The Role of CD8+ CD40L+ T Cells in the Formation of Germinal Centers in Rheumatoid Synovitis

This information is current as Ulf G. Wagner, Paul J. Kurtin, Andrea Wahner, Marc of October 1, 2021. Brackertz, Daniel J. Berry, Jörg J. Goronzy and Cornelia M. Weyand J Immunol 1998; 161:6390-6397; ; http://www.jimmunol.org/content/161/11/6390 Downloaded from

References This article cites 32 articles, 18 of which you can access for free at: http://www.jimmunol.org/content/161/11/6390.full#ref-list-1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on October 1, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1998 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Role of CD8؉ CD40L؉ T Cells in the Formation of Germinal Centers in Rheumatoid Synovitis1

Ulf G. Wagner,*§ Paul J. Kurtin,‡ Andrea Wahner,* Marc Brackertz,* Daniel J. Berry,† Jo¨rg J. Goronzy,* and Cornelia M. Weyand2*

In rheumatoid synovitis, can be arranged in follicular structures resembling secondary lymphoid follicles. To un- derstand the organizing principles of this ectopic lymphoid tissue, the cellular components contributing to synovial follicles were examined. In 9 of 24 synovial tissue biopsies, lymphoid aggregates were found consisting of CD4؉ T cells and CD20؉ B cells. In four of the nine patients, the follicular centers were occupied by CD23؉ CD21؉ cellular networks representing follicular dendritic cells involved in germinal center reactions. In five patients, CD23؉ cells were absent from the centers of the aggregates, suggesting that fully developed germinal centers are generated in only a subset of patients. To identify factors involved in the regulation of the synovial microarchitecture, cell populations contributing to the follicles were quantified by digital image analysis of immu- Downloaded from nostained tissue and by flow cytometry of tissue-derived lymphocytes. Proportions of CD4؉, CD20؉, and CD68؉ cell subsets were surprisingly invariant, irrespective of the presence or absence of CD23؉ follicular dendritic cells. Instead, tissue biopsies with CD23؉ germinal center-like regions could be distinguished from those with CD23؊ - aggregates by a fourfold increase in the frequency of tissue-infiltrating CD8؉ T cells, a fraction of which expressed CD40 ligand (CD40L). The data suggest a previously unsuspected role of CD8؉ lymphocytes in modulating germinal center formation and raise the possibility that CD8؉ ؉

CD40L T cells are involved in aggravating pathologic immune responses in rheumatoid synovitis. The Journal of Immunology, http://www.jimmunol.org/ 1998, 161: 6390–6397.

n rheumatoid arthritis (RA3), the synovial membrane of af- The follicular formations adopted by T cells and B cells in rheu- fected joints undergoes a series of changes that result in the matoid synovium have microscopic characteristics reminiscent of I formation of a proliferative tissue exhibiting invasive and germinal centers (GCs) (4, 5). GCs are specialized microanatom- destructive features. Although the factors that initiate rheumatoid ical structures that are required for the generation of high-affinity synovitis have not been identified, recruitment of T cells, B cells, Abs and the selection of memory B cells in response to Ags and into the membrane, hypertrophy, and hyperpla- (6–8). Immunization with Ag induces activation of T cells and B sia of synovial lining cells and a shift in the phenotype and func- cells in the T cell-rich areas of secondary lymphoid tissues with by guest on October 1, 2021 tion of synovial fibroblasts are accepted as principal events (1, 2). cognate- and costimulation-dependent expansion of Ag-specific In general, lymphocytes are infrequent in the hyperplastic lining lymphocytes (9). Activated T cells and B cells from these popu- layer, which is composed of a mixture of highly activated macro- lations migrate into adjacent B cell zones to form GCs (10). GCs phage-like cells and proliferating resident fibroblast-like cells. arise when B cells accumulate among the processes of follicular Lymphocytes migrate into the sublining tissue, where they accu- dendritic cells (FDCs) and undergo intense proliferation, apopto- mulate around blood vessels or infiltrate into the stroma. Evidence sis, and V(D)J gene hypermutation. The current paradigm holds suggests that the process of tissue infiltration is organized, because that somatically mutated cell surface Igs are probed on Ags bound the lymphocytes acquire a defined structural organization. T cells to FDCs and that B cells producing high-affinity Abs are selected and B cells are often organized into follicular structures that can to survive, while B cells that fail to recognize Ag die by apoptotic resemble lymphoid follicles. Besides their pseudofollicular orga- programmed cell death (6, 11). Only 10% of GC cells are CD3ϩ nization, tissue-infiltrating lymphocytes may also be involved in T cells, virtually all of which express the CD4 marker (12). CD8ϩ forming typical granulomas containing giant cells (3). T cells account for Ͻ3% of the T cells. Although they represent a minority of GC cells, CD4ϩ T cells are requisite for GC formation. In murine studies, CD4ϩ T cells migrating into the GC bear ␣␤ Departments of *Medicine, Division of Rheumatology, †Orthopedic Surgery, and ‡Laboratory Medicine and Pathology, Mayo Clinic and Foundation, Rochester, MN TCRs reactive to the Ag driving the GC reaction (13, 14). 55905; and §Department of Medicine IV, University of Leipzig, Leipzig, Germany Molecular mechanisms involved in GC formation are beginning Received for publication April 13, 1998. Accepted for publication July 21, 1998. to be understood. Patients with X-linked immunodeficiency caus- The costs of publication of this article were defrayed in part by the payment of page ing hyper-IgM production do not generate secondary Ab re- charges. This article must therefore be hereby marked advertisement in accordance sponses, lack memory B cells, and possess lymph nodes with pri- with 18 U.S.C. Section 1734 solely to indicate this fact. mary follicles lacking GC reactions (15). The molecular defect 1 Supported in part by grants from the National Institutes of Health (RO1 AR41974 and RO1 AR42527), by a National Arthritis Foundation Biomedical Science grant underlying this syndrome has been localized to the CD40 ligand (AF #16), and by the Mayo Foundation. U.G.W. was the recipient of fellowships from (CD40L) gene (16, 17). According to the current model, CD40- the Deutscher Akademischer Austauschdienst (DAAD) and the Stiftung Familie Klee. CD40L interactions are critical for the recruitment of GC precur- 2 Address correspondence and reprint requests to Dr. C. M. Weyand, Mayo Clinic, sors, with CD40Lϩ T cells and interdigitating dendritic cells pro- 200 First Street SW, Rochester, MN 55905; E-mail address: weyand.cornelia@ mayo.edu viding signals to B cell precursors in extrafollicular areas (18–20). 3 B cell memory generation appears to be regulated by the interac- Abbreviations used in this paper: RA, rheumatoid arthritis; CD40L, CD40 ligand; ϩ ϩ FDC, follicular dendritic cell; GC, germinal center. tion of CD40 centrocytes with CD40L intrafollicular T cells

Copyright © 1998 by The American Association of Immunologists 0022-1767/98/$02.00 The Journal of Immunology 6391

Table I. Demographics of the study population ME); and anti-IgD (A093, 1:1000, Dako). Anti-CD40L (anti-gp39, Ancell, Bayport, MN) was used at a dilution of 1:500, and slides were incubated overnight at 4°C. Secondary species-specific Abs were applied for 30 min Disease Age at Disease Rheumatoid Duration at room temperature followed by detection with the streptavidin-biotin Patient Sex Onset (yr) Factor (yr) Biopsy Site complex immunoperoxidase or alkaline phosphatase technique. For microscopy, 3-amino-9-ethylcarbazole (AEC; Sigma, St. Louis, MO) or 1 Female 59 ϩ 11 Elbow 3,3Ј-diaminobenzidine tetrahydrochloride (DAB; Sigma) substrate solu- 2 Female 59 Ϫ 5 Knee tions were used. For digital fluorescence imaging, the Vectastain ABC 3 Female 26 ϩ 12 Hip and alkaline phosphatase substrate kit I (Vector Red, Vector Laboratories, 4 Male 75 Ϫ 4 Knee Burlinghame, CA) were used for detection. For two-color immunohisto- 5 Female 37 ϩ 38 Knee chemistry, slides were washed for 10 min in 1% Triton X-100 in PBS and 6 Female 54 ϩ 2 Wrist blocked with 5% normal goat serum before adding the second Ab. Sec- 7 Male 64 ϩ 10 Knee tions were counterstained with hematoxylin and permanently mounted 8 Female 17 ϩ 1 Shoulder in Cytoseal-280 (Stephens Scientific, Riverdale, NJ). 9 Male 48 ϩ 10 Elbow Digital imaging Fluorescence-stained tissue sections from serial sections were scanned us- ing a confocal microscope (LSM310, Carl Zeiss, Oberkochen, Germany) (21, 22). Other molecules implicated in GC reactions include the with an argon-krypton laser, excitation 568 nm. Fluorescence signals for integrin, VLA-4, on lymphocytes interacting with VCAM-1 on each Ab were translated into pseudocolors. For analysis of spatial relation- FDCs, possibly facilitating the adherence of B cells to the FDC ships, images from the immunohistochemical analysis for CD68, CD4, CD8, CD20, and CD23 expression were overlaid. Correct positioning of reticulum (23). Additionally, mice deficient in either lymphotoxin Downloaded from ␣ ␣ the images was ensured by placing the digitized scans on a light micro- (24), TNF- (25), or the type-I TNFR (26) fail to develop sec- scope background that showed tissue landmarks and cellular infiltrates. The ondary lymphoid follicles and GCs. In mice lacking expression of area stained with each Ab was calculated by an image analysis program either CD28 or its ligands CD80 and CD86, B cells accumulate in (KS 400, Kontron Elektronik, Munich, Germany) as a percentage of the the lymphoid follicle following antigenic challenge (27, 28). How- total area of the lymphoid aggregate. ever, these cells fail to undergo proliferative expansion, do not initiate GCs, and do not acquire somatic mutations, emphasizing Tissue digestion and flow cytometry http://www.jimmunol.org/ the importance of CD28-CD80/CD86 interactions in GC reactions. Fresh synovial tissue was cut into 2- to 4-mm3 pieces and washed once in The goal of the current study was to examine the follicular struc- complete medium (RPMI 1640, 10% FCS, 200 ␮M L-glutamine, 100 U/ml ␮ 3 tures formed by tissue-infiltrating lymphocytes in rheumatoid sy- penicillin, and 100 g/ml streptomycin). Then, 1 cm of tissue was incu- bated in 10 ml digestion solution (0.05 M HEPES buffer, 3 mg/ml type 1A novitis, with special emphasis on the cellular components required collagenase, 1 mg/ml hyaluronidase, and 0.1 mg/ml type IV deoxyribonu- for GC generation. Immunohistochemical studies combined with clease I in RPMI 1640) at 37°C for 30–45 min. Lymphocytes were isolated digital image analysis demonstrated that in only one-half of the using a Ficoll density-gradient and incubated with FITC-labeled anti- patients with follicular synovitis did T cell-B cell follicles contain CD40L (Ancell) and phycoerythrin-conjugated anti-CD4 or anti-CD8 Abs (Becton Dickinson) for 30 min at 4°C in the dark. all of the cellular components of classical GCs. In the remaining ϩ patients, CD23 FDCs were lacking in T cell-B cell aggregates Statistical analysis by guest on October 1, 2021 that otherwise had normal proportions of intrafollicular CD4ϩ T cells and CD20ϩ B cells, suggesting an aborted GC reaction. Sur- All statistical analyses were performed using SigmaStat software (Jandel, San Raffael, CA). Student’s t test and the nonparametric Mann-Whitney prisingly, patients with and without typical GCs could be distin- test were used where appropriate. guished based on the presence of CD8ϩ T cells. GC formation was ϩ associated with increased recruitment of CD8 T cells to both the Results border of T cell-B cell follicles as well as to interfollicular areas. Heterogeneity of follicular structures in rheumatoid synovium Classical GC reactions occurred in patients who had an expansion of CD40L-expressing CD8ϩ T cells. Because CD40-CD40L in- To identify factors involved in regulating the microarchitecture of teraction is critically involved in GC reactions, CD8ϩ CD40Lϩ T rheumatoid synovitis, synovial tissue biopsies from 24 consecutive cells may be able to support the generation of pathologic immune patients with unequivocal RA and active synovitis at the time of responses in rheumatoid synovium. collection were screened by standard light microscopy. Nine tis- sues had prominent lymphoid follicles in the sublining stroma and Materials and Methods were selected for further analysis. Besides having follicular orga- Patients nization, all of these samples also had diffuse mononuclear infil- trates and collections of small lymphocytes surrounding the cap- Synovial tissue specimens were obtained from patients who underwent illaries. Scattered lymphocytes were found in the interfollicular synovectomy or joint replacement surgery. Patients enrolled into this study fulfilled the diagnostic criteria for RA and had active disease at the time of zones. Clinical characteristics of the nine patients enrolled into the tissue biopsy (Table I). Tissue sections were stained with hematoxylin and study are given in Table I. The group included six females and eosin, and specimens with follicular lymphoid aggregates were selected for three males. Seven of the nine patients produced rheumatoid fac- further studies. tor. Disease duration ranged from 1–38 yr. Specimens were har- vested from the shoulder, elbow, wrist, hip, or knee. Immunohistochemistry analysis demonstrated that the lymphoid Synovial tissue was embedded and frozen in O.C.T. compound (Miles, aggregates in the synovium were composed of CD20ϩ B cells Elkhart, IN) and was cut into 5-␮m sections and mounted onto gelatin- ϩ ϩ ϩ Ϫ intermingled with CD3 CD4 CD45RO T cells. In some tissue coated slides. Slides were air dried, fixed in acetone, and stored at 80°C. ϩ ϩ Sections were fixed in 1% paraformaldehyde, blocked with 5% normal goat sections, a ring of CD4 T cells was arranged around the CD20 serum (Life Technologies, Grand Island, NY), and incubated for 30 min at B cells. Often CD4ϩ T cells and CD20ϩ B cells were not spatially room temperature with the following Abs: anti-CD4 (Leu-3a, dilution separated but formed a mixed population (Fig. 1). CD8ϩ T cells 1:100, Becton Dickinson, San Jose, CA); anti-CD8 (Leu-2a, 1:5, Becton Dickinson); anti-CD20 (L26, 1:50, Dako, Carpenteria, CA); anti-CD21 were a minor population. Anti-CD23 Abs reacted with cells in the (1F8, 1:200, Dako); anti-CD23 (MHM6, 1:50, Dako); anti-CD68 (KP1, center of T cell-B cell aggregates and less frequently with cells in 1:250, Dako); anti-MIB-1 (KI-67, 1:60, Coulter Immunotech, Westbrook, the mantle zone of some of the T cell-B cell clusters. However, not 6392 CD8ϩ T CELLS IN RHEUMATOID SYNOVITIS Downloaded from http://www.jimmunol.org/

FIGURE 1. Architecture and cellular components of T cell-B cell aggregates in rheumatoid synovitis. Standard histomorphology was used to identify by guest on October 1, 2021 synovial specimens with prominent and demarcated lymphoid aggregates, and cellular components were identified by immunohistochemistry. Synovial tissue serial sections were stained with anti-CD4, anti-CD20, and anti-CD8 Abs and developed with immunoperoxidase-coupled secondary Abs and 3-amino-9-ethylcarbazole (AEC) as the substrate. CD4ϩ T cells (B) and CD20ϩ B cells (C) were the major cellular components of the aggregates. CD8ϩ T cells were mostly arranged in the periphery of the T cell-B cell aggregates in 21 of the 24 samples; in three samples, CD8ϩ T cells were also interspersed throughout the follicular structure (D). A, Hematoxylin and eosin. Original magnification ϫ200.

all of the follicular structures analyzed in this study contained cen- were occasionally seen in the outer layers of the follicular struc- trally located CD23-expressing cells. As shown in Fig. 2, essen- tures. To assure that the absence of centrally localized CD23ϩ tially two different types of T cell-B cell aggregates were found; in CD21ϩ cells was not a sampling artifact, serial sections of the one, anti-CD23-specific Abs revealed networks of centrally lo- regions of interest were examined. These studies failed to reveal cated CD23-expressing cells, consistent with the presence of evidence for FDCs in the center of the T cell-B cell clusters. As FDCs; in the other, the centralized CD23 staining pattern was dis- shown in Fig. 2, the absence of CD23ϩ FDC networks was asso- tinctly absent. ciated with a low frequency of MIB-1ϩ cells in the T cell-B cell To investigate whether tissues without CD23ϩ cells in the cen- aggregates. Also, IgDϩ B cells were dispersed throughout the fol- ter of the follicles expressed an unusual FDC phenotype, adjacent licles without the central clearing observed in the CD23ϩ tissue sections were stained with anti-CD21, anti-MIB-1, and anti- follicles. IgD. Representative results are shown in Fig. 2. These experiments In summary, these studies provided evidence that follicular or- revealed that lymphoid aggregates in rheumatoid synovium can be ganizations in the rheumatoid synovium can involve all of the categorized into two subsets. Aggregates resembling classical GCs cellular components and possess the microanatomical arrangement had central CD23ϩ networks. MIB-1ϩ cells were frequent, occu- typical of classical GCs. However, only a subset of T cell-B cell pying the center of the cell clusters with only scattered cells aggregates fulfilled the criteria for GCs, whereas the others were throughout the remainder of the follicle. IgD was expressed on the characterized by a lack of central FDCs and a lack of B cell pro- majority of B cells, but was absent from the most centrally local- liferation and differentiation. These two forms of follicular struc- ized B cells, consistent with phenotypic changes that result as these tures were not encountered side-by-side within the same tissue but cells participate in a GC reaction. In contrast, five of the nine were, instead, mutually exclusive. Individual patients either had synovial biopsies contained prominent and demarcated T cell-B fully developed GCs or atypical follicles, suggesting a critical role cell aggregates that characteristically lacked central CD23ϩ and of host factors in the formation of ectopic lymphoid tissue in the CD21ϩ cells. In these tissues, irregularly distributed CD23ϩ cells synovium. The Journal of Immunology 6393 Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021

FIGURE 2. Morphology and phenotype of cell populations contributing to two different types of synovial T cell-B cell aggregates. Synovial tissue sections were stained with anti-CD23 (A and B), anti-IgD (C and D), and anti-MIB-1 (E and F) Abs. Immunoperoxidase technique with 3-amino-9- ethylcarbazole (AEC) substrate for anti-IgD and 3,3Ј-diaminobenzidine tetrahydrochloride (DAB) substrate for anti-MIB-1 was used. Anti-CD23 was detected with a biotinylated secondary Ab followed by streptavidin-conjugated alkaline phosphatase and Vector Red as the substrate. Expression of CD23ϩ cellular networks distinguished two types of synovial T cell-B cell aggregates. CD23ϩ follicles, which were found in four of the nine patients (A), contained a high frequency of proliferating cells (anti-MIB-1, E), and IgD was lost from the centralized B cells (C), compatible with a GC reaction. In CD23 aggregates (B), IgD-expressing B cells (D) were diffusely arranged, and proliferating cells (F) were absent. Original magnification ϫ400.

Formation of GCs in the synovium is associated with the aggregate lacking CD23ϩ cells are shown in Fig. 3. The area recruitment of high numbers of CD8ϩ T cells stained with each of the specific Abs was calculated and expressed as a percentage of the total area occupied by the follicular struc- To identify potential factors influencing the formation of either ϩ typical GCs or CD23Ϫ T cell-B cell aggregates, the representation ture. The data from the four patients with CD23 follicles and the Ϫ of different cell populations in tissues of both categories was quan- five patients with CD23 T cell-B cell aggregates are summarized tified by digital image analysis. Compound images were con- in Table II. structed of consecutive tissue sections analyzed by confocal mi- This quantitative approach demonstrated that the contribution of croscopy. These images allowed for detailed inspection of the each cell type to the follicular structures was surprisingly invari- spatial relationship of different cell populations. Representative ex- able. In CD23-containing GCs, the follicle consisted of 32.8% amples of a follicle with centrally located CD23ϩ FDCs and a T-B CD4ϩ T cells and 30.9% CD20ϩ B cells by area. CD23-expressing 6394 CD8ϩ T CELLS IN RHEUMATOID SYNOVITIS

Table II. Representation of different cell cypes in synovial lymphoid follicles

Follicle Areaa

Patient CD4 CD20 CD68 CD23 CD8 CD20/CD4

Tissues with CD23ϩ GC-like follicles 1 39.3 35.5 6.0 2.7 16.6 1.0 2 35.1 39.2 5.3 4.5 15.9 1.2 3 31.3 27.9 6.0 10.4 24.4 0.9 4 25.7 20.9 25.8 14.7 13.1 0.9 Mean 32.8 30.9 10.8 8.1 17.5 1.0 Tissues with CD23Ϫ T cell-B cell aggregates 5 45.8 37.4 15.1 0.4 1.4 0.8 6 40.7 45.7 10.7 0.7 2.2 1.1 7 43.3 33.5 10.0 2.4 10.9 0.9 8 42.6 48.5 2.5 0.8 5.7 1.1 9 44.6 43.5 3.8 0.9 7.2 1.0 Downloaded from Mean 43.4 41.7 8.4 1.0 5.5 1.0

a Results are the mean of two to three distinct follicles per patient and are ex- pressed as the percentage of the total area occupied by the follicle.

area. Lymphoid aggregates lacking centrally localized FDCs were composed of 43.4% CD4ϩ T cells and 41.7% CD20ϩ B cells. http://www.jimmunol.org/ CD68ϩ macrophages were again a minor population (mean area of 8.4%). CD8ϩ T cells only represented 5.5% of the total area. The ratio of the two major cell populations in the follicles, CD4ϩ T cells and CD20ϩ B cells, was well-maintained among the patients and was essentially identical when classical GCs and CD23 ag- gregates were compared. The contribution of CD8ϩ T cells to the follicles clearly distinguished classical GCs and CD23 aggregates ϭ

( p 0.04). by guest on October 1, 2021 So far, CD8ϩ T cells have not been considered a cell population of relevance in GC formation. Therefore, it was surprising to find a large number of CD8ϩ T cells at the outer edge of typical GCs. To further examine the possibility that variations in the frequency of synovial CD8ϩ T cells could be correlated with tissue organi- zation, the total number of CD8ϩ T cells in the perifollicular re- gions, the interfollicular zones, and the perivascular cell accumu- lations was counted. As shown in Fig. 4, synovial tissue from patients with GCs contained an average of 68 CD8ϩ T cells per high-power field. In patients without GCs, CD8ϩ tissue-infiltrating T cells were much less frequent (16 cells per high-power field, p Ͻ 0.001). These data indicated that the representation of CD8ϩ T cells was variable among patients and was closely associated with FIGURE 3. Compound images of CD23ϩ and CD23Ϫ T cell-B cell ag- the emergence of GCs. gregates in rheumatoid synovitis. Serial tissue sections were stained with anti-CD4, anti-CD8, anti-CD20, anti-CD68, and anti-CD23 Abs and de- Tissue-infiltrating CD8ϩ T cells in patients with GC formation veloped with Vector Red substrate. Fluorescence-stained tissue sections express CD40L were scanned with a confocal microscope and fluorescence signals were ϩ ϩ translated into pseudocolors. Compound images were generated by over- To further examine the role of synovial CD4 and CD8 T cells, laying consecutive tissue sections. CD4ϩ T cells are represented in light mononuclear cells were isolated from freshly harvested synovial ϩ ϩ ϩ blue, CD20 B cells in dark blue, CD8 T cells in red, CD68 cells in tissue biopsies. Tissues from three patients with GC formation and ϩ ϩ green, and CD23 cells in yellow. Except for the CD23 cell network in three patients with CD23Ϫ aggregates were analyzed by flow cy- the center of A, the distribution of CD20-, CD4-, CD8-, and CD68-positive ϩ ϩ tometry. As expected, the majority of CD3 synovial tissue cells cells is similar in the CD23 aggregate (A) compared with the CD23 ϩ expressed CD4, but CD8 cells were also quite common. Synovial aggregate (B). tissue specimens with GCs contained an average of 42% CD3ϩ CD8ϩ T cells. Tissues lacking CD23ϩ GC-like regions had sig- nificantly lower numbers of CD3ϩ CD8ϩ T cells (23%, p Ͻ cells contributed 8.1% and CD68ϩ macrophages accounted for 0.005). Because CD40-CD40L interactions are critical in GC for- 10.8% of the area. Unexpectedly, CD8ϩ T cells were frequent in mation, both CD4ϩ and CD8ϩ T cells were analyzed for the ex- these centers. They were found predominantly at the outer edge of pression of CD40L. The fraction of CD4ϩ CD40Lϩ cells was very the demarcated follicle and accounted for 17.5% of the follicular similar in both patient subsets (7.6% vs 8.9% of T lymphocytes). The Journal of Immunology 6395

FIGURE 4. Frequencies of CD8ϩ T cells in synovial tissue from pa- tients with CD23ϩ GC-like follicles and CD23Ϫ T cell-B cell aggregates. Synovial tissue sections from the nine RA patients were immunostained with anti-CD8 Ab. The number of CD8ϩ T cells in five arbitrarily selected high-power fields (hpf) was counted and is shown as mean Ϯ SD. Patients Downloaded from with CD23ϩ GC-like aggregates in the synovium had significantly higher numbers of CD8ϩ T cells (p Ͻ 0.001) infiltrating the tissue.

However, a difference emerged for the percentage of CD8ϩ T cells

with up-regulated CD40L expression (Fig. 5). In synovium with http://www.jimmunol.org/ CD23-lymphoid aggregates, CD8ϩ CD40Lϩ T cells accounted for only 1.2% of T lymphocytes. The CD8ϩ CD40Lϩ subset was ex- panded three- to ninefold in specimens with typical GCs. All three ϩ ϩ ϩ patients had a significant proportion of CD40L-expressing CD8ϩ FIGURE 6. Arrangement of CD8 CD40L T cells around CD23 GCs in rheumatoid synovitis. The tissue distribution of CD40L-expressing cells (3.8–10.8% of T lymphocytes, p Ͻ 0.05). ϩ CD8 T cells in the synovium was determined by two-color immunohis- Two-color immunohistochemistry with anti-CD8- and anti- ϩ tochemistry/immunofluorescense analysis on the same tissue section. A, CD40L-specific Abs was used to localize CD8 cells with up- CD40L expression was detected by immunoperoxidase and 3,3Ј-diamino- regulated CD40L expression. In specimens from patients with benzidine tetrahydrochloride (DAB); B, Immunofluorescense was used to Ϫ ϩ ϩ CD23 T cell-B cell follicles, CD8 CD40L cells were distinctly identify CD8ϩ T cells (biotinylated secondary Ab, streptavidin-conjugated by guest on October 1, 2021 infrequent and multiple tissue sections had to be searched to iden- alkaline phosphatase, and Vector Red as substrate). Double-positive CD8ϩ tify an occasional cell with this phenotype. In patients with fully CD40Lϩ cells were arranged in a circle around the edge of the GC. Orig- inal magnification ϫ400.

developed GCs, CD8ϩ CD40Lϩ cells were found in a perifollic- ular distribution, whereas CD8ϩ T cells in the interfollicular zones and in the perivascular regions remained negative for CD40L (Fig. 6).

Discussion The formation of follicle-like structures in rheumatoid synovitis has been cited as evidence for the involvement of a T cell-B cell response in disease pathogenesis (4). Lymphocytes are a compo- nent of most inflammatory infiltrates, and their presence does not prove a fundamental role in pathology. The synovial membrane is primarily not a lymphoid tissue and the generation of ectopic lym- phoid architecture is indeed a highly significant finding. The data presented here confirm that the conditions in rheumatoid synovitis FIGURE 5. Expression of CD40L on synovial tissue CD8ϩ T cells and are unique and that the synovial microenvironment is able to sup- ϩ CD4 T cells. Tissue-infiltrating lymphocytes were purified from synovial port the formation of follicles that have all of the characteristics of tissue specimens from six patients with active rheumatoid synovitis. T ϩ functional GCs. The more important finding of our study relates to cell-B cell follicles with CD23 FDCs were found in three patients. In the the distinction between patients who have classical GCs in the remaining three patients, synovial tissue T cell-B cell aggregates lacked ϩ ϩ synovium and those who generate T cell-B cell aggregates that centralized CD23 cells. Expression of CD40L on CD8 T cells and CD4ϩ T cells was assessed by two-color flow cytometry. The fraction of lack the typical features of B cell proliferation and down-regula- CD4ϩ T cells expressing CD40L was similar in all samples, independent tion of IgD cell surface expression. This suggests that not all of the of the type of T cell-B cell clusters. In patients with CD23ϩ GC-like struc- signals required for GC organization are present in some individ- tures, the frequency of CD8ϩ CD40Lϩ T cells was increased when com- uals and that the search for specific mediators of such differences pared with patients with CD23Ϫ T cell-B cell aggregates (p Ͻ 0.05). could provide important insight into synovial immune responses. 6396 CD8ϩ T CELLS IN RHEUMATOID SYNOVITIS

Lymphoid aggregates without GC reaction lacked networks of attempt to down-regulate the GC reactions; or 2) CD8ϩ T cells CD23ϩ CD21ϩ FDCs. More surprisingly, we found that the num- assume helper cell functions in the synovium. In the first model, ber of tissue-infiltrating CD8ϩ T cells was a predictor for the ab- CD4ϩ CD40L T cells would be the primary initiating cells of GC sence or presence of fully developed GCs. As CD8ϩ T cells have formation and the recruitment of CD8 T cells would be a second- not been implicated in contributing to GC formation, it would be ary event. Although this interpretation cannot be excluded, a pri- expected that patients with a low representation of CD8ϩ T cells marily inhibitory function of the CD8 T cells is difficult to recon- in the infiltrates would be particularly successful in establishing cile with the expression of CD40L. Also, if CD8ϩ T cells, through synovium GCs. The opposite was the case, suggesting an hitherto a yet unknown mechanism, could shut down GC reactions, burned- unsuspected contribution of CD8ϩ T cells to rheumatoid synovitis. out GCs with pale cells arranged in an onionskin fashion should be Ab responses to protein Ags begin in the T cell-rich zones of found. Such structures were not encountered in the CD8ϩ T cell- secondary lymphoid tissues where interdigitating dendritic cells, T rich tissues. Rather, the frequency of tissue-infiltrating CD8ϩ T cells, and B cells make contact. However, this encounter is insuf- cells was an excellent predictor for typically structured GCs. In ficient for the maturation of humoral immune responses and for the support of the second model, Cronin et al. have reported that IL- development of memory B cells. T cell-B cell interaction in GCs 4-producing CD8ϩ T cells can provide B cell help (34). We have must provide unique signals that allow for B cell proliferation, no direct evidence to support this interpretation, but the expression positive selection based on affinity of the expressed Ig, and emer- of CD40L would be highly suggestive of a possible helper function gence of memory B cells. Therefore, the formation of these highly for this cell subset. This issue will have to be addressed by iso- specialized centers in the synovial membrane cannot be overem- lating these CD8ϩ T cells and assessing their functional profile in phasized, and it stresses the contribution of T cells, B cells, and T cell-B cell interactions. The quantification of cell populations Downloaded from specific Ag to the pathologic events defining this syndrome. contributing to the GCs indicated that CD4ϩ T cells were less Deficiency of GC formation is a hallmark of patients with severe abundant in typical CD23ϩ follicles than in CD23 aggregates. This X-linked immunodeficiency with hyper-IgM syndrome (29). Care- again would support the interpretation that CD8ϩ T cells might be ful studies of these patients have revealed the crucial role of CD40 able to assume helper cell functions under special circumstances. signaling in GC formation and T cell-dependent Ag responses (30, In summary, our studies on the composition of follicular struc- 31). Mutations of the CD40L gene not only produce the clinical tures formed in rheumatoid synovial tissue revealed a critical role syndrome, which is characterized by increased occurrence of op- for CD8ϩ T cells in regulating the organization of ectopic lym- http://www.jimmunol.org/ portunistic infections and severe neutropenia, but is also associated phoid tissue. Formation of synovial tissue GCs coincided with the with failure to produce GC reactions (29). Comparison of synovial recruitment of CD8ϩ T cells and the up-regulation of CD40L on tissue CD23Ϫ T cell-B cell aggregates with lymphoid follicles in these cells. Evidence is emerging that RA is not a single disease CD40L-deficient patients raised the possibility of shared features. entity but includes several distinct variants of polyarthritis. The Lymph node biopsies from patients with mutated CD40L lacked restriction of GC formation to a subset of RA patients raises the GCs or expressed scant GC reactions while the architecture and possibility that variations in CD8ϩ T cell function represent dis- cellular distribution in the paracortical areas was normal. Interest- ease-risk factors and contribute to the heterogeneity of the rheu-

ingly, these patients displayed quantitative and qualitative abnor- matoid disease process. Investigating the regulation and function by guest on October 1, 2021 malities of FDCs, most importantly, poor expression of CD21 and of synovial tissue CD8ϩ T cells could eventually provide clues on CD23. It has been suggested that the loss of FDCs and their phe- factors critically involved in establishing a specialized microenvi- notypic abnormalities in these patients might result in poor Ag ronment for chronic immune responses and in modulating inflam- trapping and inefficient rescue of B cells from apoptosis. Alter- matory pathways in the synovial membrane. ations in the function of FDCs or differences in the recruitment of this specialized cell type to the synovial tissue environment could Acknowledgments therefore explain why some individuals do not develop complete We thank James Tarara and Steve Ziesmer for excellent technical assis- GC reactions in this ectopic site. tance and James Fulbright and Toni L. Higgins for editorial assistance. A role for CD40L in rheumatoid synovitis was suggested by the ϩ ϩ intriguing finding that the frequencies of synovial CD8 CD40L References cells correlated with the formation of GC-like follicular structures. CD40Lϩ T cells have recently attracted attention because several 1. Harris, E. D. 1997. Rheumatoid Arthritis. W. B. Saunders Co., Philadelphia, p. 3. 2. Van Boxel, J. A., and S. A. Paget. 1975. Predominantly T-cell infiltrate in rheu- reports have associated increased and prolonged expression of matoid synovial membranes. N. Engl. J. Med. 293:517. CD40L with systemic erythematosus (32). Because produc- 3. Klimiuk, P. A., J. J. Goronzy, Bjo¨rnsson J., R. D. Beckenbaugh, and tion of a multitude of autoantibodies is a characteristic feature of C. M. Weyand. 1997. Tissue cytokine patterns distinguish variants of rheumatoid synovitis. Am. J. Pathol. 151:1311. patients with systemic lupus erythematosus, defects in mechanisms 4. Kurosaka, M., and M. Ziff. 1983. Immunoelectron microscopic study of the dis- underlying T cell-dependent B cell activation are obvious candi- tribution of T cell subsets in rheumatoid synovium. J. Exp. Med. 158:1191. 5. Schroder, A. E., A. Greiner, C. Seyfert, and C. Berek. 1996. Differentiation of B dates for disease-risk factors. CD40L expression has been reported cells in the nonlymphoid tissue of the synovial membrane of patients with rheu- to occur on synovial tissue T cells in RA (33). While it could be matoid arthritis. Proc. Natl. Acad. Sci. USA 93:221. easily imagined that up-regulation of CD40L has a place in chron- 6. Kelsoe, G. 1996. Life and death in germinal centers (redux). Immunity 4:107. 7. Gulbranson-Judge, A., M. Casamayor-Palleja, and I. C. MacLennan. 1997. Mu- ically persistent CD4 T cell-B cell responses, a contribution of tually dependent T and B cell responses in germinal centers. Ann. NY Acad. Sci. ϩ CD8 T cells has so far not been described. 815:199. CD8ϩ T cells were fourfold more frequent in the tissue of pa- 8. MacLennan, I. C. 1994. Germinal centers. Annu. Rev. Immunol. 12:117. ϩ 9. Jacob, J., J. Przylepa, C. Miller, and G. Kelsoe. 1993. In situ studies of the tients with GCs. In these patients, CD8 T cells accounted for a primary immune response to (4-hydroxy-3- nitrophenyl)acetyl. III. The kinetics significant proportion of the cells in T cell-B cell follicles. Often, of V region mutation and selection in germinal center B cells. J. Exp. Med. ϩ 178:1293. they accumulated at the outer edge of the cell clusters. CD8 T 10. Zheng, B., S. Han, and G. Kelsoe. 1996. T helper cells in murine germinal centers cells were not only represented more frequently in direct associa- are antigen-specific emigrants that downregulate Thy-1. J. Exp. Med. 184:1083. tion with the follicles, they were also more numerous in the inter- 11. Martinez-Valdez, H., C. Guret, O. de Bouteiller, I. Fugier, J. Banchereau, and follicular areas. Two models could explain this unexpected find- Y. J. Liu. 1996. Human germinal center B cells express the apoptosis-inducing ϩ genes Fas, c-myc, P53, and Bax but not the survival gene bcl-2. J. Exp. Med. ing: 1) CD8 T cells migrate into the synovial membrane in an 183:971. The Journal of Immunology 6397

12. Kelsoe, G. 1996. The germinal center: a crucible for selection. Se- 24. Fu, Y. X., H. Molina, M. Matsumoto, G. Huang, J. Min, and D. D. Chaplin. 1997. min. Immunol. 8:179. Lymphotoxin-␣ (LT␣) supports development of splenic follicular structure that is 13. Fuller, K. A., O. Kanagawa, and M. H. Nahm. 1993. T cells within germinal required for IgG responses. J. Exp. Med. 185:2111. centers are specific for the immunizing antigen. J. Immunol. 151:4505. 25. Pasparakis, M., L. Alexopoulou, V. Episkopou, and G. Kollias. 1996. Immune 14. Zheng, B., W. Xue, and G. Kelsoe. 1994. Locus-specific somatic hypermutation and inflammatory responses in TNF ␣-deficient mice: a critical requirement for in germinal centre T cells. Nature 372:556. TNF ␣ in the formation of primary B cell follicles, follicular dendritic cell net- 15. Callard, R. E., R. J. Armitage, W. C. Fanslow, and M. K. Spriggs. 1993. CD40 works and germinal centers, and in the maturation of the humoral immune re- ligand and its role in X-linked hyper-IgM syndrome. Immunol. Today 14:559. sponse. J. Exp. Med. 184:1397. 16. Allen, R. C., R. J. Armitage, M. E. Conley, H. Rosenblatt, N. A. Jenkins, N. G. 26. Matsumoto, M., S. Mariathasan, M. H. Nahm, F. Baranyay, J. J. Peschon, and Copeland, M. A. Bedell, S. Edelhoff, C. M. Disteche, and D. K. Simoneaux. D. D. Chaplin. 1996. Role of lymphotoxin and the type I TNF receptor in the 1993. CD40 ligand gene defects responsible for X-linked hyper-IgM syndrome. formation of germinal centers. Science 271:1289. Science 259:990. 27. Ferguson, S. E., S. Han, G. Kelsoe, and C. B. Thompson. 1996. CD28 is required 17. DiSanto, J. P., J. Y. Bonnefoy, J. F. Gauchat, A. Fischer, and B. de Saint. 1993. for germinal center formation. J. Immunol. 156:4576. CD40 ligand mutations in x-linked immunodeficiency with hyper-IgM. Nature 28. Borriello, F., M. P. Sethna, S. D. Boyd, A. N. Schweitzer, E. A. Tivol, D. Jacoby, 361:541. T. B. Strom, E. M. Simpson, G. J. Freeman, and A. H. Sharpe. 1997. B7-1 and 18. Kawabe, T., T. Naka, K. Yoshida, T. Tanaka, H. Fujiwara, S. Suematsu, N. B7-2 have overlapping, critical roles in immunoglobulin class switching and Yoshida, T. Kishimoto, and H. Kikutani. 1994. The immune responses in CD40- germinal center formation. Immunity 6:303. deficient mice: impaired immunoglobulin class switching and germinal center 29. Facchetti, F., C. Appiani, L. Salvi, J. Levy, and L. D. Notarangelo. 1995. Im- formation. Immunity 1:167. munohistologic analysis of ineffective CD40-CD40 ligand interaction in lym- 19. Noelle, R. J., M. Roy, D. M. Shepherd, I. Stamenkovic, J. A. Ledbetter, and A. phoid tissues from patients with X-linked immunodeficiency with hyper-IgM. Aruffo. 1992. A 39-kDa protein on activated helper T cells binds CD40 and Abortive germinal center cell reaction and severe depletion of follicular dendritic transduces the signal for cognate activation of B cells. Proc. Natl. Acad. Sci. USA cells. J. Immunol. 154:6624. 89:6550. 20. Xu, J., T. M. Foy, J. D. Laman, E. A. Elliott, J. J. Dunn, T. J. Waldschmidt, J. 30. Gray, D., P. Dullforce, and S. Jainandunsing. 1994. Memory B cell development Elsemore, R. J. Noelle, and R. A. Flavell. 1994. Mice deficient for the CD40 but not germinal center formation is impaired by in vivo blockade of CD40-CD40 ligand. [Published erratum appears in 1994 Immunity 1:613.] Immunity 1:423. ligand interaction. J. Exp. Med. 180:141. Downloaded from 21. Foy, T. M., J. D. Laman, J. A. Ledbetter, A. Aruffo, E. Claassen, and R. J. Noelle. 31. Han, S., K. Hathcock, B. Zheng, T. B. Kepler, R. Hodes, and G. Kelsoe. 1995. 1994. gp39-CD40 interactions are essential for germinal center formation and the Cellular interaction in germinal centers: roles of CD40 ligand and B7-2 in es- development of B cell memory. J. Exp. Med. 180:157. tablished germinal centers. J. Immunol. 155:556. 22. Van den Eertwegh, A. J., R. J. Noelle, M. Roy, D. M. Shepherd, A. Aruffo, J. A. 32. Desai-Mehta, A., L. Lu, R. Ramsey-Goldman, and S. K. Datta. 1996. Hyperex- Ledbetter, W. J. Boersma, and E. Claassen. 1993. In vivo CD40-gp39 interactions pression of CD40 ligand by B and T cells in human lupus and its role in patho- are essential for thymus-dependent humoral immunity. I. In vivo expression of genic autoantibody production. J. Clin. Invest. 97:2063. CD40 ligand, cytokines, and antibody production delineates sites of cognate T-B 33. MacDonald, K. A., Y. Nishioka, P. E. Lipsky, and R. Thomas. 1997. Functional cell interactions. J. Exp. Med. 178:1555. CD40 ligand is expressed by T cells in rheumatoid arthritis. J. Clin. Invest. 100:

23. Tanaka, H., S. Saito, H. Sasaki, H. Arai, T. Oki, and N. Shioya. 1994. Morpho- 2404. http://www.jimmunol.org/ logical aspects of LFA-1/ICAM-1 and VLA4/VCAM-1 adhesion pathways in 34. Cronin, D. C., R. Stack, and F. W. Fitch. 1995. IL-4-producing CD8ϩ T cell human lymph nodes. Pathol. Int. 44:268. clones can provide B cell help. J. Immunol. 154:3118. by guest on October 1, 2021