Sponsored by

RIKEN Research Center for Allergy and Immunology

Produced by the

Science/AAAS

Custom Publishing Office

PROGRESS INIMMUNOLOGYFrom Basic Discoveries to Medical Innovation

To learn more, visit aaas.org/plusyou/project2061

AAAS is here – promoting universal science literacy.

In 1985, AAAS founded Project 2061 with the goal of helping all Americans become literate in science, mathematics, andtechnology. With its landmark publications Science for All Americans and Benchmarks for Science Literacy, Project 2061 set outrecommendations for what all students should know and be able to do in science, mathematics, and technology by the time theygraduate from high school. Today, many of the state standards in the United States have drawn their content from Project 2061.

Every day Project 2061 staff use their expertise as teachers, researchers, and scientists to evaluate textbooks and assessments,create conceptual strand maps for educators, produce groundbreaking research and innovative books, CD-ROMs, and profes-sional development workshops for educators, all in the service of achieving our goal of universal science literacy.

As a AAAS member, your dues help support Project 2061 as it works to improve science education. If you are not yet a AAASmember, join us. Together we can make a difference.

Editors: Tianna Hicklin, Ph.D. and Sean Sanders, Ph.D.; Proofing: Bob French; Design: Tom Wight, MidAtlantic PublishingServices.

Cover image provided courtesy of the Laboratory for Mucosal Immunity, RCAI RIKEN. This image has been digitally altered fromoriginal version for publication purposes.

© 2012 by The American Association for the Advancement of Science. All rights reserved. September 26, 2012.

2

Human health is undeniablya topic of intense researchfocus today. It’s an issuethat affects each one of uspersonally—including ourfriends, family, and col-leagues—while also hav-ing a global impact. We areconstantly bombarded byenvironmental insults thatcan affect our health, fromthe relatively innocuous ir-ritants that cause allergiesto potentially deadly virus-es, parasites, and bacteria.Our most effective defenseagainst these microscopicinvaders is our immunesystem. However, this de-fense mechanism can also turn against itself,causing a range of autoimmune diseases. De-spite rapid progression in the field of immu-nology over the past few decades, scientistsare still searching for a deeper understandingof the intricacies of the immune system andthe relationships between host and patho-gen. New discoveries hold the potential foruncovering novel strategies to strengthen ourimmunological defense mechanisms and pre-vent their malfunctioning.

One institution that has focused on unravelingsome of the remaining questions in immunol-ogy is the RIKEN Research Center for Allergyand Immunology (RCAI). Since 2001, RCAI haspublished numerous research papers that havesignificantly advanced our knowledge in differentareas of immunology. In this booklet, we inviteyou to read a sampling of RCAI publications that

have shed light on topics such as immune celldevelopment and differentiation, the protectiverole of probiotics, immune system homeostasis,and the role of immune cells in cancer and auto-immune diseases.

Many challenges lie ahead for immunologists asthe field becomes increasingly multidisciplinaryand as advancing life-science technologies ena-ble larger and larger data sets to be collected andanalyzed. We are encouraged to see institutionssuch as RCAI look toward the future and bringtogether top-notch researchers from aroundthe world to interpret important new findings inimmunology, draw from one another’s expertise,and drive immunology forward to find new waysof improving human health.

Tianna Hicklin, Ph.D. and Sean Sanders, Ph.D.Editors, Science/AAAS Custom Publishing Office

DEEPENING OUR UNDERSTANDING IN IMMUNOLOGY

Credit:©istockphoto.com/nicolas_

2

3

The RIKEN Research Center for Allergy andImmunology (RCAI) has been one of Japan’s pre-mier strategic natural science research institutessince its establishment in 2001.

Two key principles have guided the philosophy anddirection of RCAI research in its first decade: 1)creating new paradigms in basic immunology and2) bridging basic immunology findings and medi-cal innovation. RCAI scientists explore many areasof immunology, including molecular imaging, im-mune system development, cell signaling, geneticand epigenetic regulation, mucosal immunity, tumorimmunity, and the development of new therapiesand diagnostic tools. Since the center’s inception,RCAI researchers have had a consistently strongtrack record, continuously publishing papers in dis-tinguished scientific journals; more than 30 percentof our papers published yearly are found in journalswith an impact factor greater than 10. Moreover, thenumber of citations per paper has been consistentlyabove 40. RCAI’s immunology research is highlyranked, with a high citation index score, and compares favorably with other premier immunol-ogy institutions around the world. This Science reprint collection booklet summarizes some ofRCAI’s major research findings over the last decade and introduces a selection of RCAI paperspublished in Science, Science Signaling, and Science Translational Medicine.

The immune system is critical for maintaining homeostasis in the body, is a key target for devel-oping therapeutics, and is important for preventing a variety of diseases. RCAI scientists havemade many discoveries in basic immunological research using in vitro and ex vivo techniquesand animal models; moreover, our researchers have also seen significant progress with transla-tional research studies, including the development of “humanized” mice, preclinical work on acedar pollen allergy vaccine, and clinical trials investigating a natural killer T (NKT) cell therapyfor cancer. In its second decade, RCAI has focused on developing ways to relate basic researchfindings to understanding the human immune system. In 2011, RCAI launched the MedicalImmunology World Initiative (MIWI), a new human immunology consortium of laboratories thatuses humanized mice. MIWI will serve as a worldwide network for the next generation of humanimmunology research by virtue of its focus on integrative medical immunology.

Building on the groundwork laid by RCAI, RIKEN has decided to establish a new research centerfor integrative medical science that will launch in April 2013. The research conducted duringRCAI’s first 11 years has opened the door to new directions in human immunology, and thecurrent endeavors of RCAI investigators continue to spearhead global immunology research,connect scientists around the world, and improve human well-being.

Masaru Taniguchi, M.D., Ph.D.DirectorRIKEN Research Center for Allergy and Immunology

THE RIKEN RESEARCH CENTER FOR ALLERGY AND

IMMUNOLOGY CELEBRATES ITS FIRST DECADECredit:PhotocourtesyofRIKEN

RCAI.

4

Immunology is a field of discoveries, inventions, and breakthroughs. From antibodies to vaccines, the products ofimmunology research have transformed basic science and clinical practice. Today, immunologists are addressingmajor medical and scientific challenges such as the global rise in inflammatory disease, autoimmunity, and allergy; thecomplex interaction between genes and environment; and the need for individualized and effective immunotherapies.Immunology researchers are using powerful, high throughput techniques and a variety of in vivo models to addressthese challenges. Immunology continues to be a dynamic field of research that provides tools and inspiration to clinicaland basic research.

THE INTERCONNECTION, INTERDEPENDENCE, AND

INFLUENCE OF THE IMMUNE SYSTEM

Immunology has a rich history of groundbreaking dis-coveries, from Edward Jenner’s successful smallpoxvaccinations in the 18th century to Susumu Tonegawa’s

Nobel Prize-winning explanation of antibody gene re-combination 200 years later. Immunology research in the21st century is discovering that the innate and adaptiveimmune systems are more intricately connected thanpreviously thought. In fact, the entire immune system isnow viewed as a complex communication hub, coordi-nating interactions between the external world of patho-gens, commensal microorganisms, and allergens, andour internal physiology from organs to genes. This modelof interdependency underlies current immunology re-search, whether the topic is early immune development,gene-environment interactions, receptors and signaling,or clinical therapies.A great deal of current immunology research is driven

by the global rise in rates of autoimmune disorders, al-lergy, and inflammation-related diseases. An explanationfor this global pattern, says Harald Renz, professor ofLaboratory Medicine, Department of Clinical Chemistryand Molecular Diagnostics, Philipps University, Marburg,Germany, is the hygiene hypothesis—that early exposureto a variety of antigens builds a stronger, better-function-ing adult immune system. A hyperhygienic environmentin early life inhibits the development of a proper balancebetween the actions of T effector and T regulatory cells.“We have an immune system to fight infections, and ifyou don’t test the immune system with exposure topathogens, you lose homeostasis, resulting in immunedysregulation,” says Joan M. Goverman, chair of theDepartment of Immunology, University of Washington,Seattle.In spring 2012, experimental findings using an animal

model provided support for the hygiene hypothesis.Richard Blumberg, chief of the Division of Gastroen-terology, Hepatology and Endoscopy of Brigham and

Women’s Hospital in Boston and colleagues found thatgerm-free mice are more susceptible to inflammatoryconditions resembling colitis and asthma and traced thissusceptibility to increased invariant natural killer T cells(NKT). Early exposure to the normal commensal microor-ganisms of mice resulted in normal NKT cell levels, lead-ing the researchers to conclude that “age-sensitive con-tact with commensal microbes” established appropriatemicrobial tolerance. Normalization of the immune systemoccurred only when microbial exposure was early in life,and not if it occurred in adulthood.

Investigating Earlier and BroaderInfluencesIn fact, research on the early immune system is find-

ing that effects that are prenatal or even earlier can influ-ence immune system development. For example, mousemodels show that maternal and even grandmaternalexposures to nutrients, pollutants such as cigarettesmoke, and infectious agents such as bacteria can in-fluence immune development. These transgenerationalinfluences are explained by epigenetic effects, whichare heritable changes in the accessibility of DNA to tran-scription factors through modifications to nucleotides orhistone proteins. For instance in mice, certain concen-trations of the methyl donor folate in the maternal dietresulted in epigenetic effects to genes linked to allergicairway disease. This effect was transmitted to a secondgeneration. Renz’s group found that prenatal exposure tothe soil bacterium Acinetobacter lwoffi protected againstasthma in a mouse model, and was associated with epi-genetic effects on cytokine genes such as interferon-gamma and interleukin-4. The next steps in immunol-ogy research are identifying the specific agents, and thetiming and type of exposure that give protective effectsagainst later immune dysregulation. Researchers are alsolooking for the targets of early environmental exposure,

5

such as stem and progenitor immune system cells. “Thisis a field to watch,” says Rafeul Alam, head of allergy andimmunology at National Jewish Health and professor ofmedicine at University of Colorado Denver. Says Alam,“Work on epigenetics so far has just scratched the sur-face. This is fertile ground for research.”Immunology researchers are also broadening their

definition of the environment. They are finding that fac-tors affecting immune development include “food anddietary habits, obesity and metabolic syndrome, psycho-logical factors such as stress, and neuroendocrine inter-actions,” says Renz. “A big area is microbial exposure,including microbes that normally colonize the mucosaof the respiratory system, gut, and skin. These microbeshave a strong impact early in life beyond the risk of al-lergy to risk of conditions such as type I diabetes, multi-ple sclerosis, neurodegenerative disease, schizophrenia,and depression.”Research on commensal gut microbes has found that

colonization of the gastrointestinal tract, mainly by spe-cies in the phyla Firmicutes and Bacteroidetes, beginsimmediately after birth. In mice, intestinal colonizationpromotes development of immune responses includingproduction of antimicrobial peptides, differentiation ofhelper T cells, development of T regulatory cells, and se-cretion of IgA. This is a symbiotic relationship that regu-lates growth of commensal bacteria that aid in digestionand produce vitamins and other nutrients for the host.Crosstalk clearly occurs between the microbiota and theimmune system, and current research is investigatingboth sides of this communication. Active research areasinclude studying how immune development and main-tenance are affected by specific species and strains ofmicrobes, particular compositions of the microbiome,and the amount of microbiome diversity. In humans, themain method for this type of research is sampling andcharacterizing the human microbiome, for example byhigh throughput pyrosequencing in the U.S. NationalInstitutes of Health Human Microbiome Project. Analysisis underway to discover associations between the pres-ence of specific microbes, longitudinal changes in thehuman microbiome, and conditions ranging from obesityto autoimmune disease. This will determine which mi-crobes are our friends and which are our foes—the basictask of the immune system.

Self and NonselfDistinguishing between pathogens and commensals

and self and nonself is a fundamental function of the im-mune system. When this functionality goes awry, the re-sult can be autoimmunity. Even a correctly functioningimmune system can cause transplant rejection or graftvs. host disease. Substantial research with the goal ofmodulating the immune system has elucidated many ofthe mechanisms by which antibodies, T cell receptors,major histocompatibility complexes, and other immunesystem elements determine which cells to attack andwhich to tolerate. Recent attention has focused on pat-tern recognition receptors, particularly the toll-like recep-tors (TLRs). These transmembrane and intracellular pro-teins, with the help of adaptor proteins, bind to foreignmolecules such as bacterial lipopolysaccharides or viralRNAs and signal pathogen invasion. Discovery of the im-mune functions of TLRs was awarded the 2011 NobelPrize for Physiology or Medicine.Activation of TLRs by ligand binding initiates kinase-

dependent signal transduction cascades that lead togene expression responses mediated by transcriptionfactors such as NFKB. Professor Alam says that a cur-rent area of research is investigating just how the TLRpathways lead to the activation of inflammation, a topicthat is particularly urgent given the global health burdenof chronic inflammatory conditions such as diabetes andinflammatory bowel disease. Alam says we need addi-tional research on how pathogen signals reach and trig-ger intracellular TLRs such as TLR7 and TLR9. Althoughwe have a lot to learn from TLRs, he says, future researchshould also focus on other receptors connected toinflammation such as tyrosine kinase-associated recep-tors and other RNA- and DNA-sensing receptors. Nucleicacid-sensing receptors are particularly important in un-derstanding autoimmune disease since many of theseare an inappropriate reaction to RNA or DNA.Much research on the response to receptor signaling

has focused on changes in gene transcription, but Alamadvises more attention to downstream gene expressioneffects such as alternative splicing, microRNA regulation,and protein modification. “It’s simple logic,” he says.“Higher mammals have fewer genes in their genomesthan other organisms, so posttranscriptional and post-translational modifications have a bigger impact in creat-ing different phenotypes.”

6

Directing the Immune System for TherapyAs our understanding of the internal and external

communications and responses of the immune systemgrows, the list of potential therapeutic targets and tech-nologies expands. Development of monoclonal anti-bodies, for which César Milstein, Niels Kaj Jerne, andGeorges J. F. Köhler shared the 1984 Nobel Prize inPhysiology or Medicine revolutionized basic researchby providing tools for precisely identifying, isolating, ormarking cellular components. Monoclonal antibody tech-nology has led to highly specific therapies such as thebreast cancer treatment trastuzumab (Herceptin). Thesuccess of Herceptin and similar therapies has launcheddozens of biotechnology companies and clinical trials.Goverman, who studies autoimmune diseases such asmultiple sclerosis, says, “Cancer therapy is driving thedevelopment of therapy for autoimmune diseases andother immune disorders.” For example, a monoclonalantibody therapy used to treat malignant B-cell lympho-mas has shown some promise with multiple sclerosisand diabetes. Learning about its mechanism of actionhas also informed multiple sclerosis research.On the cell-based side, in 2010, the Seattle company

Dendreon received U.S. Food and Drug Administrationapproval for their prostate cancer treatment, sipuleucel-T (Provenge). The therapy stimulates the patient’s ownperipheral blood mononuclear cells to launch a specificattack against cancerous cells. Commercialization ofProvenge remains rocky, but the clinical proof-of-con-cept has motivated additional cell-based therapy devel-opment. Poul Sørensen, senior vice president of strategicdevelopment, Stallergenes, Paris, and board member ofthe Danish Graduate School of Immunology, says thesetherapies are promising, but the double-edged swordin any type of cell-based immunotherapy is the inher-ent plasticity of immune cells. We are learning how eas-ily immune cells change function, he says, for exampleconverting from anti-inflammatory to proinflammatory.This cellular flexibility is a challenge in large-scale cellculturing, the method which Dendreon must use in theirProvenge therapy. However, immune cell plasticity alsoexplains immunotherapy successes, for example treat-ment for specific allergies. Says Sørensen, “We’re learn-ing we can reprogram the immune system with lastingeffects. This has the potential to lead to treatment break-throughs for other diseases.”

Cancer immunotherapy is “coming of age,” saysJoseph Murphy, director of Cancer Therapeutics andImmunology, Southern Research Institute, Birmingham,Alabama, citing more than 40 different immunotherapiescurrently in testing in over 60 clinical trials. Murphy, whowrote an extensive 2010 review of cancer immunothera-py, says that current cancer immunotherapy is based onresearch showing that cancerous cells create an immunemicroenvironment that suppresses antitumor activity. Hesays that new cancer therapies will be “based on a so-phisticated knowledge of immune-suppressive cells, sol-uble factors, and signaling pathways designed to breaktolerance and reactivate antitumor immunity to inducepotent, long-lasting responses.” Examples are highlyspecific, cell-free vaccines and other treatments basedon tumor-associated antigenic peptides identified from apatient’s resected tumor. An integrated, multipronged im-mune response is required to eliminate cancer, so com-binatorial approaches are likely to be the most effective.Future therapies, Murphy says, “will simultaneously ad-dress the immunosuppressive, angiogenic, invasive, andhypoxic nature of cancer.”Other therapies are also developing in multiple direc-

tions. Rachna Shah, an allergist with Gottlieb MemorialHospital, Loyola University Health System in Maywood,Illinois notes that studies on complementary and alter-native therapies such as traditional Chinese medicine totreat allergies are becoming more common. “Both pa-tients and providers are interested in seeing if these treat-ments can make a difference in symptoms,” she says. Inconventional therapy, says Shah, oral immunotherapy isa major trend, with recent trials showing promise in sub-lingual therapy for egg, peanut, and pollen allergy.Research on the pathogenesis of infectious diseas-

es such as influenza and SARS could contribute to thedevelopment of immunomodulating therapies. Influenzaand other viral infections can induce a cytokine stormin which strong production of complement factors andproinflammatory cytokines brings excessive numbers ofinflammatory cells to the lungs, ultimately impairing lungfunction. Learning to modulate immune system commu-nications in this situation will have clinical implicationsbeyond infectious disease.

7

Credit:©istockphoto.com/jeangill

8

HIGHLIGHTS OF RCAI RESEARCH AND PROGRAMS: 2001–2012

Creating New Paradigms1) Immune System DevelopmentAs the immune system develops, the progeny of

multipotent hematopoietic stem cells (HSCs) become in-creasingly restricted in their lineage potential and acquirespecialized functions. Scientists have long believed thatlymphoid lineage precursors lack the ability to matureinto myeloid lineage cells; however, Hiroshi Kawamoto’sresearch showing that early thymic progenitors can giverise to myeloid cells has called this dogma into question(1). His group later discovered that the transcription fac-tor Bcl11b determines the T cell lineage (see full paperon page 15) (2).Ichiro Taniuchi’s group provided further insight into

what drives T cell differentiation. They found that theRunx complex represses the expression of the transcrip-tion factor Th-POK during thymocyte differentiation, al-lowing the development of CD8+ cytotoxic T cells ratherthan CD4+ helper cells (see full paper on page 19) (3, 4).

2) Immune Cell SignalingImmune cells use receptor-mediated signaling at

multiple stages in their development/differentiation todirect lineage choices and acquisition of effector func-tions. Changes in expression levels or posttranslationalmodification—such as phosphorylation—of signalingmolecules propagate signals within immune cells.Tomohiro Kurosaki’s group has dissected the molecularevents in B cell signaling. They identified Stim1/Stim2as critical for calcium (Ca2+) influx, a key event in manysignaling pathways (5, 6), and the Ser/Thr kinase extra-cellular signal-regulated kinase (ERK) as essential forearly B cell development and plasma cell differentiation(see abstract on page 23) (7, 8).A surprising discovery was made by Toshio Hirano

and his RCAI team, when they found that not only Ca2+,but also zinc (Zn2+) ions can act as signaling molecules.

They determined that Zn2+ signaling plays a role in cellmigration, dendritic cell activation, allergic reactions, andconnective tissue formation (9–12). Further, they iden-tified two types of Zn2+ signaling: early, which involvesa “Zn2+ wave,” similar to that observed during Ca2+ in-flux, and late, which is induced by the expression of Zn2+

transporters (13).Visualization of signaling events at the single-molecule

level is now possible using highly sensitive fluorescencemicroscopes developed by RCAI’s Makio Tokunaga (14).Takashi Saito observed the dynamic movement of sig-naling molecules during T cell activation and discoveredthat activation is initiated by small complexes called “Tcell receptor (TCR) microclusters,” in which TCRs andvarious signaling proteins are assembled (15, 16). Thegroup could also visualize T cell activation modulated bycostimulatory molecules. These findings have providedkey insights into how the spatiotemporal patterns ofsignaling events affect the initiation, maintenance, andregulation of T cell activation (17, 18).

3) Mucosal ImmunityThe adaptive immune system has two separate but in-

terrelated components: systemic and mucosal. Much ofclassical immunology focuses on the systemic compart-ment; however, protection at mucosal surfaces is alsoessential for survival. The “lymph-node equivalent” in theintestine are Peyer’s patches (PPs), which are separatedfrom the intestinal contents by a single layer of special-ized epithelial cells, among which are antigen-samplingM cells that deliver antigen to PPs to initiate mucosal im-mune responses, but the mechanisms by which M cellsdifferentiate and carry out this function have not beenfully elucidated.Hiroshi Ohno and his colleagues discovered that a

transcription factor, Spi-B, is essential for M-cell differ-

Since its establishment in 2001, the Research Center for Allergy and Immunology (RCAI) has been at the cutting edgeof immunology research and has focused on two goals: creating new paradigms in basic immunology and developingmedical innovations. During the 11-year directorship of Masaru Taniguchi, RCAI researchers have made a big impacton understanding how the immune system works. Although basic research has always been a focus and strength ofRCAI, the center has also developed programs with a translational component. Here, a sampling of the pioneeringstudies and programs that have contributed to RCAI’s mission are briefly described.

9

entiation (19) and that glycoprotein-2 (GP-2) serves asa bacterial uptake receptor (20). The intestinal immunesystem has intimate and important interactions with thebacterial flora in the gut. Using integrated multi-“omics”approaches, Ohno’s team studied the “probiotic” effectof these interactions. They discovered that in a mousemodel, bacteria from the genus Bifidobacterium pro-duced acetate that prevented Shiga toxin generated byenterohemorrhagic E. coli O157 from entering the blood-stream, thereby protecting the mice (21).Moreover, a group led by Sidonia Fagarasan investi-

gated the role of immunoglobulin A (IgA) in maintainingthe symbiotic balance between gut microbiota and thehost immune system (22). In collaboration with ShoheiHori, they demonstrated that environmental cues in thegut promote the selective differentiation of IgA-support-ing T follicular B helper (TFH cells) from Foxp3+ T cells(see abstract on page 25) (23). Fagarasan’s group alsorecently found that skewing of the microbiota, which re-sults from IgA with reduced bacteria-binding capacity,drives the generalized activation of B and T cells in micethat develop B cell-dependent autoimmune disease (seefull paper on page 11) (24).

4) Systems ImmunologyRCAI is actively creating new immunology fields that

are integrated with other scientific fields as well as newtechnologies. One example is systems immunology,which fuses biology and mathematics.RCAI’s Mariko Okada, in collaboration with Boris

Kholodenko (University College Dublin), has used math-ematical modeling to tackle the question of how cellfate is determined. Epidermal Growth Factor (EGF) andHeregulin (HRG) both share extracellular signal-regulat-ed kinase (ERK) signaling pathways, but induce distinctcell fates; EGF induces cell division, while HRG induc-es differentiation. The group found that EGF and HRGinduce either transient or sustained cytoplasmic ERKactivity, respectively, and that discrimination of thesesignals controls cell fate (25).New imaging technologies have the potential to

become important tools in systems biology since theinformation obtained can be digitized and quantified. Astep in this direction was taken by Takaharu Okada, whoutilized two-photon microscopy to track B cells by moni-

toring the expression of Bcl6 with a surrogate fluorescentmarker. His team could observe, track, and quantify thein vivo migration of Bcl6-expressing antigen-specific Bcells to the germinal center, a process dependent uponinteraction with Bcl6-expressing TFH cells (26).

—Takashi Saito, Deputy Director, RIKEN RCAI

Developing Innovations in BiomedicalSciences1) Humanized MiceRCAI scientists are working to develop innovative ex-

perimental models for investigating the human immunesystem and disease. RCAI researchers FumihikoIshikawa, Haruhiko Koseki, and Osamu Ohara, in col-laboration with Leonard Shultz (Jackson Laboratory),have generated immunodeficient mice (NOD/SCID/IL-2rgKO (NSG)) that express a human class I antigen.When newborn mice were transplanted with humanhematopoietic stem cells (HSCs), human cytotoxic Tcells successfully developed and became functional inthese “humanized” mice (27). More recently, this groupcreated an improved recipient mouse (hSCF Tg NSG)that expresses human stem cell factor (SCF), a cytokinethat is critical for the maintenance of stem and progeni-tor cell activities (28). Ishikawa also identified a humanleukemia stem cell (LSC) by transplanting human acutemyeloid leukemia (AML) cells into NSG mice. The LSCspropagate the leukemia, but are typically in a dormantstate and therefore resistant to cell cycle-dependentchemotherapy (29–31). His group is currently identify-ing genes differentially expressed between normal HSCsand LSCs to identify potentially new therapeutic targetsfor AML (see abstract on page 24).

2) Natural Killer T Cell TherapyA clinical trial of a natural killer T (NKT) cell targeted

therapy for patients with advanced non-small-cell lungcancer is being conducted by RCAI’s Masaru Taniguchi,in collaboration with Chiba University. In this trial, autolo-gous dendritic cells pulsed with the NKT cell-activatingligand α-galactosylceramide were administered intrave-nously. Sixty percent of treated patients (10/17) showeda significantly longer median survival compared to the

10

untreated group (29.3 versus 4.6 months). Based onthese promising results, expanded trials of the NKT celltherapy are planned, and this treatment is now offeredas a new therapeutic option under the advanced medi-cal care assessment system approved by the JapaneseMinistry of Health, Labor and Welfare. Thus, this treatmentwill be covered in part by national health insurance (32).

3) Primary Immunodeficiency Clinical ArchiveTo promote awareness and improve treatment op-

tions for Primary Immunodeficiency (PID), RCAI hasbeen expanding its collaborative network with the clini-cal study group from 13 Japanese universities, the In-stitute of Bioinformatics in India, and the Kazusa DNAResearch Institute, supported in part by the JeffreyModellFoundation. The number of patient samples deposited inthe Japanese clinical archive, “PIDJ,” which was estab-lished by RCAI, continues to increase each year. RCAI isalso working toward establishing an Asian PID networkand has established an integrated bioinformatics plat-form for PID called RAPID (rapid.rcai.riken.jp) and a PIDmutation annotation server, Mutation@A Glance (rapid.rcai.riken.jp/mutation).

—Toshitada Takemori, Research Coordinator, RIKENRCAI

Toward the Future1) Medical Immunology World Initiative (MIWI)To understand the process of disease develop-

ment and to identify critical events that change humanhealth, RCAI established an interdisciplinary biomedi-cal research platform and international consortium ofresearch groups that share a common interest in hu-man immunology. For this purpose, RCAI launched thezMedical ImmunologyWorld Initiative (MIWI) as a new hu-man immunology platform that includes humanized mice.Nine institutions are currently affiliated with MIWI: theImmunology Frontier Research Center of OsakaUniversity (IFReC), the U.S. National Institutes of Health(NIH), the Institute of Medical Science, the Universityof Tokyo (IMSUT), two departments from ZurichUniversity, INSERM/Necker Hospital, the PasteurInstitute, Imperial College London, and RCAI. The goal of

MIWI is to use integrative immunological approaches toobtain fundamental knowledge about the human immunesystem and the underlying mechanisms of disease de-velopment and to discover new principles for diagnosisand treatment.

2) Young Chief Investigator Program (YCI)RCAI launched the new Young Chief Investigator (YCI)

Program to provide a career path for young investiga-tors (under 40 years old) conducting multidisciplinaryresearch that bridges immunology with other fields. AYCI will head an independent laboratory but will haveaccess tomentorswhoarespecialists in relatedfields.Fiveresearchers have been selected thus far to conductpioneering work in the following new areas: stemcells and aging reversal, development of the interfacebetween integrative biology and mathematics, mucosalflora and epigenetic analysis, immune regeneration, andmulti system-wide analysis.

—Shigeo Koyasu, Deputy Director, RIKEN RCAI

References1. H. Wada et al., Nature 452, 768 (2008).2. T. Ikawa et al., Science 329, 93 (2010).3. R. Setoguchi et al., Science 319, 822 (2008).4. S. Muroi et al., Nature Immunol. 9, 1113 (2008).5. Y. Baba et al., Nature Immunol. 9, 81 (2008).6. M. Matsumoto et al., Immunity 34, 703 (2011).7. T. Yasuda et al., Immunity 28, 499 (2008).8. T. Yasuda et al., Sci. Signal. 4, ra25 (2011).9. S. Yamashita et al., Nature 429, 298 (2004).10. H. Kitamura et al., Nature Immunol. 7, 971 (2006).11. K. Nishida et al., J. Exp. Med. 206, 1351 (2009).12. T. Fukada et al., PLoS ONE, 3, e3642 (2008).13. T. Hirano et al., Adv. Immunol. 97, 149 (2008).14. M. Tokunaga, N. Imamoto, K. Sakata-Sogawa, Nature Methods 5, 159 (2008).15. T. Yokosuka et al., Nature Immunol. 6, 1253 (2005).16. A. Hashimoto-Tane et al., Immunity 34, 919 (2011).17. T. Yokosuka et al. Immunity 29, 589 (2008).18. T. Yokosuka et al., J. Exp. Med. 209, 1201 (2012).19. T. Kanaya et al., Nature Immunol. 13, 729 (2012).20. K. Hase et al., Nature 462, 226 (2009).21. S. Fukuda et al., Nature 469, 543 (2011).22. S. Fagarasan et al., Science 298, 1424 (2002).23. M. Tsuji et al., Science 323, 1488 (2009).24. S. Kawamoto et al., Science 336, 485 (2012).25. T. Nakakuki et al., Cell 141, 884 (2010).26. M. Kitano et al., Immunity 34, 961 (2011).27. L. D. Shultz et al., Proc. Natl. Acad. Sci. U.S.A. 107, 13022 (2010).28. S. Takagi et al., Blood 119, 2768 (2012).29. F. Ishikawa et al., Nature Biotechnol. 25, 1315 (2007).30. Y. Saito et al., Science Translational Medicine 2, 17ra9 (2010).31. Y. Saito et al., Nature Biotechnol. 28, 275 (2010).32. S. Motohashi et al., J. Immunol. 182, 2492 (2009).

HIGHLIGHTS OF RCAI RESEARCH AND PROGRAMS: 2001–2012

11

turnover of IgAs at their inductive or residentialplaces. Most of the IgA+ B cells are generatedin the GCs of Peyer’s patches (PPs) and arriveat the LP as plasmablasts [B220–IgA+MHCII+

(major histocompatibility class II–positive), here-after, PBs] (13). In the LP, PBs down-regulate theB cell receptor and MHCII and differentiate intoplasma cells (B220– IgA+/lowMHCIIlowCD138+,hereafter, PCs) (fig. S1F). The proliferation of

(fig. S1A). Thus, PD-1 deficiency perturbs thebalance of bacterial communities in the gut.

The frequencies and the absolute numbers ofIgA-producing cells in the lamina propia (LP)were comparable in Pdcd1–/– and WT mice (fig.S1, B, C, and F). Flow cytometric analyses offecal bacteria revealed, however, that the propor-tion of bacteria coated with IgA (bound IgAs)was reduced in Pdcd1–/– mice (10) (Fig. 1B). Incontrast, the concentration of free IgA in in-testinal secretions was higher in Pdcd1–/– than inWT mice (Fig. 1C). To assess the features of LPIgA, we sequenced the immunoglobulin heavychain (IgH) genes in single sorted IgA-producingcells. Both WT and Pdcd1–/– mice had a diverseIgA repertoire, yet Pdcd1–/– mice had an en-richment of IgA-producing cells with the IgHlocus belonging to non-VH1 family genes (Fig.1D). This difference in the IgA repertoire may bethe result of the altered composition of the mi-croflora in Pdcd1–/– mice. Indeed, as signs ofantigen-mediated selection in their IgH genes,about 90% of the sequences from WT andPdcd1–/– mice had mutations and high ratiosof replacement (R) to silent (S) mutations incomplementarity-determining region 1 (CDR1)and CDR2, compared with those in frameworkregions 1 to 3 (FWR1-3) (fig. S1, D and E).However, the calculated affinity maturation index(11, 12) was lower in IgA-producing cells fromLP of Pdcd1–/– mice (Fig. 1E). Together, the re-sults indicate that alterations of IgA compartmentin Pdcd1–/– mice have an impact on symbioticrelations between host and commensal bacteria inthe gut.

The altered repertoire of the IgA plasmacells in gut could result from changes in the

The Inhibitory Receptor PD-1Regulates IgA Selection and BacterialComposition in the GutShimpei Kawamoto,1* Thinh H. Tran,1,2* Mikako Maruya,1* Keiichiro Suzuki,1,3 Yasuko Doi,1

Yumi Tsutsui,1 Lucia M. Kato,1,4 Sidonia Fagarasan1†

Immunoglobulin A (IgA) is essential to maintain the symbiotic balance between gut bacterialcommunities and the host immune system. Here we provide evidence that the inhibitoryco-receptor programmed cell death–1 (PD-1) regulates the gut microbiota through appropriateselection of IgA plasma cell repertoires. PD-1 deficiency generates an excess number of T follicularhelper (TFH) cells with altered phenotypes, which results in dysregulated selection of IgAprecursor cells in the germinal center of Peyer’s patches. Consequently, the IgAs produced inPD-1–deficient mice have reduced bacteria-binding capacity, which causes alterations of microbialcommunities in the gut. Thus, PD-1 plays a critical role in regulation of antibody diversificationrequired for the maintenance of intact mucosal barrier.

1Laboratory for Mucosal Immunity, Research Center for Al-lergy and Immunology, RIKEN Yokohama 1-7-22, Tsurumi,Yokohama 230-0045, Japan. 2Department of Biochemistry,Hanoi Medical University, 1st Ton That Tung, Hanoi, Vietnam.3AK project, Graduate School of Medicine, Kyoto University,Yoshida Sakyo-ku, Kyoto 606-8501, Japan. 4Department ofImmunology and Genomic Medicine, Graduate School ofMedicine, Kyoto University, Yoshida Sakyo-ku, Kyoto 606-8501, Japan.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:sidonia-f@rcai.riken.jp

Theprimary function of immunoglobulin A(IgA) is to maintain homeostasis at mu-cosal surfaces. Intestinal IgA production

occurs via both T helper cell–dependent and in-dependent pathways (1). The diversification ofIgA repertoire by somatic hypermutation (SHM),however, takes place mostly in specialized mi-croenvironments called germinal centers (GCs),in which B cell interaction with T follicular helper(TFH) cells induces the expression of activation-induced cytidine deaminase (AID) (2, 3). TFHcells express high amounts of the inhibitory co-receptor programmed cell death–1 (PD-1) (4).PD-1 deficiency leads to species-specific, anti-body-mediated autoimmune diseases (5–7). Notethat the incidence of diseases in PD-1–deficientmice varies among mouse colonies and dependson AID (8).

We investigated whether PD-1 regulates mi-crobial communities and IgA production in thegut. Although the total number of bacteriacultured from the lumen of the small intestinewas comparable between PD-1–deficient mice(Pdcd1–/–) and wild-type (WT) mice (4.8 × 108

and 4.7 × 108 bacteria per g of intestinal content,respectively), Pdcd1–/– mice had a 93 to 95%reduction in the number of anaerobic bacteriacompared withWTmice (average 2.86 × 108 and0.18 × 108 in WTand Pdcd1–/– mice, respective-ly) (Fig. 1A). The total numbers of “healthy”bacteria, such as Bifidobacterium and Bacte-

roides (9) were not detectable or markedlyreduced in Pdcd1–/– mice. In contrast, bacteriaof the Enterobacteriaceae family, which wereminor representatives in the WT mice, wereincreased about 400-fold inPdcd1–/–mice. Theseresults were confirmed by 16S ribosomal RNA(rRNA) gene pyrosequencing of cecal contents

IgA-producing cells in LP was assessed using fluo-rescent ubiquitination-based cell-cycle indicator(Fucci) mice, in which the cells in S-G2-M phaseof the cell cycle are fluorescently marked (14).Similar to humans, most of the proliferating cellswere PBs (13, 15), and the frequency of PBs inthe cell cycle was comparable for Pdcd1–/– miceand WT mice (fig. S1F).

We next determined the turnover of IgA-producing cells in LP based on 5-bromo-2′-deoxyuridine (BrdU) incorporation. Mice werecontinuously fed BrdU for 10 days, and thenBrdU was removed from their drinking water.Ten days after the initiation of BrdU administra-tion, there were significantly more BrdU+ PBs inthe LP of Pdcd1–/– mice than in those of WTmice (Fig. 1F). Five days after the removal ofBrdU, however, the frequency of BrdU+ PBswasreduced by half in Pdcd1–/– mice, whereas thefrequency of BrdU+ PBs in WT mice remainedunchanged (Fig. 1F). In contrast, the frequency ofBrdU-labeled PCs did not differ significantlybetween Pdcd1–/– and WT mice (Fig. 1F). Thehigher turnover of Pdcd1–/– PBs may be theresult of increased apoptosis. Cell death was as-sessed by detection of caspase activation. Com-pared with WT mice, we observed a significantincrease in frequency of caspasehi PBs and PCsin the LP of Pdcd1–/– mice, but this diminishedafter antibiotic treatment (Fig. 1G and fig. S1G).Together, these results indicate enhanced com-mensal-driven turnover of IgA-producing cells inLP of Pdcd1–/– mice.

Next, we examined the PPs of Pdcd1–/–mice.Compared with WT mice, Pdcd1–/– mice hadsignificantly more peanut agglutinin–positive(PNA+) Fas+ GC B cells and IgA+ B cells inPPs (Fig. 2, A to C). Although we observed amodest increase in the frequency of light-zone Bcells in the PP GCs of Pdcd1–/– mice (16) (fig.S2A), quantitative real-time fluorescence poly-merase chain reaction analyses of several impor-tant B cell markers for dark-zone and light-zoneGC B cells did not show significant differencesbetween Pdcd1–/– mice and WT mice (fig. S2B).A comparable proportion of IgA+ GC cells fromPdcd1–/– and WT mice were in cell cycle, asassessed by Fucci or a 12-hour BrdU pulse test(17) (fig. S2, A and C). The pulse-chase BrdUexperiment revealed considerable differences indynamics of IgA GCs in Pdcd1–/– mice, howev-er. Namely, in both Pdcd1–/– and WT mice, thefrequency of GC and IgA+ B cells that incor-porated BrdU reached a plateau of 95 and 80%,respectively, 5 days after the initiation of BrdUlabeling (Fig. 2D). However, 5 days after the

12

A B

ND

Aerobic

Anaerobic

103 104 105 106 107 108 109

No. bacteria/g small intestine content

Bacteroidaceae

Eubacterium

Lactobacillus

Staphylococcus

Enterobacteriaceae

Bifidobacterium

Streptococcus

***

*

*

C

VH1

VH2

VH3

VH4

VH5

VH6

VH7

VH8

VH9

VH10

VH11

VH12

VH13

VH14

n=141 n=165

RCDR/Stotal+RCDR

E

F

D

G

FreeIgA(μg/ml)

0

50

100

150

200

250

0

25

50

75***

*

Pdcd1-/-WT

Pdcd1 -/-WT

IgAcoated

bacteria(%) Pdcd1-/-WT

72%55%

8%

16%

12%

PCPB

**

*

LPcaspasehi (%)

0

1

2

3

4

*

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0 5 10 15 20 250

20

40

60

80

100

45 60

Chase

Days

**

**

LPBrdU+(%)

BrdU

Pdcd1-/-WT

WT PB

WT PCPdcd1-/- PB

Pdcd1-/- PC

Pdcd1-/-WT

Fig. 1. The gut microbial community com-position and the repertoires of LP-residentIgA-producing cells are altered in Pdcd1–/–

mice. (A) Small intestine microbiota composition in 3-month-old, specific pathogen–free (SPF) Pdcd1–/–and WT mice. Con-tents of the entire small intestine were pooled, and bacteriawere identified with standard microbiological methods. Datafrom two experiments, n = 3 mice per group, are shown. ND,not detected. (B) The percentages of fecal bacteria coated withIgA as determined by flow cytometry and (C) free IgA concen-trations in the feces as determined by enzyme-linked immuno-sorbent assay (ELISA). Data points represent individual mice.(D) IgH V family usage and (E) the affinity maturation of IgA-producing cells from the LP of Pdcd1–/–and WT mice. RCDR, replacement inCDR1 and CDR2; Stotal, silent mutations in both CDRs and in frameworkregions 1 to 3 (FWR1-3) (12). The number of sequences analyzed (pooled fromthree mice per group) is indicated. (F) Frequency of BrdU+ and (G) Caspasehi

IgA PBs and PCs from the LP of Pdcd1–/– and WT mice as determined by flow

cytometry. Note that some of the BrdU+ PCs may be also derived fromperitoneal B1 cells. Data shown are combined from two independent sets ofexperiments. Mean T SEM (n = 3 to 5 mice per group). Two-tailed unpairedStudent’s t test was used for all statistical analyses; ***P < 0.001; **P < 0.01,*P < 0.05.

A

AIDBCL6 PD-1

LZ

DZ

LZ

DZ

B

D

n=77

n=71WT GCPdcd1-/- GCWT IgA+

Pdcd1-/- IgA+

0 5 10 15 20 250

20

40

60

80

100

45 60Days

BrdU Chase

PPBrdU+(%)

CD3 AID

Pdcd1-/-

WT 5

7.2

B220

IgA

B220+

10

14

Fas

PNA

C

Pdcd1 -/-

WT

E

Pdcd1 -/-

WT

Cellnumberx104

PNA+ Fas+ B220+ IgA+

0

25

50

75

100*

0

20

40

60

80

100

120*

Pdcd1 -/-WT

Fig. 2. Increased numbers and enhanced dynamics of GC and IgA+ B cells in PPsof Pdcd1–/– mice. (A) Representative sections of the PPs and (B) flow cytometricprofiles of PP cells from WT and Pdcd1–/– mice stained as indicated to reveal thestructure and characteristics of GCs. Scale bars, 100 mm. LZ (light-zone AIDlow)and DZ (dark-zone AIDhi) are shown. (C) Absolute numbers of indicated B cellpopulations isolated from PPs of WT and Pdcd1–/– mice. Means T SEM (n = 16mice per group). (D) Frequency of BrdU+ GCB cells (PNA+Fas+) and B220+IgA+ B

cells from PPs of Pdcd1–/– and WT mice as determined by flow cytometry. Dataare combined from two independent sets of experiments. Means T SEM (n = 3 to5 mice per group). (E) Representative charts of clonal diversity, calculated asfrequency of transcripts of a given in-frame VH-(N)-D-(N)-JH rearrangement of GCIgA+ B cells in PPs of Pdcd1–/– andWTmice. The number of productive sequencescompared is indicated. Two-tailed unpaired Student’s t test was used for allstatistical analyses; **P < 0.01; *P < 0.05.

13

493305

709

66.454.8

87.7

85.353.8

114

11.5

6.8

30.239

15.5

42.219.3

21.4

715541

1010

CD4

TCRβ

ICOS

CXCR5 BCL6 IRF4

Pdcd1-/-

WTA

Pdcd1-/-WT

B

C

D

IL-21

IL-10

TFH CXCR5-

IFN- γ

IL-17

30

1.53.4

65

18

1.44.3

77

7.5

0.456.8

85 1

0.0120.11

99

1.6

00.18

98

1.1

0.0460.046

99

1

0.0260.17

99 14

0.536.6

79

TFH CXCR5-

Pdcd1 -/-

WT

IFN-γ

**

*

TFH CXCR5-

Expression(relativetoGapdh)

IL-21E

Cellnumberx104

TFH

CXCR5+

CXCR5-

*****

*

0

10

20

30

40

50

60

RatioTFH/GC

0.0

0.1

0.2

0.3

0.4

Pdcd1-/-WT

0

1

2

3

0

1

2

3

4

5

Fig. 3. Expansion of activated T cells and skewed cytokine profiles of TFHcells in PP GCs of Pdcd1–/– mice. (A) Representative flow cytometric profilesand plots of PP cells from WT and Pdcd1–/– mice stained for the indicatedmarkers. Numbers in the graphs indicate the geometric mean fluorescenceintensity of BCL6 and IRF4 in the corresponding color-coded T cell subsetgates. (B) Absolute numbers of major T cell populations and (C) the ratio ofTFH cells to GC B cells in the PPs of Pdcd1–/– and WT mice. Means T SEM (n =

at least 5 mice per group). (D) Flow cytometric profiles and (E) mRNA ex-pression of indicated cytokines in PP T cell subsets. Numbers representpercentage of cells in the quadrants or gates. Relative amounts of mRNAnormalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are shown.Means T SEM of at least two independent experiments. Two-tailed unpairedStudent’s t test was used for all statistical analyses; ***P < 0.001; **P < 0.01,*P < 0.05.

removal of BrdU, the frequency of labeled GCand IgA+ B cells was reduced by half in Pdcd1–/–

mice, whereas most of the BrdU+ cells remainedin the PPs in WT mice (Fig. 2D). The significantdrop in BrdU+ cells in PPs of Pdcd1–/– micecould be the result of increased cell death in situ.However, there were no differences in the fre-quency of caspasehi GCs and IgA+ cells forWT and Pdcd1–/– mice (fig. S2D). The data sug-gest that the increased turnover of IgA+ B cellsin PPs of Pdcd1–/– mice may be because theypass through the GCs more rapidly. Also, theincreased number of GC B cells in Pdcd1–/–

mice likely results from more inflow of B cellsinto GCs. Indeed, the frequency and numbersof IgD+ cells with an early GC phenotype(B220+IgD+IgA+/–GL7+Fas+CCR6+) (18) werehigher in PP of Pdcd1–/– mice (fig. S2, E andF). Taken together, our results suggest thatPD-1 deficiency is associated with an increasedimport and export of B cells into the GC. Suchan abbreviated transit time likely results in im-paired clonal selection and expansion of IgA+

B cells in GCs of PPs, as shown by the reduction

of the frequency of clonally related IgA sequences[identical VH-(N)-D-(N)-JH where VH is the var-iable region of the immunoglobulin heavy chain,N is a randomnucleotide, D is the diversity region,and JH is the joining region of the immunoglobulinheavy chain] in Pdcd1–/– mice compared withWT mice (Fig. 2E).

We then asked why PD-1 deficiency causessuch changes in the dynamics of GC B cells inPPs. It is well established that the number andactivation status of T cells—and TFH cells inparticular—play a fundamental role in shapingGC biology. In fact, B cells compete for T cellhelp before entry into GCs, as well as withinGCs, and deregulation of TFH cells leads to inap-propriate GC B cell selection and humoral auto-immunity (16–19). Thus, we assessed TFH cellsin the PPs of Pdcd1–/– mice. The frequency andabsolute number of CD4+ T cells, includingCXCR5hiICOShi (high levels of ICOS, protein-inducible costimulator) TFH cells, were signifi-cantly increased in PPs of Pdcd1–/– mice (Fig. 3,A and B, and Fig. 2A). The ratio of TFH to GCB cells in Pdcd1–/– mice was twice as high as

that in WT mice (Fig. 3C). The expression ofBCL6, a transcription factor required for TFHdifferentiation, was higher in all CD4+ T cell sub-sets, including the TFH cells in PPs of Pdcd1–/–

mice (Fig. 3A). In contrast, the expression ofinterferon regulatory factor 4 (IRF4), which isrequired for the production of interleukin-21(IL-21), the cytokine that promotes growth anddifferentiation of IgA+ B cells into PCs (19, 20),was reduced in TFH as well as CXCR5hiICOSint

cells from Pdcd1–/– mice. Indeed, TFH cells fromPdcd1–/– mice produced less IL-21 comparedwith those fromWTmice, as previously reported(21) (Fig. 3, D and E). Furthermore, the propor-tion of TFH cells producing interferon-g (IFN-g)was increased in PPs of Pdcd1–/– mice.

In order to better characterize how TFH cellsmay contribute to the altered GC responses ob-served in Pdcd1–/– mice, we decided to evalu-ate the ability of PD-1–deficient TFH cells tosupport IgA generation in gut in vivo. Thus, PPCXCR5hiICOShi TFH cells were isolated fromPdcd1–/– and WT mice and were adoptivelytransferred into T cell–deficient (lacking the

/

14

finity maturation compared with those inducedupon transfer of TFH cells (Fig. 4, D and E). Theseobservations suggest that a significant fraction ofPCs in Pdcd1–/– mice may be generated with thehelp provided by TFH cells derived from Foxp3+

T cells.Taken together, the results indicate that (i) the

quality of gut IgAs critically depends on thenumber and the nature of TFH cells in PPs; (ii)PD-1 deficiency interferes with the selection ofB cells in GCs; and (iii) in addition to selectionin PP GCs, IgAs appear to undergo a second,commensal-driven selection in the LP. Thus, theLP may serve not only as reservoir of PC butalso as a site where proliferating PB are reselectedto fit the geographical distribution of dynamicbacterial communities along the intestine.

In Pdcd1–/– mice, as in SHM-defectiveAIDG23S mice (23), the gut microbiota inducehyperactivation of the systemic immune sys-tem (fig. S4, A and B). Indeed, along with Tand B cell hyperplasia, we found that serum

18 24

18 4.4

Pdcd1-/- TFHWT TFH

IgA

B220

Fas

PNA (B220+ gate)

A

LP

PPB

CD3 AID DAPI

WT TFH

Pdcd1-/-Foxp3+WT Foxp3+

IgA DAPI

5.7 9.2PP

0.77

0.62

2.1

Cd3e-/-transferred with:

- Pdcd1-/- TFH

Cd3e-/-transferred with:

PPTFHPP CD4+T

TFH Foxp3+

Cd3e-/- transferred with:

** **

0

20

40

60

80

0

5

10

15

20

TFH Foxp3+

Cellnumberx104

E

Cd3e-/- transferred with:

0.0

0.1

0.2

0.3

0.4

0.5

***

TFH Foxp3+-RCDR/Stotal+RCDR

Pdcd1-/-WT

DC

0

20

40

60

80

100

120

Cd3e-/- transferred with:

FecalIgA(μg/ml)

0

2

4

6

8

10

***Mutations/VHgene

TFH Foxp3+-TFH Foxp3+-

F**

Fig. 4. TFH cells from Pdcd1–/– mice are impaired in their ability to support the generation of IgA plasmacells in gut. (A) Representative flow cytometric profiles of cells isolated from PPs and LP stained for theindicated markers. Numbers represent percentage of cells in the indicated gates. (B) Sections of the PPsand small intestine stained as indicated and (C) fecal IgA levels as determined by ELISA from the Cd3e–/–

mice 3 months after reconstitution with TFH cells and Foxp3+ T cells from WT and Pdcd1–/– mice. Scalebars, 100 mm. (D) Absolute number of somatic mutations in VH genes and (E) the affinity maturation ofthe IgA-producing cells from the LP of Cd3e–/– mice at 3 months after the reconstitution with the T cellsindicated. The number of sequences analyzed: WT, TFH = 171; Pdcd1–/–, TFH = 175; WT Foxp3+, T = 153;Pdcd1–/– Foxp3+, T = 177. (F) Total numbers of indicated cells from the PPs of Cd3e–/– mice at 3 monthsafter the reconstitution with the T cells indicated. Means T SEM (n = 2 to 3 mice per group). Two-tailedunpaired Student’s t test was used for all statistical analyses; ***P < 0.001; **P < 0.01.

CD3e subunit, Cd3e–/–) mice. Injection of TFH

cells from both WT and Pdcd1–/– mice intoCd3e–/– mice reconstituted the GC B cells andB220+IgA+ B cells in PPs (Fig. 4, A and B).However, Cd3e–/– mice transferred with PD-1–deficient TFH cells generated very few IgA-secreting cells in the LP compared with miceinjected with WT TFH cells (Fig. 4, A to C). Thefew IgAs induced by Pdcd1–/– TFH cells hadincreased turnover (which mirrored the intactPdcd1–/– mice) and a comparable number ofmutations in their VH genes and affinity matura-tion index with those induced with the help ofWT TFH cells (Fig. 4, D and E, and fig. S3A).Similar to what we observed in Pdcd1–/– mice,the absolute number of PP TFH cells was muchhigher in Cd3e–/– mice that received Pdcd1–/–

TFH cells than in mice injected with WT TFHcells, and the Pdcd1–/– TFH cells maintained theirskewed cytokine profile (Fig. 4F, and fig. S3, Band C). The results indicate that TFH cells inPdcd1–/– mice preclude the generation of IgA-secreting cells in gut. On the other hand, bothPD-1–deficient and sufficient Foxp3+ T cells,which, as previously reported (22), did not ex-pand much but converted into IL-21–producingTFH cells in PPs, supported the generation of LPIgAs after the transfer into Cd3e–/– mice (Fig. 4,B, C, and F; and fig. S3, D and E). It was in-teresting that the IgAs induced by Foxp3+ Tcellshad significantly fewer mutations and low af-

6. H. Nishimura, M. Nose, H. Hiai, N. Minato, T. Honjo,Immunity 11, 141 (1999).

7. H. Nishimura et al., Science 291, 319 (2001).8. T. Okazaki et al., J. Exp. Med. 208, 395 (2011).9. R. E. Ley et al., Proc. Natl. Acad. Sci. U.S.A. 102, 11070

(2005).10. L. A. van der Waaij, P. C. Limburg, G. Mesander,

D. van der Waaij, Gut 38, 348 (1996).11. U. Hershberg, M. Uduman, M. J. Shlomchik, S. H. Kleinstein,

Int. Immunol. 20, 683 (2008).12. Materials and methods are available as supplementary

materials on Science Online.13. H. E. Mei et al., Blood 116, 5181 (2010).14. A. Sakaue-Sawano et al., Cell 132, 487 (2008).15. S. Yuvaraj et al., J. Immunol. 183, 4871 (2009).16. G. D. Victora et al., Cell 143, 592 (2010).17. C. D. Allen, T. Okada, H. L. Tang, J. G. Cyster,

Science 315, 528 (2007).18. T. A. Schwickert et al., J. Exp. Med. 208, 1243 (2011).19. C. G. Vinuesa, S. G. Tangye, B. Moser, C. R. Mackay,

Nat. Rev. Immunol. 5, 853 (2005).

from Pdcd1–/–mice contained antibodies specificfor components of commensal bacteria, whichindicated a breach of normal mucosal-systemiccompartmentalization (fig. S4, A and C) (24). Onthe basis of the data presented here, we proposethat the skewed gut microbial communities thatresult from the dysregulated selection of IgAsdrive the expansion of self-reactive B and Tcells. Our studies have implications for howmodulation of PD-1 expression promotes toler-ance or uncontrolled immune reactions leadingto autoimmunity.

References and Notes1. S. Fagarasan, S. Kawamoto, O. Kanagawa, K. Suzuki,

Annu. Rev. Immunol. 28, 243 (2010).2. J. Jacob, G. Kelsoe, K. Rajewsky, U. Weiss, Nature 354,

389 (1991).3. M. Muramatsu et al., Cell 102, 553 (2000).4. N. M. Haynes et al., J. Immunol. 179, 5099 (2007).5. T. Okazaki, T. Honjo, Trends Immunol. 27, 195 (2006).

22. M. Tsuji et al., Science 323, 1488 (2009).23. M. Wei et al., Nat. Immunol. 12, 264 (2011).24. E. Slack et al., Science 325, 617 (2009).

Acknowledgments: We thank T. Honjo, T. Okazaki,T. Okada, I. Taniuchi, O. Kanagawa, N. Minato, andS. Casola for inspiring discussions, suggestions, and criticalcomments; M. Miyajima and D. Sutherland for criticallyreading of the manuscript; and Y. Murahashi and T. Kajifor technical assistance. The data reported in this paperare tabulated in the main paper and the supplementarymaterials. Supported by Grants-in-Aid for ScientificResearch in Priority Areas (S.F.) and by RIKEN President’sDiscretionary Fund (S.F.) and Special Postdoctoral ResearchersProgram (S.K.).

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6080/485/DC1Materials and MethodsFigs. S1 to S4References (25–34)

9 December 2011; accepted 16 March 201210.1126/science.1217718

20. H. Kwon et al., Immunity 31, 941 (2009).21. K. L. Good-Jacobson et al., Nat. Immunol. 11, 535

(2010).

9 December 2011; accepted 16 March 2012Published 27 April 201210.1126/science.1217718

6. H. Nishimura, M. Nose, H. Hiai, N. Minato, T. Honjo,Immunity 11, 141 (1999).

7. H. Nishimura et al., Science 291, 319 (2001).8. T. Okazaki et al., J. Exp. Med. 208, 395 (2011).9. R. E. Ley et al., Proc. Natl. Acad. Sci. U.S.A. 102, 11070

(2005).10. L. A. van der Waaij, P. C. Limburg, G. Mesander,

D. van der Waaij, Gut 38, 348 (1996).11. U. Hershberg, M. Uduman, M. J. Shlomchik, S. H. Kleinstein,

Int. Immunol. 20, 683 (2008).12. Materials and methods are available as supplementary

materials on Science Online.13. H. E. Mei et al., Blood 116, 5181 (2010).14. A. Sakaue-Sawano et al., Cell 132, 487 (2008).15. S. Yuvaraj et al., J. Immunol. 183, 4871 (2009).16. G. D. Victora et al., Cell 143, 592 (2010).17. C. D. Allen, T. Okada, H. L. Tang, J. G. Cyster,

Science 315, 528 (2007).18. T. A. Schwickert et al., J. Exp. Med. 208, 1243 (2011).19. C. G. Vinuesa, S. G. Tangye, B. Moser, C. R. Mackay,

Nat. Rev. Immunol. 5, 853 (2005).

References and Notes1. S. Fagarasan, S. Kawamoto, O. Kanagawa, K. Suzuki,

Annu. Rev. Immunol. 28, 243 (2010).2. J. Jacob, G. Kelsoe, K. Rajewsky, U. Weiss, Nature 354,

389 (1991).3. M. Muramatsu et al., Cell 102, 553 (2000).4. N. M. Haynes et al., J. Immunol. 179, 5099 (2007).5. T. Okazaki, T. Honjo, Trends Immunol. 27, 195 (2006).

22. M. Tsuji et al., Science 323, 1488 (2009).23. M. Wei et al., Nat. Immunol. 12, 264 (2011).24. E. Slack et al., Science 325, 617 (2009).

Acknowledgments: We thank T. Honjo, T. Okazaki,T. Okada, I. Taniuchi, O. Kanagawa, N. Minato, andS. Casola for inspiring discussions, suggestions, and criticalcomments; M. Miyajima and D. Sutherland for criticallyreading of the manuscript; and Y. Murahashi and T. Kajifor technical assistance. The data reported in this paperare tabulated in the main paper and the supplementarymaterials. Supported by Grants-in-Aid for ScientificResearch in Priority Areas (S.F.) and by RIKEN President’sDiscretionary Fund (S.F.) and Special Postdoctoral ResearchersProgram (S.K.).

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6080/485/DC1Materials and MethodsFigs. S1 to S4References (25–34)

9 December 2011; accepted 16 March 201210.1126/science.1217718

20. H. Kwon et al., Immunity 31, 941 (2009).21. K. L. Good-Jacobson et al., Nat. Immunol. 11, 535

(2010).

9 December 2011; accepted 16 March 2012Published 27 April 201210.1126/science.1217718

9 December 2011; accepted 16 March 2012Published 27 April 201210.1126/science.1217718

15

An Essential DevelopmentalCheckpoint for Production of theT Cell LineageTomokatsu Ikawa,1 Satoshi Hirose,2 Kyoko Masuda,1 Kiyokazu Kakugawa,1 Rumi Satoh,1

Asako Shibano-Satoh,1 Ryo Kominami,2 Yoshimoto Katsura,1,3 Hiroshi Kawamoto1*

In early T cell development, progenitors retaining the potential to generate myeloid and naturalkiller lineages are eventually determined to a specific T cell lineage. The molecular mechanismsthat drive this determination step remain unclarified. We show that, when murine hematopoieticprogenitors were cultured on immobilized Notch ligand DLL4 protein in the presence of a cocktailof cytokines including interleukin-7, progenitors developing toward T cells were arrested and thearrested cells entered a self-renewal cycle, maintaining non-T lineage potentials. Reducedconcentrations of interleukin-7 promoted T cell lineage determination. A similar arrest and self-renewal of progenitors were observed in thymocytes of mice deficient in the transcription factorBcl11b. Our study thus identifies the earliest checkpoint during T cell development and shows thatit is Bcl11b-dependent.

Tcells are generated from multipotent he-matopoietic stem cells through a seriesof differentiation steps. The first step in

this pathway is the generation of progenitors thathave lost erythroid/megakaryocyte potential butretain the capacity to generate other hemopoieticcells, including myeloid, T, and B cells (1–6). Weand others have recently identified the next stage,in which the T cell progenitors have lost B cellpotential but are still able to generate myeloid cells,dendritic cells (DCs), and natural killer (NK) cells(7, 8). Therefore, the most critical step for devel-opment of the T cell lineage is now thought to beat the point where myeloid potential is terminated.

We sought to identify the step at which pro-genitors become fully committed to the T celllineage and what regulates this transition. Areliable way to substantiate that a given step iscritical for the development of a lineage is to

demonstrate developmental arrest at the stagebefore that step under particular conditions. In thecase of B cell differentiation, deletion of E2a,Ebf, or Pax5 genes leads to an early develop-mental arrest before formation of a functionalIgH chain gene; these arrested B cell progenitorsundergo self-renewal and remain B lineage un-committed, with the potential to develop alongother lineages, includingmyeloid and Tcell (9–11).This case illustrates that such a critical develop-mental checkpoint exists at the step when un-committed B cell progenitors become determinedto the B cell lineage. Unlike the B cell lineage, todate no such checkpoint has been identified forthe T cell lineage before the initiation of TCRgene rearrangement.

As T cell progenitors develop, they proceedthrough developmental stages referred to as DN1to DN4 (double-negative CD4–CD8–) that can betracked by surface phenotype. The DN2 stagecan be subdivided into two stages based on trans-genic green fluorescent protein (GFP) expressioncontrolled by the proximal lck (plck) promoter(lck is a src family kinase selectively expressedby T cells). GFP– cells retain non-T lineagepotential, including that for myeloid cells, DCs,and NK cells, whereas the latter stage GFP+ cellsare determined to the T cell lineage (7, 12). We

designate these two stages DN2mt (myeloid-T)and DN2t (T-lineage determined) and term thestep between these stages the DN2-determinationstep. This determination step is thought to be thefirst critical checkpoint in Tcell development (13).

We cultured lineage-negative (Lin–) c-kit+

Sca-1+ (LKS) cells from 13 days post-coitum (dpc)murine fetal liver with immobilized Delta-like 4(DLL4) protein in the presence of the cytokinesSCF (stem cell factor), Flt3L (FMS-like tyrosinekinase ligand), and interleukin (IL)–7 (fig. S1).After 7 days of culture, cells remained at the DNstage (Fig. 1A, left panel), whereas in the controlgroup, where cells were cultured with TSt-4 stro-mal cells expressing DLL4 (TSt-4/DLL4), gen-eration of CD4+CD8+ double-positive (DP) cellswas observed (fig. S2). Upon closer analysis onDN cells generated in the feeder-free condition,we observed that these cells resembled DN2mtcells (maintained c-kithighCD25+) (Fig. 1A, rightpanel) and thus named them FFDN2 cells (feeder-free-cultured DN2-like cells). By several criteria,the FFDN2 cells appeared identical to DN2mtcells: (i) they gave rise to authentic ab T cellswhen transferred to a TSt-4/DLL4 stromal cocul-ture system (fig. S3, A and B); (ii) they retained thepotential to produce macrophages (Fig. 1B), NKcells, and DCs (fig. S3, C and D); (iii) intracellularT cell receptor (TCR) b chain protein was notexpressed (Fig. 1C); and (iv) their gene expressionprofiles were similar to those of DN2mt cells (Fig.1D and fig. S4). Furthermore, GFP expressionwas not observed in FFDN2 cells generated fromprogenitors isolated from plck-GFP mice (Fig.1E). It is unlikely that this arrest is due to the fail-ure of TCR gene rearrangement because enforcedexpression of a functional TCRb chain gene didnot prevent the developmental arrest (fig. S5).FFDN2 cells could not generate B cells (fig. S3E),indicating that dedifferentiation to more primitiveprogenitors did not occur in this culture system.Ofnote, FFDN2 cells showed an almost unlimited invitro expansion (Fig. 1F), while essentially main-taining c-kit and CD25 expression (Fig. 1G) and adevelopmental potential comparable to that offreshly isolated DN2mt cells (fig. S6). Cells in thec-kit+CD25+ fraction possessed the potential tomaintain long-term culture, because long-term cul-ture could be maintained by using c-kit+CD25+

cells at the time of passage (fig. S7). Such self-renewal capacity, together with our other results,

1Laboratory for Lymphocyte Development, RIKEN ResearchCenter for Allergy and Immunology, Yokohama 230-0045, Japan.2Department of Molecular Genetics, Graduate School of Medicaland Dental Sciences, Niigata University, Niigata 951-8510,Japan. 3Division of Cell Regeneration and Transplantation,Advanced Medical Research Center, Nihon University School ofMedicine, Tokyo 173-8610, Japan.

*To whom correspondence should be addressed. E-mail:kawamoto@rcai.riken.jp

1Laboratory for Lymphocyte Development, RIKEN ResearchCenter for Allergy and Immunology, Yokohama 230-0045, Japan.2Department of Molecular Genetics, Graduate School of Medicaland Dental Sciences, Niigata University, Niigata 951-8510,Japan. 3Division of Cell Regeneration and Transplantation,Advanced Medical Research Center, Nihon University School ofMedicine, Tokyo 173-8610, Japan.

*To whom correspondence should be addressed. E-mail:kawamoto@rcai.riken.jp

demonstrate developmental arrest at the stagebefore that step under particular conditions. In thecase of B cell differentiation, deletion of E2a,Ebf, or Pax5 genes leads to an early develop-mental arrest before formation of a functionalIgH chain gene; these arrested B cell progenitorsundergo self-renewal and remain B lineage un-committed, with the potential to develop alongother lineages, includingmyeloid and Tcell (9–11).This case illustrates that such a critical develop-mental checkpoint exists at the step when un-committed B cell progenitors become determined

to the B cell lineage. Unlike the B cell lineage, todate no such checkpoint has been identified forthe T cell lineage before the initiation of TCRgene rearrangement.

As T cell progenitors develop, they proceedthrough developmental stages referred to as DN1to DN4 (double-negative CD4–CD8–) that can betracked by surface phenotype. The DN2 stagecan be subdivided into two stages based on trans-genic green fluorescent protein (GFP) expressioncontrolled by the proximal lck (plck) promoter(lck is a src family kinase selectively expressedby T cells). GFP– cells retain non-T lineagepotential, including that for myeloid cells, DCs,and NK cells, whereas the latter stage GFP+ cellsare determined to the T cell lineage (7, 12). Wedesignate these two stages DN2mt (myeloid-T)and DN2t (T-lineage determined) and term thestep between these stages the DN2-determinationstep. This determination step is thought to be thefirst critical checkpoint in Tcell development (13).

We cultured lineage-negative (Lin–) c-kit+

Sca-1+ (LKS) cells from 13 days post-coitum (dpc)murine fetal liver with immobilized Delta-like 4(DLL4) protein in the presence of the cytokinesSCF (stem cell factor), Flt3L (FMS-like tyrosinekinase ligand), and interleukin (IL)–7 (fig. S1).After 7 days of culture, cells remained at the DNstage (Fig. 1A, left panel), whereas in the controlgroup, where cells were cultured with TSt-4 stro-mal cells expressing DLL4 (TSt-4/DLL4), gen-eration of CD4+CD8+ double-positive (DP) cellswas observed (fig. S2). Upon closer analysis onDN cells generated in the feeder-free condition,we observed that these cells resembled DN2mtcells (maintained c-kithighCD25+) (Fig. 1A, rightpanel) and thus named them FFDN2 cells (feeder-free-cultured DN2-like cells). By several criteria,the FFDN2 cells appeared identical to DN2mtcells: (i) they gave rise to authentic ab T cellswhen transferred to a TSt-4/DLL4 stromal cocul-

ture system (fig. S3, A and B); (ii) they retained thepotential to produce macrophages (Fig. 1B), NKcells, and DCs (fig. S3, C and D); (iii) intracellularT cell receptor (TCR) b chain protein was notexpressed (Fig. 1C); and (iv) their gene expressionprofiles were similar to those of DN2mt cells (Fig.1D and fig. S4). Furthermore, GFP expressionwas not observed in FFDN2 cells generated fromprogenitors isolated from plck-GFP mice (Fig.1E). It is unlikely that this arrest is due to the fail-ure of TCR gene rearrangement because enforcedexpression of a functional TCRb chain gene didnot prevent the developmental arrest (fig. S5).FFDN2 cells could not generate B cells (fig. S3E),indicating that dedifferentiation to more primitiveprogenitors did not occur in this culture system.Ofnote, FFDN2 cells showed an almost unlimited invitro expansion (Fig. 1F), while essentially main-taining c-kit and CD25 expression (Fig. 1G) and adevelopmental potential comparable to that offreshly isolated DN2mt cells (fig. S6). Cells in thec-kit+CD25+ fraction possessed the potential tomaintain long-term culture, because long-term cul-ture could be maintained by using c-kit+CD25+

cells at the time of passage (fig. S7). Such self-renewal capacity, together with our other results,indicated that the DN2-determination step maybe a critical checkpoint for T cell development.

To investigate the molecular mechanisms ofT cell lineage determination, we searched for anenvironmental cue that could drive the arrestedcells through the DN2-determination step. Aftertesting various cytokines and Notch ligand con-ditions in the feeder-free culture system, we foundthat FFDN2 cells initiate differentiation when theconcentration of IL-7 is reduced on day 7 of cul-ture (10 ng/ml to 1 ng/ml). In this induction sys-tem, GFP+ DN3 cells appear on day 3 after IL-7reduction (Fig. 2A). These cells did not expressmyeloid-lineage transcription factors PU.1 (Sfpi1)and C/EBPa, whereas T cell lineage–associatedgenes such as lck, Tcf1, pTa, and Bcl11b weremarkedly up-regulated (Fig. 2B). Notably, cellsin these cultures developed up to the abTCR-expressing CD4+CD8+DP stage (Fig. 2C and figs.S8 and S9). Although the kinetics of DP cell growthwas delayed compared with that in the TSt-4/DLL4feeder cell culture system, the final yield of DP cellswas nearly identical (fig. S10). The DP cells gen-erated by reducing the concentration of IL-7 ap-peared to be authentic DP cells, because they giverise to CD4 and CD8 single-positive (SP) cellswhen transferred to a fetal thymus organ culturesystem (fig. S11). These results demonstrated thatabTCR+ cells can be generated from prethymicprogenitors in a “feeder-free” culture system andthat the TCRb-selection, which is thought to serveas the critical checkpoint for preTCR formation inprogenitors, does not require additional environ-mental factors in this feeder-free culture system.

Often transcription factors regulate cell lineagedetermination steps. Among genes up-regulatedby our induction system, we focused on Bcl11b,a Tcell lineage–specific transcription factor origi-

17

expressing CD4+CD8+DP stage (Fig. 2C and figs.S8 and S9). Although the kinetics ofDP cell growthwas delayed compared with that in the TSt-4/DLL4feeder cell culture system, the final yield of DP cellswas nearly identical (fig. S10). The DP cells gen-erated by reducing the concentration of IL-7 ap-peared to be authentic DP cells, because they giverise to CD4 and CD8 single-positive (SP) cellswhen transferred to a fetal thymus organ culture

system (fig. S11). These results demonstrated thatabTCR+ cells can be generated from prethymicprogenitors in a “feeder-free” culture system andthat the TCRb-selection, which is thought to serveas the critical checkpoint for preTCR formation inprogenitors, does not require additional environ-mental factors in this feeder-free culture system.

Often transcription factors regulate cell lineagedetermination steps. Among genes up-regulated

by our induction system, we focused on Bcl11b,a Tcell lineage–specific transcription factor origi-nally identified as a tumor suppressor in murineT cell lymphoma (14). Bcl11b-deficient mice ex-hibit impaired thymocyte development aroundthe DN3 to immature SP stage because of an in-ability to rearrange the Vb to Db gene segments(15). We carefully reexamined the phenotype offetal thymus cells fromBcl11b-deficient mice andfound that, at 18 dpc, there was a developmentalarrest at the DN2 stage (Fig. 3A). Despite this, theabsolute number of DN2 cells was not increased(Fig. 3B), indicating that self-renewing expansionis not so prominent in vivo, a difference that couldbe due to the limited niche space in the thymus forearly progenitors. We cultured these DN2 cellsunder TSt-4/DLL4 conditions, which can supportT cell differentiation up to the DP stage. In suchcultures, Bcl11b−/− cells continued to proliferateeven after 4 weeks, maintaining their DN2 surfacephenotype (Fig. 3C). Similar to FFDN2 cells,Bcl11b−/−DN2 cells exhibited features of DN2mtcells, including the potential to develop intomacro-phages and NK cells (Fig. 3D), and loss of B cellpotential (fig. S12).

Bcl11b deficiency is lethal around the neo-natal period (15). To investigate whether the de-velopmental arrest of Bcl11b−/− progenitors isseen in the adult thymus, where T cells are con-tinuously generated, we produced chimeric miceby transferring Bcl11b−/− fetal liver cells intoirradiated B6Ly5.1 congenic mice. At 8 weeksafter transfer, we observed nearly complete de-velopmental arrest at the DN2 stage, with only afew DP cells (Fig. 3E). Similar to ex vivo fetalthymocytes of Bcl11b−/− mice and culturedBcl11b−/− DN2 cells, the arrested DN2 cellswere equivalent to DN2mt cells. There was noincrease in thymic B cells in the recipients of theBcl11b−/− fetal liver cells (fig. S13), indicatingthat the Bcl11b−/−DN2 cells that developed in thethymus did not dedifferentiate into more primitiveprogenitors in vivo.

The similar stage of arrest in the DLL4/IL-7cultures and in the Bcl11b−/−mice suggested thatthe arrest in the cultures may be due to a failure toup-regulate Bcl11b. To examine this possibility,we retrovirally transduced Bcl11b cDNA into fe-tal liver LKS cells and cultured these transducedcells under DLL4/IL-7 conditions. The Bcl11b-transduced cells could give rise toDN3 cells evenin the presence of a high concentration of IL-7 (Fig.4A), and TCRb gene rearrangement was enhanced(Fig. 4B), whereas myeloid-lineage–associatedgenes were suppressed (Fig. 4C), demonstratingthat Bcl11b expression eliminated the DN2 arrestthat occurred in the DLL4/IL-7 cultures.

As has been reported (16), the absence ofBcl11b had a severe impact on the generation ofthymic abT cells, whereas there was little effecton the generation of gd T cells (fig. S14A). Thesame is true for cells generated in the DLL4/IL-7cultures (fig. S14B). These results suggested thatthe segregation to the gdT cell lineage occursbefore the DN2mt stage, although the possibility

Bcl11b+/+

CD8

CD4

CD25

c-kit

A BBcl11b+/- Bcl11b-/-

CD8

CD4

TCR

TCR

CD25

c-kit

Bcl11b+

/+Bcl11b-/-

TN Gated12.3 82.7

3.51.52

51.3

0.39

0.59 4.4

887.08

1.73 5.4

2.5290.3

5.13

4.04

0.44 84.4

2.3512.8

C

ECD25

c-kit

totalx105

DN2

+/-0

30

10

20

0

2

4

6

8

+/+ -/-

x105

2.6 73.4

4.7

0.7 66.3

11.4

0.63 0.26

3.8

6.0 73.4

16.6

2.6 31.0

49.9 4.016.5

2.6 16.6

55.125.7

CD45.2

CD3

NK1.1

Mac1

F4/80

NK cellsM

D

+/-

+/+ -/-

Fig. 3. Bcl11b is essential for T cell lineage determination. (A) Flow cytometric analysis of fetalthymocytes from Bcl11b−/− mice. Profiles are shown for CD4 versus CD8 of 18 dpc fetal thymocytes, andc-kit versus CD25 of cells gated in upper panels, from the indicated mice. For each group, more than fivemice were individually analyzed, and representative profiles are shown. (B) Absolute numbers of totalthymocytes and DN2 cells in 18 dpc fetuses of the indicated Bcl11b genotypes. More than five mice wereindividually analyzed for each group, and the mean + SD is shown. (C) Flow cytometric analysis of c-kitversus CD25 of fetal liver LKS cells from Bcl11b−/− mice cultured on TSt-4/DLL4 stromal cells for 30 days.Data are representative of three independent experiments. (D) Generation of macrophages and NK cellsfrom cultured Bcl11b−/− fetal liver cells. The c-kit+ CD25+ cells shown in (C) were cultured (200 cells perwell) for 7 days with TSt-4 cells in the presence of M-CSF (left panel) or IL-15 (right panel) and analyzed formacrophage and NK cell markers by flow cytometry. Data are representative of three independentexperiments. (E) Fetal liver cells from Bcl11b+/+ or Bcl11b−/− mice (Ly5.2) were transferred into lethallyirradiated mice (Ly5.1). Flow cytometric profiles of reconstituted thymocytes of recipient mice 8 weeksafter transfer are shown. In right panels, profiles of cells gated on CD3–CD4–CD8– [triple negative (TN)]fraction are shown. For each group, more than five mice were individually analyzed, and representativedata are shown.

nally identified as a tumor suppressor in murineT cell lymphoma (14). Bcl11b-deficient mice ex-hibit impaired thymocyte development aroundthe DN3 to immature SP stage because of an in-ability to rearrange the Vb to Db gene segments(15). We carefully reexamined the phenotype offetal thymus cells fromBcl11b-deficient mice andfound that, at 18 dpc, there was a developmental

arrest at the DN2 stage (Fig. 3A). Despite this, theabsolute number of DN2 cells was not increased(Fig. 3B), indicating that self-renewing expansionis not so prominent in vivo, a difference that couldbe due to the limited niche space in the thymus forearly progenitors. We cultured these DN2 cellsunder TSt-4/DLL4 conditions, which can supportT cell differentiation up to the DP stage. In such

/

cultures, Bcl11b−/− cells continued to proliferateeven after 4 weeks, maintaining their DN2 surfacephenotype (Fig. 3C). Similar to FFDN2 cells,Bcl11b−/−DN2 cells exhibited features of DN2mtcells, including the potential to develop intomacro-phages and NK cells (Fig. 3D), and loss of B cellpotential (fig. S12).

Bcl11b deficiency is lethal around the neo-natal period (15). To investigate whether the de-velopmental arrest of Bcl11b−/− progenitors isseen in the adult thymus, where T cells are con-tinuously generated, we produced chimeric miceby transferring Bcl11b−/− fetal liver cells intoirradiated B6Ly5.1 congenic mice. At 8 weeksafter transfer, we observed nearly complete de-velopmental arrest at the DN2 stage, with only afew DP cells (Fig. 3E). Similar to ex vivo fetalthymocytes of Bcl11b−/− mice and culturedBcl11b−/− DN2 cells, the arrested DN2 cellswere equivalent to DN2mt cells. There was noincrease in thymic B cells in the recipients of theBcl11b−/− fetal liver cells (fig. S13), indicatingthat the Bcl11b−/−DN2 cells that developed in thethymus did not dedifferentiate into more primitiveprogenitors in vivo.

The similar stage of arrest in the DLL4/IL-7cultures and in the Bcl11b−/−mice suggested thatthe arrest in the cultures may be due to a failure toup-regulate Bcl11b. To examine this possibility,we retrovirally transduced Bcl11b cDNA into fe-tal liver LKS cells and cultured these transducedcells under DLL4/IL-7 conditions. The Bcl11b-transduced cells could give rise toDN3 cells evenin the presence of a high concentration of IL-7 (Fig.4A), and TCRb gene rearrangement was enhanced(Fig. 4B), whereas myeloid-lineage–associatedgenes were suppressed (Fig. 4C), demonstratingthat Bcl11b expression eliminated the DN2 arrestthat occurred in the DLL4/IL-7 cultures.

As has been reported (16), the absence ofBcl11b had a severe impact on the generation ofthymic abT cells, whereas there was little effecton the generation of gd T cells (fig. S14A). Thesame is true for cells generated in the DLL4/IL-7cultures (fig. S14B). These results suggested thatthe segregation to the gdT cell lineage occursbefore the DN2mt stage, although the possibilitystill remains that the gd T cells that had been gen-erated from “leaky” DN3 cells underwent com-pensatory proliferation.

The developmental steps just after the forma-tion of preTCR (DN3 stage) and ab TCR (DPstage) serve as critical checkpoints (16, 17), andcells that fail to pass these points succumb toapoptotic cell death. In contrast, the arrestedprogenitors at the DN2-determination step enter aself-renewal cycle. The appearance of self-renewing progenitors among Bcl11b−/− thymo-cytes may explain the previous findings thatloss-of-function mutations in the Bcl11b gene arefrequently observed in murine T cell lymphomasinduced by g irradiation (14) and that chromo-somal aberration disrupting Bcl11b gene wasidentified in human T cell acute lymphoblastic

Bcl11b+/+

CD8

CD4

CD25

c-kit

A BBcl11b+/- Bcl11b-/-

CD8

CD4

TCR

TCR

CD25

c-kit

Bcl11b+

/+Bcl11b-/-

TN Gated12.3 82.7

3.51.52

51.3

0.39

0.59 4.4

887.08

1.73 5.4

2.5290.3

5.13

4.04

0.44 84.4

2.3512.8

C

ECD25

c-kit

totalx105

DN2

+/-0

30

10

20

0

2

4

6

8

+/+ -/-

x105

2.6 73.4

4.7

0.7 66.3

11.4

0.63 0.26

3.8

6.0 73.4

16.6

2.6 31.0

49.9 4.016.5

2.6 16.6

55.125.7

CD45.2

CD3

NK1.1

Mac1

F4/80

NK cellsM

D+/-

+/+ -/-

Fig. 3. Bcl11b is essential for T cell lineage determination. (A) Flow cytometric analysis of fetalthymocytes from Bcl11b−/− mice. Profiles are shown for CD4 versus CD8 of 18 dpc fetal thymocytes, andc-kit versus CD25 of cells gated in upper panels, from the indicated mice. For each group, more than fivemice were individually analyzed, and representative profiles are shown. (B) Absolute numbers of totalthymocytes and DN2 cells in 18 dpc fetuses of the indicated Bcl11b genotypes. More than five mice wereindividually analyzed for each group, and the mean + SD is shown. (C) Flow cytometric analysis of c-kitversus CD25 of fetal liver LKS cells from Bcl11b−/− mice cultured on TSt-4/DLL4 stromal cells for 30 days.Data are representative of three independent experiments. (D) Generation of macrophages and NK cellsfrom cultured Bcl11b−/− fetal liver cells. The c-kit+ CD25+ cells shown in (C) were cultured (200 cells perwell) for 7 days with TSt-4 cells in the presence of M-CSF (left panel) or IL-15 (right panel) and analyzed formacrophage and NK cell markers by flow cytometry. Data are representative of three independentexperiments. (E) Fetal liver cells from Bcl11b+/+ or Bcl11b−/− mice (Ly5.2) were transferred into lethallyirradiated mice (Ly5.1). Flow cytometric profiles of reconstituted thymocytes of recipient mice 8 weeksafter transfer are shown. In right panels, profiles of cells gated on CD3–CD4–CD8– [triple negative (TN)]fraction are shown. For each group, more than five mice were individually analyzed, and representativedata are shown.

18

leukemia cases (18) because the acquisition of sor we speculate that Bcl11b directly suppresses

A recent study demonstrated that Bcl11b isexpressed in the T cell–like lymphoid cells oflamprey (21). Because a Bcl11b homolog has notbeen found in animals other than vertebrates(fig. S16), we propose that Bcl11b arose in phy-logeny to construct a new lineage distinct fromthe preexisting innate type killer cells.

References and Notes1. Y. Katsura, H. Kawamoto, Int. Rev. Immunol. 20,

1 (2001).2. M. Lu, H. Kawamoto, Y. Katsube, T. Ikawa, Y. Katsura,

J. Immunol. 169, 3519 (2002).3. Y. Katsura, Nat. Rev. Immunol. 2, 127 (2002).4. J. Adolfsson et al., Cell 121, 295 (2005).5. C. V. Laiosa, M. Stadtfeld, T. Graf, Annu. Rev. Immunol.

24, 705 (2006).6. H. Kawamoto, Y. Katsura, Trends Immunol. 30, 193

(2009).7. H. Wada et al., Nature 452, 768 (2008).8. J. J. Bell, A. Bhandoola, Nature 452, 764 (2008).9. S. L. Nutt, B. Heavey, A. G. Rolink, M. Busslinger, Nature

401, 556 (1999).10. T. Ikawa, H. Kawamoto, L. Y. Wright, C. Murre, Immunity

20, 349 (2004).11. J. M. Pongubala et al., Nat. Immunol. 9, 203 (2008).12. K. Masuda et al., J. Immunol. 179, 3699 (2007).13. E. V. Rothenberg, J. E. Moore, M. A. Yui, Nat. Rev.

Immunol. 8, 9 (2008).14. Y. Wakabayashi et al., Biochem. Biophys. Res. Commun.

301, 598 (2003).15. Y. Wakabayashi et al., Nat. Immunol. 4, 533 (2003).16. H. J. Fehling, H. von Boehmer, Curr. Opin. Immunol. 9,

263 (1997).17. Y. Takahama, Nat. Rev. Immunol. 6, 127 (2006).18. G. K. Przybylski et al., Leukemia 19, 201 (2005).19. M. P. McCormack et al., Science 327, 879 (2010).20. Q. Yu, B. Erman, J. H. Park, L. Feigenbaum, A. Singer,

J. Exp. Med. 200, 797 (2004).21. P. Guo et al., Nature 459, 796 (2009).22. The authors are grateful to C. Murre, T. Kadesch,

Y. Agata, and S. Yamasaki for providing us with reagentsand protocols, and to P. Burrows for critical reading ofthe manuscript. This work was partially supported byGrant-in-Aid for Young Scientists (A) from the Ministry ofEducation, Science, Sports, and Culture, Japan.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/329/5987/93/DC1Materials and MethodsFigs S1 to s16

leukemia cases (18), because the acquisition ofself-renewal capacity is regarded as the first stepin leukemia development. In this context, a sim-ilar outcome was recently observed when Lmo2,

signals in the thymus. It is quite likely that thereduction in IL-7 signaling is a physiologicalmediator of this step, because the IL-7R is dra-matically down-regulated at the transition fromthe DN2 to the DN3 stage (20). Considering thatBcll1b is thought to be a transcriptional repres-sor, we speculate that Bcl11b directly suppressesmyeloid-lineage–associated genes, such as PU.1 orC/EBPa, and that such suppression is critical fordifferentiation toward the T cell fate.

A recent study demonstrated that Bcl11b isexpressed in the T cell–like lymphoid cells oflamprey (21). Because a Bcl11b homolog has notbeen found in animals other than vertebrates(fig. S16), we propose that Bcl11b arose in phy-

a known oncogene, was overexpressed in thymo-cytes and caused the cells to enter a self-renewalcycle in vivo (19).

The present study thus defines a Bcl11b-driven checkpoint at which T cell progenitorsterminate non-T-lineage potential in order to be-come determined to the T cell lineage (fig. S15).Our finding that Bcl11b up-regulation can betriggered by an extrinsic cue, diminished IL-7,suggests that progression through the DN2-determination step is instructed by environmental

logeny to construct a new lineage distinct fromthe preexisting innate type killer cells.

References and Notes1. Y. Katsura, H. Kawamoto, Int. Rev. Immunol. 20,

1 (2001).2. M. Lu, H. Kawamoto, Y. Katsube, T. Ikawa, Y. Katsura,

J. Immunol. 169, 3519 (2002).3. Y. Katsura, Nat. Rev. Immunol. 2, 127 (2002).4. J. Adolfsson et al., Cell 121, 295 (2005).5. C. V. Laiosa, M. Stadtfeld, T. Graf, Annu. Rev. Immunol.

24, 705 (2006).6. H. Kawamoto, Y. Katsura, Trends Immunol. 30, 193

(2009).7. H. Wada et al., Nature 452, 768 (2008).8. J. J. Bell, A. Bhandoola, Nature 452, 764 (2008).9. S. L. Nutt, B. Heavey, A. G. Rolink, M. Busslinger, Nature

401, 556 (1999).10. T. Ikawa, H. Kawamoto, L. Y. Wright, C. Murre, Immunity

20, 349 (2004).11. J. M. Pongubala et al., Nat. Immunol. 9, 203 (2008).12. K. Masuda et al., J. Immunol. 179, 3699 (2007).13. E. V. Rothenberg, J. E. Moore, M. A. Yui, Nat. Rev.

Immunol. 8, 9 (2008).14. Y. Wakabayashi et al., Biochem. Biophys. Res. Commun.

301, 598 (2003).15. Y. Wakabayashi et al., Nat. Immunol. 4, 533 (2003).16. H. J. Fehling, H. von Boehmer, Curr. Opin. Immunol. 9,

263 (1997).17. Y. Takahama, Nat. Rev. Immunol. 6, 127 (2006).18. G. K. Przybylski et al., Leukemia 19, 201 (2005).19. M. P. McCormack et al., Science 327, 879 (2010).20. Q. Yu, B. Erman, J. H. Park, L. Feigenbaum, A. Singer,

J. Exp. Med. 200, 797 (2004).21. P. Guo et al., Nature 459, 796 (2009).22. The authors are grateful to C. Murre, T. Kadesch,

Y. Agata, and S. Yamasaki for providing us with reagentsand protocols, and to P. Burrows for critical reading ofthe manuscript. This work was partially supported byGrant-in-Aid for Young Scientists (A) from the Ministry ofEducation, Science, Sports, and Culture, Japan.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/329/5987/93/DC1Materials and MethodsFigs. S1 to s16References

2 March 2010; accepted 12 May 201010.1126/science.1188995

erated from “leaky” DN3 cells underwent com-pensatory proliferation.

The developmental steps just after the forma-tion of preTCR (DN3 stage) and ab TCR (DPstage) serve as critical checkpoints (16, 17), andcells that fail to pass these points succumb toapoptotic cell death. In contrast, the arrestedprogenitors at the DN2-determination step enter aself-renewal cycle. The appearance of self-renewing progenitors among Bcl11b−/− thymo-cytes may explain the previous findings thatloss-of-function mutations in the Bcl11b gene arefrequently observed in murine T cell lymphomasinduced by g irradiation (14) and that chromo-somal aberration disrupting Bcl11b gene wasidentified in human T cell acute lymphoblastic

cytes and caused the cells to enter a self-renewalcycle in vivo (19).

The present study thus defines a Bcl11b-driven checkpoint at which T cell progenitorsterminate non-T-lineage potential in order to be-come determined to the T cell lineage (fig. S15).Our finding that Bcl11b up-regulation can betriggered by an extrinsic cue, diminished IL-7,suggests that progression through the DN2-determination step is instructed by environmentalsignals in the thymus. It is quite likely that thereduction in IL-7 signaling is a physiologicalmediator of this step, because the IL-7R is dra-matically down-regulated at the transition fromthe DN2 to the DN3 stage (20). Considering thatBcll1b is thought to be a transcriptional repres-

leukemia cases (18), because the acquisition ofself-renewal capacity is regarded as the first stepin leukemia development. In this context, a sim-ilar outcome was recently observed when Lmo2,

signals in the thymus. It is quite likely that thereduction in IL-7 signaling is a physiologicalmediator of this step, because the IL-7R is dra-matically down-regulated at the transition fromthe DN2 to the DN3 stage (20). Considering thatBcll1b is thought to be a transcriptional repres-sor, we speculate that Bcl11b directly suppressesmyeloid-lineage–associated genes, such as PU.1 orC/EBPa, and that such suppression is critical fordifferentiation toward the T cell fate.

A recent study demonstrated that Bcl11b isexpressed in the T cell–like lymphoid cells oflamprey (21). Because a Bcl11b homolog has notbeen found in animals other than vertebrates(fig. S16), we propose that Bcl11b arose in phy-

a known oncogene, was overexpressed in thymo-cytes and caused the cells to enter a self-renewalcycle in vivo (19).

The present study thus defines a Bcl11b-driven checkpoint at which T cell progenitorsterminate non-T-lineage potential in order to be-come determined to the T cell lineage (fig. S15).Our finding that Bcl11b up-regulation can betriggered by an extrinsic cue, diminished IL-7,suggests that progression through the DN2-determination step is instructed by environmental

2 March 2010; accepted 12 May 2010Published 2 July 201010.1126/science.1188995

2 March 2010; accepted 12 May 2010Published 2 July 201010.1126/science.1188995

19

Repression of the TranscriptionFactor Th-POK by Runx Complexesin Cytotoxic T Cell DevelopmentRuka Setoguchi,1*† Masashi Tachibana,1* Yoshinori Naoe,1* Sawako Muroi,1,2

Kaori Akiyama,1,2 Chieko Tezuka,1 Tsukasa Okuda,3 Ichiro Taniuchi1,2‡

Mouse CD4+CD8+ double-positive (DP) thymocytes differentiate into CD4+ helper-lineagecells upon expression of the transcription factor Th-POK but commit to the CD8+ cytotoxic lineage in itsabsence. We report the redirected differentiation of class I–restricted thymocytes into CD4+CD8–

helper-like T cells upon loss of Runx transcription factor complexes. A Runx-binding sequence within theTh-POK locus acts as a transcriptional silencer that is essential for Th-POK repression and for developmentof CD8+ T cells. Thus, Th-POK expression and genetic programming for T helper cell development areactively inhibited by Runx-dependent silencer activity, allowing for cytotoxic T cell differentiation.Identification of the transcription factors network in CD4 and CD8 lineage choice provides insight intohow distinct T cell subsets are developed for regulating the adaptive immune system.

The peripheral T cell repertoire is formedafter developing thymocytes have under-gone a series of developmental selection

processes. CD4+CD8+ double-positive (DP) thy-

mocytes undergo positive selection through Tcell receptor (TCR) interaction with major histo-compatibility complex (MHC) proteins. Thisgives rise to two functionally distinct subsets:

CD4+CD8– helper and CD4–CD8+ cytotoxic Tcells. Cells expressing MHC class II–restrictedTCRs differentiate into the helper lineage andcease CD8 expression, whereas cells expressingclass I–restricted TCRs differentiate into the cyto-toxic linage and silence CD4 expression (1–3).Recently, gain or loss of function of the BTB/POZdomain–containing zinc finger transcription factor,Th-POK, revealed that its expression is essentialand sufficient for development of helper-lineagecells (4, 5).

1Laboratory for Transcriptional Regulation, RIKEN ResearchCenter for Allergy and Immunology, 1-7-22 Suehiro-cho,Tsurumi-ku, Yokohama 230-0045, Japan. 2Precursory Re-search for Embryonic Science and Techonology (PRESTO),Japan Science and Technology Agency (JST), 4-1-8 Honcho,Kawaguchi, Saitama 332-0012, Japan. 3Kyoto PrefecturalUniversity of Medicine, Kawaramachi-Hirokoji, Kamigyo-ku,Kyoto 602-8566, Japan.

*These authors contributed equally to this work.†Present address: Department of Immunology, University ofWashington, 1959 NE Pacific Street, Seattle, WA 98195–7650, USA.‡To whom correspondence should be addressed. E-mail:taniuchi@rcai.riken.jp

A

B

C

D

0.1

1

10

0.1

1

10

100

cellnumber(x106 )

cellnumber(x104 )

2 3 412 3 41

21

3

32

5

31

4

IL-4 IFN-γ

42

2

20

CD154

class I-restrictedCD4+CD8-

CD4+

CD8+

cont.

R1Δ446/Δ446:R3f/f:Cd4

IIoII+ IIoII+60

38

1

95

0

4 88

3

6

4

93

1

CD8

CD4

65

34

1

95

1

3 67

0

31

9

88

2

mature thymocytes(HSA- TCRβhigh)

mature CD4+CD8-thymocytes

LN TCRβ+ cells

CD4+CD8- TCRβ+splenocytes

R1+/+:R3+/+:Cd4 R1f/f:R3+/+:Cd4 R1+/f:R3f/f:Cd4 R1f/f:R3f/f:Cd4

R1+/+:R3f/f:Cd4

65

34

1

82

11

6 34

62

3 78

1

20 96

1

3

13

85

2 74

2

24 74

17

5

CD8

CD4

R1 Δ446/Δ446:R3+/+:Cd4

R1 Δ446/Δ446:R3f/f:Cd4

R1 +/Δ446:R3f/f:Cd4

Fig. 1. Differentiation of class I–restricted cells into CD4+CD8– helper-likecells by loss of Runx complex function. (A) CD4 and CD8 expression in lymphnode abT cells from mice with indicated genotypes. (B) CD4 and CD8expression in mature thymocytes and LN TCRb+ T cells either in the presence(II+) or absence (II°) of I-A MHC class II molecules. (C) Cell numbers of maturethymocytes and splenocytes showing CD4+CD8– abT cells in class II+ controlmice (lane 1), class II° control mice (lane 2), class II+ Runx1D446/D446:Runx3f/f:Cd4

mice (lane 3), and class II° Runx1D446/D446:Runx3f/f:Cd4 mice (lane 4). Error barsindicate standard deviation. (D) Expression of CD154 at 42 hours after in vitroTCR stimulation of control CD4+, CD8+, and class I–restricted CD4+CD8– cells.Intracellular staining of IL-4 and IFN-g analyzed at 6 hours after re-stimulationof cells that were cultured for 5 days after initial TCR stimulation. Numbers inthe plots in (A), (B), and (D) indicate the percentage of cells in each quadrantor region.

lineage–

2020

Runx transcription factor complexes are com-posed of heterodimers for one of three Runxproteins and their obligatory non–DNA-bindingpartner, Cbfb protein (6). Because of the em-bryonic or neonatal lethality of mice deficient forany of Runx family genes, we used the Cre/loxP-mediated conditional gene inactivation (7) toclarify Runx complex function in silencing of

the Cd4 gene (8) and recently reported that thecombined inactivation of Runx1 and Runx3 inDP thymocytes resulted in a dramatic loss ofCD8+ T cells (9). Runx proteins possess a con-served Val-Trp-Arg-Pro-Tyr (VWRPY) motif atthe C-terminal end, allowing the recruitment ofthe Groucho/TLE co-repressor proteins to theirtarget genes (10, 11). To test whether VWRPY-

dependent repression might be involved in theloss of CD8+ T cells, we introduced the Runx1D446

allele (12) that generates a mutant Runx1 pro-tein lacking the VWRPY motif on a Runx3-deficient background (Runx3f/f:Cd4 mice) (13).A marked reduction of splenic CD8+ T cells inRunx1D446/D446:Runx3f/f:Cd4 mice (Fig. 1A andfig. S1) indicated that VWRPY-dependent repres-

Fig. 2. De-repressionof Th-POK by loss ofRunx complex func-tion. (A and B) Rela-tive Th-POK expressionabundances (normal-ized to hprt) in sortedCD69– DP thymocytes(A) from wild-type (lane1), Runx1f/f:Cd4 (lane 2),Runx1D446/D446 (lane3), Runx3f/f:Cd4 (lane4), Runx1f/f:Runx3f/f:Cd4(lane 5), Runx1D446/D446:Runx3 f/f:Cd4 (lane 6), Cbfb f/f:lck (lane 7), and Cbfb f/f:Cd4(lane 8) mice and in CD4+ and CD8+ peripheral T cells in mice of theindicted genotype (B). One representative result out of three experiments isshown. Lane 9 in (A) indicates Th-POK expression in control CD4+CD8– SP

thymocytes. (C) Relative Th-POK expression abundances after reconstitutionof Runx complex function. Purified CD4+CD8– and CD4+CD8int cells fromCbfb f/f:Cd4 mice were transduced with control retroviral vector (GFP) orvector encoding Cbfb (Cbfb).

A B C

4+ 4+8+ 8+ 8+ 4+8int4+ 4+

Wt R1f/f:Cd4 R3f/f:Cd4 Cbfβf/f:Cd4

0

20

80

40

60

100

120

4+8-GFP Cbfβ GFP Cbfβ

4+8int

0

40

160

80

120

0

20

80

40

60

0

20

40

60500

600

700100

Relative

Th

-PO

Kexpression

(arbitraryunits)

Relative

Th

-PO

Kexpression

(arbitraryunits)

Relative

Th

-PO

Kexpression

(arbitraryunits)

1 2 3 4 5 6 7 8 9

CD4+ CD8+

Splenocytes

CD69-DP CD4+CD8int CD4+SP CD8+SP

Thymocytes

Thymocytes

H

5.4 kb

GFP

GFP

15.5 kb

RV

RV

3.6 kb

10 kb Th-POK

LenepATG TAA

1 2

6.1 kb

Kp E47

a

b

c

d

e

f

g

A

RBS

Δ674 bp

GFP

GFP

GFP

H

562 bp

GFP

593 bp

Ia Ib II III

B

C

DP CD4+SP CD8+SP 4+T 8+T

RBS1

RBS2

UP1

TCRβ

1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

1. Input

2. controlIgG

3. anti-Cbfβ2Ab

GFP

3/3

3/3

1/1

3/6

2/2

2/2

XX

Splenocytes

1/1

Runx sites mutated

387180 148

4500 53 52 8

72852 51 70 50

012 (203)26 (610) 00 21

023 (380)0 9 22 (624) 0

2547161 1226

1 60 83 47 80 36

f

a

b

c

d

e

g

Fig. 3. Identification and characterization of RBSs at the Th-POK locus. (A)The structure of the murine Th-POK locus is shown at the top. Circles rep-resent putative Runx motifs, with those in red indicating evolutionarilyconserved Runx motifs. Black boxes represent exons, and each green barrepresents the signal intensity of an individual oligonucleotide probe in aChIP-on-chip experiment. Blue boxes represent RBSs. The maps for eachreporter transgene construct (Tg-a to Tg-g) are indicated. The restriction sitesshown are Eco47III (E47), EcoRV (RV), HindIII (H), KpnI (Kp), and XhoI (X). (B)ChIP experiment showing binding of Runx complexes to RBS-1 and RBS-2 in

the indicated cell subsets. The regions at 1 kb upstream of exon Ia (UP1) andthe TCRb enhancer (TCRb) were used as negative and positive controls,respectively. (C) Histograms showing the GFP expression in the indicated Tcell subsets from representative transgenic founder for each construct. Thedashed line indicates nontransgenic littermate control. Numbers in thehistogram indicate the percentage of GFP+ cells, and numbers in parenthesisindicate mean fluorescent intensity of GFP in GFP+ cells. The numbers oftransgenic founders expressing GFP among the total transgenic founders areindicated at right.

21

sion by Runx1 was involved in the generation ofCD8+ T cells. Because the leaky CD4–CD8+

subset that escaped Cre-mediated recombination(9) was less apparent in Runx1D446/D446:Runx3f/f:Cd4mice (Fig. 1A), we used these mice for furtheranalyses.

Potentially, the loss of CD8+ T cells couldoccur either by a developmental block of classI–restricted cells or by a redirection of class I–restricted cells toward the CD4+CD8– lineage.To determine whether CD4+CD8– cells that emergein Runx mutant mice are class II–restricted orredirected class I–restricted cells, we crossedRunx1D446/D446:Runx3f/f:Cd4 mice onto a MHCclass II–deficient background (14). Althoughthere was a marked decrease in CD4+CD8– Tcell numbers in control class II–deficient mice,the predominance of CD4+CD8– T cells persistedin class II–deficient Runx1D446/D446:Runx3f/f:Cd4mice in both the thymus and the periphery(Fig. 1, B and C). These results indicated thatthe absence of Runx complexes forced the ma-jority of class I–restricted cells to differentiateinto CD4+CD8– T cells.

We next examined the functional propertiesof these CD4+CD8– cells. One of the character-istic features of CD4+ helper-lineage T cells is theearly induction of CD154, the ligand for CD40,after TCR stimulation (15) and the productionof interleukin-4 (IL-4). These were observed incontrol CD4+ T cells as well as in class I–restrictedCD4+CD8– cells, but not in control CD8+ T cells(Fig. 1D). In contrast, although high interferon-g(IFN-g) production was detected in control CD8+

T cells, it was absent in both wild-type CD4+ Tcells and in class I–restricted CD4+CD8– cells

(Fig. 1D). We conclude from these results thatclass I–restricted CD4+CD8– cells that developin Runx mutant animals are functionally helper-like T cells.

Because ectopic expression of Th-POK hasbeen shown to redirect class I–restricted cells tobecome CD4+CD8– cells (4, 5), we measuredthe expression of Th-POK in several Runx mu-tant mice, including a strain in which the Cbf bgene is conditionally inactivated by either a Lck-Cre or a Cd4-Cre transgene (13). Consistentwith a previous report (4), Th-POK expressionwas not detected in control CD69– DP thymo-cytes. In contrast, a 40-fold increase in Th-POKtranscript abundances was detected in CD69–

DP thymocytes in which Runx complexes weredisrupted either by combined Runx1 mutationswith a Runx3 deficiency or by loss of Cbfbprotein (Cbfbf/f:Lck mice) (Fig. 2A). A modestTh-POK de-repression by inactivation of Runx1alone indicated a redundant function of Runx3in the repression of Th-POK.

Although Th-POK mRNA was undetectablein control CD8+ T cells and in CD8+ T cellsdeficient for Runx1 or Runx3, it was present inCbfb-deficient CD4+CD8int T cells (Fig. 2B)that still developed in Cbf b f/f:Cd4 mice becauseof the slow turnover of Cbfb protein after inac-tivation of the Cbf b gene (13). We therefore nextexamined whether Th-POK repression could berestored in these CD4+CD8int cells upon re-expression of Cbfb protein. Purified CD4+CD8–

and CD4+CD8int cells were transduced with aretroviral vector encoding Cbfb or with an emptyvector control. In these experiments, expres-sion of Th-POK was markedly reduced upon re-

expression of Cbfb in CD4+CD8int cells, withno detectable effect in CD4+CD8– cells (Fig. 2C).These results suggest that Runx-mediated Th-POKrepression operates in peripheral CD8+ T cells.

To understand mechanisms underlying Runx-mediated repression of Th-POK, we examinedwhether Runx complexes directly associate withthe Th-POK locus. Using a ChIP-on-chip (ChIPindicates chromatin immunoprecipitation) ap-proach with an antibody against Cbfb2, we de-tected two regions occupied by Runx complexeswithin the Th-POK locus. Distal and proximalRunx-binding sequences (RBS-1 and RBS-2, re-spectively) are located ~3.1 kb upstream and~7.4 kb downstream of exon Ia (Fig. 3A) and con-tain two or one conserved Runx motifs, respec-tively (Fig. 3A and figs. S2 and S3). By usingChIP analysis in T cell subsets, we confirmed anassociation between Runx complexes and thesetwo regions (Fig. 3B). However, binding of Runxcomplexes to RBS-1 and RBS-2 was detected inboth Th-POK–expressing and nonexpressingcells, revealing that the binding of Runx com-plexes to these regions did not correspond withTh-POK repression.

To better understand the functional activ-ities of RBS-1 and RBS-2 in light of these re-sults, we performed transgenic reporter assays.A 15.5-kb genomic fragment encompassingthe RBSs and exons Ia and Ib was linked to agreen fluorescent protein (GFP) reporter trans-gene cassette (Tg-a in Fig. 3A). In all trans-genic mouse founders obtained with Tg-a,GFP expression was first detected in post-selection CD4+CD8int thymocytes and was up-regulated in CD4+ SP thymocytes, remaining

Fig. 4. Essential require-ment of the Th-POK si-lencer for developmentof CD8+ T cells. (A) Sche-matic structure of theTh-POK locus and tar-geted alleles Th-POKSD,Th-POKGFP, and Th-POKGFP:SD. Exons and loxPsequences are indicatedas black boxes and blacktriangles, respectively.(B) CD4 and CD8 ex-pression in lymph nodeabT cells from wild-type (+/+), Th-POKSD

heterozygous (SD/+),and Th-POK hemizygoustransgenic (Th-POK Tg)mice. (C) Relative Th-POKexpression abundancesin sorted CD69– DPthymocytes showing de-repression of Th-POKupon deletion of the Th-POK silencer. (D) GFPexpression from the Th-POKGFP and Th-POKGFP:SD alleles in indicated thymocyte subsets. Dashed and bold lines indicate GFP expression in control mice and Th-POK+/GFP (+/GFP) or Th-POK+/GFP:SD (+/GFP:SD) mice, respectively. The numbers in parenthesis indicate mean fluorescent intensity of GFP in total CD69– DP thymocytes.

CD69-CD4+CD8+ CD4+CD8- CD4-CD8+

(6.79)

(17.28)

64.3 0.7

34.30.7

96.4 0.9

0.81.9

93.2 1.8

3.41.5

A B

DC

TCRβ+ LN cells+/+ SΔ/+ Th-POK Tg

CD8

CD4

CD69- DP thymocytes

+/+ SΔ/+ Cbfβf/f:Lck

+/GFP

+/GFP:SΔ

GFP

Th-POK

Th-POKSΔ

Th-POKGFP

Th-POKGFP:SΔ

ATG TAAIa IbRBS-1 RBS-2

GFP

GFP

Relative

Th

-PO

Kexpression

(arbitraryunits)

0

10

20

40

30

22

high in splenic CD4+ T cells, whereas it wasalmost undetectable in splenic CD8+ T cells(Fig. 3C). The 15.5-kb fragment thus containsthe major cis-regulatory regions that direct ex-pression of Th-POK in the helper lineage.

To further narrow down the critical Th-POKregulatory regions, we deleted either 5′ or 3′ se-quences as well as RBS-1 from the 15.5-kbfragment. Whereas RBS-2 (fig. S3) was found tobe required for positive transcriptional regulatoryactivity (as in the Tg-c and Tg-d constructs),deletion of a 674-bp fragment of RBS-1 (Tg-b)resulted in GFP expression both in CD4+ helper-lineage and in CD8+ cytotoxic-lineage cells,indicating that RBS-1 is a transcriptional silencerrequired to repress the reporter gene in CD8lineage cells. Efficient repression of GFP in CD8+

T cells by a 562-bp fragment of RBS-1 (fig. S2)in the context of Tg-e construct required Runxmotifs (Tg-f and Tg-g) (Fig. 3C), consistentwith Runx-dependent activity of RBS-1 silencer.

To examine the physiological function of theRBS-1 silencer, we deleted the 674-bp KpnI-Eco47III sequences from the Th-POK locus byhomologous recombination in embryonic stem(ES) cells (Fig. 4A and fig. S4). Deletion ofRBS-1 from one Th-POK allele led to the lossof peripheral CD8+ T cells (Fig. 4B) and to theTh-POK de-repression in CD69– DP thymo-cytes (Fig. 4C). We further investigated Th-POKde-repression by using mice in which the codingsequence for Th-POK was replaced with the gfpgene (Th-POKGFP locus). GFP expression inmice heterozygous for Th-POKGFP allows us toexamine expression of Th-POK at the single-celllevel. Although GFP expression from the Th-POKGFP locus was not detected in CD69– DPthymocytes, deletion of RBS-1 (Th-POKGFP:SD

locus in Fig. 4A) resulted in uniform de-repressionof GFP in CD69– DP thymocytes, followed by

high GFP expression in both helper- andcytotoxic-lineage mature thymocytes (Fig. 4D).

Our results reveal that helper lineage–specificexpression of Th-POK is regulated by the RBS-1silencer, whose activity depends on binding ofRunx complexes. We therefore refer to RBS-1 asthe Th-POK silencer (fig. S5). The association ofRunx complexes with the Th-POK silencer incells expressing Th-POK indicates that specific-ity of silencer activity is not regulated at the levelof Runx complex binding. Additional moleculesthat interact with Runx factors bound to the Th-POK silencer may therefore have a central rolein regulating Th-POK silencer activity.

The antagonistic interplay between primarylineage-determining factors is often observed whentwo opposing fates are induced in progenitorcells (16, 17). Th-POK was recently described asan inhibitor of Runx-dependent Cd4 silencer ac-tivity (18), consistent with an antagonistic inter-play between these two factors. Identification ofTh-POK and Runx complex target genes willhelp to further unravel the transcription factorsnetwork regulating lineage specification of DPthymocytes.

Uniform de-repression of Th-POK in CD69–

DP thymocytes upon deletion of the Th-POK si-lencer indicates that silencer-mediated Th-POKrepression operates in all pre-selection DP thymo-cytes. It is therefore possible that TCR signalsafter engagement of MHC class II result inantagonism of Th-POK silencer activity andthus induce Th-POK expression. Given that sus-tained class II–specific TCR signals are thoughtto be necessary for specification of the helperlineage (19–21), reversal of silencer-mediatedTh-POK repression may require class II–specificTCR signals during a specified time window.Our results suggest that a mechanism regulatingTh-POK silencer activity acts as a sensor to dis-

17 October 2007; accepted 20 December 2007Published 8 February 200810.1126/science.1151844

Th-POK silencer activity acts as a sensor to dis-tinguish qualitative differences in TCR signal-ing. Further studies on the regulatory pathwaysof Th-POK repression will shed light on howsignals initiated by external stimuli are con-

d i i i h ll l

tinguish qualitative differences in TCR signal-ing. Further studies on the regulatory pathwaysof Th-POK repression will shed light on howsignals initiated by external stimuli are con-verted into genetic programs in the cell nucleus.

6. Y. Ito, Genes Cells 4, 685 (1999).7. H. Gu, Y. R. Zou, K. Rajewsky, Cell 73, 1155 (1993).8. I. Taniuchi et al., Cell 111, 621 (2002).9. T. Egawa, R. E. Tillman, Y. Naoe, I. Taniuchi, D. R. Littman,

J. Exp. Med. 204, 1945 (2007).10. D. Levanon et al., Proc. Natl. Acad. Sci. U.S.A. 95, 11590

(1998).11. B. D. Aronson, A. L. Fisher, K. Blechman, M. Caudy,

J. P. Gergen, Mol. Cell. Biol. 17, 5581 (1997).12. M. Nishimura et al., Blood 103, 562 (2004).13. Y. Naoe et al., J. Exp. Med. 204, 1749 (2007).14. M. J. Grusby, R. S. Johnson, V. E. Papaioannou,

L. H. Glimcher, Science 253, 1417 (1991).15. M. Roy, T. Waldschmidt, A. Aruffo, J. A. Ledbetter,

R. J. Noelle, J. Immunol. 151, 2497 (1993).16. E. V. Rothenberg, Nat. Immunol. 8, 441 (2007).17. S. H. Orkin, Nat. Rev. Genet. 1, 57 (2000).18. K. F. Wildt et al., J. Immunol. 179, 4405 (2007).19. K. Yasutomo, C. Doyle, L. Miele, C. Fuchs, R. N. Germain,

Nature 404, 506 (2000).20. X. Liu, R. Bosselut, Nat. Immunol. 5, 280 (2004).21. S. D. Sarafova et al., Immunity 23, 75 (2005).22. We are grateful to D. R. Littman and W. Ellmeier for

critical reading of the manuscript. This work was supportedby grants from PRESTO, JST. The accession number formouse Th-POK silencer is EU371956 in GenBank.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/319/5864/822/DC1Material and MethodsFigs. S1 to S5References

References and Notes1. A. Singer, R. Bosselut, Adv. Immunol. 83, 91 (2004).2. W. Ellmeier, S. Sawada, D. R. Littman, Annu. Rev. Immunol.

17, 523 (1999).3. T. K. Starr, S. C. Jameson, K. A. Hogquist, Annu. Rev.

Immunol. 21, 139 (2003).

References and Notes1. A. Singer, R. Bosselut, Adv. Immunol. 83, 91 (2004).2. W. Ellmeier, S. Sawada, D. R. Littman, Annu. Rev. Immunol.

17, 523 (1999).3. T. K. Starr, S. C. Jameson, K. A. Hogquist, Annu. Rev.

Immunol. 21, 139 (2003).4. X. He et al., Nature 433, 826 (2005).5. G. Sun et al., Nat. Immunol. 6, 373 (2005).6. Y. Ito, Genes Cells 4, 685 (1999).7. H. Gu, Y. R. Zou, K. Rajewsky, Cell 73, 1155 (1993).8. I. Taniuchi et al., Cell 111, 621 (2002).9. T. Egawa, R. E. Tillman, Y. Naoe, I. Taniuchi, D. R. Littman,

J. Exp. Med. 204, 1945 (2007).10. D. Levanon et al., Proc. Natl. Acad. Sci. U.S.A. 95, 11590

(1998).11. B. D. Aronson, A. L. Fisher, K. Blechman, M. Caudy,

J. P. Gergen, Mol. Cell. Biol. 17, 5581 (1997).12. M. Nishimura et al., Blood 103, 562 (2004).13. Y. Naoe et al., J. Exp. Med. 204, 1749 (2007).14. M. J. Grusby, R. S. Johnson, V. E. Papaioannou,

L. H. Glimcher, Science 253, 1417 (1991).15. M. Roy, T. Waldschmidt, A. Aruffo, J. A. Ledbetter,

R. J. Noelle, J. Immunol. 151, 2497 (1993).16. E. V. Rothenberg, Nat. Immunol. 8, 441 (2007).17. S. H. Orkin, Nat. Rev. Genet. 1, 57 (2000).18. K. F. Wildt et al., J. Immunol. 179, 4405 (2007).19. K. Yasutomo, C. Doyle, L. Miele, C. Fuchs, R. N. Germain,

Nature 404, 506 (2000).20. X. Liu, R. Bosselut, Nat. Immunol. 5, 280 (2004).21. S. D. Sarafova et al., Immunity 23, 75 (2005).22. We are grateful to D. R. Littman and W. Ellmeier for

critical reading of the manuscript. This work was supportedby grants from PRESTO, JST. The accession number formouse Th-POK silencer is EU371956 in GenBank.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/319/5864/822/DC1Material and MethodsFigs. S1 to S5References

23

24

25

Preferential Generation of Follicular B Helper

T Cells from Foxp3+ T Cells in Gut Peyer’s

PatchesMasayuki Tsuji,1* Noriko Komatsu,2* Shimpei Kawamoto,1,4* Keiichiro Suzuki,1 Osami Kanagawa,3 Tasuku Honjo,4

Shohei Hori,2† Sidonia Fagarasan1†

Most of the immunoglobulin A (IgA) in the gut is generated by B cells in the germinal centers of Peyer’s patches through a

process that requires the presence of CD4+ follicular B helper T (TFH) cells. The nature of these TFH cells in Peyer’s patches

has been elusive. Here, we demonstrate that suppressive Foxp3+CD4+ T cells can differentiate into TFH cells in mouse

Peyer’s patches. The conversion of Foxp3+ T cells into TFH cells requires the loss of Foxp3 expression and subsequent

interaction with B cells. Thus, environmental cues present in gut Peyer’s patches promote the selective differentiation of

distinct helper T cell subsets, such as TFH cells.

1 Laboratory for Mucosal Immunity, RIKEN, Yokohama 1-7-22, Tsurumi, Yokohama, 230-0045, Japan. 2 Research Unit for Immune Homeostasis, RIKEN,

Yokohama 1-7-22, Tsurumi, Yokohama, 230-0045, Japan. 3 Laboratory for Autoimmune Regulation, Research Center for Allergy and Immunology, RIKEN,

Yokohama 1-7-22, Tsurumi, Yokohama, 230-0045, Japan. 4Department of Immunology and Genomic Medicine, Kyoto University, Graduate School of Medicine,

Sakyo-ku, Kyoto 606-8501, Japan.

*These authors contributed equally to this work. †To whom correspondence should be addressed. E-mail: shohei@rcai.riken.jp (S.H.); sidonia-f@rcai.riken.jp (S.F.)

SCIENCE | VOL 323 | 13 MARCH 2009 | PAGES 1488-1492

A Critical Role for the Innate Immune Signaling

Molecule IRAK-4 in T Cell ActivationNobutaka Suzuki,1 Shinobu Suzuki,1* Douglas G. Millar,2† Midori Unno,1 Hiromitsu Hara,1 Thomas Calzascia,2 Sho

Yamasaki,1 Tadashi Yokosuka,1 Nien-Jung Chen,3 Alisha R. Elford,2 Jun-ichiro Suzuki,4‡ Arata Takeuchi,1 Christine

Mirtsos,3 Denis Bouchard,3 Pamela S. Ohashi,2 Wen-Chen Yeh,3§ǁ Takashi Saito1ǁIRAK-4 is a protein kinase that is pivotal in mediating signals for innate immune responses. Here, we report that IRAK-4

signaling is also essential for eliciting adaptive immune responses. Thus, in the absence of IRAK-4, in vivo T cell responses

were significantly impaired. Upon T cell receptor stimulation, IRAK-4 is recruited to T cell lipid rafts, where it induces

downstream signals, including protein kinase CΘ activation through the association with Zap70. This signaling pathway was

found to be required for optimal activation of nuclear factor κB. Our findings suggest that T cells use this critical regulator ofinnate immunity for the development of acquired immunity, suggesting that IRAK-4 may be involved in direct signal cross

talk between the two systems.

1Laboratory forCellSignaling,RIKENResearchCenter forAllergyand Immunology, 1-7-22Suehiro-cho,Tsurumi-ku,YokohamaCity,Kanagawa230-0045, Japan.2 Institute for Breast Cancer Research, University Health Network, 620 University Avenue, Suite 706, Toronto, Ontario M5G 2C1, Canada. 3Advanced Medical

Discovery Institute, University Health Network and Department of Medical Biophysics, University of Toronto, 620 University Avenue, Suite 706, Toronto, Ontario

M5G 2C1, Canada. 4Department of Molecular Genetics, Chiba University Graduate School of Medicine, 1-8-1 Inohana, Chuo-ku, Chiba City, Chiba 260-8670,

Japan.

*Present address: Nihon Schering K. K. Research Center, BMA 3F, 1-5-5, Minatojima-minami-machi, Chuo-ku, Kobe 650-0047, Japan. † Present address:

Faculty of Life Sciences, University of Manchester, C-2259 Michael Smith Building, Oxford Road, Manchester M13 9PT, UK.

‡ Present address: Faculty of Pharmacy, Musashino University, 1-1-20 Shin-machi, Nishitokyo-shi, Tokyo 202-8585, Japan. § Present address: Amgen San

Francisco, 1120 Veterans Boulevard, South San Francisco, CA 94080, USA.

ǁTo whom correspondence should be addressed. E-mail: saito@rcai.riken.jp (T.S.); wyeh@uhnres.utoronto.ca (W.-C.Y.)

SCIENCE | VOL 311 | 31 MARCH 2006 | PAGES 1927-1932

26

Autoimmune Disease and Impaired Uptake of

Apoptotic Cells in MFG-E8–Deficient MiceRikinari Hanayama,1 Masato Tanaka,1,4 Kay Miyasaka,1,5 Katsuyuki Aozasa,2 Masato Koike,3 Yasuo Uchiyama,3

Shigekazu Nagata1,5,6*

Apoptotic cells expose phosphatidylserine and are swiftly engulfed by macrophages. Milk fat globule epidermal growth factor

(EGF) factor 8 (MFG-E8 ) is a protein that binds to apoptotic cells by recognizing phosphatidylserine and that enhances

the engulfment of apoptotic cells by macrophages. We report that tingible body macrophages in the germinal centers

of the spleen and lymph nodes strongly express MFG-E8. Many apoptotic lymphocytes were found on the MFG-E8—/—tingible body macrophages, but they were not efficiently engulfed. The MFG-E8—/— mice developed splenomegaly, with the

formation of numerous germinal centers, and suffered from glomerulonephritis as a result of autoantibody production. These

data demonstrate that MFG-E8 has a critical role in removing apoptotic B cells in the germinal centers and that its failure

can lead to autoimmune diseases.

1Department of Genetics, 2Department of Pathology, and 3Department of Cell Biology and Neuroscience, Osaka University Medical School, 2-2 Yamada-oka,

Suita, Osaka 565-0871, Japan. 4 Laboratory for Innate Cellular Immunity, RIKEN Research Center for Allergy and Immunology, Yokohama, Kanagawa 230-

0045, Japan. 5 Laboratory of Genetics, Integrated Biology Laboratories, Graduate School of Frontier Biosciences, Osaka University, Osaka 565-0871, Japan.6Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation, Osaka 565-0871, Japan.

*To whom correspondence should be addressed. E-mail: nagata@genetic.med.osaka-u.ac.jp.

SCIENCE | VOL 304 | 21 MAY 2004 | PAGES 1147-1150

Innate Antiviral Responses by Means of TLR7-

Mediated Recognition of Single-Stranded RNASandra S. Diebold,1 Tsuneyasu Kaisho,2,3 Hiroaki Hemmi,2 Shizuo Akira,2,4 Caetano Reis e Sousa1*

Interferons (IFNs) are critical for protection from viral infection, but the pathways linking virus recognition to IFN induction

remain poorly understood. Plasmacytoid dendritic cells produce vast amounts of IFN-α in response to the wild-type influenza

virus. Here, we show that this requires endosomal recognition of influenza genomic RNA and signaling by means of Toll-like

receptor 7 (TLR7) and MyD88. Single-stranded RNA (ssRNA) molecules of nonviral origin also induce TLR7-dependent

production of inflammatory cytokines. These results identify ssRNA as a ligand for TLR7 and suggest that cells of the innate

immune system sense endosomal ssRNA to detect infection by RNA viruses.

1 Immunobiology Laboratory, Cancer Research UK, London Research Institute, London WC2A 3PX, UK. 2 Department of Host Defense, Research Institute

for Microbial Diseases, Osaka University, Yamadaoka 3-1, Suita City, Osaka 565-0871, Japan. 3 RIKEN Research Center for Allergy and Immunology, 1-7-22

Suehiro-cho, Tsurumi-ku, Yokohama City, Kanagawa 230-0045, Japan. 4Akira Innate Immunity Project, ERATO, Japan Science and Technology Corporation,

Osaka 565-0871, Japan.

*To whom correspondence should be addressed. E-mail: caetano@cancer.org.uk

SCIENCE | VOL 303 | 5 MARCH 2004 | PAGES 1529-1531

27

28

Critical Roles of Activation-Induced Cytidine

Deaminase in the Homeostasis of Gut FloraSidonia Fagarasan,1,2*Masamichi Muramatsu,1* Keiichiro Suzuki,1 Hitoshi Nagaoka,1 Hiroshi Hiai,3 Tasuku Honjo1†

Activation-induced cytidine deaminase (AID) plays an essential role in class switch recombination (CSR) and somatic

hypermutation (SHM) of immunoglobulin genes. We report here that deficiency in AID results in the development of

hyperplasia of isolated lymphoid follicles (ILFs) associated with a 100-fold expansion of anaerobic flora in the small intestine.

Reduction of bacterial flora by antibiotic treatment of AID—/— mice abolished ILF hyperplasia as well as the germinal center

enlargement seen in secondary lymphoid tissues. Because an inability to switch to immunoglobulin A on its own does

not lead to a similar phenotype, these results suggest that SHM of ILF B cells plays a critical role in regulating intestinal

microflora.

1 Department of Medical Chemistry, 3 Department of Pathology and Biology of Disease, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku,

Kyoto 606-8501, Japan. 2RIKEN Research Center for Allergy and Immunology, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan.

*These authors contributed equally to this work. †To whom correspondence should be addressed. E-mail: honjo@mfour.med.kyoto-u.ac.jp

SCIENCE | VOL 298 | 15 NOVEMBER 2002 | PAGES 1424-1427

29

There’s only oneGalileo Galilei

Career advice I Job postings I Job Alerts I Career Forum I Crafting resumes/CVs I Preparing for interviews

For your career in science, there’s only one

Careers

ScienceCareers.org

orn in 1564, Galileo Galilei once contemplated a career in the priesthood. It’s perhaps fortunatefor science that upon the urging of his father, he instead decided to enroll at the University of

Pisa. His career in science began with medicine and from there he subsequently went on to becomea philosopher, physicist, mathematician, and astronomer, for which he is perhaps best known. Hisastronomical observations and subsequent improvements to telescopes built his reputation as aleading scientist of his time, but also led him to probe subject matter counter to prevailing dogma.His expressed views on the Earth’s movement around the sun caused him to be declared suspectof heresy, which for some time led to a ban on the reprinting of his works.

Galileo’s career changed science for all of us and he was without doubt a leading light in thescientific revolution, which is perhaps why Albert Einstein called him the father of modern science.

Want to challenge the status quo and make the Earth move? At Sciencewe are here to help youin your own scientific career with expert career advice, forums, job postings, and more— all for free.For your career in science, there’s only one Science. Visit Science today at ScienceCareers.org.

B

RIKEN Research Center for Allergy and Immunology

http://www.rcai.riken.jp/english/

1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama City, Kanagawa, 230-0045, Japan

Tel : 045-503-9111

Fax : 045-503-9113

Email : yokohama@riken.jp

RIKEN 2012-056 RIKEN—Celebrating 100 Years in 2017