of March 26, 2018.This information is current as

Lesions of Autoimmune Gastritis13-Producing Stromal Network in ChronicSupported by CXC Chemokine Ligand Th1-Biased Tertiary Lymphoid Tissue

Gonda and Akira ShimizuTomoya Katakai, Takahiro Hara, Manabu Sugai, Hiroyuki

http://www.jimmunol.org/content/171/8/4359doi: 10.4049/jimmunol.171.8.4359

2003; 171:4359-4368; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/171/8/4359.full#ref-list-1

, 17 of which you can access for free at: cites 37 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2003 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Th1-Biased Tertiary Lymphoid Tissue Supported by CXCChemokine Ligand 13-Producing Stromal Network in ChronicLesions of Autoimmune Gastritis1

Tomoya Katakai,* Takahiro Hara,* Manabu Sugai,* Hiroyuki Gonda,* † and Akira Shimizu2*†

Secondary lymphoid tissue is developmentally programmed and characterized by well-ordered compartmentalization of lympho-cyte subsets and specialized stromal cells supporting the tissue architecture. By contrast, tertiary lymphoid tissue is defined as thatinduced in ectopic sites by inflammation, although its immunological role is largely unknown. In this study, we characterize thelymphoid tissue induced in the chronic lesion of murine autoimmune gastritis (AIG). Within the lymphoid cluster in the gastricmucosa, there is a clear segregation of T and B cells. Follicle-like B cell areas are always located on the luminal side of the mucosa,while T cells are located in the basal part. A typical lymphoid reticular network and follicular dendritic cells support the structure.Importantly, complement receptor 1� follicular dendritic cells within the follicle express a B cell homing chemokine, CXCchemokine ligand 13. The number and size of the clusters correlate with the age of the mice and the serum autoantibody titer,suggesting the functional importance of the clusters in local Ab production, although involvement of the autoantibody in thedisease progression is still unclear. AIG gastric lesions are known to constitute a Th1-biased, memory T cell-dependent immu-nomicroenvironment. The expression pattern of cytokines, including lymphotoxin-�, and chemokines in the AIG stomach isconsistent with this observation. Taken together, these facts suggest that, during the chronic phase of autoimmunity, long-lastinglymphocyte infiltration probably induces a unique tertiary lymphoid tissue that has a function distinct from that of regional lymphnodes. These neolymphoid tissues may maintain the local self reactivity supporting the vicious cycle of Th1-type reaction as wellas autoantibody production. The Journal of Immunology, 2003, 171: 4359–4368.

C hronic inflammation is a kind of long-lasting immunereaction against pathogens, tumors, or self components,characterized by an accumulation of lymphocytes within

local tissue. This situation is almost always accompanied by thefunctional destruction of the tissue. Infiltrating lymphocytes arelikely to originate from secondary lymphoid tissue, such as drain-ing lymph nodes, which have been activated in an Ag-specific orbystander fashion (1). A continuous Ag supply from the targettissue may be a driving force in maintaining the chronic phase ofimmune reactions. Especially in autoimmune diseases, self com-ponents released by the destructive inflammation within the targetorgan may further stimulate specific effector cells and trigger avicious cycle. However, it is largely unknown whether the extrav-asated lymphocyte subsets are randomly located or following somerules within the tissue. Whether the infiltrating lymphocytes stay inthe lesion for a long period or are replaced quickly is also unclear.There may be recirculation of lymphocytes between regionallymph nodes and the inflamed tissue. It has been reported thatlymphoid structures resembling secondary lymphoid tissue de-velop at the inflamed site in several diseases (2–7), and such ec-

topically induced lymphoid tissue was defined as tertiary lymphoidtissue (8).

In this study, we found organized lymphoid clusters in thechronic lesions in a murine experimental model for autoimmunegastritis (AIG).3 The location of T cells, B cells, reticular networks(RN), and follicular dendritic cells (FDCs) in the cluster was an-alyzed by multicolor immunohistochemistry and confocal micros-copy. Cytokine and chemokine expression in gastric mucosa (GM)and its relationship to autoantibody production in AIG were alsoassessed. These analyses revealed a unique feature of the lymphoidcluster, indicating that it is similar to, but clearly different fromsecondary lymphoid tissue in several respects, including the ab-sence of typical germinal centers (GCs), high endothelial cellmarkers, naive T cells, Th2 cells, and Th2-type factors. We con-clude that this gastric lymphoid tissue is a novel type of mucosallymphoid tissue developed in the chronically Th1-biased autoim-mune environment in the stomach of AIG.

Materials and MethodsMice

BALB/c mice were purchased from Japan SLC (Shizuoka, Japan). Themice were maintained at the facility in Center for Molecular Biology andGenetics, Kyoto University. AIG was induced by neonatal thymectomy ofBALB/c mice 3 days after birth and diagnosed by ELISA for the detectionof autoantibody in sera, as described previously (9–11). Six to thirty-twoweeks after thymectomy, mice suffering from AIG were used for experiments.

*Center for Molecular Biology and Genetics, Kyoto University, and †TranslationalResearch Center, Kyoto University Hospital, Kyoto, Japan

Received for publication February 19, 2003. Accepted for publication August12, 2003.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported in part by Grants-in-Aid from the Ministry of Education,Science, Sports, and Culture of Japan.2 Address correspondence and reprint requests to Dr. Akira Shimizu, 53 Shogoin-Kawa-hara-cho, Sakyo-ku, Kyoto 606-8507, Japan. E-mail address: shimizu@virus.kyoto-u.ac.jp

3 Abbreviations used in this paper: AIG, autoimmune gastritis; CCL, CC chemokineligand; CR1, complement receptor-1; CXCL, CXC chemokine ligand; d3-Tx, day 3thymectomy; FDC, follicular dendritic cell; GC, germinal center; GLN, gastric lymphnode; GM, gastric mucosa; HEV, high endothelial venule; LT-�, lymphotoxin-�;PECAM, platelet endothelial cell adhesion molecule; PNA, peanut agglutinin; PNAd,peripheral node adressin; RN, reticular network; SLC, secondary lymphoid tissuechemokine.

The Journal of Immunology

Copyright © 2003 by The American Association of Immunologists, Inc. 0022-1767/03/$02.00

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Immunohistochemistry

Glandular stomachs and lymph nodes were isolated from the mice, em-bedded in OTC compound (Sakura Finetechnical, Tokyo, Japan), and fro-zen using liquid nitrogen. Cryosections (10–20 �m) were fixed with coldacetone for 5 min, treated with 0.05% Tween 20-PBS containing 1% BSAand 5% mouse serum, and stained with the following Abs: biotin anti-H�/K�-ATPase (1H9) (12), biotin anti-CD3, FITC anti-B220, anti-complement receptor 1 (CR1), anti-peripheral node adressin (PNAd), anti-syndecan-1 (BD PharMingen, San Diego, CA), anti-platelet endothelialcell adhesion molecule-1 (PECAM-1; Caltag, Burlingame, CA), anti-ER-TR7 (Serotec, Oxford, U.K.), anti-Mac-1 (Immunotech, Luminy, France),biotin anti-CXC chemokine ligand 13 (CXCL13), biotin anti-CC chemo-kine ligand 21 (CCL21) (R&D Systems, Minneapolis, MN), FITC anti-mouse IgM, PE anti-mouse IgD (Southern Biotechnology, Birmingham,AL), or FITC anti-mouse IgG (Jackson ImmunoResearch Laboratories,West Grove, PA). PE anti-rat IgG Ab (Caltag), biotin anti-mouse IgM Ab(BD PharMingen), PE streptavidin, and APC streptavidin (MolecularProbes, Eugene, OR) were used as the secondary reagents. GC B cells weredetected with FITC peanut agglutinin (PNA; Sigma-Aldrich, St, Louis,MO). After staining, sections were mounted using PermaFluor (Shandon,Pittsburgh, PA) and examined by confocal laser-scanning microscopy(MRC-1024; Bio-Rad, Osaka, Japan). The digital images obtained wereanalyzed using Adobe Photoshop software (Adobe Systems, San Jose,CA). To produce a view of the whole lymph node, images at differentpositions were assembled into a single image by combining adjacent im-ages on using this software.

Preparation of GM-infiltrating cells and flow cytometric analysis

Mononuclear cells infiltrating into the GM were collected, as describedpreviously (10, 11). In brief, 10% FCS-RPMI medium was injected into thesubmucosa of the stomach, which were then cut into small pieces to releaseinfiltrating cells. After removing large tissue debris by passing throughcotton in a funnel, the cells were used for flow cytometry. Isolated cellswere stained with FITC anti-Thy-1.2, PE anti-CD4, FITC anti-CD8 (BDBiosciences, Mountain View, CA), biotin anti-B220 (Caltag), FITC anti-CD5, FITC anti-Mac-1, or FITC anti-B220 (BD PharMingen) Ab, and thenanalyzed on a FACSCalibur flow cytometer using CellQuest software (BDBiosciences).

RT-PCR

Semiquantitative RT-PCR analysis was performed, as described previ-ously, with some modifications (11, 13). In brief, total RNA was preparedusing TRIzol (Life Technologies, Gaithersburg, MD). Oligo(dT)12–18-primed cDNA was synthesized from 2 �g of RNA using Superscript IIreverse transcriptase (Life Technologies). Five-fold serial dilutions ofcDNA were analyzed by PCR in 10-�l reactions with ExTaq DNA poly-merase (Takara, Otsu, Japan) and the following specific primer pairs:IFN-�, 5�-TGAACGCTACACACTGCATCTTGG-3� and 5�-CGACTCCTTTTCCGCTTCCTGAG-3�; IL-4, 5�-ATGGGTCTCAACCCCCAGCTAGT-3� and 5�-GCTCTTTAGGCTTTCCAGGAAGTC-3�; lymphotoxin-�(LT-�), 5�-CATCTTGCCCTCACCCTCTA-3� and 5�-AAACGCTTCTTCTTGGCTCG-3�; CCL4 (macrophage-inflammatory protein-1�), 5�-CAGCTCTGTGCAAACCTAAC-3� and 5�-TCAGTTCAACTCCAAGTCAC-3�;CCL5(RANTES),5�-TCTGAGACAGCACATGCATC-3�and5�-CCTAGCTCATCTCCAAATAG-3�; CXCL9 (monokine induced by IFN-�), 5�-CCAACACAGTGACTCAATAG-3� and 5�-TTATGTAGTCTTCCTTGAACG-3�;CXCL10 (IFN-inducible protein-10), 5�-AGACATCCCGAGCCAACCTT-3�and 5�-GTTAAGGAGCCCTTTTAGAC-3�; CCL17 (thymus- and activation-regulated chemokine), 5�-TCTGCTTCTGGGGACTTTTC-3� and 5�-GTTCGCCTGTAGTGCATAAG-3�; CCL22 (macrophage-derived chemokine),5�-CTGGTCATTAGACACCTGAC-3� and 5�-CCCTAGGACAGTTTCTGGAG-3�; CCL19 (Epstein-Barr virus-induced molecule 1 ligand chemokine),5�-GCACACAGTCTCTCAGGCTC-3� and 5�- CTCTCTTCTGGTCCTTGGTT-3�; CCL21 (secondary lymphoid tissue chemokine (SLC)), 5�-AGCTATGTGCAAACCCTGAG-3� and 5�-TCATAGGTGCAAGGACAAGG-3�; CXCL13 (B-lymphocyte chemoattractant), 5�-TTGAACTCCACCTCCAGGCA-3� and 5�-CTTCAGGCAGCTCTTCTCTT-3�; CXCL12 (stromal-derived factor-1), 5�-AAACCAGTCAGCCTGAGC TAC-3� and 5�-TTACTTGTTTAAAGCTTTCTC-3�; GAPDH, 5�-CCATCACCATCTTCCAGGAG-3� and 5�-CCTGCTTCACCACCTTCTTG-3�. Values werestandardized relative to that for GAPDH as an internal control. The numberof PCR cycles was as follows: 32 (IFN-�), 30 (IL-4), 30 (LT-�), 30(CCL4), 27 (CCL5), 28 (CXCL9), 28 (CXCL10), 30 (CCL17), 30(CCL22), 30 (CCL19), 30 (CCL21), 30 (CXCL13), 29 (CXCL12), and 21(GAPDH).

ResultsAIG develops in BALB/c mice subjected to thymectomy at 3 daysafter the birth (d3-Tx), and is an experimental model for the CD4�

T cell-mediated autoimmune disorder induced at high incidence(�50–80%) in mice that undergo d3-Tx (9–11). Disease-bearingmice (AIG mice) exhibit a characteristic profile of chronic inflam-mation in the stomach, with features such as the thickening of theGM due to epithelial hyperplasia (Fig. 1, A and B), selective lossof parietal cells (Fig. 1C), lymphocyte infiltration into the laminapropria (Fig. 1B), and the production of autoantibodies againstparietal cells (12) (data not shown). In addition, the gastric lymphnode (GLN) located at the lesser curvature is also enlarged (Fig.1A). Notably, we have observed that the infiltrating lymphocytesfrequently form clearly isolated clusters that are located at thebasal part of the lamina propria adjacent to the muscularis muco-sae, and are frequently associated with blood vessels (Fig. 1B,arrowheads; Fig. 2A, arrows).

FIGURE 1. Histopathology of AIG in BALB/c mice. A, Gross photo-graphs of dissected stomach and associated regional lymph node (GLN,arrow) from normal (left) and AIG (right) mice. The lower image showsthe luminal surface view of the glandular stomach. B and C, Sections ofGM from normal (left) and AIG (right) mice stained with hematoxylin (B),or anti-H�/K�-ATPase Ab detecting the location of parietal cells (C), re-spectively. Arrowheads indicate the lymphocytic infiltration forming alarge lesional lymphoid cluster.

4360 TERTIARY LYMPHOID TISSUE IN AIG

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Lymphoid tissue-like structure of the lymphoid clusters

Inside secondary lymphoid tissue such as lymph nodes and thespleen, T and B lymphocytes are each located in separate regionsknown as the T cell area of the paracortex (detected with anti-CD3Ab) and the B cell follicle (detected with anti-B220 Ab), respec-tively (Fig. 3A). Two types of stromal cells of mesenchymal originsupport these structures. A RN constructed of fibroblastic reticularcells covers a large part of the lymph node with a complex mesh-work that is stained with ER-TR7 Ab (Fig. 3B), while follicles andGCs are backed by FDC networks that can be detected with anti-CR1 (CD35) Ab (Fig. 3C). In contrast to those from normal ani-mals, GLNs from AIG mice become greatly hypertrophic, withdramatically expanded RN, and develop many GCs, reflecting ex-tensive lymphocyte activation, especially of B cells.

To determine whether any lymphoid structure is organized atthese focal lesions or not, we performed confocal fluorescenceimmunohistochemical analysis of the sections of GM from AIGmice (Fig. 2). To our surprise, a large fraction of lymphocyticclusters (�100 �m), but not of relatively scattered infiltrates, hadclear segregation of the T and B cells. Each cluster had a singlefollicle-like B cell aggregate and associated T cell area. It is worthnoting that the B cell follicle within a lymphoid cluster was alwayslocated on the luminal side of the mucosa, while the T cells werepositioned at the relatively basal part. The characteristic lymphoidRN was distributed precisely at the cluster. Moreover, we detectedan FDC marker, CR1, at the B cell area, indicating that a follicularstructure like that in secondary lymphoid tissue had indeed developed

in these clusters. Mac-1� myeloid cells were distributed around theclusters (Fig. 2C). No typical lymphocytic infiltration or lymphoidstructure was observed in normal mouse GM (Fig. 1B).

Development of mature B cell follicles requires a homeostaticchemokine, CXCL13 (B-lymphocyte chemoattractant), expressedby stromal cells (14, 15). Gene targeting has revealed the criticalimportance of this chemokine for the organization of normal fol-licular architecture (14). Therefore, we next assessed whetherCXCL13 was expressed in the B cell area of the clusters in GMlesions (Fig. 4, A and B). As a positive control experiment, anti-CXCL13 polyclonal Ab was shown to stain a part of the CR1�

FDC network in a thin, filamentous pattern in the follicles of theGLNs from AIG mice. Control goat Ab showed some backgroundstaining, but no structural pattern. Strikingly, the CXCL13 expres-sion could also clearly be detected on the CR1� FDC-like networkon the follicles of GM lymphoid clusters. However, some of theseclusters exhibited very weak or undetectable signals, probably dueto CXCL13 expression levels below the limit of experimental sen-sitivity (data not shown). Despite a formation of obvious follicularstructure, typical GCs (as detected by PNA) were rarely observedeven in the largest lymphoid clusters (Fig. 5A).

We previously showed that almost all T cells infiltrating into theGM of AIG mice are memory cells (11). Homing of naive T cellsto LNs is primarily mediated by the expression of PNAd, carbo-hydrate ligands for L-selectin, and a chemokine, CCL21 (SLC), onhigh endothelial venules (HEV) (16–18). HEVs in the GLN fromAIG mice expressed these molecules strongly (Fig. 5B), and naive

FIGURE 2. Typical tissue architecture of the lymphoid clusters in the GM lesions. Serial frozen sections of stomach isolated from AIG mice were stainedwith Abs against CD3 (T cells), B220 (B cells), PECAM-1 (endothelial cells), ER-TR7 (reticular fibroblast), CR1 (follicular dendritic cells), or Mac-1(myeloid cells), and analyzed by confocal imaging. A, Transverse view of GM containing three isolated lymphoid clusters (arrows). B, Serial transversesections of a cluster. Clear T/B segregation, associated blood vessel (arrowhead), and lymphoid RN (arrow) are shown. C, Three independent clusters inapical-basal plane sections.

4361The Journal of Immunology

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

T cells are present in the GLN to some extent (11). In contrast, wedetected no substantial signals for PNAd or SLC in the GM lym-phoid tissues, suggesting that the absence of these factors at thelesional endothelial cells prevents the migration of naive T cellsinto the GM.

Taken together, these findings demonstrate that the architectureof the lymphoid clusters developed in the GM is well organizedand embodies some aspects of secondary lymphoid tissue. We con-clude that these clusters are kind of tertiary lymphoid tissues in-duced by chronic inflammation. Because the animals used in thisstudy were maintained under conventional conditions, some infec-tious agent(s) might have influenced the development of lymphoidtissue in the AIG animals. However, even under specific pathogen-free conditions, we have observed obviously clustered lympho-cytic infiltration in the GM lesion (data not shown), and the clus-ters had the same features of tertiary lymphoid tissue as shownabove.

Number and size of GM lymphoid tissue, and correlation withautoantibody titer in sera

The number of lymphoid clusters observed in the sections alongthe greater curvature of the stomach wall in AIG mice increasedgradually as the age of the mice increased (Table I). In addition,the size of each cluster was larger (data not shown) and the mat-uration of the FDC network as determined by CR1 or CXCL13staining was more advanced in older AIG mice than in youngerones (Table I). These findings support the idea that the long-term

retention of activated lymphocytes in peripheral tissue enables thedevelopment of organized lymphoid architecture. We also foundthat the number of the clusters and the titer of anti-GM autoanti-body in the serum were significantly correlated (Fig. 6). Therefore,the lymphoid cluster development in the GM lesion should be avaluable hallmark for evaluating AIG progression. These findingsalso suggest that tertiary lymphoid tissues in the target organ playa pivotal role in the autoantibody production by self-specific Bcells and in disease progression, although the definitive evidencesare as yet absent.

Immune cell subsets in the GM of AIG mice

To assess the composition of the immune cell subsets infiltratinginto chronic GM lesions, we checked several markers on isolatedcells and compared the GM, GLNs, and PBLs from normal andAIG mice by flow cytometry (Fig. 7). As expected, AIG micesuffered from T cell lymphopenia caused by the neonatal thymec-tomy, resulting in a marked decrease in the cells positive for T cellmarkers such as Thy-1.2, CD4, and CD8 in blood as well as in theGLN. By contrast, B cells were markedly increased, especially inthe GLNs of AIG mice, reflecting the vigorous proliferation of Bcells in addition to T lymphopenia. Interestingly, the cell popula-tion in the inflamed GM exhibited some similarity to that in normallymph nodes as to T cell-B cell ratio, although the CD4� andCD8� composition was slightly different and Mac-1� cells con-stituted a substantial fraction of the cells within the GM of AIGmice. A striking difference was observed between the cell popu-lations in the GM and the GLN in AIG mice, suggesting that theongoing immune reactions in these two sites are quite different.We could not isolate enough leukocytes from the GM of normalmice to analyze the cell populations (see Fig. 8, normal GMprofiles).

It seemed possible that peritoneal B-1 cells migrate to the GMand form the follicle-like B cells area of the lymphoid clusters,because B-1 cells are highly sensitive to CXCL13 attraction (19–21). We checked this point by staining GM cells with B-1 markers(Fig. 8). Typical B-1 cells in the peritoneal cavity showed theB220�CD5�Mac-1� phenotype in normal controls, while a largepart of the B220� cells in the GLN were CD5�Mac-1� conven-tional B (B-2) cells, and this was also the case in AIG mice. In theGM of AIG mice, most B220� cells expressed neither CD5 norMac-1, thus indicating that the follicular structure in the GM iscomposed of conventional B cells derived from the circulation. Bcells within the GM follicle were actually IgM�IgD� follicular-type B cells comparable to those in secondary lymphoid organs(Fig. 5C). In contrast, IgG�syndecan-1�B220low plasma cellswere distributed outside the follicles and scattered beneath the gas-tric epithelial layer (Fig. 5C, arrows). This situation again corre-sponds to that in the secondary organ in which plasma cells arelocated in extrafollicular sites, especially in the medula or red pulp(22).

Cytokine and chemokine expression in the GM of AIG mice

Lymphocyte migration or homing is regulated by various chemo-kines (23, 24), and therefore we next examined the expressionprofile of chemokines by RT-PCR analysis (Fig. 9). GLNs fromboth normal and AIG mice showed substantial expression of allchemokines assessed. As expected, CXCL13 expression wasmarkedly augmented in the lesional GM compared with normalcontrols. In addition, we observed a slight augmentation of theCCL21 signal in AIG GM by RT-PCR compared with normal GM,in which weak basal expression was constantly detected (Fig. 9),although we could not detect CCL21 in the GM lymphoid clustersby immunohistochemistry (Fig. 5B). This may indicate that the

FIGURE 3. Tissue architecture of GLN as assessed by confocal fluo-rescence immunohistochemistry. Serial frozen sections of GLN isolatedfrom normal and AIG mice were stained with several Abs to detect CD3 (Tcells), B220 (B cells) (A), ER-TR7 (reticular fibroblasts) (B), or CR1 (fol-licular dendritic cells) (C). All images are shown at the same magnification.

4362 TERTIARY LYMPHOID TISSUE IN AIG

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

CCL21 induction due to chronic inflammation outside the lym-phoid cluster, probably on the tissue lymphatic endothelial cellsthat constitutively produce CCL21 (25), occurs in the GM of AIGmice. Another homeostatic chemokine, CCL19, was weak or un-detectable, while CXCL12 was constantly detected in normal andAIG GM. The signals of inflammatory chemokines such as CCL4,CCL5, CXCL9, and CXCL10, all of which have been shown to beinvolved in Th1 cell migration (23), were readily detected frominflamed GM, while normal GM exhibited only faint signals. Incontrast, much weaker signals compared with those from normalGLN standards were detected for Th2 attractants, CCL17 andCCL22, from the GM of both normal and AIG mice.

Gastric lesions in this disease model have been shown to con-stitute a Th1-type-biased, memory T cell-dependent microenviron-ment, as expected for the expression patterns of IFN-� and IL-4(11) (Fig. 9). This situation is markedly different from that in re-

gional lymph nodes, in which Th1- and Th2-type reactions arecoactivated. In addition, LT-�, a Th1 cytokine and key factor forthe development of lymphoid organs (26, 27), is easily detected inthe AIG GM, suggesting that the activated lymphocytes constitu-tively supply LT-� throughout the chronic phase of gastric inflam-mation. Therefore, the pattern of cytokine and chemokine expres-sion at each site in AIG mice well reflects effector cell localization.These findings suggest that, in contrast to the regional lymph node,the gastric lymphoid tissue of AIG mouse has the unique feature ofsupporting only a Th1-type immunomicroenvironment.

DiscussionTertiary lymphoid tissue is defined as a lymphoid structure in-duced in ectopic tissue, while the formation of secondary lymphoidtissue is developmentally programmed (2, 8, 28). In the intestine,several kinds of predetermined lymphoid tissues, such as Peyer’s

FIGURE 4. CXCL13 expression in FDCs at the GM follicle. A, Serial frozen sections of GLNs and GM from AIG mice were stained with severalcombinations of Abs and analyzed by confocal imaging. B, CXCL13� FDC network on B cell follicle in GM (upper panels) and higher magnification viewfocused on the cluster stained with CR1, CXCR13, or ER-TR7 (lower panels). C, Schematic representation of a typical GM lymphoid tissue. B, B cell area(follicle); T, T cell area; FRC, fibroblastic reticular cell; BV, blood vessel; EP, epithelial cell layer; LP, lamina propria; MM, muscularis mucosae; PC,plasma cell; SM, submucosa.

4363The Journal of Immunology

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

patches, isolated lymphoid follicles, and crypt patches, are ob-served (28–30). However, under normal conditions, there is noremarkable lymphoid organization in the stomach. In this study,

we found an organized lymphoid tissue associated with chronicinflammation in the GM of AIG mice. Although the sizes are rel-atively small (100–500 �m in diameter), each lymphoid cluster

Table I. Summary of tertiary lymphoid tissue development in the GM of AIG mice

Age(wk) n

Lymphoid Tissue with T/Ba Lymphoid Tissue with CR1 Lymphoid Tissue with CXCL13

Miceb No./sec.c Miceb No./sec.c Miceb No./sec.c

6 5 3 (60%) 0.8 � 0.4 2 (40%) 0.8 � 0.5 0 (0%) 0 � 010–13 5 4 (80%) 2.8 � 1.5 2 (40%) 1.3 � 1.0 2 (40%) 1.3 � 1.515 4 3 (75%) 4.0 � 2.6 3 (75%) 1.7 � 0.6 3 (75%) 1.3 � 0.620 5 4 (80%) 4.0 � 2.4 4 (80%) 1.6 � 1.1 3 (60%) 1.3 � 1.024 4 3 (75%) 6.0 � 1.7 3 (75%) 3.6 � 1.5 3 (75%) 3.3 � 1.228–32 4 4 (100%) 9.4 � 6.6 4 (100%) 5.9 � 5.2 4 (100%) 4.1 � 3.6

a T/B indicates the clear segregation of T and B cells.b Number of mice in which one or more of the indicated types of tertiary lymphoid tissue were observed in the sections.c Number � SD of indicated tertiary lymphoid tissue in the section.

FIGURE 5. A, Absence of typicalGC morphology in GM follicle. Serialfrozen sections of GLNs and GMsfrom AIG mice were stained withFITC-conjugated PNA and Absagainst several markers and analyzedby confocal imaging. Mature GCswere easily detected at follicles inGLNs (arrow). B, Absence of PNAdand CCL21 (SLC) expression on en-dothelial cells associated with GMtertiary lymphoid tissue. Serial sec-tions were stained with Abs againstendothelial cell marker PECAM-1 orHEV marker PNAd or SLC. Asterisksindicate the positions of GM follicleson the serial sections. No remarkablesignals for PNAd or CCL21 were de-tected in the GM lymphoid tissue (ar-rows). Arrowheads show some non-specific staining of PNAd on gastricepithelial cells. C, Extrafollicular dis-tribution of IgG�syndecan-1� plasmacells in the GM lesions (arrow). Serialsections were stained with Absagainst Ig subclasses and a plasmacell marker, syndecan-1.

4364 TERTIARY LYMPHOID TISSUE IN AIG

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

had a clear B cell follicle and T cell area with apical-basal polarity.Typical lymphoid stromal structures, such as a reticular meshworkand FDC network expressing CXCL13, support the cluster. Al-though the overall architecture of these lymphoid clusters wasquite similar to that of secondary lymphoid tissues, several crucialdifferences were evident. First, the development of typical GC inthe GM lymphoid follicles was quite rare. Second, naive T cellswere virtually absent in the GM lesions, a finding supported by thelack of detectable expression of HEV markers (which should berequired for naive T cell homing) on endothelium associated withthe lymphoid clusters. Third, Th2 cell itself as well as the expres-sion of Th2-type chemokines are negligible in the GM.

Ectopic lymphoid tissues have been reported in various diseases(2–7). Most cases exhibit obvious B cell follicular structures withor without GC; however, T-B cell compartmentalization is lessdefined. Using multicolor confocal microscopy, we clearly showedT-B segregation within the GM lymphoid cluster. Therefore, sim-ilarly to within secondary lymphoid tissue, there were distinct sub-compartments inside tertiary lymphoid tissue. Interestingly, GMfollicles were always positioned on the luminal side of the mucosaas compared with the T cell area, suggesting that B cells and/orFDCs have some preference for the epithelial layer side.

Stromal cells expressing CXCL13 at the ectopic follicle havealso been reported in human rheumatoid arthritis and Helicobacterpylori-induced follicular gastritis (3, 6), suggesting an importantrole of CXCL13 in the development and maintenance of the fol-licular structure in peripheral tissues. CXCL13 as well as lympho-toxins regulate secondary lymphoid tissue development and theproper organization of the lymphoid structures (14, 15, 26, 27).Forced expression of these factors in nonlymphoid tissue caninduce neolymphoid genesis, and the constitutive production ofone of these factors is sufficient for triggering the program of thelymphoid tissue development (31–33). It has also been shown thatLT-� plays a crucial role in the expression of CXCL13 from stro-mal cells (34). In the same way, LT-� from long-lasting activatedlymphocytes in chronic lesions should induce CXCL13 productionby stromal cells and the formation of follicles or other associatedlymphoid architecture. Indeed, in the AIG model, LT-� expressionis easily detected in the GM. In addition, as the number and size of

FIGURE 6. Correlation between the number of tertiary lymphoid tissueand the serum autoantibody titer. Titers of serum anti-GM autoantibodieswere determined by ELISA as the greatest dilution at which the absorbanceat 450 nm was �0.1. Lymphoid tissue lesion numbers were determined bycounting the CD3�/B220� clusters observed in the sections along thegreater curvature of the glandular stomach wall from the cardia to thepylorus. Data from various animals (age 6–24 wk) were plotted on asemilogarithmic.

FIGURE 7. Composition of immune cell subsets in PBL, GLN, and GMfrom normal and AIG mice. Isolated cells from more than five mice (age10–20 wk) were stained with various Abs and analyzed by flow cytometry.The results are shown as the percentage of positive cells in the leukocytegate.

FIGURE 8. Infiltration of conventional B cells, but not B-1 cells, intothe GM lesions of AIG mice. Isolated cells stained with B220/CD5 orB220/Mac-1 were analyzed by flow cytometry. Results are shown as dotplots. B-1 cells (B220�CD5�, B220�Mac-1�) and B-2 cells(B220�CD5�, B220�Mac-1�) were gated, respectively, and shown as thepercentage in leukocyte gated cells.

4365The Journal of Immunology

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

GM lymphoid clusters become larger as the age of the mice in-creases, maturation of this ectopic lymphoid organization seems todepend on how long the activated lymphocytes reside in the lesion.Therefore, the findings in AIG support the above idea about thedevelopment of tertiary lymphoid tissue.

Of prime importance, the GM lymphoid tissue supports only aTh1-type response, while both Th1- and Th2-type responses areinduced in the GLN (11, 35). Chemokine expression in the GM isalso biased to target Th1-type effector cells as well as CXCL13production, but not to target Th2-type effector cells. This is well inaccordance with the subset localization in AIG mice. Recently, wehave shown that large vessels adjacent to the lymphoid foci ex-pressed P-selectin and mucosal addressin cell adhesion mole-cule-1, and possibly recruited Th1-type effector cells that showedP-selectin-ligand� and �4�7-integrinhigh phenotypes (35). Inaddition, the absence of HEV factors such as PNAd and CCL21in vessels in the GM lymphoid tissue is likely the causative thatGM-infiltrating T cells showed exclusively an activated/memory phenotype (11). Based on these findings, we speculatethat the selective expression of chemoattractants and adhesionmachinery in the GM lesion enables the gathering of onlyTh1-type-biased, activated/memory effectors, resulting in a Th1-type vicious cycle.

It is possible that the lymphoid tissue that develops in the GMof AIG mice has an important role in maintaining the autoimmunereaction with regard to autoantibody production. Pervious studieshave shown that CD4� T cells, but not CD8� T cells or autoan-tibodies from AIG animals can transfer the disease to nude mice,indicating that the initiation of self reactivity is primarily mediatedby CD4� T cells (36). However, the roles of self-reactive B cellsand autoantibodies in the disease progression of AIG are mostlyunknown. B cells in the GM are conventional type and thus de-rived from the circulating pool, rather than B-1 type, which havebeen suggested to be self reactive (18), from the peritoneal cavity.At least two B cell subpopulations infiltrated in GM: 1)IgM�IgD�IgG� cells in the GM follicles. They are similar tothose in secondary tissue follicles; therefore, they might be of na-ive phenotype. 2) Parafollicular plasma cells of IgG�syndecan-1�B220low phenotype. It is likely that the source of the autoanti-body produced in the GM of AIG or by GM-infiltrating cellsobserved in vitro (11) is the latter plasma cells. Titer of serumautoantibody and the number of GM lymphoid tissue lesions ac-tually showed a significant correlation, suggesting an importantrole of the ectopic follicle in the target tissue for local autoantibodyproduction. However, typical GC morphology was rarely observed

FIGURE 9. Cytokine and chemokine mRNA profiles expressed in GLN and GM. To detect transcripts, semiquantitative RT-PCR analysis was per-formed. Five-fold serial dilutions of cDNA and reaction products produced without reverse transcriptase (RT�) were amplified with specific primer pairsand separated by electrophoresis. The gel images obtained are shown as the black-white reversed form. Two examples of representative AIG mice werecompared with a normal mouse.

4366 TERTIARY LYMPHOID TISSUE IN AIG

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

in the lymphoid tissue of the AIG stomach. This could be ex-plained in part by the absence of Th2 cells in the lesion, becauseof the critical importance of a Th2 cytokine, IL-4, for mucosal GCformation (37). There is as yet no evidence supporting the notionthat these plasma cells are actually differentiated from GM follic-ular B cells de novo. Therefore, we cannot explain the role of theGM follicular B cells at present.

IgG2a and IgG1 are the major autoantibody classes in sera (11)(unpublished data). These autoantibodies are thought to be pro-duced in the GLN, GM, and spleen of AIG mice. Both Th1 andTh2 cells are actually activated in the GLN and spleen (11), andtherefore there is a consistency between the Th subsets and Igsubclasses as a whole in this disease model. If a fraction of theplasma cells in the GM are from GM follicles in which they havematured without making clear GCs, their Ig subclass should bepredominantly the Th1-supported IgG2a because of an extremelyTh1-biased GM environment. However, it is also likely that theIgG1 plasma cells as well as IgG2a plasma cells that differentiatedat other sites can migrate into the GM and produce IgG1 autoan-tibody. We have actually observed the production of both IgG1and IgG2a autoantibodies from isolated GM cells in vitro (unpub-lished data). We think that in fact plasma cells may be derivedfrom GM follicles as well as GLN.

Concerning the formation of GC, a notable finding was reportedby Oshima et al. (7). They reported that a follicular gastritis char-acterized by large GCs in the GM was induced by the infection ofneonatally thymectomized BALB/c mice with H. pylori. Thepathophysiological features of H. pylori-infected AIG mice seemto be different from those of noninfected, AIG-bearing animals inseveral respects. For instance, H. pylori-infected AIG mice exhib-ited a decrease of the gland atrophy, preservation of parietal cells,and the augmentation of IL-4 expression in the GM (7). Based onthese findings, we can postulate that preformed ectopic lymphoidtissue in AIG mice is a crucial prerequisite for the subsequentdevelopment of severe follicular gastritis and gastric GC inducedby H. pylori infection. H. pylori-derived bacterial component(s)may alter the immunological character of the GM lymphoid tissueand the activation level of B cells with respect to the induction ofmature GC. This offers the promise that GC formation in H. pylori-induced follicular gastritis could be prevented by controlling theimmunological state of the tertiary lymphoid tissue.

The detailed immunological function of tertiary lymphoid tissuein the disease progression and persistence is still largely unknown.From our recent findings, additional questions emerge. Do dy-namic immune reactions, such as autoantigen presentation fromdendritic cells to Th1 cells, resulting in their activation and pro-liferation, and help for B cells to induce the affinity maturation ofIg, occur within the GM lymphoid tissue? Why does GM lymphoidtissue support only Th1-type reactions mediated by activated/memory T cells, although the tissue architecture is quite similar tothat of secondary lymphoid tissue, which supports both Th1- andTh2-type reactions, as well as naive T cells? What is the truesignificance of the differences in the immune reactions that occurin the secondary and the tertiary lymphoid tissue?

Fortunately, the d3-Tx AIG model has several advantages fortertiary lymphoid tissue studies. For instance, because the GMlymphoid structure appears to rather strictly follow some rules, wewill be able to compare in more detail the tertiary structure vs thesecondary structure, including the behavior of immune cell sub-sets, and to evaluate subtle changes in different disease conditions.Moreover, kinetic studies of the development and maturation oftertiary lymphoid tissue could also be performed. Therefore, novelfindings from the AIG model will give some clues to answering theabove questions.

AcknowledgmentsWe thank Dr. T. Masuda for useful discussions; Dr. T. Honjo for use of aconfocal microscope; Drs. K. Tashiro and N. Kanazawa for PCR primers;and T. Ohfuji for excellent technical assistance.

References1. Zinkernagel, R. M., S. Ehl, P. Aichele, S. Oehen, T. Kundig, and H. Hengartner.

1997. Ag localization regulates immune responses in a dose- and time-dependentfashion: a geographical view of immune reactivity. Immunol. Rev. 156:199.

2. Hjelmstrom, P. 2001. Lymphoid neogenesis: de novo formation of lymphoidtissue in chronic inflammation through expression of homing chemokines. J. Leu-kocyte Biol. 69:331.

3. Takemura, S., A. Braum, C. Crowson, P. J. Kurtin, R. H. Cofield, W. M.O’Fallon, J. J. Goronzy, and C. M. Weyand. 2001. Lymphoid neogenesis inrheumatoid synovitis. J. Immunol. 167:1072.

4. Chvatchko, Y., M. H. Kosco-Vilbois, S. Herren, J. Lefort, and J.-Y. Bonnefoy.1996. Germinal center formation and local Ig E (IgE) production in the lung afteran airway antigenic challenge. J. Exp. Med. 184:2353.

5. Yoneyama, H., K. Matsuno, Y. Zhang, M. Murai, M. Itakura, S. Ishikawa,G. Hasegawa, M. Naito, H. Asakura, and K. Matsushima. 2001. Regulation by che-mokines of circulating dendritic cell precursors, and the formation of portal tract-associated lymphoid tissue, in a granulomatous liver disease. J. Exp. Med. 193:35.

6. Mazzucchelli, L., A. Blaser, A. Kappeler, P. Scharli, J. A. Laissue, M. Baggiolini,and M. Uguccioni. 1999. BCA-1 is highly expressed in Helicobacter pylori-induced mucosa-associated lymphoid tissue and gastric lympoma. J. Clin. Invest.104:R49.

7. Oshima, C., K. Okazaki, Y. Matushima, M. Sawada, T. Chiba, K. Takahashi,H. Hiai, T. Katakai, S. Kasakura, and T. Masuda. 2000. Induction of folliculargastritis following postthymectomy autoimmune gastritis in Helicobacter pylori-infected BALB/c mice. Infect. Immun. 68:100.

8. Picker, L. J., and E. C. Butcher. 1992. Physiological and molecular mechanismsof lymphocyte homing. Annu. Rev. Immunol. 10:561.

9. Kojima, A., O. Taguchi, and Y Nishizuka. 1980. Experimental production ofpossible autoimmune gastritis followed by macrocytic anemia in athymic nudemice. Lab. Invest. 42:389.

10. Katakai, T., Y. Agata, A. Shimizu, C. Ohshima, A. Nishio, M. Inaba,S. Kasakura, K. J. Mori, and T. Masuda. 1997. Structure of the TCR expressedon a gastritogenic T cell clone, II-6, and frequent appearance of similar clono-types in mice bearing autoimmune gastritis. Int. Immunol. 9:1849.

11. Katakai, T., K. J. Mori, T. Masuda, and A. Shimizu. 1998. Differential localiza-tion of Th1 and Th2 cells in autoimmune gastritis. Int. Immunol. 10:1325.

12. Mori, Y., K. Fukuma, Y. Adachi, K. Shigeta, R. Kannagi, H. Tanaka, M. Sakai,K. Kuribayashi, H. Uchino, and T. Masuda. 1989. Parietal cell autoantigens in-volved in neonatal thymectomy-induced murine autoimmune gastritis: studiesusing monoclonal autoantibodies. Gastroenterology 97:364.

13. Agata, Y., T. Katakai, S.-K. Ye, M. Sugai, H. Gonda, T. Honjo, K. Ikuta, andA. Shimizu. 2001. Histone acetylation determines the developmentally regulatedaccessibility for TCR � gene recombination. J. Exp. Med. 193:873.

14. Gunn, M. D., V. N. Ngo, K. M. Ansel. E. H. Ekland, J. G. Cyster, andL. T. Williams. 1998. A B cell-homing chemokine made in lymphoid folliclesactivates Burkitt’s lymphoma receptor-1. Nature 391:799.

15. Ansel, K. M., V. N. Ngo, P. L. Hyman, S. A. Luther, R. Forster, J. D. Sedgwick,J. L. Browning, M. Lipp, and J. G. Cyster. 2000. A chemokine-driven positivefeedback loop organizes lymphoid follicles. Nature 406:309.

16. Yeh, J.-C., N. Hiraoka, B. Petryniak, J. Nakayama, L. G. Ellies, D. Rabuka,O. Hindsgaul, J. D. Marth, J. B. Lowe, and M. Fukuda. 2001. Novel sulfatedlymphocyte homing receptors and their control by a core 1 extension �1,3-N-acetylglucosaminyltransferase. Cell 105:957.

17. Nakano, H., S. Mori, H. Yonekawa, H. Nariuchi, A. Matsuzawa, and T. Kakiuchi.1998. A novel mutant gene involved in T-lymphocyte-specific homing into pe-ripheral lymphoid organs on mouse chromosome 4. Blood 91:2886.

18. Gunn, M. D., S. Kyuwa, C. Tam, T. Kakiuchi, A. Matsuzawa, L. T. Williams, andH. Nakano. 1999. Mice lacking expression of secondary lymphoid organ che-mokine have defects in lymphocyte homing and dendritic cell localization.J. Exp. Med. 189:451.

19. Ishikawa, S., T. Sato, M. Abe, S. Nagai, N. Onai, H. Yoneyama, Y. Zhang, T. Suzuki,S. Hashimoto, T. Shirai, et al. 2001. Aberrant high expression of B lymphocytechemokine (BLC/CXCL13) by CD11b�CD11c� dendritic cells in murine lupus andpreferential chemotaxis of B1 cells toward BLC. J. Exp. Med. 193:1393.

20. Ansel, K. M., R. B. S. Harris, and J. G. Cyster. 2002. CXCL13 is requred for B1cell homing, natural Ab production, and body cavity immunity. Immunity 16:67.

21. Fagarasan, S., N. Watanabe, and T. Honjo. 2000. Generation, expansion, migra-tion and activation of mouse B1 cells. Immunol. Rev. 176:205.

22. Hargreaves, D. C., P. L. Hyman, T. T. Lu, V. N. Ngo, A. Bidgol, G. Suzuki, Y.-R.Zou, D. R. Littman, and J. G. Cyster. 2001. A coordinated change in chemokineresponsiveness guides plasma cell movements. J. Exp. Med. 194:45.

23. Sallusto, F., A. Lanzavecchia, and C. R. Machay. 1998. Chemokines and che-mokine receptors in T cell priming and Th1/Th2-mediated responses. Immunol.Today 19:568.

24. Ansel, K. M., and J. G. Cyster. 2001. Chemokine in lymphopoiesis and lymphoidorgan development. Curr. Opin. Immunol. 13:172.

25. Kriehuber, E., S. Breiteneder-Geleff, M. Groeger, A. Soleiman, S. F. Schoppmann,G. Stingl, D. Kerjaschki, and D. Maurer. 2001. Isolation and characterization ofdermal lymphatic and blood endothelial cells reveal stable and functionally special-ized cell lineages. J. Exp. Med. 194:797.

4367The Journal of Immunology

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

26. De Togni, P., J. Goellner, N. H. Ruddle, P. R. Streeter, A. Fick, S. Mariathasan,S. C. Smith, R. Carlson, L. P. Shornick, and J. Strauss-Schoenberger. 1994.Abnormal development of peripheral lymphoid organs in mice deficient in lym-photoxin. Science 264:703.

27. Koni, P. A., R. Sacca, P. Lawton, J. L. Browning, N. H. Ruddle, andR. A. Flavell. 1997. Distinct roles in lymphoid organogenesis for lymphotoxins� and � revealed in lymphotoxin �-deficient mice. Immunity 6:491.

28. Fu, Y.-X., and D. D. Chaplin. 1999. Development and maturation of secondarylymphoid tissues. Annu. Rev. Immunol. 17:399.

29. Hamada, H., T. Hiroi, Y. Nishiyama, H. Takahashi, Y. Masunaga, S. Hachimura,S. Kaminogawa, H. Takahashi-Iwanaga, T. Iwanaga, H. Kiyono, et al. 2002.Identification of multiple isolated lymphoid follicles on the antimesenteric wall ofthe mouse small intestine. J. Immunol. 168:57.

30. Kanamori, Y., K. Ishimura, M. Nanno, K. Maki, K. Ikuta, H. Nariuchi, andH. Ishikawa. 1996. Identification of novel lymphoid tissue in murine intestinalmucosa where clusters of c-kit� IL-7R� Thy1� lympho-hematopoietic progen-itors develop. J. Exp. Med. 184:1449.

31. Kratz, A., A. Campos-Neto, M. S. Hanson, and N. H. Ruddle. 1996. Chronic in-flammation caused by lymphotoxin is lymphoid neogenesis. J. Exp. Med. 183:1461.

32. Hjelmstrom, P., J. Fjell, T. Nakagawa, R. Sacca, C. A. Cuff, and N. H. Ruddle.2000. Lymphoid tissue homing chemokines are expressed in chronic inflamma-tion. Am. J. Pathol. 156:1133.

33. Luther, S. A., T. Lopez, W. Bai, D. Hanahan, and J. G. Cyster. 2000. BLCexpression in pancreatic islets causes B cell recruitment and lymphotoxin-depen-dent lymphoid neogenesis. Immunity 12:471.

34. Ngo, V. N., H. Korner, M. D. Gunn, K. N. Schmidt, D. S. Riminton,M. D. Cooper, J. L. Browning, J. D. Sedgwick, and J. G. Cyster. 1999. Lym-photoxin �/� and TNF are required for stromal cell expression of homing che-mokines in B and T cell areas of the spleen. J. Exp. Med. 189:403.

35. Katakai, T., K. J. Mori, T. Masuda, and A. Shimizu. 2002. Selective accumula-tion of type 1 effector cells expressing P-selectin ligand and/or �4�7-integrin atthe lesions of autoimmune gastritis. Int. Immunol. 14:167.

36. Gleeson, P. A., B.-H. Toh, and I. R. van Driel. 1996. Organ-specific autoimmu-nity induced by lymphopenia. Immunol. Rev. 149:97.

37. Vajdy, M., M. H. Kosco-Vilbois, M. Kopf, G. Kohler, and N. Lycke. 1995.Impaired mucosal immune responses in IL 4-targeted mice. J. Exp. Med.181:41.

4368 TERTIARY LYMPHOID TISSUE IN AIG

by guest on March 26, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from