Clinical ImmunologyVol. 102, No. 3, March, pp. 237–248, 2002

Colitis-Related Public T Cells Are Selected in the Colonic LaminaPropria of IL-10-Deficient Mice

Ichiro Takahashi,* Jennifer Matsuda,† Laurent Gapin,† Hilde DeWinter,† Yasuyuki Kai,*Hiroshi Tamagawa,* Mitchell Kronenberg,† and Hiroshi Kiyono*

*Department of Mucosal Immunology, Research Institute for Microbial Diseases, Osaka University, 3-1 Yamadaoka, Suita-Osaka565-0871, Japan; and †Division of Developmental Immunology, The La Jolla Institute for Allergy and Immunology,

IL-10 is an important regulatory cytokine in the mu-cosal immune system, as supported by the fact that micedeficient in IL-10 spontaneously develop Crohn’s dis-ease-like colitis. An aberrant, Th1-driven CD4� T-cell re-sponse to enteric bacteria seems to be important in thepathogenesis of this murine colitis. However, no specificbacteria or bacterial products have been identified, andwhether the colitis is mediated by the activation of CD4�

T cells that recognize specific peptide–MHC complexesis controversial. In this study, we analyzed the TCR�chain complementarity determining region 3 lengthspectratype of colonic CD4� T cells isolated from dis-eased IL-10-deficient mice by using the Immunoscopetechnique. Screening of the diseased interleukin-10-de-ficient mice resulted in a restricted clonotype in TCR V�13 and 14 subfamilies of colonic CD4� T cells. In contrast,a Gaussian distribution of clonotype of individual TCRV� subsets was observed in CD4� T cells from the periph-eral lymphoid tissues. Although individual variability inthe disease-related response was also noted in other IL-10-deficient mice maintained in La Jolla and Osaka, per-haps because of different stages of the disease, geneticbackground, or the housing environment, colitis-relatedpublic clones seemed to be shared in all the diseasedmice tested. To address whether public clones were in-volved, we determined the DNA sequence of the clones.Public motifs were shared in colonic CD4� T cells fromdifferent background interleukin-10-deficient mice withcolitis. The frequently found motifs were SXDWG andSATGNYAEQ. These motifs were not seen in the periph-eral lymphoid tissues of diseased mice as well as thecolon of nondiseased mice. Thus, the common motif maybe related to a public gut-derived antigen, which couldbe important for the development of pathogenic CD4� Tcells in this inflammatory bowel disease (IBD) model.The selection of V�-J� usage is perhaps stochastic inindividual mice; however, the epigenetic generation ofSXDWG motif by the recombination machinery and se-lection for this motif in the gut environment could beimportant for triggering IBD. © 2002 Elsevier Science (USA)

237

INTRODUCTION

Inflammatory bowel disease (IBD)1 is a chronic, pos-sibly noninfectious, inflammation limited to the largebowel (ulcerative colitis) or occurring anywhere alongthe gastrointestinal tract (Crohn’s disease); the formeris a relatively superficial, ulcerative inflammation,while the latter is a transmural, granulomatous in-flammation. The major hypothesis concerning thepathogenesis of IBD is that the diseases are due to anabnormal and uncontrolled immune response to one ormore normally occurring gut constituents (1–3). Thishypothesis is based on the concept that immune ho-meostasis in the mucosal immune system relies on adelicate balance between the ability to react to poten-tial gut pathogens and the ability not to react withcommon ubiquitous gut constituents (food antigensand bacterial flora). If the second arm of the balancemechanism, known as oral tolerance, is disturbed, andthe mucosal system reacts with one or more commonantigens in the mucosal environment, a condition is setup leading to chronic enterocolitis.

Evaluation of T-cell receptor diversity and clonality,in subjects who are going to be receiving vaccine orsuffering from T-cell-mediated autoimmune disease, iscritical for fully understanding the molecular basis ofvaccine development and pathogenesis of autoimmunedisease. In this regard, we and others have been study-ing the pathogenesis of IBD produced in TCR � chain-deficient mice (4, 5). It was shown that Th2-biasedCD4� T cells expressing a TCR �-chain homodimerwith a limited number of TCR V� subfamilies wereincreased in the diseased mice. Furthermore, clonotypeanalysis of the �� T cells isolated from the diseasedcolon by using PCR-based single-strand conformationpolymorphism (PCR-SSCP) analysis exhibited mono-

1 Abbreviations used: aa, amino acid; CDR3, complementarity de-termining region 3; IBD, inflammatory bowel disease; IEL, intraepi-thelial lymphocyte; KO, knock-out; LP, lamina propria; MLN, mes-enteric lymph nodes; PCR-SSCP, PCR single-strand conformation

10355 Science Center Drive

doi:10.1006/clim.2001.5166, available online at http://www.idealibra

n Diego, California 92121

om on

, Sa

ry.c

polymorphism.

1521-6616/02 $35.00© 2002 Elsevier Science (USA)

All rights reserved.

clonal or oligoclonal expansions. Thus, it was stronglysuggested that the pathogenic �� T cells were clonallyaccumulated upon stimulation with gut-derived spe-cific antigens. In addition to this PCR–SSCP analysis,another sensitive and elegant tool for the evaluation ofTCR repertoire diversity and clonal expansion is the“Immunoscope,” originally developed by Kourilsky’slaboratory at the Pasteur Institute (6–8).

IL-10 is well known as an anti-inflammatory cyto-kine that down-regulates macrophage and Th 1 cellfunction (9). The importance of IL-10 as a regulatorycytokine in the mucosal immune system is supportedalso by the evidence that mice genetically deficient forIL-10 spontaneously develop human Crohn’s disease-like IBD with age (10). Another important finding con-cerning the regulatory role of IL-10 was clearly shownby the finding that cotransfer of a CD4�, CD45 RBhigh

T-cell population isolated from IL-2 promoter-drivenIL-10 transgenic mice, with the same pathogenic sub-set of T cells isolated from normal mice, significantlyprevented the onset and reduced the severity of thecolitis in the recipient SCID mice (11). Additional exper-iments by Powrie and co-workers have shown that theprevention of colitis by CD4�, CD45 RBlow T cells requiresIL-10 (12). The chronic enterocolitis which developed inthe IL-10-deficient mice is mediated by CD4� T cellsproducing IFN-� (13). In addition, adoptive transfer ofTh1-biased CD4� T cells into SCID mice caused the coli-tis (14). It was also noted that germ-free IL-10 knock-out(KO) mice have no colitis, and resident enteric bacteriaare necessary for the development of colitis in the mice(15). Although the involvement of the T cells for thedevelopment of colitis is clear, it is not known whetherparticular antigens from bacteria or other microorgan-isms play a role in disease induction.

To this end, several major questions in the IL-10-deficient murine colitis model remain to be elucidated:(i) Do the TCR repertoires of the mice with IBD sharepublic specificities? (ii) Are the TCR repertoires uniquein the colon? (iii) Where does the activation of T cellsoccur? To address these questions, an analysis of theTCR repertoire in murine colitis provides a powerfulopportunity to characterize the pathogenic T-cell pop-ulation (antigen-specificity, publicity, and distribution)in a manner not possible in studies on human patients(16). In this study we have analyzed TCR diversity ofthe pathogenic CD4� T cells in the colon of IL-10-deficient mice to assess the dissemination of the Ag-reactive T cells and the potential role of common bac-terial constituents in colitis pathogenesis.

MATERIALS AND METHODS

Mice

IL-10-deficient mice with two different backgrounds,C57BL/6 � 129Ola mixed or C57BL/6, were purchased

from The Jackson Laboratory (Bar Harbor, ME) andmaintained at the La Jolla Institute for Allergy andImmunology and Osaka University, respectively. TheC57BL/6 � 129Ola mixed-background IL-10-deficientmice were housed under specific pathogen-free condi-tions, whereas the C57BL/6-background IL-10-defi-cient mice were maintained under conventional condi-tions free of specific pathogens. All procedures wereapproved by the Animal Care and Utilization Commit-tees at the respective institutions. Ages of the mice atthe time of analysis are given under Results and Dis-cussion.

Isolation of Lamina Propria Lymphocytes

Mucosal lymphocytes were isolated according to apublished method (4, 5). Briefly, the large intestinewas removed and carefully cleaned from the mesen-tery, opened longitudinally, and flushed of fecal con-tent. The removed intestine was cut into 0.5-cm pieces,transferred into 50-ml Erlenmeyer flasks, and shakenthree times at 200 rpm for 20 min each time at 37°C inHBSS without Ca2� or Mg2� and containing 1 mM DTT(Sigma, St. Louis, MO) to remove intestinal epithelialcells and intraepithelial lymphocytes. To isolate lam-ina propria lymphocytes, the remaining intestinal tis-sue was minced, transferred to 20-ml scintillation glassvials, and shaken for 30 min at 37°C in RPMI supple-mented with 2% FBS containing collagenase at 0.5mg/ml (Nitta Gelatin, Osaka, Japan). The cell suspen-sions were collected and passed through a glass fiber.The pelleted cells were passed over a Percoll gradientand processed as described (4). For purification of co-lonic CD4� T cells, the mononuclear cells were sortedwith a FACSVantage flow cytometer (Becton Dickin-son) after double labeling. PE-labeled anti-CD4 (L3T4)and FITC-labeled anti-TCR� (H57-597) (PharMingen,San Diego, CA) were used in combination.

mRNA Extraction and cDNA Synthesis

mRNA was prepared using the QuickPrep micromRNA purification kit (Pharmacia). The full quantityof mRNA obtained was used for single-strand cDNAsynthesis. The mRNA was denatured for 10 min at70°C and then incubated with oligo(dT)15 (5 �M),dNTPs (1 �M each), RNasin (40 U; Promega, Madison,WI), and AMV reverse transcriptase (2 U; Roche,Mannheim, Germany) in the supplier’s buffer at 42°Cfor 1 h, followed by incubation at 72°C for 10 min.

PCR Amplifications, Primer Extensions, and DataAnalysis

PCR amplification was conducted in 50 �l using 1/40of the cDNA with 2 U of Taq polymerase (Perkin–

238 TAKAHASHI ET AL.

Elmer, Foster City, CA) in the supplier’s buffer. Senseoligonucleotides specific for each of the 23 V� genesand antisense oligonucleotides for C� have been de-scribed (16). Forty cycles of PCR were conducted in aPerkin–Elmer 9600 Automate, with each cycle consist-ing of 94°C for 45 s, 60°C for 45 s, and 72°C for 45 s.Each PCR product was then used as a template forextension, or run-off, reactions with oligonucleotideslabeled with a fluorescent tag. Fluorescent primersused included an internal C� primer and primers spe-cific for each of the 12 J�s (16). The fluorescent run-offproducts generated varied in size depending oncomplementarity determining region 3 (CDR3) length.Run-off products were subjected to capillary electro-phoresis in an automated DNA sequencer (Applied Bio-systems, Foster City, CA), and CDR3 size distributionand signal intensities were analyzed with GeneScansoftware (Perkin–Elmer). The patterns observed con-tained up to eight size peaks, each spaced by threenucleotides, corresponding to in-frame transcripts. Thearea under each peak was proportional to the quantityof TCR transcript of the corresponding CDR3 length inthe sample.

Cloning and Sequencing of Selected V�-J�Rearrangements

Amplified PCR products were diluted 1/100 withH2O, and 1 �l was used as a template for amplificationof selected V�-J� rearrangements. PCR was performedwith the reagents and quantities described above, us-ing a sense oligonucleotide specific for the V�-chain ofinterest and an antisense oligonucleotide specific forthe J�-chain of interest. Twenty-five cycles, each at94°C for 45 s, 60°C for 45 s, and 72°C for 45 s, werecompleted in a GeneAmp 9600 PCR System (Perkin–Elmer). PCR products were analyzed on a 2% agarosegel stained with ethidium bromide to monitor the qual-ity and quantity of the reaction products. Each V�-J�amplified product was then shotgun cloned with thepGEM-T Easy TA cloning kit (Promega, Madison, WI).Resulting colonies were randomly selected, and cul-tures were grown. Plasmid DNA was isolated frombacteria cultures using Wizard Plus Miniprep kits(Promega). Sequencing reactions were performed withan ABI Prism dRhodamine Terminator Cycle Sequenc-ing Ready reaction kit (Perkin–Elmer) and analyzed onan automated sequencer.

RESULTS AND DISCUSSION

Immunoscope Analysis of TCR V� Clonotype RevealedClonal Expansion of Pathogenic CD4� T Cellsin Diseased Colon

The first series of experiments was aimed at identi-fying expanded populations of CD4�, TCR V�� cells

that might be associated with IBD in IL-10 KO miceand to determine whether there is a unique TCR rep-ertoire in the large intestine, the primary site of in-flammation. To this end, RNAs extracted from FACS-purified CD4� T cells (100,000 cells) in the coloniclamina propria (LP) and mesenteric lymph node (MLN)of a diseased IL-10 KO mouse with anal prolapse(C57BL/6 and 129Ola mixed background) that wasmaintained at the animal facility of the La Jolla Insti-tute for Allergy and Immunology (La Jolla IL-10-defi-cient mouse) were reverse-transcribed into cDNA. Ali-quots were amplified by PCR with one of the V�1-V�20primers and a C�-specific primer. The PCR productswere then visualized by carrying out a run-off PCRwith 6-FAM-labeled C�5 primer, and the fluorescentrun-off products were loaded onto an ABI 310 GeneticAnalyzer and analyzed using GeneScan software,which allowed determination of the fluorescence inten-sity of each band as well as its actual size. The absenceof specific stimulation appears as a Gaussian-like dis-tribution of the CDR3 sizes for a given V�-J� run-off.Specific antigenic stimulation leads to the expansion ofsingle peaks.

As seen in Fig. 1, a typical Gaussian distribution ofCDR3 lengths of individual TCR V�-C� PCR productswas observed in the peripheral lymphoid tissue (MLN)of La Jolla IBD mouse 1, suggesting a polyclonal rep-ertoire of T cells away from the site of disease. On theother hand, the TCR V�3, V�5.2, V�8.1, V�13, V�14,and V�18-C� PCR products in colonic CD4� T cellsexhibited a bias of single peak of CDR3 length, whilethe TCR V�10 and V�12-C� product and the V�8.2-C�PCR product had two and three predominant CDR3peaks in each case; this bias suggested a clonal expan-sion of the colonic CD4� T cells induced by stimulationwith a particular antigen.

Expansions of a single predominant peak in the co-lonic LP were observed also in two other IL-10-defi-cient mice (La Jolla mice 2 and 3) with IBD that alsohad been maintained at La Jolla (Fig. 2). In La JollaIBD mouse 2, the TCRV�12 and V�14-C� PCR prod-ucts exhibited a single peak of CDR3 length, while theTCR V�5.2 and V�10-C� PCR products were composedof two or three predominant peaks. In La Jolla IBDmouse 3, the TCR V�4, V�10, V�12, and V�14-C�products showed a predominant bias of CDR3 segment.These results strongly suggest that potentially patho-genic CD4� T cells were antigen-specifically andclonally infiltrated in the colon of IL-10 KO mice withcolitis. It was interesting that there was some sharingof predominant CDR3 lengths (e.g., V� 14) among thedifferent diseased IL-10-deficient mice bred in La Jolla.

The identical expanded peaks in the colonic LP werealso found in the MLN T cells bearing a TCR� chainexhibiting a Gaussian distribution of CDR3 segments(Fig. 1). Thus, it is not certain that any expanded peaks

239EPIGENETIC GENERATION OF COLITIS-RELATED PUBLIC T CELLS

that we have observed are really LP lymphocyte-spe-cific, as opposed to being found in MLN and perhapselsewhere. For example, the IL-10-deficient La Jollamouse 1 has a predominant 9-amino-acid (aa) peak forV� 8.2 in the colonic C� run-off; there also is a 9-aapeak in the MLN. Although it is a small peak, it could

represent the same clone. There is a limited recircula-tion of intestinal intraepithelial lymphocytes, and to alesser extent LP lymphocytes (17, 18), such that once Tcells are found in the intestine, they probably tend tostay there. Because migration outward, i.e., from theintestine to the systemic lymphoid tissue, is unlikely,the finding of common expanded peaks in the intestine(colonic LP) and MLN leads to the hypothesis that theinitial T cell activation may occur outside the colonicLP. Perhaps dendritic cells in the mucosa systemicallytransport bacterial or other relevant gut-derived anti-gens (19–21). On the other hand, it was shown that asubset of LP lymphocytes enter the recirculating lym-phocyte pool, e.g., thoracic duct, while over 70% of theLP lymphocytes are composed of the continual entry ofcells from the common lymphocyte pool (17, 22).

Development of Pathogenic CD4� T Cells withOligoclonal/Public Clones in DifferentIL-10-Deficient Mice with IBD

We further performed the spectratype analysis of theTCR V� repertoire by using IL-10 KO mice with adifferent genetic background (C57BL/6) and from con-ventional housing conditions free of specific pathogensin order to address two issues: (i) Are there commonclones in different IL-10-deficient mice? (ii) Are normalLP lymphocytes as oligoclonal as are those in IL-10-deficient mice? Thus, we purified CD4� T cells by FACSfrom IL-10 KO mice (Osaka-1 and Osaka-2) with coli-tis, IL-10 KO mice (Osaka-3 and Osaka-4) withoutIBD, and normal C57BL/6 mice. Three sets of immu-noscope analysis were done by using C�, J� 2.1, and J�2.4 run-offs. We chose the J�2.1 and J�2.4 from 12kinds of J� subfamilies based on our previous studythat J�2.1 and J�2.4 were dominantly used in thepathogenic T-cell clones in TCR-� chain-deficient micewith IBD (5). In addition, when dominant subsets(TCR V�6, V�8.1/8.2, and V�14) of the CD4� T cellsisolated from the colonic LP of diseased Osaka IL-10-deficient mice were analyzed for the CDR3 TCR� dis-tribution with all 12 J� primers, J�2.1- and J�2.4-C�run-offs preferentially gave clonal skewing peaks (datanot shown). This was the additional rationale for theselection of particular J�2.1 and J�2.4 in the followingrun-off analysis.

Although the results were complicated due to indi-vidual variability in the disease progression, severaldisease-related public clones were present in the colonsof IL-10-deficient Osaka mouse 1 and Osaka mouse 2with IBD (Fig. 3). Distinct colitis-related private cloneswere noted also in the individual diseased mice (Fig. 3).

A Gaussian distribution of CDR3 lengths was ob-served in all the V� and C� run-off products of thecolonic as well as peripheral MLN CD4� T cells iso-lated from IL-10-deficient Osaka mouse 1 with IBD,

FIG. 1. Immunoscope analysis of TCR V� clonotype (CDR3� sizedistribution) of CD4� T cells isolated from colonic LP and MLN ofIL-10-deficient mouse 1 with IBD, which was maintained underspecific pathogen-free conditions at the La Jolla Institute for Allergyand Immunology. CD4� T cells were purified from the colonic LP andMLN of IL-10-deficient mouse 1 by cell sorting using FACSvantage.cDNA from the FACS-purified CD4� T cells was subjected to PCRusing particular V�- and C�-specific primers followed by a run-offwith a nested fluorescent C�-specific primer. The CDR3� size distri-bution was analyzed with the GeneScan software program. Depictedare the CDR3 profiles from mouse 1 for selected V�-C� PCR ampli-fication with the indicated V� primers. The intensity of fluorescenceis presented in arbitrary units as a function of CDR3 length in aminoacids, and a CDR3 length of particular aa is indicated. The peaks arespaced by 3 nucleotides.

240 TAKAHASHI ET AL.

which was maintained in a conventional animal facil-ity (Fig. 3). However, when particular J� run-offs wereperformed, the Gaussian distribution of CDR3 sizes forthe colonic V�-C� run-off did not reflect the true diver-sity of the response in the IL-10 KO Osaka mouse 1with IBD. Clonal expansion of the CDR3 region wasobserved in the colonic V�10-J�2.1 and V�13-J�2.1run-off of Osaka mouse 1. The predominant peaks boreTCR lengths of 9 aa, in the CDR3 region. In addition,when a J�2.4 run-off was performed, a single predom-inant CDR3 region peak was detected for the V�8.2and V�13 subsets of colonic CD4� T cells from Osakamouse 1, composed of 10- and 6-amino-acid CDR3 re-gion lengths, respectively. The paucity of clones in J�run-offs for the colonic CD4� T cells was also observedin Osaka IL-10 KO mouse 2 with IBD. Clonal skewingof CDR3 length was noted in the colonic V�8.1-J�2.1 (9aa), V�8.2-J�2.1 (9 aa), V�13-J�2.1 (9 aa), V�14-J�2.1(6 aa), V�13-J�2.4 (9 aa), and V�14-J�2.4 (9 aa).

There seemed to be a tendency for the colonic LPlymphocytes of IL-10-deficient mice (Osaka 1 andOsaka 2) with IBD to have reduced diversity comparedto LP lymphocytes of the IL-10-deficient mice withoutIBD (Osaka-3 and Osaka-4) and wild-type C57BL/6mice. No obvious clonal expansion was seen for nondis-eased IL-10-deficient as well as wild-type C57BL/6mice in the CDR3 size profile for each V�-C� PCRproduct. In addition, analysis of J� run-offs of colonic

CD4� T cells in nondiseased IL-10-deficient andC57BL/6 mice confirmed their polyclonality, as theCDR3 size distribution for each J� primer extensionreaction was also Gaussian. However, expansion ofsome V�-J� combinations was occasionally observed incolonic LP of IL-10 KO Osaka mice without IBD andwild-type C57BL/6 mice (such as the V�14-J�2.4 com-bination in IL-10 KO Osaka-3 and the V�10-J�2.4combination in C57BL/6 mice). These scattered expan-sions were not disease-related since they did not havesequence homology to the disease-related clone (seebelow). Thus, they may reflect individual responses toantigens irrelevant to IBD, although their role in IBDis not excluded.

Another interesting finding, and the most striking, isthe sharing of predominant CDR3 lengths between thetwo different diseased IL-10 KO Osaka mice for theJ�2.1 and J�2.4 run-offs. For J�2.1, there is concor-dant V�13 (9 amino acids). For J�2.4, there are sharedpeaks of V�13 (9 aa). This seems to be a higher degreeof sharing or public specificities than was seen in theCD4�, CD45RBhigh T cell transfer model (16). It is con-ceivable that a model that takes months to developIBD, such as IL-10 KO mice, might select for publicclones less than in the more acute model of disease thatresults from CD4�, CD45RBhigh cell transfer (16). Fur-thermore, in animals with an intact immune system,there is probably more regulation by other T cells and

FIG. 2. CDR3� size distribution among colonic CD4� T cells isolated from different IL-10-deficient mice with IBD. C57BL/6 � 129Olamixed-background IL-10-deficient mice with IBD were maintained under specific pathogen-free conditions at the La Jolla Institute forAllergy and Immunology. The Immunoscope analysis was performed as described in the legend of Fig. 1. Depicted are the CDR3 profiles fromIL-10-deficient mice 1, 2, and 3 for selected V�-C� PCR amplification with the indicated V� primers. Arrows indicate expansions discussedin the text. The intensity of fluorescence is presented in arbitrary units as a function of CDR3 length in amino acids, and a CDR3 length ofparticular aa is indicated. The peaks are spaced by 3 nucleotides.

241EPIGENETIC GENERATION OF COLITIS-RELATED PUBLIC T CELLS

242 TAKAHASHI ET AL.

selection for particular clones (23–27). When T cells areplaced into an immune-deficient animal, they first un-dergo homeostatic proliferation to fill the “empty”space of the immune organs (28). Thus, the CD45RBhigh

transfer model would give a different result from spon-taneous models, including the IL-10-deficient mice.

In contrast, there is a relatively low level of sharingof the TCR repertoire in unimmunized inbred mice(29), a finding which might be due to stochastic ele-ments in the antigen receptor gene rearrangement andselection processes. In addition to differences in thenaive repertoire, small differences in the time of en-counter with antigen may be critical to the preferentialexpansion of particular T cell clones (30).

Determination of the DNA Sequence of the PredominantClones That Were Shared in DiseasedIL-10-Deficient Mice

There seemed to be a shared skewing of the CDR3TCR length among the different diseased IL-10 micebred in La Jolla (B6 and 129 mixed background) andOsaka (B6 background). To this end, we selected theexpanded V�-J��, CD4� colonic T-cell populations aswell as corresponding MLN T-cell populations as can-didates for IBD-associated expansions of particular Tcells at the onset of the disease. The populations wereas follows: (i) La Jolla IBD mouse 1; V�12-J�2.4 andV�13-J�2.1 in colonic and MLN CD4�T cells. (ii) LaJolla mouse 2; V�10-J�2.1, V�11-J�2.1, and V�12-J�2.4 in colonic and MLN CD4� T cells. (iii) La JollaIBD mouse 3; V�8.2-J�2.4, V�10-J�2.1, V�12-J�2.4,and V�14-J�2.4 in colonic and MLN CD4� T cells. (iv)Osaka IBD mouse 1; V�8.2-J�2.4, V�13-J�2.1, andV�14-J�2.4 in colonic and MLN CD4� T cells. (v)Osaka IBD mouse 2; V�8.1-J�2.1, V�8.2-J�2.4, V�13-J�2.1, V�13-J�2.4, and V�14-J�2.4 in colonic andMLN CD4� T cells. (vi) Osaka IL-10 mouse 3 withoutIBD; V�10-J�2.1, V�13-J�2.1, and V�14-J�2.4 in co-lonic and MLN CD4� T cells. (vii) Osaka IL-10 mouse 4without IBD; V�8.2-J�2.4, V�13-J�2.1, and V�14-J�2.4 in colonic and MLN CD4� T cells. We then de-termined DNA sequences to identify public clones, i.e.,clones shared between Osaka and La Jolla IL-10 KOmice with IBD. Eight to 36 clones were analyzed forDNA sequence for each amplified V�-J� combination.

As shown in Figs. 4A and 4B, some public CDR3motifs were shared between colonic CD4� T cells iso-lated from IL-10-deficient mice with IBD maintainedin Osaka and those maintained in La Jolla. The mostfrequently shared motif was SXDWG, which wasshared in the La Jolla IBD mouse 1 (V�13-J�2.1) at afrequency of 12/12 (Fig. 4A), La Jolla IBD mouse 2(V�10-J�2.1) at 4/8 (Fig. 4A), La Jolla IBD mouse 3(V�10-J�2.1) at 14/24 (Fig. 4A), Osaka IBD mouse 1(V�8.2-J�2.4 at 3/7 and V�14-J�2.4 at 5/14; Fig. 4B),and Osaka IBD mouse 2 (V�13-J�2.4 at 2/23 and V�14-J�2.4 at 1/21; Fig. 4B). Another interesting public mo-tif was SATGNYAEQ, which was shared in the OsakaIBD mouse 1 (V�13-J�2.1) at 30/33 (Fig. 4B) andOsaka mouse 2 (V�13-J�2.1) at 12/27 (Fig. 4B).

These shared motifs were not seen in the peripherallymphoid tissues, i.e., MLN of either the diseased LaJolla or Osaka mice (Figs. 4A and 4B). Finally, it wasimportant to confirm that the motifs were not seen inthe nondiseased IL-10 KO mice in order to determinewhether the motifs were truly IBD-specific. As shownin Fig. 4C, the SXDWG and SATGNYAEQ motifs werenot seen in the colons and MLN of nondiseased IL-10-deficient Osaka mouse 3 and Osaka mouse 4. Thus, thecommon motifs might see a public gut-derived antigenwhich is important for the development of pathogenicCD4� T cells in this colitis model. In terms of an SX-DWG motif using a flexible V�-J� rearrangement, itmay be that an epitope drives the onset of IBD. In thisregard, it was reported that mice lacking the TCR V� 5chain retained a conformational determinant by using“shuffled TCR � chain” (31). In our case, the selection ofV�-J� usage is perhaps stochastic in individual mice;however, the epigenetic generation of the SXDWG mo-tif in different V�-D�-J� combinations by recombina-tion machinery followed by selection in the gut envi-ronment could be crucial for the triggering of IBD.

It is possible that the public motifs SXDWG and (or)SATGNYAEQ are derived from commensal bacterialantigens. In support of this view, it has been reportedthat IL-10-deficient mice housed under germ-free con-ditions remained healthy, while those housed underconventional and SPF conditions developed colitis afterweaning (10, 32). In addition, the data obtained from aCD4�, CD45RBhigh T-cell adoptive transfer model withreduced flora-bearing SCID mice suggest that bacteria

FIG. 3. Comparison of CDR3� size distribution of CD4� T cells isolated from colonic lamina propria and mesenteric lymph nodes ofIL-10-deficient mice with IBD with those of IL-10-deficient mice without IBD and wild-type C57BL/6 mice. C57BL/6 background IL-10-deficient mice were maintained under conventional conditions at the animal facility of Osaka University. The Immunoscope analysis wasperformed as described in the legend of Fig. 1. Depicted are the CDR3 profiles for selected V�-C�, V�-J�2.1, and V�-J�2.4 PCR amplificationswith the indicated V� primers. One representative of two different mice is shown both in IL-10-deficient mice without IBD and in C57BL/6mice. E indicates expansions discussed in the text. G and PG mean Gaussian-type and pseudo-Gaussian-type distribution of the CDR TCR� length, respectively. The intensity of fluorescence is presented in arbitrary units as a function of CDR3 length in amino acids, and a CDR3length of particular aa is indicated. The peaks are spaced by 3 nucleotides.

243EPIGENETIC GENERATION OF COLITIS-RELATED PUBLIC T CELLS

FIG. 4. (A–C) Identical clones can be found throughout the colonic LP of different IL-10-deficient mice with IBD. Translated amino acidsequences of expanded clones in colonic LP and their counterparts in respective MLN are shown. Amplifications were done using theindicated V� and J� primers from the indicated IL-10-deficient mice with or without IBD. Primers pairs were chosen based on fragmentlength analysis showing the presence of a predominant V�-J� CDR3� length in colonic LP. However, the V�-J� amplification products wereshotgun-cloned without any selection for rearrangements of the predominant size. The predominant or expanded sequence is indicated, andthe detection frequencies are determined by the number of sequences with the listed CDR3 sequence out of the total number of sequencesgenerated for the particular V�-J� combination in the mouse indicated.

244 TAKAHASHI ET AL.

or bacterial products likely are important in the morerapid kinetics of homing of pathogenic T cells to thelarge intestine, their subsequent expansion, and theenhanced susceptibility of the large intestine to inflam-mation (33). Thus, additional experiments will be re-

quired to determine what particular antigens from bac-teria or other microorganisms mediate the developmentof the pathogenic T cells and drive the onset of the colitis.Otherwise, because bacteria also can stimulate inflam-matory cytokines and chemokine secretion by intestinal

FIG. 4—Continued

245EPIGENETIC GENERATION OF COLITIS-RELATED PUBLIC T CELLS

epithelial cells (34), the presence of a higher bacterialload in the large intestine may provide a non-Ag condi-tioning factor that makes the intestinal tissue more sus-ceptible to colonizing by T lymphocytes with a variety ofspecificities.

It is intriguing that public clones were present in theIL-10-deficient mice from both Osaka and La Jolla.Although the frequency of the public motif found in thediseased sites was variable, it is still possible that anidentical/common antigen is involved in the murinemodel. If one can identify these clones from the periph-eral blood as well (using CDR3-specific primers), onecould predict the time to disease onset by clonal anal-ysis. Furthermore, it is important to determine

whether this clone drives disease onset or whether thisclonal expansion is a product of the disease. If theclonal expansion of a public clone is causative, transferof this clone should induce the disease in recipientanimals.

Analyses of human IBD have provided some evi-dence for shared clones in peripheral blood CD4� Tlymphocytes of identical twins concordant for Crohn’sdisease (35) and for shared sequence motifs in acti-vated CD4� and/or CD8� T cells of the intestinal mu-cosa from unrelated individuals with IBD (36). Similarto the results from our analysis of diseased mice, how-ever, in all these cases private or individual TCR expan-sions greatly outnumber the public clones identified.

FIG. 4—Continued

246 TAKAHASHI ET AL.

Although the bacterial microflora may be importantin disease pathogenesis (37), not all bacteria have thesame capacity to induce inflammation. Some bacteriathat are considered to be probiotics, such as the variousLactobacillus species, prevent colitis from occurring inIL-10-deficient animals living under specific pathogen-free conditions (38). On the other hand, Helicobacterhepaticus caused colitis when introduced into patho-gen-free IL-10-deficient mice and not in immune-com-petent controls (39).

Sutton et al. (40) recently isolated a novel microbialI2 antigen by using representational difference analy-sis. Bacterial 12 antigen was specifically localized tomucosal lesions in human Crohn’s disease. In vitroCD4� T-cell proliferation and cytokine assays revealedthat there were memory CD4� T-cell responses to bac-terial I2 antigen in murine IBD models, including IL-10-deficient and TCR �-chain-deficient mice (41). Al-though those workers have not done repertoire studieson I2-reactive T cells, it will be of interest to learnwhether I2-reactive T cells have the public motifsSXDWG or SATGNYAEQ.

ACKNOWLEDGMENTS

This study was supported, in part, by Grants-in-Aid for ScientificResearch from the Ministry of Education, Science, Sports, and Cul-ture of Japan, and the Ministry of Health and Welfare of Japan. Theauthors thank Dr. William R. Brown (Denver VA Medical Center) forhis helpful comments and editorial assistance.

REFERENCES

1. Powrie, F., T cells in inflammatory bowel disease: Protective andpathogenic roles. Immunity 3, 171–174, 1995.

2. Blumberg, R. S., Saubermann, L. J., and Strober, W. Animalmodels of mucosal inflammation and their relation to humaninflammatory disease. Curr. Opin. Immunol. 11, 648–656, 1999.

3. DeWinter, H., Cheroutre, H., and Kronenberg, M., The yin andyang of T cells in intestinal inflammation: Pathogenic and pro-tective roles in a mouse colitis model. Am. J. Pathol. 276, G1317–1321, 1999.

4. Takahashi, I., Iijima, H., Katashima, R., Itakura, M., andKiyono, H., Clonal expansion of CD4� TCR ��� T cells in TCR�-chain deficient mice by gut-derived antigens. J. Immunol. 162,1843–1850, 1999.

5. Mizoguchi, A., Mizoguchi, E., Saubermann, L. J., Higaki, K.,Blumberg, R. S., and Bhan, A. K., Limited CD4 T-cell diversityassociated with colitis in T-cell receptor � mutant mice requiresa T helper 2 environment. Gastroenterology 119, 983–995, 2000.

6. Cibotti, R., Cabaniols, J. P., Pannetier, C., Delarbre, C., Vergnon,I., Kanellopoulos, J. M., and Kourilsky, P., Public and private V�T cell receptor repertoires against hen egg white lysozyme (HEL)in nontransgenic versus HEL transgenic mice. J. Exp. Med. 180,861–872, 1994.

7. Musette, P., Bequet, D., Delarbre, C., Gachelin, G., Kourilsky, P.,and Dormont, D., Expansion of a recurrent V�5.3� T-cell popu-lation in newly diagnosed and untreated HLA-DR2 multiplesclerosis patients. Proc. Natl. Acad. Sci. USA 93, 12461–12466,1996.

8. Pannetier, C., Even, J., and Kourilsky, P., T-cell repertoire di-versity and clonal expansions in normal and clinical samples.Immunol. Today 16, 176–181, 1995.

9. Moore, K. W., O’Garra, A., deWaal Malefyt, R., Vieira, P., andMossman, T. R., Interleukin-10. Annu. Rev. Immunol. 11, 165–190, 1993.

10. Kuhn, R., Lohler, J., Rennnick, D., Rajewsky, K., and Muller, W.,Interleukin-10-deficient mice develop chronic enterocolitis. Cell75, 263–274, 1993.

11. Hagenbaugh, A., Sharma, S., Dubinett, S. M., Wei, S. H. Y.,Aranda, R., Cheroutre, H., Fowell, D. J., Binder, S., Tsao, B.,Locksley, R. M., Moore, K. W., and Kronenberg, M., Alteredimmune responses in interleukin 10 transgenic mice. J. Exp.Med. 185, 21011–21010, 1997.

12. Asseman, C., Mauze, S., Leach, M. W., Coffman, R. L., andPowrie, F., An essential role for interleukin 10 in the function ofregulatory T cells that inhibit intestinal inflammation. J. Exp.Med. 190, 995–1004, 1999.

13. Berg, D. J., Davidson, N., Kuhn, R., Muller, W., Menon, S.,Holland, G., Thompson-Snipes, L., Leach, M. W., and Rennick,D., Enterocolitis and colon cancer in interleukin-10-deficientmice are associated with aberrant cytokine production and CD4�

TH1-like responses. J. Clin. Invest. 98, 1010–1020, 1996.14. Davidson, N. J., Leach, M. W., Fort, M. M., Thompson-Snipes, L.,

Kuhn, R., Muller, W., Berg, D. J., and Rennick, D. M., T helpercell 1-type CD4� T cells, but not B cells, mediate colitis ininterleukin 10-deficient mice. J. Exp. Med. 184, 241–251, 1996.

15. Sellon, R., Tonkonogy, S., Schultz, M., Dieleman, L. A., Gren-ther, W., Balish, E., Rennick, D. M., and Sartor, R. B., Residententeric bacteria are necessary for development of spontaneouscolitis and immune system activation in interleukin-10-deficientmice. Infect. Immun. 66, 5224–5231, 1998.

16. Matsuda, J., Gapin, L., Sydra, B. C., Byrne, F., Binder, S.,Kronenberg, M., and Aranda, R., Systemic activation and anti-gen-driven oligoclonal expansion of T cells in a mouse model ofcolitis. J. Immunol. 164, 2797–2806, 2000.

17. Poussier, P., Edouard, P., Lee, C., Binnie, M., and Julius, M.,Thymus-independent development and negative selection of Tcells expressing T cell receptor �� in the intestinal epithelium:Evidence for distinct circulation patterns of gut- and thymus-derived T lymphocytes. J. Exp. Med. 176, 187–199, 1992.

18. Camerini, V., Panwala, C., and Kronenberg, M., Regional spe-cialization of the mucosal immune system: Intraepithelial lym-phocytes of the large intestine have a different phenotype andfunction than those of the small intestine. J. Immunol. 151,1765–1776, 1993.

19. Iwasaki, A., and Kelsall, B. L. Freshly isolated Peyer’s patch, butnot spleen, dendritic cells produce interleukin 10 and induce thedifferentiation of T helper type 2 cells. J. Exp. Med. 190, 229–239, 1999.

20. Iwasaki, A., and Kelsall, B. L., Localization of distinct Peyer’spatch dendritic cell subsets and their recruitment by chemokinesmacrophage inflammatory protein (MIP)-3�, MIP-3�, and sec-ondary lymphoid organ chemokine. J. Exp. Med. 191, 1381–1394, 2000.

21. Cook, D. N., Prosser, D. M., Forster, R., Zhang, J., Kuklin, N. A.,Abbondanzo, S. J., Niu, X-D., Chen, S-C., Manfra, D. J.,Wiekowski, M. T., Sullivan, L. M., Smith, S. R., Greenberg,H. B., Narula, S. K., Lipp, M., and Lira, S. A., CCR6 mediatesdendritic cell localization, lymphocyte homeostasis, and immuneresponses in mucosal tissue. Immunity 12, 495–503, 2000.

22. Arstila, T., Arstila, T. P., Calbo, S., Selz, F., Malassis-Seris, M.,Vassalli, P., Kourilsky, P., and Guy-Grand, D., Identical T cellclones are located within the mouse gut epithelium and lamina

247EPIGENETIC GENERATION OF COLITIS-RELATED PUBLIC T CELLS

propria and circulate in the thoracic duct lymph. J. Exp. Med.191, 823–834, 2000.

23. Groux, H., O’Garra, A., Bigler, M., Rouleau, M., Antonenko, S.,de Vries, J. E., and Roncarolo, M. G., A CD4� T-cell subsetinhibits antigen-specific T-cell responses and prevents colitis.Nature 389, 737–742, 1997.

24. Kronenberg, M., and Cheroutre, H., Do mucosal T cells preventintestinal inflammation? Gastroenterology 118, 974–977, 2000.

25. Poussier, P., Ning, T., Chen, J., Banerjee, D., and Julius, M.,Intestinal inflammation observed in IL-2R/IL-2 mutant mice isassociated with impaired intestinal T lymphopoiesis. Gastroen-terology 118, 880–891, 2000.

26. Boismenu, R., Natural and pharmacological strategies for theprotection of mucosal surfaces. Mucosal Immunol. Update 8,5–9, 2000.

27. Saubermann, L. J., Beck, P., DeJong, Y. P., Pitman, R. S., Ryan,M. S., Kim, H. S., Exley, M., Snapper, S., Balk, S. P., Hagen,S. J., Kanauchi, O., Motoki, K., Sakai, T., Terhorst, C., Koezuka,Y., Podolsky, D. K., and Blumberg, R. S., Activation of naturalkiller T cells by �-galactosylceramide in the presence of CD1dprovides protection against colitis in mice. Gastroenterology 119,119–128, 2000.

28. Ernst, B., Lee, D-S., Chang, J. M., Sprent, J., and Surh, C. D.,The peptide ligands mediating positive selection in the thymuscontrol T cell survival and homeostatic proliferation in the pe-riphery. Immunity 11, 173–181, 1999.

29. Bousso, P., Casrouge, A., Altman, J. D., Haury, M., Kanellopou-los, J., Abastado, J. P., and Kourilsky, P., Individual variationsin the murine T cell response to a specific peptide reflect vari-ability in naive repertoires. Immunity 9, 169–178, 1998.

30. Bousso, P., Levraud, J. P., Kourilsky, P., and Abastado, J. P.,The composition of a primary T cell response is largely deter-mined by the timing of recruitment of individual T cell clones. J.Exp. Med. 189, 1591–1600, 1999.

31. Mallick-Wood, C. A., Lewis, J. M., Richie, L. I., Owen, M. J.,Tigelaar, R. E., and Hayday, A. C., Conservation of T cell recep-tor conformation in epidermal �� cells with disrupted primaryV� gene usage. Science 279, 1729–1733, 1998.

32. Song, F., Ito, K., Denning, T. L., Kuninger, D., Papacon-stantinou, J., Gourley, W., Klimpel, G., Balish, E., Hokanson, J.,and Ernst, P. B., Expression of the neutrophil chemokine KC inthe colon of mice with enterocolitis and by intestinal epithelial

cell line: Effects of flora and proinflammatory cytokines. J. Im-munol. 162, 2275–2280, 1999.

33. Aranda, R., Sydra, B. C., McAllister, P. L., Binder, S. W., Yang,H. Y., Targan, S. R., and Kronenberg, M., Analysis of intestinallymphocytes in mouse colitis mediated by transfer of CD4�,CD45RBhigh T cells to SCID recipients. J. Immunol. 158, 3464–3473, 1997.

34. Jung H. C., Eckman, L., Yang, S. K., Panja, A., Fierer, J.,Morzycka-Wroblewska, E., and Kagnoff, M. F., A distinct arrayof proinflammatory cytokines is expressed in human colon epi-thelial cells in response to bacterial invasion. J. Clin. Invest. 95,55–65, 1995.

35. Probert, C. S., Chott, A., Turner, J. R., Saubermann, L. J.,Stevens, A. C., Bodinaku, K., Elson, C. O., Balk, S. P., andBlumberg, R. S., Persistent clonal expansions of peripheral bloodCD4 lymphocytes in chronic inflammatory bowel disease. J. Im-munol. 157, 3183–3191, 1996.

36. Saubermann, L. J., Probert, C. S., Christ, A. D., Chott, A.,Turner, J. R., Stevens, A. C., Balk, S. P., and Blumberg, R. S.,Evidence of T cell receptor �-chain patterns in inflammatory andnoninflammatory bowel disease states. Am. J. Physiol. 276,G613–G621, 1996.

37. Cong, Y., Brandwein, S. L., McCabe, R. P., Lazenby, A., Birken-meier, E. H., Sundberg, J. P., and Elson, C. O. CD4� T cellsreactive to enteric bacterial antigens in spontaneously coliticC3H/HeJBir mice: Increased T helper cell type 1 response andability to transfer disease. J. Exp. Med. 187, 855–864, 1998.

38. Madsen, K. L., Doyle, J. S., Jewell, L. D., Tavernini, M. M., andFedorak, R. N., Lactobacillus species prevents colitis in interleu-kin 10 gene-deficient mice. Gastroenterology 116, 1107–1114,1999.

39. Kullberg, M. C., Ward, M. J., Gorelick, P. L., Caspar, P., Hieny,S., Cheever, A., Jankovic, D., and Sher, A., Helicobacter hepati-cus triggers colitis in specific-pathogen-free interleukin-10 (IL-10)-deficient mice through an IL-12-and gamma-interferon-de-pendent mechanism. Infect. Immun. 66, 5157–5166, 1998.

40. Sutton, C. L., Kim, J., Yamane, A., Dalwadi, H., Wei, B., Land-ers, C., Targan, S. R., and Braun, J., Identification of a novelbacterial sequence associated with Crohn’s disease. Gastroenter-ology 119, 23–31, 2000.

41. Dalwadi, H. N., Wei, B., Kronenberg, M., Sutton, C. L., andBraun, J., The Crohn’s disease-associated 12 bacterial antigen isthe target of a CD4� T cell response in normal and colitic mice.FASEB J. 14, A975, 2000.

Received November 9, 2001; accepted December 14, 2001

248 TAKAHASHI ET AL.