but instead
depends on Gi, consistent with previous work (Cronshaw
et al., 2004).
Mutant CCR4 promotes ATLL expansion

in the presence of ligand
Lastly, we tested whether the acquisition of CCR4 mutations
by ATLL cells imparts a selective growth advantage relative
to cells with WT CCR4. After shRNA-mediated knockdown
of endogenous CCR4 expression, ED40515(+) cells were
transduced with a retrovirus expressing CCR4-WT together
with GFP or with a retrovirus expressing only CCR4-Q330*
(Fig. 4 A, Exp. 1). After puromycin selection of infected cells,
these two cell populations were mixed in equal numbers and
cultured with or without CCL22 for 12 d. The ratio of GFP/
CCR4-Q330*expressing cells versus GFP+/CCR4-WT
expressing cells was monitored every 4 d by FACS (Fig. 4 B,
2501
Figure 4. Growth advantage of CCR4-Q330*

reconstituted ATLL cells. (A) Schematic of the competitive
growth assay. ED40515(+) cells depleted of endogenous
CCR4 expression by RNA interference were transduced with
the indicated CCR4-expressing vectors or mock vectors. After
puromycin selection, two transduced populations, one expressed GFP and the other did not, were mixed equally and
co-cultured for 12 d in vitro. The ratio of the two populations was determined by FACS every 4 d. (B) CCR4-Q330*
transduced cells have a competitive growth advantage.
CCL22 was added every 2 d at 50 ng/ml. The ratios of the
two populations were normalized to the value at day 0.
Growth curves represent the mean of eight replicates obtained from four biologically independent experiments SEM.
*, P < 0.05; ***, P < 0.001.
Exp. 1). Without CCL22, the ratio of the two populations

did not change. In contrast, in the presence of CCL22, CCR4Q330*reconstituted cells preferentially expanded in a timedependent manner. To confirm this finding, we performed
a GFP-swapping experiment in which the GFP+ cells expressed CCR4-Q330* and the GFP cells expressed CCR4WT (Fig. 4 A, Exp. 2). Again, CCR4-Q330*reconstituted
cells had a selective growth advantage in the presence of
CCL22 (Fig. 4 B, Exp. 2). As a negative control, we used
empty vectors to create GFP+ and GFP populations and
observed no change in the GFP+/GFP ratio in the presence
or absence of CCL22 (Fig. 4, A and B, Exp. 3).
In conclusion, the present study demonstrates that mutations in CCR4 are a frequent genetic event in ATLL and provides a mechanistic rationale for their selection in this cancer.
Mutant CCR4 isoforms enhanced migration of ATLL cells
toward CCL17 and CCL22 and additionally promoted PI3K/
AKT activation in response to ligand engagement, leading us
to propose two nonmutually exclusive hypotheses regarding
the role of CCR4 mutations in ATLL pathogenesis. First,
enhanced migration of ATLL cells along a CCL17/CCL22
gradient might allow them to access a favorable microenvironmental niche that could support their proliferation and/or
survival. Lymph nodes and skin are plausible ATLL niches
because CCL17 and CCL22 are known to be produced by
dendritic cells and M2-phenotype macrophages in lymph nodes
and also by Langerhans cells and venules in skin (Campbell et al.,
2502
1999; Vissers et al., 2001; Vulcano et al., 2001; Chong et al.,

2004). Previous studies showed that monocyte and M2phenotype macrophages can promote ATLL cell growth through
cellcell interaction and by paracrine mechanisms (Chen
et al., 2010; Komohara et al., 2013). Thus, the chemotaxis
promoted by CCR4 mutations in ATLL might set up favorable interactions between ATLL cells and immune bystander
cells and/or disrupt homeostatic mechanisms that keep the
growth of these cells in check. Recently, WHIM syndromelike CXCR4 somatic mutations have been detected in the
malignant cells of patients with Waldenstrms macroglobulinemia, and these mutations correlated with bone marrow involvement as well as chemotherapy drug resistance
(Roccaro et al., 2014; Treon et al., 2014). Given that CCR4
mutations and WHIM syndromelike CXCR4 mutations
share similar biological properties in terms of chemotaxis and
ATK activation (Cao et al., 2014; Treon et al., 2014), it will
be interesting to determine whether CCR4 mutations affect
the frequency of lymph node or skin involvement in ATLL or
the clinical course.
A second hypothesis raised by our findings is that CCR4
mutations may be selected in ATLL for their ability to
enhance signaling downstream of this chemokine receptor,
including activation of the PI3K/AKT signaling pathway.
During physiological CCR4 signaling, ligand desensitization
occurs in part because CCR4 is down-modulated from the cell
surface, a process which is disrupted by the CCR4 mutations
in ATLL. Perhaps as a result, ATLL cells bearing CCR4 mutant isoforms displayed prolonged PI3K/AKT activation in
response to ligand. Given the importance of PI3K/AKT signaling to both cellular metabolism and survival, this enhanced
PI3K/AKT response might provide a selective advantage for
ATLL cells. Indeed, in our long-term competitive growth
assay, ATLL cells expressing mutant CCR4 outgrew cells
with WT CCR4 in the presence of CCR4 ligand. Finally, it
is conceivable that both hypotheses detailed above may pertain. Specifically, the ability of CCR4 mutants to increase
chemotaxis toward CCR4 ligands would expose them to higher
ligand concentrations, which might contribute to their growth
and/or survival.
Our findings provide a rationale to test whether inhibition of CCR4 signaling might have therapeutic potential for
patients with ATLL. Either a CCR4 inhibitor or a PI3K
inhibitor might be considered (Pease and Horuk, 2014).
The anti-CCR4 monoclonal antibody KW-0761, which is
showing promising results in clinical trials (Yamamoto et al.,
2010; Ishida et al., 2012), was designed to promote antibodydependent cellular cytotoxicity of ATLL cells. This antibody
only inhibits chemotaxis weakly (Ishida et al., 2006), and its
effect on PI3K/AKT signaling has not been evaluated. To
improve the therapy of ATLL, our findings would support
the development of a therapeutic anti-CCR4 antibody
that both inhibits CCR4 signaling and mediates antibodydependent cellular cytotoxicity.
CCL17/TARC and recombinant human CCL22/MDC were purchased from

R&D Systems. Human recombinant IL2 was obtained from Roche. Anti
human CCR4 antibodies (clone 1G1) conjugated to PE and Alexa Fluor 647
fluors were purchased from BD. The following antibodies were purchased
from Cell Signaling Technology: Akt, P-Akt (Ser473; D9E), P-Akt (S473)
Alexa Fluor 647 (D9E), and Alexa Fluor 647conjugated isotype control
antibody (DA1E). Antirabbit IgG-HRP antibody was purchased from GE
Healthcare. CD8a(Lyt 2) microbeads were purchased from Miltenyi Biotec.
BKM120 was purchased from Selleck Chemicals. PTX was purchased from
Sigma-Aldrich. Big Dye Terminator v1.1 was purchased from Applied Biosystems. Human gamma globulin fraction II was purchased from ICN Biomedicals
Inc. Paraformaldehyde was purchased from Electron Microscopy Sciences.
Sanger DNA resequencing. Genomic DNA of CCR4 exon 2 region

was amplified by PCR using the primers CCR4-E2F, 5-CTTCCCCT
CATTAGCTGCTTCTGGTTG-3; and CCR4-E2R, 5-CCTGAC
ACTGGCTCAGGAATCTCTTAC-3. The purified PCR products were
sequenced using a Big Dye Terminator v1.1 cycle sequencing kit and
analyzed on an ABI 3730 Genetic Analyzer (Applied Biosystems). The
following primers were used for sequencing: CCR4-E2.1F, 5-CCTTC
CTGGCTTTCTGTTCAGCACTTG-3; and CCR4-E2.1R, 5-TGA
TTTCCAGGGAGCTGAGAACCTTCC-3.
Experimental design. All experiments presented have been repeated at

least two times, and consistent results were obtained. Data are depicted as
means SEM. Statistical comparisons were made using the Students t test.
P < 0.05 was considered statistically significant.
Patient samples and cell lines. Written informed consent was obtained in
accordance with the Declaration of Helsinki and was approved by the Investigational Review Board of the National Cancer Institute (NCI). Some samples
were obtained before cytotoxic chemotherapy, whereas others were taken after
treatment (Table S1). PBMCs were isolated from ATLL patients by FicollHypaque. In cases where the ATLL cells were <70% of the mononuclear cells,
the ATLL cells were purified by an initial negative selection magnetic column
method (Miltenyi Biotec) to enrich for CD4+ cells followed by a positive
selection on a CD25 column. The resultant population consisted of 7095%
CD4+CD25+ ATLL cells. The ATLL cell lines were provided by the following researchers: M. Maeda (Kyoto University, Kyoto, Japan; ED40515(+),
ED40515(), ED41214(+), ED41214C(), ATL43T(+), ATL43Tb(),
ATL55T(+), and ATL42T(+)),Y.Yamada (Nagasaki University, Nagasaki, Japan;
ST1, KOB, KK1, and LMY1), T. Hata (Nagasaki University; ST1), T. Naoe
(Nagoya University, Nagoya, Japan; ATN1), and N. Arima (Kagoshima University, Kagoshima, Japan; Su9T01). 32D (a variant of the mouse 32D myeloid cell line engineered to express the human IL-2 receptor subunit) was
also used. IL2-dependent cell lines (ED40515(+), ED41214(+), ATL43T(+),
ATL55(+), ATL42T(+), KOB, KK, LMY1, and 32D) were cultured with
RPMI/10% FCS/penicillin-streptomycin with 100 IU/ml human recombinant IL2. Other lines were cultured with RPMI/10% FCS/penicillinstreptomycin. ED40515(+) and KOB were engineered to express ecotropic
retroviral receptors and TET repressor for transduction of retroviral shRNA
vectors and expression vectors as previously described (Schmitz et al., 2012).
Reagents and antibodies. Doxycycline, puromycin, ethidium bromide, and
Lipofectamine 2000 were purchased from Invitrogen. Recombinant human
RNA-seq. RNA was extracted using the AllPrep kit (QIAGEN) and
sequencing libraries were prepared using the TruSeq RNA sample Prep kit v2
(Illumina) according to the manufacturers instructions. Paired-end 108-bp
read sequencing was performed on a HiSeq 2000 system (Illumina). The mapping of paired-end reads and the extraction of putative single nucleotide variants (SNVs) were performed as previously described (Schmitz et al., 2012). In
brief, paired-end reads were mapped to the RNA sequences in the RefSeq
database (NCBI build 37) using the Burrows-Wheeler Aligner (BWA) software
with default parameters. Reads that failed to map to RefSeq were mapped
to RNA sequences in the Ensembl database. The remaining unmapped reads
were mapped to the human genome assembly (NCBI build 37). Mutant SNV
calls were declared if more than two reads were mutated and the ratio of mutant reads versus total coverage was >20%. SNVs that corresponded to single
nucleotide polymorphisms in the dbSNP database (build #132), the 1000
Genomes database (May 2011 release), and the NHLBI GO Exome Sequencing Project (ESP) database (ESP5400 December 2011 release) were excluded.
RNA-seq data were submitted to the NCBI Short Read Archive (SRA;
accession no. SRP042199).
Sanger RNA resequencing. RNA was DNase-digested and reverse transcribed using the Omniscript RT kit (QIAGEN). The CCR4 coding region
was amplified by PCR using primers CCR4-E2F and CCR4-E2R. The
purified PCR products were cloned using the TOPO XL PCR Cloning kit
(Invitrogen). Sequencing was performed as described in the Sanger DNA
resequencing section.
CCR4 mutation search of public databases. cBioPortal and COSMIC
were used.
Retroviral vectors and retroviral transduction. CCR4-WT and CCR4Q330* cDNA was PCR amplified from ED40515(+) cDNA using the primers HindIII-CCR4-S, 5-GGAAGCTTTTGAAAGGCACCGGGTC-3;
and CCR4-TAG-BamHI-AS, 5-CGGGATCCCTACAGAGCATCATGGAGAT-3. The amplified fragments were cloned into HindIIBamHI sites
of doxycycline-inducible pRetroCMV/TO/puro and pRetroCMV/TO/PG
vectors. MSCV-CCR4-WT-ires-huKO or MSCV-CCR4-Q330*-ires-GFP
was made by inserting HindIII (blunted)XhoI fragments from pRetroCMV/
TO/CCR4-WT-puro or pRetroCMV/TO/CCR4-Q330*-puro vectors
into EcoRI(blunted)XhoI sites of MSCV-ires-huKO or MSCV-ires-GFP
(provided by S. Tsuzuki, Aichi Cancer Center Research Institute, Nagoya,
Japan). shRNA targeting the 3 UTR of CCR4 was cloned into doxycyclineinducible pRSMX-puro vector with these two annealed oligos (shRNA target
2503
sequences are underlined): shCCR4-4A11_F, 5-GATCCCGAATGAAGTTGTAGGTAATTTCAAGAGAATTACCTACAACTTCATTCTTTTT-3;

and shCCR4-4A11_R, 5-AGCTAAAAAGAATGAAGTTGTAGGTA
ATTCTCTTGAAATTACCTACAACTTCATTCGG-3. pBMN-shCCR4ires-Lyt2 was made by insertion of XbaIBsmI fragment from pRSMXshCCR4-puro vector into XbaIBsmI sites of pBMN-ires-Lyt2 (provided by
G. Nolan, Stanford University, Stanford, CA). Retroviruses for ED40515(+)
and KOB were produced by cotransfection of 293T cells with helper plasmids
expressing gag-pol genes and an ecotropic pseudotyping env gene and a retroviral vector using Lipofectamine 2000 as described previously (Schmitz et al.,
2012). For 32D, retroviruses were produced by transfection of PlatE with retroviral vectors. Target cells were infected with the retroviruses with 8 g/ml
polybrene with spin infection at 1,290 g for 1.5 h. pRetroCMV/TO-puro,
pRetroCMV/TO-PG, and pRSMX-infected cells was purified by using
2 g/ml puromycin. pBMN-ires-Lyt2infected cells were purified by MACS
mouse CD8a (Lyt2) microbeads (Miltenyi Biotech). Doxycycline was used at
40 ng/ml for induction of shRNA and cDNA.
Chemotaxis assay. HTS Transwell 96-well permeable supports with an
8.0-m pore polyester membrane (Corning) were used for chemotaxis assays
of ED40515(+) and KOB. 235 l of culture medium containing the indicated dose of chemokine was added to the lower chamber, followed by the
addition of the Transwell inserts and the addition of cells in growth medium
(75 l) at a concentration of 5 105 cells/ml into the upper chamber. ChemoTx
(5 M pore; Neuro Probe) was used for chemotaxis assays of 32D. 300 l
of chemokine-containing medium was added to the lower chamber, and
55 l of cells in growth media (2 106 cells/ml) were added to the upper
chamber. After incubation for 2 h at 37C in 5% CO2, the upper chamber
was removed and 50 l of 1 PBS with a 1:100 dilution of SPHERO particles (SpheroTECH) and 1:2,500 dilution of 10 mg/ml ethidium bromide
was added into each lower chamber. The numbers of cells and SPHERO
particles in lower chamber were quantified with a FACSCalibur (BD), and
the numbers of cells that had migrated in each condition were normalized
based on the SPHERO particle count.
Flow cytometry for CCR4 surface staining. Cells were washed with
PBS containing 200 g/ml of human gamma globulin (Human Gamma
Globulin Fraction II; ICN Biomedicals Inc.) twice and incubated for 30 min
on ice with antihuman CCR4-PE or CCR4Alexa Fluor 647 (1G1; BD)
or control isotype IgG. After washing with FACS buffer (PBS/1% FCS)
twice, the cells were analyzed on a FACSCalibur (BD). For the CCR4 internalization assay, cells were exposed to CCL22 for varying lengths of time
and were then treated for 1 min at 37C to an acid buffer (pH 3.0, consisting
of 50 mM glycine and 100 mM NaCl) to remove cell-bound CCL22. The
cells were washed twice and then stained with antihuman CCR4Alexa
Fluor 647 on ice for 30 min, fixed with 2% paraformaldehyde (Electron
Microscopy Sciences) on ice for 10 min, and analyzed on a FACSCalibur.
Immunoblot analysis. Cells were washed and resuspended in 2 SDS
sample buffer at 106 cells/100 l and boiled for 5 min. Samples were separated on Novex 412% Tris-Glycine gel (Invitrogen) and transferred to a
PVDF membrane (Immobilon-P; EMD Millipore). Proteins were detected
with the following primary antibodies: p-AKT(S473) (D9E) and AKT (5G3;
Cell Signaling Technology). An antirabbit IgG-HRP antibody (GE Healthcare) was used as the secondary antibody.
AKT phospho-flow analysis. After incubation with 50 ng/ml CCL22 for
the indicated time, cells were fixed in 2% paraformaldehyde for 10 min at
room temperature and then permeabilized with cold methanol at 20C
overnight. Cells were washed twice in FACS buffer and stained with p-Akt
(S473)Alexa Fluor 647 (D9E) or isotype control (DA1E; Cell Signaling
Technology) for 20 min at room temperature. Cells were washed and resuspended in FACS buffer for analysis on a FACSCalibur.
Online supplemental material. Table S1, included as a separate Excel
file, shows Sanger sequencing analysis of exon 2 of CCR4 (NM_005508.4)
2504
in 53 cases of ATLL samples. Online supplemental material is available at

http://www.jem.org/cgi/content/full/jem.20140987/DC1.
We thank the following researchers for providing ATLL cell lines: Michiyuki Maeda,
Yasuaki Yamada, Tomoko Hata, Tomoki Naoe, and Naomichi Arima. We also thank
Richard N. Bamford, Liyanage P. Perera, Ryan Young, Art Shaffer, and Michele
Ceribelli for helpful discussions.
This research was supported by the Intramural Research Program of the
National Institutes of Health, National Cancer Institute, Center for Cancer Research.
Roland Schmitz was supported by the Dr. Mildred Scheel Stiftung fr
Krebsforschung (Deutsche Krebshilfe).
Submitted: 23 May 2014
Accepted: 13 November 2014
REFERENCES
Arisawa, K., M. Soda, S. Endo, K. Kurokawa, S. Katamine, I. Shimokawa, T.

Koba, T. Takahashi, H. Saito, H. Doi, and S. Shirahama. 2000.
Evaluation of adult T-cell leukemia/lymphoma incidence and its
impact on non-Hodgkin lymphoma incidence in southwestern Japan.
Int. J. Cancer. 85:319324. http://dx.doi.org/10.1002/(SICI)10970215(20000201)85:3<319::AID-IJC4>3.0.CO;2-B
Balabanian, K., B. Lagane, J.L. Pablos, L. Laurent, T. Planchenault,
O. Verola, C. Lebbe, D. Kerob, A. Dupuy, O. Hermine, et al. 2005.
WHIM syndromes with different genetic anomalies are accounted for
by impaired CXCR4 desensitization to CXCL12. Blood. 105:2449
2457. http://dx.doi.org/10.1182/blood-2004-06-2289
Campbell, J.J., G. Haraldsen, J. Pan, J. Rottman, S. Qin, P. Ponath, D.P. Andrew,
R. Warnke, N. Ruffing, N. Kassam, et al. 1999. The chemokine receptor
CCR4 in vascular recognition by cutaneous but not intestinal memory
T cells. Nature. 400:776780. http://dx.doi.org/10.1038/23495
Campo, E., S.H. Swerdlow, N.L. Harris, S. Pileri, H. Stein, and E.S. Jaffe.
2011. The 2008 WHO classification of lymphoid neoplasms and beyond: evolving concepts and practical applications. Blood. 117:5019
Cao, Y., Z.R. Hunter, X. Liu, L. Xu, G. Yang, J. Chen, C.J. Patterson, N.
Tsakmaklis, S. Kanan, S. Rodig, et al. 2014. The WHIM-like CXCR4S338X
somatic mutation activates AKT and ERK, and promotes resistance
to ibrutinib and other agents used in the treatment of Waldenstroms
Macroglobulinemia. Leukemia. In press. http://dx.doi.org/10.1038/
leu.2014.187
Chen, J., M. Petrus, B.R. Bryant, V.P. Nguyen, C.K. Goldman, R. Bamford,
J.C. Morris, J.E. Janik, and T.A. Waldmann. 2010. Autocrine/paracrine
cytokine stimulation of leukemic cell proliferation in smoldering and
chronic adult T-cell leukemia. Blood. 116:59485956. http://dx.doi
.org/10.1182/blood-2010-04-277418
Chong, B.F., J.E. Murphy, T.S. Kupper, and R.C. Fuhlbrigge. 2004. E-selectin,
thymus- and activation-regulated chemokine/CCL17, and intercellular
adhesion molecule-1 are constitutively coexpressed in dermal microvessels: a foundation for a cutaneous immunosurveillance system. J. Immunol.
172:15751581. http://dx.doi.org/10.4049/jimmunol.172.3.1575
Cronshaw, D.G., C. Owen, Z. Brown, and S.G. Ward. 2004. Activation of
phosphoinositide 3-kinases by the CCR4 ligand macrophage-derived
chemokine is a dispensable signal for T lymphocyte chemotaxis. J. Immunol.
172:77617770. http://dx.doi.org/10.4049/jimmunol.172.12.7761
DAmbrosio, D., C. Albanesi, R. Lang, G. Girolomoni, F. Sinigaglia, and C.
Laudanna. 2002. Quantitative differences in chemokine receptor engagement generate diversity in integrin-dependent lymphocyte adhesion. J. Immunol. 169:23032312. http://dx.doi.org/10.4049/jimmunol
.169.5.2303
Diaz, G.A. 2005. CXCR4 mutations in WHIM syndrome: a misguided
immune system? Immunol. Rev. 203:235243. http://dx.doi.org/10.1111/
j.0105-2896.2005.00226.x
Elliott, N.E., S.M. Cleveland, V. Grann, J. Janik, T.A. Waldmann, and U.P. Dav.
2011. FERM domain mutations induce gain of function in JAK3 in
adult T-cell leukemia/lymphoma. Blood. 118:39113921. http://dx.doi
.org/10.1182/blood-2010-12-319467
Imai, T., M. Baba, M. Nishimura, M. Kakizaki, S. Takagi, and O.Yoshie. 1997.

The T cell-directed CC chemokine TARC is a highly specific biological
ligand for CC chemokine receptor 4. J. Biol. Chem. 272:1503615042.
http://dx.doi.org/10.1074/jbc.272.23.15036
Imai, T., D. Chantry, C.J. Raport, C.L. Wood, M. Nishimura, R. Godiska,
O. Yoshie, and P.W. Gray. 1998. Macrophage-derived chemokine is a
functional ligand for the CC chemokine receptor 4. J. Biol. Chem.
273:17641768. http://dx.doi.org/10.1074/jbc.273.3.1764
Iqbal, J., D.D. Weisenburger, T.C. Greiner, J.M. Vose, T. McKeithan,
C. Kucuk, H. Geng, K. Deffenbacher, L. Smith, K. Dybkaer, et al.
International Peripheral T-Cell Lymphoma Project. 2010. Molecular
signatures to improve diagnosis in peripheral T-cell lymphoma and prognostication in angioimmunoblastic T-cell lymphoma. Blood. 115:1026
Ishida, T., A. Utsunomiya, S. Iida, H. Inagaki, Y. Takatsuka, S. Kusumoto,
G. Takeuchi, S. Shimizu, M. Ito, H. Komatsu, et al. 2003. Clinical significance of CCR4 expression in adult T-cell leukemia/lymphoma: its
close association with skin involvement and unfavorable outcome. Clin.
Cancer Res. 9:36253634.
Ishida, T., T. Ishii, A. Inagaki, H.Yano, H. Komatsu, S. Iida, H. Inagaki, and R.
Ueda. 2006. Specific recruitment of CC chemokine receptor 4-positive
regulatory T cells in Hodgkin lymphoma fosters immune privilege.
Cancer Res. 66:57165722. http://dx.doi.org/10.1158/0008-5472.
CAN-06-0261
Ishida, T., T. Joh, N. Uike, K. Yamamoto, A. Utsunomiya, S. Yoshida,
Y. Saburi, T. Miyamoto, S. Takemoto, H. Suzushima, et al. 2012.
Defucosylated anti-CCR4 monoclonal antibody (KW-0761) for relapsed
adult T-cell leukemia-lymphoma: a multicenter phase II study. J. Clin.
Oncol. 30:837842. http://dx.doi.org/10.1200/JCO.2011.37.3472
Kawai, T., U. Choi, N.L. Whiting-Theobald, G.F. Linton, S. Brenner, J.M.
Sechler, P.M. Murphy, and H.L. Malech. 2005. Enhanced function
with decreased internalization of carboxy-terminus truncated CXCR4
responsible for WHIM syndrome. Exp. Hematol. 33:460468. http://
dx.doi.org/10.1016/j.exphem.2005.01.001
Komohara, Y., D. Niino, Y. Saito, K. Ohnishi, H. Horlad, K. Ohshima,
and M. Takeya. 2013. Clinical significance of CD163+ tumor-associated
macrophages in patients with adult T-cell leukemia/lymphoma. Cancer
Sci. 104:945951. http://dx.doi.org/10.1111/cas.12167
Luttrell, L.M., and R.J. Lefkowitz. 2002. The role of beta-arrestins in the
termination and transduction of G-protein-coupled receptor signals.
J. Cell Sci. 115:455465.
Matsuoka, M., and K.T. Jeang. 2007. Human T-cell leukaemia virus type 1
(HTLV-1) infectivity and cellular transformation. Nat. Rev. Cancer.
7:270280. http://dx.doi.org/10.1038/nrc2111
Pease, J.E., and R. Horuk. 2014. Recent progress in the development of

antagonists to the chemokine receptors CCR3 and CCR4. Expert
Opin Drug Discov. 9:467483. http://dx.doi.org/10.1517/17460441
.2014.897324
Roccaro, A.M., A. Sacco, C. Jimenez, P. Maiso, M. Moschetta, Y. Mishima,
Y. Aljawai, I. Sahin, M. Kuhne, P. Cardarelli, et al. 2014. C1013G/
CXCR4 acts as a driver mutation of tumor progression and modulator
of drug resistance in lymphoplasmacytic lymphoma. Blood. 123:4120
Schmitz, R., R.M. Young, M. Ceribelli, S. Jhavar, W. Xiao, M. Zhang, G.
Wright, A.L. Shaffer, D.J. Hodson, E. Buras, et al. 2012. Burkitt lymphoma pathogenesis and therapeutic targets from structural and functional genomics. Nature. 490:116120. http://dx.doi.org/10.1038/
nature11378
Treon, S.P., Y. Cao, L. Xu, G. Yang, X. Liu, and Z.R. Hunter. 2014. Somatic
mutations in MYD88 and CXCR4 are determinants of clinical presentation and overall survival in Waldenstrom macroglobulinemia. Blood.
123:27912796. http://dx.doi.org/10.1182/blood-2014-01-550905
Vissers, J.L., F.C. Hartgers, E. Lindhout, M.B. Teunissen, C.G. Figdor,
and G.J. Adema. 2001. Quantitative analysis of chemokine expression by dendritic cell subsets in vitro and in vivo. J. Leukoc. Biol. 69:
785793.
Vulcano, M., C. Albanesi, A. Stoppacciaro, R. Bagnati, G. DAmico, S. Struyf,
P.Transidico, R. Bonecchi, A. Del Prete, P. Allavena, et al. 2001. Dendritic
cells as a major source of macrophage-derived chemokine/CCL22 in vitro
and in vivo. Eur. J. Immunol. 31:812822. http://dx.doi.org/10.1002/15214141(200103)31:3<812::AID-IMMU812>3.0.CO;2-L
Yamagishi, M., and T. Watanabe. 2012. Molecular hallmarks of adult
T cell leukemia. Front Microbiol. 3:334. http://dx.doi.org/10.3389/fmicb
.2012.00334
Yamamoto, K., A. Utsunomiya, K. Tobinai, K. Tsukasaki, N. Uike, K. Uozumi,
K.Yamaguchi,Y.Yamada, S. Hanada, K. Tamura, et al. 2010. Phase I study
of KW-0761, a defucosylated humanized anti-CCR4 antibody, in relapsed patients with adult T-cell leukemia-lymphoma and peripheral
T-cell lymphoma. J. Clin. Oncol. 28:15911598. http://dx.doi.org/
10.1200/JCO.2009.25.3575
Yoshie, O. 2005. Expression of CCR4 in adult T-cell leukemia. Leuk. Lymphoma.
46:185190. http://dx.doi.org/10.1080/10428190400007607
Yoshie, O., R. Fujisawa, T. Nakayama, H. Harasawa, H. Tago, D. Izawa, K.
Hieshima,Y. Tatsumi, K. Matsushima, H. Hasegawa, et al. 2002. Frequent
expression of CCR4 in adult T-cell leukemia and human T-cell leukemia
virus type 1-transformed T cells. Blood. 99:15051511. http://dx.doi.org/
10.1182/blood.V99.5.1505
2505
Cancer research solutions

Advancing tumor immunology one cell at a time
MACS Sample Preparation
MACS Cell Separation
MACS Flow Cytometry
t 1SFTFSWFEDFMMJOUFHSJUZBOE
TVSGBDFFQJUPQFT
t 5IFMBSHFTUQPSUGPMJPGPSUVNPS
JNNVOFDFMMJTPMBUJPO
t #SPBESBOHFPGJOTUSVNFOUTBOE
SFBHFOUTGPSZPVSOFFET
t (FOUMFZFUFGGFDUJWFUVNPS
EJTTPDJBUJPO
t *TPMBUFEDFMMTBMXBZTDPNQBUJCMF
XJUIEPXOTUSFBNBQQMJDBUJPO
t 3FBEZUPVTFLJUTBOEJOUVJUJWF
TPGUXBSFGPSFBTFPGVTF
t 'MFYJCMFLJUTGPSBOZUVNPSFOUJUZ
t 5IFNPTUDJUFEDFMMTFQBSBUJPO
UFDIOPMPHZ
t 4FOTJUJWFBOBMZTJTPGMPX
GSFRVFODZDFMMT
Tumor Dissociation
Cell isolation or depletion
Tumor characterization
macscancer.com
6OMFTTPUIFSXJTFTQFDJDBMMZJOEJDBUFE .JMUFOZJ#JPUFDQSPEVDUTBOETFSWJDFTBSFGPSSFTFBSDIVTFPOMZBOEOPUGPSUIFSBQFVUJDPSEJBHOPTUJDVTF."$4JTBSFHJTUFSFE
USBEFNBSLPG.JMUFOZJ#JPUFD(NC)$PQZSJHIU.JMUFOZJ#JPUFD(NC)"MMSJHIUTSFTFSWFE

JEM Immunity-and-Cancer Special Issue PDF

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

JEM Immunity-and-Cancer Special Issue PDF

Hochgeladen von

Copyright:

Verfügbare Formate

THE

IMMUNITY AND CANCER

AVANTIS NEW PROBES

Avanti Number 900401

Avanti Number 900404

pacFA Glucosyl Ceramide

Avanti Number 900405

pacFA Galactosyl Ceramide

Avanti Number 900407

Avanti Number 900408

FIRST CHOICE FOR

PHOSPHOLIPIDS, SPHINGOLIPIDS, DETERGENTS

CGMP LIPIDS FOR PHARMACEUTICAL PRODUCTION

Selected Articles February 2015

Engagement of the ICOS pathway markedly enhances efficacy of CTLA-4 blockade in

phone (212) 327-8575

Interplay of host microbiota, genetic

The preferential localization of some neoplasms in

The Rockefeller University Press $30.00

In the intestine, the microbiota is in close

Microbiota and serrated polyps | Bongers et al.

has also been suggested to play a role in the pathogenesis

Interestingly, we found a significant enrichment of the

Figure 2. Alterations in the microbiota affect formation

As expected, none of the mice on continued antibiotic treat

treated a group of animals at 35 wk of age with antibiotics for

A specific cecal microbiota is necessary

although at a significantly reduced frequency (17 vs. 63%;

Microbiota and serrated polyps | Bongers et al.

SPs belonged to the order of Clostridiales. To test whether

Figure 5. Bacterial infiltration SPs

The inflammatory response in SPs

cells from SPs and surrounding unaffected cecal tissue and

bacterium (Fig. 7 B and Table S6). Analysis of cytokine and

Figure 6. Clostridial probe signal in antibiotic-treated HBUS mice.

of an HB-EGFprocessing metalloproteinase. These find

partial adoption of a Taconic cecal microbiome by the off

age we observed SPs in rederived HBUS mice, likely reflect

Detailed analysis of the SP-associated microbiome indi

functionally related to the changes in Claudin-2 expression, as

Microbiota and serrated polyps | Bongers et al.

randomization (coin flip). Histological analysis was performed blinded to the

eubacterial oligonucleotide probe ([AminoC6 + Alexa Fluor 594] 5-GCT

Arthur, J.C., E. Perez-Chanona, M. Mhlbauer, S. Tomkovich, J.M. Uronis,

Lei, Z., T. Maeda, A. Tamura, T. Nakamura, Y. Yamazaki, H. Shiratori, K.

Microbiota and serrated polyps | Bongers et al.

Engagement of the ICOS pathway markedly

of Immunology and 2Department of Genitourinary Medical Oncology, The University of Texas

Cytotoxic T lymphocyte antigen-4 (CTLA-4) blockade with a monoclonal antibody yields

Harnessing T cell responses to eradicate tumors

The Rockefeller University Press $30.00

in patients with melanoma, ovarian, prostate, renal

shed ICOS (detection range >0.1 ng/ml). Mice were given

B16/F10 cells or ICOSL-positive B16/F10 cells (IVAX).

poor activation status and low frequency of ICOS expression

of B16/F10 challenge (Fig. 3, AC) or a more established

Teff cells, a cellular vaccine expressing ICOSL can trigger the

Figure 4. Stimulation of the ICOS pathway also improved memory

Changed balance of antitumor Teff cells

Figure 5. Combination of IVAX and CTLA-4

Figure 6. Combination of IVAX and CTLA-4

efficacy observed with the combination therapy of anti

JEM Vol. 211, No. 4

JEM Vol. 211, No. 11