T Cells + Regulatory CD8 CD103 Is a Marker for Alloantigen-Induced

CD103 Is a Marker for Alloantigen-Induced Regulatory CD8 + T Cells Elena Uss, Ajda T. Rowshani, Berend Hooibrink, Neubury M. Lardy, René A. W. van Lier and Ineke J. M. ten Berge This information is current as of September 26, 2021. J Immunol 2006; 177:2775-2783; ; doi: 10.4049/jimmunol.177.5.2775 http://www.jimmunol.org/content/177/5/2775 Downloaded from

References This article cites 51 articles, 24 of which you can access for free at: http://www.jimmunol.org/content/177/5/2775.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• 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 September 26, 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 © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

CD103 Is a Marker for Alloantigen-Induced Regulatory CD8؉ T Cells

Elena Uss,1*§ Ajda T. Rowshani,§ Berend Hooibrink,† Neubury M. Lardy,‡ Rene´A. W. van Lier,* and Ineke J. M. ten Berge§

␣ ␤ The E 7 integrin CD103 may direct lymphocytes to its ligand E-cadherin. CD103 is expressed on T cells in lung and gut and on -allograft-infiltrating T cells. Moreover, recent studies have documented expression of CD103 on CD4؉ regulatory T cells. Ap proximately 4% of circulating CD8؉ T cells bear the CD103 molecule. In this study, we show that the absence or presence of CD103 was a stable trait when purified CD103؊ and CD103؉CD8؉ T cell subsets were stimulated with a combination of CD3 and CD28 mAbs. In contrast, allostimulation induced CD103 expression on ϳ25% of purified CD103؊CD8؉ T cells. Expression of CD103 on alloreactive cells was found to be augmented by IL-4, IL-10, or TGF-␤ and decreased by addition of IL-12 to MLCs. ؉ ؉ The alloantigen-induced CD103 CD8 T cell population appeared to be polyclonal and retained CD103 expression after restimu- Downloaded from lation. Markedly, in vitro-expanded CD103؉CD8؉ T cells had low proliferative and cytotoxic capacity, yet produced considerable amounts of IL-10. Strikingly, they potently suppressed T cell proliferation in MLC via a cell-cell contact-dependent mechanism. -Thus, human alloantigen-induced CD103؉CD8؉ T cells possess functional features of regulatory T cells. The Journal of Immu nology, 2006, 177: 2775–2783.

ϩ ncreasing evidence exists for a protective role for regulatory Although in the past so-called CD8 T suppressor cells were http://www.jimmunol.org/ T cells (Tregs)2 in autoimmune diseases, allergic diseases, supposed to play a regulatory role in autoimmune diseases, trans- I and allograft rejection, where they control the potential dam- plantation, and in protection against cancer (11, 12), only over the aging activity of effector Th1 cells, Th2 cells, or CTLs (1–3). last 10 years has this concept re-emerged (13–15). Until then, ef- Tregs can be classified into natural (constitutive) and induced reg- forts to understand the cellular and molecular mechanisms under- ϩ ulatory cells. To the latter category belong CD4 T regulatory 1 lying CD8 T cell-mediated immunosuppression were hampered by ϩ ϩ (Tr1) cells and CD4 Th3 cells, as well as regulatory CD8 T difficulties in isolating these cells and by a lack of defining mark- cells (4). ers. The cell surface marker profile described for CD8ϩ Tregs, ϩ ϩ Natural CD4 CD25 Tregs have first been described by Sak- such as CD28Ϫ, CD45RClow, and CTLA-4ϩ indicates more that by guest on September 26, 2021 aguchi et al. (5) as thymic-derived cells that have multiple immu- these cells are in an “activated” or “memory” state than that they noregulatory properties, including active suppression of self-Ag- are associated with a regulatory function. CD8ϩ Tregs are reported reactive T cells, promotion of tolerance to allogeneic bone marrow to mediate Ag-specific suppression by production of the cytokines grafts, and suppression of antitumor immune reactivity. ␤ ϩ ϩ IL-10 and/or TGF- and/or by a direct inhibitory action on den- CD4 CD25 T cells have a low proliferative capacity after allo- dritic cells (16). geneic or polyclonal stimulation and express CTLA4, which de- ␣ ␤ The E 7 integrin CD103 has initially been described to be livers a negative signal for T cell activation. Their mechanism of expressed on both murine and human CD8ϩ T lymphocytes lo- suppression includes cell-cell contact, but secreted cytokines such calized in intestine, bronchoalveolar fluid, and allograft tissues as TGF-␤ and IL-10 may also play a role (6). Expression of the (17–19). An important function of this molecule appears to be transcription factor Foxp3, a critical regulator of CD4ϩCD25ϩ directing lymphocytes to their ligand E-cadherin, expressed on ep- Treg cell development and function, seems the best marker to iden- ithelial cells (20, 21). Although the CD103 molecule can be ex- tify natural CD4ϩ Tregs (7). Two subsets of inducible CD4ϩ pressed on alloactivated, graft-infiltrating lymphocytes, this seems Tregs have been recognized: Ag-specific Tr1 cells that secrete not a prerequisite for the cytotoxic function of these cells as in large amounts of IL-10 and Th3 cells, mainly producing TGF-␤ vitro cytotoxicity against alloantigens exerted by sorted (8–10). CD103ϪCD8ϩ T cells equaled that of sorted CD103ϩCD8ϩ T cells (22). Moreover, although CD8ϩ T cells from CD103 knock-

*Department of Experimental Immunology and †Department of Cell Biology and out mice cannot reach the renal epithelial cells, they show normal Histology, Academic Medical Center, Amsterdam, The Netherlands; ‡HLA-Diagnos- cytotoxic capacity (23). Recently, CD103 was shown to be a target tics Laboratory, Sanquin, Amsterdam, The Netherlands; and §Division of Nephrol- of FoxP3 (24) and was found to be expressed on CD4ϩ Tregs (25, ogy, Department of Internal Medicine, Academic Medical Center, Amsterdam, The ϩ ϩ Netherlands 26). Whether CD103 CD8 T cells can also exert regulatory Received for publication December 5, 2005. Accepted for publication May 17, 2006. functions is currently unknown. ϩ ϩ The costs of publication of this article were defrayed in part by the payment of page In this study, we show that human CD103 CD8 T cells can be charges. This article must therefore be hereby marked advertisement in accordance induced by stimulation with alloantigens in vitro. These cells pos- with 18 U.S.C. Section 1734 solely to indicate this fact. sess potent suppressive activity in MLC, which is cell-cell contact 1 Address correspondence and reprint requests to Dr. Elena Uss, Department of Ex- dependent. CD103ϩCD8ϩ T cells secrete IL-10 rather than IFN-␥ perimental Immunology, Academic Medical Center, University of Amsterdam, L1- 110, 1105 AZ Amsterdam, The Netherlands. E-mail address: [email protected] and maintain their phenotype after restimulation with alloantigen. 2 Thus, CD103 represents a novel marker for a subset of alloantigen- Abbreviations used in this paper: Treg, T regulatory cell; Tr1, T regulatory 1; pen/ ϩ strep, penicillin/streptomycin; Tc, T cytotoxic. induced regulatory CD8 T cells.

Materials and Methods Cell-sorting experiments Cells T cell subpopulations were purified by FACS sorting. Briefly, unstimulated PBMC were isolated from whole heparinized blood obtained from 14 PBMC or CFSE-labeled PBMC stimulated for 5 days with irradiated al- healthy donors by Ficoll-Paque density centrifugation (Pharmacia Bio- logeneic stimulator PBMC were labeled with CD103-PE, CD4-PerCP Cy5.5, and CD8-allophycocyanin. To identify alloresponsive T cells, a gate tech). The current study was approved by the local medical ethics com- low ϩ ϩ Ϫ ϩ mittee of the Academic Medical Center. was set on the CFSE population. Next, CD103 CD8 , CD103 CD8 , CD103ϩCD4ϩ, and CD103ϪCD4ϩ subsets were separated with the Aria Monoclonal Abs FACS (BD Biosciences). All sorted subsets were Ͼ95% pure. Biotinylated CD103 mAb and V␤3-FITC were purchased from Immuno- In vitro suppressor assay tech and CD103-PE from Caltag Laboratories. CD4Ϫ and CD8-PerCP Cy5.5, and CD27-, CD28-, CCR7-, IFN-␥-, IL-4-, IL-10-, streptavidin-PE, Responder PBMC were labeled with CFSE and cocultured with irradiated allogeneic stimulator PBMC. Purified alloreactive CD103ϩCD8ϩ, mIgG1-PE, and CD8-allophycocyanin were all purchased from BD Immu- Ϫ ϩ ϩ ϩ Ϫ ϩ nocytometry Systems. CD45RA-RD1 and streptavidin-allophycocyanin CD103 CD8 , CD103 CD4 , or CD103 CD4 cells from the same do- were acquired from BD Pharmingen. nor were labeled with 1,3-dichloro-9,9-dimethylacridin-2-one succinimidyl ester (DDAO-SE; Molecular Probes) and added to the MLC at different CFSE labeling regulator:responder ratios. The assay was performed in round-bottom 96- well microtiter plates. After 5 days of culture, precursor frequency was PBMC were washed once with PBS containing antibiotics (100 U/ml so- ␮ determined and cells were stained for CD103, CD8, and CD4 to determine dium penicillin (Brocades Pharma) and 100 g/ml streptomycin sulfate the phenotype of proliferating lymphocytes. Blocking Abs were used at the (Invitrogen Life Technologies)) at room temperature before staining. One following concentrations: 1 ␮g/ml anti-IL-10 (BD Pharmingen), 2 ␮g/ml microliter of CFSE (Molecular Probes; 5 ␮M stock concentration) was ␤ 7 anti-TGF (R&D Systems), which had been tested for optimal activity in added in 1 ml of PBS-penicillin/streptomycin (pen/strep)/10 cells. Cells previous experiments. Downloaded from were subsequently incubated for 12 min at 37°C. Next, cells were washed three times in PBS-pen/strep at 4°C and resuspended in 1 ml of culture Transwell assay medium. To evaluate the role of cell-cell contact in suppressive activity, 24-well Culture and stimulation of the cells plates equipped with a Transwell insert (Costar) consisting of a 200-␮l upper well separated from an 800-␮l bottom well by a 0.4-␮m microporous All culture experiments were performed in culture medium consisting of polycarbonate membrane that had not been pretreated in tissue culture IMDM (Invitrogen Life Technologies) containing 10% heat-inactivated au- medium were used. The distance between the Transwell membrane and the http://www.jimmunol.org/ tologous human serum, pen/strep, and 2-ME (0.0035% v/v; Merck). MLCs well bottom was 1 mm. All the cells were resuspended in culture medium. were performed with selected responder-stimulator combinations that gave Responder PBMC labeled with CFSE and cocultured with irradiated allo- high proliferative responses in classical MLCs, detected by [3H]thymidine geneic stimulator PBMC were plated in the bottom well of the Transwell incorporation. Responder PBMC were labeled with CFSE and cultured system at a concentration of 5 ϫ 105 cells/ml. The top well insert was with irradiated allogeneic stimulator PBMCs or with irradiated autologous ϩ ϩ Ϫ ϩ inoculated with culture medium alone or CD103 CD8 or CD103 CD8 PBMCs. When indicated, responder PBMC were stimulated in the pres- T cell population as indicated. After 5 days of culture, cells were stained ence of CD3 mAb (soluble CLB-T3/4E) and CD28 mAb (CLB-CD28/1). for CD103 and CD8 and analyzed by FACS. The precursor frequency (percentage of cells in the initial subset that has undergone one or more divisions after culture) was calculated as follows: 51Cr-release assay ⌺ n ⌺ n [ n Ն 1(Pn/2 )]/[ n Ն 0(Pn/2 )], where n is the division number that cells by guest on September 26, 2021 have gone through and Pn is the number of the cells in division n (27). MLC were performed as described previously (30). Briefly, MLC consist- To analyze the effect of cytokines on CD103 expression, MLCs were ing of responder PBMC and irradiated stimulator cells were used to gen- performed in the absence or presence of IL-7 (10 ng/ml; Strathmann), erate effector cells. After 5 days of proliferation, CD103ϩCD8ϩ and IL-15 (10 ng/ml), IL-4 (10 ng/ml), IL-12 (10 ng/ml), TNF-␣ (10 ng/ml), CD103ϪCD8ϩ alloreactive T cells were sorted and used as effectors. An TGF-␤ (1 ng/ml) (all from R&D Systems), IL-21 (50 ng/ml; Zymogenet- 8-h 51Cr-release assay was used to detect lysis of stimulator or autologous ics), IL-10 (10 ng/ml) or IL-17 (10 ng/ml) (both from Sanquin). After 5 target cells at varying E:T ratios. Lymphocytes, cultured for 5 days, were days of culture, cells were analyzed by FACSCalibur flow cytometer (BD used as target cells. The specific percentage of lysis was calculated ac- Biosciences). Responder PBMC stimulated with irradiated autologous cells cording to the formula: (ER Ϫ SR)/(MR Ϫ SR) ϫ100, where ER is the in the presence of the above-mentioned cytokines were used as controls. experimental chromium release, SR represents the spontaneous chromium Anti-TGF-␤-neutralizing mAb was purchased from R&D Systems and release of target cells in medium alone, and MR equals the maximum used in a final concentration of 2 ␮g/ml. chromium release of target cells in 5% saponin solution. In these experi- ␤ ments, the percentage of specific lysis was derived from the E:T ratio of TCR-V spectratyping 5:1. All determinations were done in duplicate. cDNA was synthesized from RNA from equal numbers of sorted Statistical analysis CD103ϩCD8ϩ and CD103ϪCD8ϩ cells. PCR was performed in single TCR-V␤ PCRs with a TCR-C␤ primer labeled with a fluorescent dye (In- The Mann-Whitney U test was used for comparison of two independent vitrogen Life Technologies) (CAG GCA CAC CAG TGT GGC-FAM) as groups of observations. The two-tailed Kruskal-Wallis test was used for described previously (28, 29). comparisons of more than two means. Values of p below 0.05 were con- sidered statistically significant. Flow cytometric analyses Immunofluorescent staining and flow cytometry were performed on day 0 Results and after 5 days of culture. A total of 3 ϫ 105 PBMC were incubated with CD103ϩCD8ϩ T cells derive from CD103ϪCD8ϩ T cells upon fluorescent-labeled conjugated mAbs (concentrations according to manu- allostimulation facturer’s instructions) for 30 min at 4°C protected from light. In some cases, this was followed by incubation with a second-step reagent (strepta- As reported previously, CD103 defines a subset of CD8ϩ T cells vidin-PE, -allophycocyanin) for 30 min at 4°C. For the cytokine staining, (31). We found that on average ϳ4% of freshly isolated CD8ϩ ϫ 6 cells were first stimulated in a 24-well plate at 1 10 cells/well in 0.5 ml A of culture medium with PMA (1 ng/ml), ionomycin (1 ␮g/ml), and mo- lymphocytes expressed the CD103 molecule (Fig. 1 ). These cells nensin (1 ␮M) (all from Sigma-Aldrich) for4hat37°C. Cells were appeared to have a primed phenotype as they lacked CD45RA, washed, and then the same staining procedure as described above was CD62L, and chemokine receptor CCR7 (data not shown). To in- performed using anti-IL-10-PE, anti-IL-4-PE, or anti-IFN-␥-PE Abs. vestigate the stability of CD103 expression on CD8ϩ T cells, we ϩ ϩ Ϫ ϩ Evaluation of cytokine production by ELISA purified CD103 CD8 and CD103 CD8 subsets from PBMC and stimulated them with a combination of CD3 and CD28 mAb Determination of cytokine concentrations in culture supernatants was performed by ELISA according to manufacturer’s protocols. Reagents for for 3 days. Both populations maintained their phenotype (Fig. 1B). IL-10, IL-2, IFN-␥ measurements were obtained from Sanquin. Reagents In contrast, in alloantigen-stimulated cultures, an up-regulation of Ϫ ϩ for TGF-␤ measurements were purchased from R&D Systems. the CD103 molecule was found on CD103 CD8 T cells since The Journal of Immunology 2777

CD103ϪCD8ϩ T cells stimulated with a combination of CD3 and CD28 mAb, and separated with a semipermeable membrane from allostimulated PBMC, up-regulated CD103 (6.8 Ϯ 2.3% of CD8ϩ T cells became CD103ϩ) (Fig. 1CII), although to a lesser extent than when they were stimulated with alloantigen in control cultures without semipermeable membrane (14.7 Ϯ 3.9% of CD103ϩCD8ϩ T cells) (Fig. 1CIII). Similar results were obtained when CD103ϪCD8ϩ T cells were stimulated with CD3/CD28 mAb and to which supernatant from a 5-day allostimulated culture was added (Fig. 1CIV). To obtain more insight into the details of alloantigen-induced up-regulation of CD103, MLCs starting with unseparated PBMC were performed (Fig. 2, A and B). Because the previous experiment showed that CD103ϩ T cells represented only a minute subset of the circulating CD8ϩ T cell pool and CD103ϪCD8ϩ T cells had a superior mitogenic response upon allostimulation, these experiments largely reﬂected de novo expression of CD103 on for- merly CD103Ϫ lymphocytes. Following stimulation with irradi- ϩ ated allogeneic cells, CD8 T cells already up-regulated CD103 at Downloaded from day 3 when 26.33 Ϯ 14.97% (mean Ϯ SD) expressed the CD103 molecule (Fig. 2C). Expression increased gradually with a peak on day 5 (53.63 Ϯ 13.81%, Fig. 2C). As we described previously, alloreactive CD8ϩ T cells bear the phenotype of effector T cells (32). Next to this, CD4ϩ cells up-regulated CD103 less signiﬁ-

cantly ( p Ͻ 0.05), because only 10.77 Ϯ 5.23% of alloreactive http://www.jimmunol.org/ CD4ϩ T cells became CD103 positive at day 5 (Fig. 2B). There- after, the expression of CD103 in both T cell subsets slightly decreased, yielding 37.77 Ϯ 16.09% CD103ϩCD8ϩ (Fig. 2C) and 8.53 Ϯ 3.71% CD103ϩCD4ϩ T cells on day 7 of culture (data not shown). Neither CD4ϩ nor CD8ϩ T cells expressed CD103 after autologous stimulation (data not shown). Based on the diminution of CFSE fluorescence, expression of CD103 on CD8ϩ alloreactive FIGURE 1. CD103ϩCD8ϩ T cells derive from CD103ϪCD8ϩ T cells T cells increased with the number of cell divisions (Fig. 2D). by guest on September 26, 2021 upon allostimulation. A, CD103 defines a subset of circulating resting ϩ ϩ CD8 T cells. The dot plots represent T cells gated on CD8 T cells. One ϩ representative experiment of nine is shown. B, CD103 is up-regulated by CD103 expression on alloreactive CD8 T cells is increased in ϩ ϩ Ϫ ϩ ␤ allostimulation. Purified CD103 CD8 and CD103 CD8 T cells were the presence of TGF- , IL-4, and IL-10 but is down-regulated labeled with CFSE and stimulated with a combination of CD3 and CD28 by IL-12 in MLCs mAb (anti-CD3/anti-CD28) for 3 days, or with alloantigen for 5 days as ϩ To assess the influence of cytokines on CD103 expression, stan- indicated. Dot plots (gated on CD8 T cells) represent the phenotype of the dard MLCs were performed in the presence or absence of different subsets in nondivided (left panel) and divided (middle panel) populations cytokines. In preliminary experiments, the optimal concentration based on CFSE dilution on days 3 and 5, respectively. The x-axis shows log fluorescence of CD8-allophycocyanin, the y-axis shows log fluorescence of of these cytokines had been determined (data not shown). Con- firming previous studies (31), addition of TGF-␤ to MLC in- CD103-PE. The right panel shows the divisions of these cells. One rep- ϩ resentative experiment of three is shown. C, Up-regulation of CD103 is creased the percentage of CD8 T cells expressing CD103 to dependent on soluble factors. Purified CD103ϪCD8ϩ T cells were labeled Ͼ90% (Fig. 3A). Only two of the other cytokines tested also had with CFSE and stimulated with a combination of CD3 and CD28 mAb for a significant effect on CD103 expression on alloreactive cells. IL-4 3 days. To determine the influence of alloantigen-induced soluble factors, increased the percentage of CD103ϩ-expressing CD8ϩ T cells Ϫ ϩ responder CD103 CD8 T cells were plated in the top insert of a Trans- from 45.41 Ϯ 10.61% to 75.69 Ϯ 21.91% (four donors, p Ͻ 0.05, 5 well system at a concentration of 2 ϫ 10 cells/200 ␮l. The bottom well Fig. 3B). Expression of CD103 on alloreactive CD8ϩ T cells stim- was inoculated with culture medium alone (I) or with PBMC cocultured ulated in the presence of IL-12 was significantly lower than in the with irradiated allogeneic stimulator PBMC (II). Purified CD103ϪCD8ϩ T control cultures (17.6 Ϯ 12.1% vs 45.41 Ϯ 10.61%). Addition of cells stimulated with alloantigen for 5 days were used as a control (III). In parallel, purified CD103ϪCD8ϩ T cells stimulated with CD3/CD28 mAb IL-10 did not have a significant effect on the frequency of T cells were cultured in the presence of the supernatant from a 5 day alloculture expressing CD103 (Fig. 3B); however, the amount of CD103 ex- (IV). Dot plots represent the phenotype of CD8ϩ T cells (one of two in- pressed per cell was clearly increased as evidenced by the higher dependent experiments). The x-axis shows log fluorescence of CD8-allo- intensity of CD103 staining (Fig. 3C). To assess the role of TGF-␤ phycocyanin; the y-axis shows log fluorescence of CD103-PE. in the elevated expression of CD103 induced by IL-4 and IL-10, MLCs were performed in the presence of both the aforementioned cytokines and a neutralizing anti-TGF-␤ mAb (final concentration 31.44 Ϯ 8.49% (mean Ϯ SD of nine donors tested) of these cells 2 ␮g/ml). As a control, MLCs were performed in the presence of acquired CD103 expression. Additionally, Fig. 1B shows that both rTGF-␤ in a final concentration of 1 ng/ml together with the anti- CD103Ϫ and CD103ϩ CD8ϩ T can respond to alloantigens, albeit TGF-␤ mAb in a final concentration of 2 ␮g/ml. Addition of the CD103Ϫ cells have a more potent proliferative response. To ad- blocking anti-TGF-␤ mAb did not influence the amount of CD103 dress the influence of soluble factors on expression of CD103 by expressed on alloreactive T cells induced by IL-4 or IL-10. Like- CD103ϪCD8ϩ T cells, we conducted a Transwell assay. wise, TGF-␤-production was not influenced by the addition of 2778 CD103ϩ ALLOANTIGEN-INDUCED CD8ϩ Tregs Downloaded from http://www.jimmunol.org/

FIGURE 2. After allogeneic stimulation in vitro, CD103 is mainly expressed on divided CD8ϩ T lymphocytes. PBMC were labeled with CFSE and cultured with irradiated allogeneic or autologous PBMC. Flow cyto- by guest on September 26, 2021 metric analysis was performed on day 5 of culture. The subset distribution was assessed by staining with CD4-PerCP Cy5.5, CD8-PerCP Cy5.5, CD27-PE, and CD103-bio-allophycocyanin. A, Dot plots represent alloreactive T cells gated on CD4ϩ T cells at days 0 and 5 of culture. Analysis was done in undivided and divided subsets. The x-axis shows log fluores- FIGURE 3. CD103 expression on alloactivated CD8ϩ T cells is in- cence of CD103-bio-allophycocyanin; the y-axis shows log fluorescence of creased when cocultured in the presence of TGF-␤, IL-4, or IL-10 and CD27-PE. B, The mean percentage of CD103ϩ cells Ϯ SD from 14 inde- decreased by IL-12. A, Expression of CD103 on alloreactive CD8ϩ T cells pendent experiments is shown. f, The mean percentage of CD103ϩ cells on day 5 of MLC, performed in the absence or presence of several cyto- within the CD4ϩ alloreactive T cells; u, the mean percentage of CD103ϩ kines. x-axis, Log fluorescence CD103-PE; filled histograms, isotype con- cells within the CD8ϩ subset. Differences in the percentage of CD103ϩ trol; bold histograms, CD8ϩ alloreactive T cells. One representative ex- cells within alloreactive CD4ϩ, respectively, CD8ϩ cells between days 0 periment of four experiments is shown. B, The bars represent the mean and 5 of culture were significant (p Ͻ 0.05), as was the difference between percentage Ϯ SD of CD103ϩCD8ϩ T cells at day 5 of MLC in the absence CD103ϩ cells in the divided CD4ϩ and CD8ϩ T cell subsets at day 5, or presence of different cytokines (four experiments). Statistical signifi- p Ͻ 0.05). C, The bars represent mean ,ء) C, Time course of alloantigen-induced CD103 cance is indicated by an asterisk .(ء) indicated by an asterisk expression. Bars represent the mean percentage of CD103ϩCD8ϩ alloreac- fluorescence intensity (MFI) of CD103 on CD103ϩCD8ϩ T cells on day 5 tive T cells Ϯ SD (y-axis) of three independent experiments as determined of MLC in the absence or presence of different cytokines. The mean per- on days 3–7 after initiation of the MLC (x-axis). D, Expression of CD103 centage Ϯ SD from four independent experiments is shown. Statistical .(p Ͻ 0.05 ,ء) on CD8ϩ alloreactive T cells increases with the number of cell divisions. significance is indicated by an asterisk The x-axis shows log fluorescence of CFSE; the y-axis shows log fluorescence of CD103-PE. The TCR V␤ repertoire of sorted CD103ϩCD8ϩ T cells was determined and compared with that of CD103ϪCD8ϩ T cells to rIL-12 at a final concentration of 10 ng/ml. (data not shown). Ad- investigate the clonal composition of the CD103ϩCD8ϩ T popu- dition of other cytokines such as IL-7, IL-15, IL-21, TNF-␣,or lation. As shown in Fig. 4A, although somewhat less diverse than IL-17 did not significantly affect CD103 expression (Fig. 3B). the CD103ϪCD8ϩ fraction, CD103ϩCD8ϩ T cells used a broad spectrum of V␤ families inferring that this population contained Functional properties of CD103ϩCD8ϩ and Ϫ ϩ multiple clones that have expanded in response to alloantigen CD103 CD8 subsets exposure. Having established that CD103ϩCD8ϩ T cells specifically expand As compared with their expression in the CD103ϪCD8ϩ T cell upon alloantigenic stimulation, subsequent experiments were per- population, expression of V␤3, V␤10, and V␤20 was significantly formed to determine the V␤ usage, stability, and functional prop- reduced in the CD103ϩCD8ϩ T cell population (Fig. 4A). The erties of this subset. lower expression of V␤3 by CD103ϩCD8ϩ T cells was confirmed The Journal of Immunology 2779

FIGURE 4. CD103ϩCD8ϩ T cells are a polyclonal population with low proliferative capacity. A, TCR V␤ repertoire of sorted CD103ϩCD8ϩ and CD103ϪCD8ϩ T cells. After allostimulation for 5 days, CD8ϩ cells were sorted into CD103ϩCD8ϩ and CD103ϪCD8ϩ subsets. The TCR V␤ repertoire of sorted CD103ϩCD8ϩ T cells was determined and ␤V,ء .compared with that of CD103ϪCD8ϩ T cells expressed differentially between CD103ϩCD8ϩ and CD103ϪCD8ϩ populations. B, The difference in V␤3 expression between CD103ϩCD8ϩ and CD103ϪCD8ϩ T cells was confirmed by FACS analysis of alloreactive T cells. Dot plots represent alloreactive CD8ϩ T cells. The x-axis shows log fluorescence of V␤3-FITC; the y-axis shows log fluo- Downloaded from rescence of CD103-PE. One representative experiment of two is shown. C and D, Analysis of stability of CD103 expression. CD103ϩCD8ϩ and CD103ϪCD8ϩ subsets, sorted after 5 days of allostimulation, were labeled with CFSE and restimulated with alloantigen. After 5 days of restimulation, FACS analysis was per-

formed. Dot plots of the left panel represent sorted http://www.jimmunol.org/ CD103ϩCD8ϩ T cells (C) and sorted CD103ϪCD8ϩ T cells (D) restimulated with alloantigen, showing maintenance of phenotype. The right panels show hardly any proliferative capacity of the CD103ϩCD8ϩ T cell subset compared with the CD103ϪCD8ϩ T cell subset. One representative experiment of three is shown. by guest on September 26, 2021

by FACS analysis (Fig. 4B). To examine the stability of CD103ϩ Finally, we investigated whether CD103ϩCD8ϩ and expression on alloantigen-expanded CD8ϩ T cells, sorted CD103ϪCD8ϩ populations differed in cytotoxic potential. After CD103ϩCD8ϩ T cells were labeled with CFSE and restimulated separation of the alloreactive cells into CD103ϩCD8ϩ and with alloantigen or a combination of CD3 and CD28 mAb. Irre- CD103ϪCD8ϩ populations, their lytic capacity against allogeneic spective of the stimulus, acquired CD103 expression appeared to target cells was tested (see Materials and Methods). As shown in be a stable trait as sorted CD103ϩCD8ϩ remained predominantly Fig. 5C, CFSElow-negCD103ϪCD8ϩ T cells are superior in cyto- CD103ϩ; purified CD103ϪCD8ϩ T cells also retained their phe- toxic activity compared with CD103ϩCD8ϩ T cells. notype (Fig. 4, C and D). On day 7 after restimulation, CD103 ϩ ϩ expression on the sorted CD103ϩ T cells subsequently went down, The CD103 CD8 subset possesses regulatory activity but no changes were observed regarding CD103 expression on the Given that the CD103ϩCD8ϩ subset was able to produce IL-10, sorted CD103Ϫ subset (data not shown). Remarkably, based on the but no IFN-␥, combined with the observation that CD103ϩCD8ϩ diminution of CFSE fluorescence, CD103ϩCD8ϩ T cells hardly T cells had low proliferative capacity, we examined the possible proliferated, whereas CD103ϪCD8ϩ cells divided up to four times regulatory capacity of this subset. Alloexpanded purified subsets after allogeneic restimulation (Fig. 4, C and D). were tested for their suppressive activity in MLCs. Responder Next, the ability of CD103ϩCD8ϩ T cells to produce inflam- PBMC were labeled with CFSE and stimulated with irradiated matory and tolerogenic cytokines was investigated. To this end, allogeneic PBMC for 5 days. Addition of CD103ϩCD8ϩ T cells, CD8ϩ T cells from MLC after 5 days of stimulation were sorted autologous to the responder PBMC, from a regulator:responder into CD103ϩ and CD103Ϫ subsets and restimulated with alloan- ratio of 1:10, considerably decreased the precursor frequency from tigen for 5 days. Production of the cytokines by the purified subsets 3.88 Ϯ 1.81% in control allostimulated cultures to 2.16 Ϯ 1.89% as measured by ELISA showed that IL-10 and TGF-␤ were pro- in the experimental cultures ( p Ͻ 0.005, Fig. 6A). Addition of duced by both subsets, whereas IFN-␥ was significantly higher in CD103ϪCD8ϩ cells to the culture did not influence the precursor the supernatants of CD103ϪCD8ϩ T cells (Fig. 5A). In accordance frequencies of allostimulated cells (Fig. 6A). Notably, with this observation, we found that production of IFN-␥, deter- CD103ϩCD4ϩ T cells also showed regulatory capacity as has been mined by FACS, predominantly correlated with the absence of reported previously (26). Added at a 1:2 ratio to the MLC, CD103 expression on T cells (53.87 Ϯ 8.03% of CD103ϪCD8ϩ CD103ϩCD4ϩ T cells significantly inhibited the proliferation of cells produced IFN-␥ vs 14.89 Ϯ 4.91% of CD103ϩCD8ϩ T cells, alloreactive cells (precursor frequency 1.03 Ϯ 0.48%) compared Fig. 5B). with control cultures (precursor frequency 3.46 Ϯ 0.45%) (Fig. 2780 CD103ϩ ALLOANTIGEN-INDUCED CD8ϩ Tregs

FIGURE 5. Functional properties of alloreactive CD103ϩCD8ϩ and CD103ϪCD8ϩsubsets. A, Cytokine profiles of the alloreactive CD103ϩCD8ϩ and CD103ϪCD8ϩ T cell subsets. Sorted alloreactive CD103ϩCD8ϩ or CD103ϪCD8ϩ T cell subsets were restimulated with alloantigen for 5 days. The production of the tolerogenic cytokines IL-10 and TGF-␤ and the inflammatory cytokine IFN-␥ by purified subsets was detected by ELISA. The bars represent mean Ϯ SD of the concentration of IL-10, IFN-␥, or TGF-␤ in the culture supernatants. The mean percentage Ϯ SD from three independent experiments is shown. Statistical sig- ,p Ͻ 0.05). B ,ء) nificance is indicated by an asterisk Intracellular cytokine expression. PBMC were labeled with CFSE and stimulated with alloantigen. After 5 days of culture, expression of the tolerogenic cytokine IL-10 and the inflammatory cytokine IFN-␥ was analyzed by FACS. Dot plots represent alloreactive T cells gated on CD8ϩ T cells. The x-axis shows log fluorescence of

CD103-bio-allophycocyanin; the y-axis shows log ﬂuo- Downloaded from rescence of IL-10- and IFN-␥-PE. One representative experiment of four is shown. C, Cytotoxic capacity of CD103ϪCD8ϩ and CD103ϩCD8ϩ alloreactive T cells. Cytotoxicity was measured by the percentage of 51Cr release, as depicted on the y-axis. Alloreactive CTL were generated in MLC after 5 days of allogeneic stim- ϩ ϩ Ϫ ϩ ulation and CD103 CD8 (ࡗ) and CD103 CD8 al- http://www.jimmunol.org/ loreactive T cells (f) were sorted and used as effectors. The cytotoxic activity of the total alloreactive lymphocyte pool is shown as Œ. Different numbers of alloreactive effector cells are depicted on the x-axis. One representative experiment of three is shown.

6B). Thus, the CD103ϩCD8ϩ subset showed a similar inhibitory responder X stimulated with Z), from which two types of regula- ϩ ϩ ϩ ϩ effect in in vitro suppressor assay as CD103 CD4 cells. To ex- tory CD103 CD8 T cells were isolated by cell sorting (XY by guest on September 26, 2021 amine whether suppressive capacity is an intrinsic characteristic of CD103ϩCD8ϩ and XZ CD103ϩCD8ϩ). Next, a suppressor assay alloactivated CD103ϩCD8ϩ T cells or whether it is a feature of all was performed in which responder PBMC (X) were labeled with CD8ϩ T cells bearing the CD103 molecule, we purified CFSE and cultured with irradiated allogeneic stimulator PBMC CD103ϩCD8ϩ T cells from freshly isolated PBMC and tested (Y) for 5 days. When XY CD103ϩCD8ϩ T cells were added in their inhibitory ability. Freshly isolated CD103ϩCD8ϩ T cells sig- several regulator:responder ratios, the alloresponse was inhibited nificantly inhibited proliferation of polyclonally stimulated PBMC by 60–80%. This alloresponse could also be inhibited in the pres- (combination of CD3 and CD28 Abs) at a regulator:responder ratio ence of XZ CD103ϩCD8ϩ T cells, although to a lower extent (Fig. 1:2 whereas CD103ϪCD8ϩ T cells did not (data not shown). 6E), indicating that suppression of the immune response by We next analyzed whether the inhibitory effect of the CD103ϩCD8ϩ cells was not dependent upon the presence of a CD103ϩCD8ϩ subset is mediated by the immunosuppressive cy- specific allopeptide in the culture. Addition of XZ CD103ϪCD8ϩ tokines IL-10 and/or TGF-␤. Addition of anti-IL-10 or anti-TGF- T lymphocytes as well as addition of XY CD103ϪCD8ϩ T cells ␤-neutralizing Abs to the in vitro suppressive assay did not affect hardly affected the precursor frequency of allostimulated cells the inhibition of proliferation of alloactivated cells by the (Fig. 6E). CD103ϩCD8ϩ subset (Fig. 6C). In previous experiments, anti- IL-10 and anti-TGF-␤ mAbs were shown to inhibit the function of Discussion IL-10 and TGF-␤, respectively (data not shown). To address the In the present study, we show that the integrin CD103 is up-reg- involvement of membrane-bound compounds, we studied the ef- ulated on human CD8ϩ T cells after stimulation by alloantigen in fect of cell-cell interaction on suppressive activity of vitro, which is dependent on soluble factors. In contrast, the up- ϩ ϩ CD103 CD8 T cells. We found that the inhibitory function of regulation of CD103 was far less after nonspecific polyclonal stim- this population depends on cell-cell contact as no suppression of T ulation, suggesting a role for monocyte-derived factors in the in- cell proliferation was observed when the regulatory cells were sep- duction of CD103 expression. CD103 appeared to be mainly arated from MLC by a Transwell insert (Fig. 6D). expressed on dividing CD8ϩ T cells, which, as we previously

ϩ ϩ showed, bear the phenotype of effector T cells (32). Up-regulation In vitro-activated CD103 CD8 T cells are Ag nonspeciﬁc in of CD103 on in vitro-alloactivated T cells was augmented by their ability to suppress T cell proliferation TGF-␤, IL-4, or IL-10 and diminished by IL-12. Secretion of the To investigate whether suppression by alloactivated tolerogenic cytokine IL-10 was slightly higher in the sorted CD103ϩCD8ϩ T cells is Ag speciﬁc, we designed the following CD103ϩCD8ϩ T cell subset than in the CD103ϪCD8ϩ T cell sub- experiments. For the generation of CD103ϩCD8ϩ regulatory cells, set, but CD103ϪCD8ϩ T cells secreted considerably higher we used PBMC from completely HLA-mismatched donors. Two amounts of IFN-␥. CD103ϩCD8ϩ T cells were polyclonal, had a MLC combinations were used (responder X stimulated with Y and low proliferative and cytotoxic capacity compared with The Journal of Immunology 2781 Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021

FIGURE 6. CD103ϩCD8ϩ T cells have a strong regulatory capacity in vitro, which is not mediated by soluble IL-10 or TGF-␤ but appeared cell-cell contact dependent and Ag-nonspecific. Purified CD103ϩCD8ϩ, CD103ϪCD8ϩ, CD103ϩCD4ϩ, and CD103ϪCD4ϩ T cells were tested for their suppressive activity in in vitro MLC. A, Responder PBMC were labeled with CFSE and cocultured with irradiated allogeneic stimulator PBMC. Purified CD103ϩCD8ϩ (u) or CD103ϪCD8ϩ (o) were added to the MLC at different regulator:responder ratios, namely 1:20, 1:10, 1:5, 1:2, and 1:1 (x-axis). As a positive control, PBMC were stimulated with alloantigen only (f). On the y-axis, the mean percentage of precursor frequencies Ϯ SD from five separate experiments is p Ͻ 0.05). B, Purified CD103ϩCD4ϩ (u) or CD103ϪCD4ϩ (o) T cells were added to ,ء) represented. Statistical significance is indicated by an asterisk MLC at a regulator:responder ratio of 1:2. As a positive control, PBMC were stimulated with alloantigen only (f). The y-axis shows the mean percentage of precursor frequencies Ϯ SD from two separate experiments. C, CFSE-labeled PBMC were cocultured in MLC together with purified CD103ϩCD8ϩ T cells (u) at a regulator:responder ratio of 1:2 in the presence or absence of neutralizing Abs against IL-10 or TGF-␤. As a positive control, PBMC were stimulated with alloantigen (f). The y-axis shows the mean percentage of precursor frequencies Ϯ SD from three independent experiments. D, Responder PBMC labeled with CFSE and cocultured with irradiated allogeneic stimulator PBMC were plated in the bottom well of a Transwell system at a concentration of 5 ϫ 105 cells/ml. The top well insert was inoculated with culture medium alone or with purified CD103ϩCD8ϩ or CD103ϪCD8ϩ T cell population as indicated. The y-axis shows precursor frequencies Ϯ SD from three independent experiments. Statistical significance is indicated by an p Ͻ 0.05). E, To evaluate Ag specificity of the CD103ϩCD8ϩ regulatory T cells, purified XY CD103ϩCD8ϩ(u), XZ CD103ϩCD8ϩ(Ⅺ), XY ,ء) asterisk CD103ϪCD8ϩ (o), and XZ CD103ϪCD8ϩ (1) T lymphocytes were added to the MLC at different regulator/responder ratios, as depicted on the x-axis (see explanation in Results). As a positive control, PBMC were stimulated with alloantigen only (f). The y-axis shows the mean percentage of precursor .(p Ͻ 0.05 ,ء) frequencies Ϯ SD from three independent experiments. Statistical significance is indicated by an asterisk

CD103ϪCD8ϩ T cells, but displayed a strong suppressive activity CD103 is the receptor for E-cadherin, which is expressed on in the MLC. This suppressive effect of CD103ϩCD8ϩ T cells ap- epithelial cells, including renal tubular epithelial cells. As such, in peared not to be mediated by IL-10 or TGF-␤, but was found to be renal allografts CD103 may direct graft-inﬁltrating lymphocytes to cell-cell contact dependent. Addition of CD103ϩCD8ϩ T cells tubular epithelial cells, which are the main target for apoptotic cell stimulated by third-party cells inhibited the proliferative capacity death during allograft rejection (33, 34). Independent from its role of alloactivated cells, though to a lesser extent than did alloanti- in migration of immune cells, CD103 is also a marker of T cell gen-speciﬁc CD103ϩCD8ϩ T cells. activation (35). In support of this, effector-memory tonsil-resident 2782 CD103ϩ ALLOANTIGEN-INDUCED CD8ϩ Tregs

CD103ϩCD8ϩ T cells were found to be more reactive to their activation of T cell subsets by specific DR alleles may play an cognate Ag than the CD103ϪCD8ϩ T cells, which permits rapid important role in alloresponses as occur in MLRs and in organ recall responses at even low Ag concentrations, suggesting a role transplantation. of CD103 not only in homing and retention of T cells at epithelial Tregs may exert their suppressive function by several mecha- sites but also in promoting T cell function (36). Whether the nisms. Regardless of their mechanism of action, Tregs may operate CD103 molecule also directly contributes to the regulatory func- both at the level of T effector cells and at the level of APCs by tion of natural Tregs (26) and CD8ϩ Tregs (this manuscript) is decreasing the expression of MHC class II and costimulatory mol- unknown. Recently, CD103 was demonstrated to be a biomarker ecules (6). Regarding CD8ϩ Tregs, IL-10 has been described to be for murine CD8ϩ suppressor T cells. These cells required IFN-␥ to involved in their capacity to suppress T cell proliferation (47). The suppress CD4ϩ T cell division (37). A putative regulatory function CD103-bearing CD8ϩ alloreactive lymphocytes that we here de- for CD103-expressing CD8ϩ-alloreactive T cells appears to ac- scribe produced IL-10 and TGF-␤, but could not be inhibited in cord with the absence of CD103ϩCD8ϩ T cells in biopsies from their immunosuppressive activity by anti-IL-10 or anti-TGF-␤ rat renal allografts undergoing unmodified acute rejection (38) and mAbs, suggesting an IL-10- and TGF-␤-independent mechanism from patients with acute rejection (39). In contrast, CD103ϩCD8ϩ of suppression. In contrast, the suppressive effect of T cells are mainly present in late allograft rejection, occurring CD103ϩCD8ϩ T cells appeared to be cell-cell contact dependent. beyond the first 6 mo after transplantation, as well as in biopsies Recent reports have re-examined a role of cell death as a mecha- with signs of chronic rejection (40). Thus, during ongoing, smoul- nism of suppression by Treg (48, 49). Grossman et al. (50) indi- dering alloactivation in a predominant Th2 microenvironment, as cates that human CD4ϩCD25ϩ Treg mediate their suppressive ef- is the case during chronic allograft rejection (41), negative regu- fects via death induced by a granzyme A-perforin-dependent Downloaded from latory signals may prevent the alloresponse from inducing an acute mechanism. Granzyme B was shown to be involved in contact- inflammatory reaction. In this respect, studies to a possible in- mediated suppression by murine CD4ϩCD25ϩ Treg (51). We volvement of CD103ϩCD8ϩ T cells in subclinical allograft rejec- demonstrate that CD103ϩCD8ϩ alloreactive T cells were less cy- tion, in which no deterioration of allograft function occurs, despite totoxic than CD103ϪCD8ϩ T cells and therefore cytotoxicity is the presence of a clear inflammatory cellular infiltrate, will be of not likely to be involved in the mechanism of suppression by ϩ ϩ particular interest (34). CD103 CD8 T cells. Further studies are needed to elucidate http://www.jimmunol.org/ Whereas about half of the in vitro alloactivated CD8ϩ T cells their mechanism of action. The absence of an appropriate cosignal, expressed the CD103 molecule, the tolerogenic cytokines TGF-␤, the presence of immature or plasmacytoid dendritic cells, or the IL-4, and IL-10 markedly increased the percentage and/or the in- presence of specific cell surface or soluble molecules may all be tensity of CD103 expression. Other in vitro studies have also in- operational (6). In contrast, high cytotoxicity of the CD103ϪCD8ϩ dicated an important role for TGF-␤ in the up-regulation of CD103 T cells might be an explanation for a slightly decreased prolifer- expression on CD8ϩ effector T cells (31). Hadley and colleagues ative capacity of allostimulated PBMC in the presence of sorted (42) demonstrated that up-regulation of CD103 expression by al- allogeneic CD103ϪCD8ϩ T cells. loreactive CD8ϩ cells occurred subsequent to entry into the graft, It is still unsolved whether Tregs need a second encounter with which was dependent on the local level of TGF-␤. The enhancing the same peptide/Ag to become functional. In this study, we dem- by guest on September 26, 2021 effect of IL-4 on CD103 expression of allogeneic-stimulated onstrate that once activated, CD103ϩCD8ϩ alloreactive Tregs are CD8ϩ T cells is in line with the described up-regulation of CD103 able to suppress T cell reactivity. It may be conceived that during on both CD4ϩ and CD8ϩ cells after in vitro stimulation by CD3 a local response, CD103 initially serves to retain alloantigen-in- mAbs in the presence of this cytokine (43). IL-4 may exert its duced CD8ϩ Treg cells at the graft site and later functions on effect on CD103 expression directly or indirectly by induction of Tregs that exert their role as soon as alloantigen is recognized. TGF-␤ production as has been described for murine T cells (44). However, we did not find a blocking effect of anti-TGF-␤ mAb on Acknowledgments the expression of CD103 induced by IL-4. Our finding of down- We thank Si-La Yong for her excellent technical support. We are grateful ϩ regulation of CD103 expression on CD8 T cells after allogeneic to Drs. Eddy Wierenga and Ester M. M. van Leeuwen for critical reading stimulation in vitro in the presence of IL-12 was also observed of the manuscript. after stimulation of T cells with CD3 mAbs in the presence of this cytokine (43). Altogether, these findings indicate that a Th2 more Disclosures than a Th1 microenvironment favors CD103 expression, which The authors have no financial conflict of interest. supports its role in late and chronic rejection more than in acute renal allograft rejection. CD103 might well be differentially ex- References ϩ pressed on T cytotoxic (Tc)1 and Tc2 CD8 T cells, where 1. Chen, C., W. H. Lee, P. Yun, P. Snow, and C. P. Liu. 2003. Induction of au- CD103-expressing CD8ϩ CTLs, possibly of the Tc2 type, are toantigen-specific Th2 and Tr1 regulatory T cells and modulation of autoimmune diabetes. J. Immunol. 171: 733–744. more involved in a regulatory function than in a cytotoxic one. 2. Akdis, M., J. Verhagen, A. Taylor, F. Karamloo, C. Karagiannidis, R. Crameri, Indeed, in agreement with data from the literature (22), we showed S. Thunberg, G. Deniz, R. Valenta, H. Fiebig, et al. 2004. Immune responses in that the cytotoxic capacity of CD103ϩCD8ϩ T cells against al- healthy and allergic individuals are characterized by a fine balance between al- Ϫ ϩ lergen-specific T regulatory 1 and T helper 2 cells. J. Exp. Med. 199: 1567–1575. loantigens is less than that of CD103 CD8 T cells, indicating no 3. Graca, L., A. Le Moine, S. P. Cobbold, and H. Waldmann. 2003. Dominant role for CD103 in allocytotoxic effector function. transplantation tolerance. Curr. Opin. Immunol. 15: 499–506. In this study, we show that expression of V␤3, V␤10, and V␤20 4. Mills, K. H. 2004. Regulatory T cells: friend or foe in immunity to infection? Nat. ϩ ϩ Rev. Immunol. 4: 841–855. was significantly reduced in the CD103 CD8 alloreactive T cell 5. Sakaguchi, S., N. Sakaguchi, M. Asano, M. Itoh, and M. Toda. 1995. Immuno- population as compared with that in the CD103ϪCD8ϩ T cell pop- logic self-tolerance maintained by activated T cells expressing IL-2 receptor ␣ ulation. Thus, not all specificities are present within the alloreac- -chains (CD25): breakdown of a single mechanism of self-tolerance causes ϩ ϩ various autoimmune diseases. J. Immunol. 155: 1151–1164. tive CD103 CD8 T cell population. Several studies have pro- 6. Vigouroux, S., E. Yvon, E. Biagi, and M. K. Brenner. 2004. Antigen-induced vided evidence for a restricted V␤ gene usage in response to DR regulatory T cells. Blood 104: 26–33. 7. Fontenot, J. D., M. A. Gavin, and A. Y. Rudensky. 2003. Foxp3 programs the synthetic peptides presented in the context of self MHC molecules, development and function of CD4ϩCD25ϩ regulatory T cells. Nat. Immunol. 4: i.e., allostimulation via the indirect pathway (45, 46). Preferential 330–336. The Journal of Immunology 2783

8. Chen, Y., V. K. Kuchroo, J. Inobe, D. A. Hafler, and H. L. Weiner. 1994. Reg- 30. Zeijlemaker, W. P., M. H. Van Oers, and V. P. Eijsvoogel. 1976. Human lym- ulatory T cell clones induced by oral tolerance: suppression of autoimmune en- phocyte subpopulations involved in MLC and CML. Scand. J. Immunol. (Suppl. cephalomyelitis. Science 265: 1237–1240. 5): 143–156. 9. Groux, H., A. O’Garra, M. Bigler, M. Rouleau, S. Antonenko, J. E. de Vries, and 31. Hadley, G. A., S. T. Bartlett, C. S. Via, E. A. Rostapshova, and S. Moainie. 1997. ϩ ␣ M. G. Roncarolo. 1997. A CD4 T-cell subset inhibits antigen-specific T-cell The epithelial cell-specific integrin, CD103 ( E integrin), defines a novel subset responses and prevents colitis. Nature 389: 737–742. of alloreactive CD8ϩ CTL. J. Immunol. 159: 3748–3756. 10. McGuirk, P., C. McCann, and K. H. Mills. 2002. Pathogen-specific T regulatory 32. Nikolaeva, N., E. Uss, E. M. van Leeuwen, R. A. van Lier, and I. J. ten Berge. 1 cells induced in the respiratory tract by a bacterial molecule that stimulates 2004. Differentiation of human alloreactive CD4ϩ and CD8ϩ T cells in vitro. interleukin 10 production by dendritic cells: a novel strategy for evasion of pro- Transplantation 78: 815–824. tective T helper type 1 responses by Bordetella pertussis. J. Exp. Med. 195: 33. Wever, P. C., J. G. Boonstra, J. C. Laterveer, C. E. Hack, F. J. van der Woude, 221–231. M. R. Daha, and I. J. Berge. 1998. Mechanisms of lymphocyte-mediated cyto- 11. Fowler, D. H., J. Breglio, G. Nagel, M. A. Eckhaus, and R. E. Gress. 1996. ϩ toxicity in acute renal allograft rejection. Transplantation 66: 259–264. Allospecific CD8 Tc1 and Tc2 populations in graft-versus-leukemia effect and 34. Rowshani, A. T., S. Florquin, F. Bemelman, J. A. Kummer, C. E. Hack, and graft-versus-host disease. J. Immunol. 157: 4811–4821. I. J. ten Berge. 2004. Hyperexpression of the granzyme B inhibitor PI-9 in human 12. Martin, P. J. 1993. Donor CD8 cells prevent allogeneic marrow graft rejection in renal allografts: a potential mechanism for stable renal function in patients with mice: potential implications for marrow transplantation in humans. J. Exp. Med. subclinical rejection. Kidney Int. 66: 1417–1422. 178: 703–712. 35. Schieferdecker, H. L., R. Ullrich, A. N. Weiss-Breckwoldt, R. Schwarting, 13. Gilliet, M., and Y. J. Liu. 2002. Generation of human CD8 T regulatory cells by H. Stein, E. O. Riecken, and M. Zeitz. 1990. The HML-1 antigen of intestinal CD40 ligand-activated plasmacytoid dendritic cells. J. Exp. Med. 195: 695–704. ϩ lymphocytes is an activation antigen. J. Immunol. 144: 2541–2549. 14. Sarantopoulos, S., L. Lu, and H. Cantor. 2004. Qa-1 restriction of CD8 sup- 36. Woodberry, T., T. J. Suscovich, L. M. Henry, M. August, M. T. Waring, A. Kaur, pressor T cells. J. Clin. Invest. 114: 1218–1221. ␣ ␤ 15. Najafian, N., T. Chitnis, A. D. Salama, B. Zhu, C. Benou, X. Yuan, C. Hess, J. L. Kutok, J. C. Aster, F. Wang, et al. 2005. E 7 (CD103) expression J. Im- M. R. Clarkson, M. H. Sayegh, and S. J. Khoury. 2003. Regulatory functions of identifies a highly active, tonsil-resident effector-memory CTL population. ϩ Ϫ munol. CD8 CD28 T cells in an autoimmune disease model. J. Clin. Invest. 112: 175: 4355–4362. 1037–1048. 37. Myers, L., M. Croft, B. S. Kwon, R. S. Mittler, and A. T. Vella. 2005. Peptide-

␥ ␤ Downloaded from 16. Chang, C. C., R. Ciubotariu, J. S. Manavalan, J. Yuan, A. I. Colovai, F. Piazza, speciﬁc CD8 T regulatory cells use IFN- to elaborate TGF- -based suppression. S. Lederman, M. Colonna, R. Cortesini, R. Dalla-Favera, and N. Suciu-Foca. J. Immunol. 174: 7625–7632. 38. Yuan, R., R. El Asady, K. Liu, D. Wang, C. B. Drachenberg, and G. A. Hadley. 2002. Tolerization of dendritic cells by TS cells: the crucial role of inhibitory ϩ ϩ receptors ILT3 and ILT4. Nat. Immunol. 3: 237–243. 2005. Critical role for CD103 CD8 effectors in promoting tubular injury fol- 17. Rihs, S., C. Walker, J. C. Virchow, Jr., C. Boer, C. Kroegel, S. N. Giri, and lowing allogeneic renal transplantation. J. Immunol. 175: 2868–2879. ␣ ␤ R. K. Braun. 1996. Differential expression of E 7 integrins on bronchoalveolar 39. Leteurtre, E., M. C. Copin, M. Labalette, C. Noel, D. Roumilhac, F. R. Pruvot, lavage T lymphocyte subsets: regulation by ␣ ␤ -integrin crosslinking and M. Lecomte-Houcke, B. Gosselin, and J. P. Dessaint. 2000. Negative immuno- 4 1 ␣ ␤ TGF-␤. Am. J. Respir. Cell Mol. Biol. 15: 600–610. histochemical detection of CD103 ( E 7 integrin) in the inﬁltrates of acute re-

18. Sarnacki, S., B. Begue, H. Buc, F. Le Deist, and N. Cerf-Bensussan. 1992. En- jection in liver and kidney transplantation. Transplantation 70: 227–229. http://www.jimmunol.org/ hancement of CD3-induced activation of human intestinal intraepithelial lym- 40. Hadley, G. A., C. Charandee, M. R. Weir, D. Wang, S. T. Bartlett, and ␤ ϩ phocytes by stimulation of the 7-containing integrin defined by HML-1 mono- C. B. Drachenbere. 2001. CD103 CTL accumulate within the graft epithelium clonal antibody. Eur. J. Immunol. 22: 2887–2892. during clinical renal allograft rejection. Transplantation 72: 1548–1555. 19. Pauls, K., M. Schon, R. C. Kubitza, B. Homey, A. Wiesenborn, P. Lehmann, 41. Waaga, A. M., M. Gasser, I. Laskowski, and N. L. Tilney. 2000. Mechanisms of ␣ ␤ T. Ruzicka, C. M. Parker, and M. P. Schon. 2001. Role of integrin E(CD103) 7 chronic rejection. Curr. Opin. Immunol. 12: 517–521. ϩ for tissue-specific epidermal localization of CD8 T lymphocytes. J. Invest. Der- 42. Wang, D., R. Yuan, Y. Feng, R. El-Asady, D. L. Farber, R. E. Gress, P. J. Lucas, matol. 117: 569–575. and G. A. Hadley. 2004. Regulation of CD103 expression by CD8ϩ T cells 20. Karecla, P. I., S. J. Bowden, S. J. Green, and P. J. Kilshaw. 1995. Recognition of responding to renal allografts. J. Immunol. 172: 214–221. E-cadherin on epithelial cells by the mucosal T cell integrin ␣ ␤7(␣ ␤ ). ␣ ␤ M290 E 7 43. Teraki, Y., and T. Shiohara. 2002. Preferential expression of E 7 integrin Eur. J. Immunol. 25: 852–856. (CD103) on CD8ϩ T cells in the psoriatic epidermis: regulation by interleukins 21. Cepek, K. L., S. K. Shaw, C. M. Parker, G. J. Russell, J. S. Morrow, D. L. Rimm, 4 and 12 and transforming growth factor-␤. Br. J. Dermatol. 147: 1118–1126. and M. B. Brenner. 1994. Adhesion between epithelial cells and T lymphocytes 44. Seder, R. A., T. Marth, M. C. Sieve, W. Strober, J. J. Letterio, A. B. Roberts, and by guest on September 26, 2021 ␣ ␤ mediated by E-cadherin and the E 7 integrin. Nature 372: 190–193. B. Kelsall. 1998. Factors involved in the differentiation of TGF-␤-producing cells 22. Hadley, G. A., E. A. Rostapshova, D. M. Gomolka, B. M. Taylor, S. T. Bartlett, from naive CD4ϩ T cells: IL-4 and IFN-␥ have opposing effects, while TGF-␤ C. I. Drachenberg, and M. R. Weir. 1999. Regulation of the epithelial cell-spe- positively regulates its own production. J. Immunol. 160: 5719–5728. ϩ Transplantation cific integrin, CD103, by human CD8 cytolytic T lymphocytes. 45. Li, J., I. Goldstein, E. Glickman-Nir, H. Jiang, and L. Chess. 2001. Induction of 67: 1418–1425. TCR V␤-specific CD8ϩ CTLs by TCR V␤-derived peptides bound to HLA-E. 23. Feng, Y., D. H. Wang, R. W. Yuan, C. M. Parker, D. L. Farber, and G. A. Hadley. J. Immunol. 167: 3800–3808. 2002. CD103 expression is required for destruction of pancreatic islet allografts ϩ 46. Goronzy, J. J., C. Xie, W. Hu, S. K. Lundy, and C. M. Weyand. 1993. Restric- by CD8 T cells. J. Exp. Med. 196: 877–886. tions in the repertoire of allospecific T cells: contribution of the ␣-helical se- 24. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell de- quence polymorphism of HLA-DR molecules. J. Immunol. 151: 825–836. velopment by the transcription factor Foxp3. Science 299: 1057–1061. 25. Banz, A., A. Peixoto, C. Pontoux, C. Cordier, B. Rocha, and M. Papiernik. 2003. 47. Gilliet, M., and Y. J. Liu. 2002. Generation of human CD8 T regulatory cells by A unique subpopulation of CD4ϩ regulatory T cells controls wasting disease, CD40 ligand-activated plasmacytoid dendritic cells. J. Exp. Med. 195: 695–704. IL-10 secretion and T cell homeostasis. Eur. J. Immunol. 33: 2419–2428. 48. Malipiero, U., K. Frei, K. S. Spanaus, C. Agresti, H. Lassmann, M. Hahne, 26. Lehmann, J., J. Huehn, M. de la Rosa, F. Maszyna, U. Kretschmer, V. Krenn, J. Tschopp, H. P. Eugster, and A. Fontana. 1997. Myelin oligodendrocyte gly- ␣ ␤ coprotein-induced autoimmune encephalomyelitis is chronic/relapsing in perforin M. Brunner, A. Scheffold, and A. Hamann. 2002. Expression of the integrin E 7 identifies unique subsets of CD25ϩ as well as CD25Ϫ regulatory T cells. Proc. knockout mice, but monophasic in Fas- and Fas ligand-deficient lpr and gld mice. Natl. Acad. Sci. USA 99: 13031–13036. Eur. J. Immunol. 27: 3151–3160. 27. He, X., C. A. Janeway, Jr., M. Levine, E. Robinson, P. Preston-Hurlburt, C. Viret, 49. Grossman, W. J., J. W. Verbsky, B. L. Tollefsen, C. Kemper, J. P. Atkinson, and and K. Bottomly. 2002. Dual receptor T cells extend the immune repertoire for T. J. Ley. 2004. Differential expression of granzymes A and B in human cytotoxic foreign antigens. Nat. Immunol. 3: 127–134. lymphocyte subsets and T regulatory cells. Blood 104: 2840–2848. 28. Schelonka, R. L., F. M. Raaphorst, D. Infante, E. Kraig, J. M. Teale, and 50. Grossman, W. J., J. W. Verbsky, W. Barchet, M. Colonna, J. P. Atkinson, and A. J. Infante. 1998. T cell receptor repertoire diversity and clonal expansion in T. J. Ley. 2004. Human T regulatory cells can use the perforin pathway to cause human neonates. Pediatr. Res. 43: 396–402. autologous target cell death. Immunity 21: 589–601. 29. Maslanka, K., T. Piatek, J. Gorski, M. Yassai, and J. Gorski. 1995. Molecular 51. Gondek, D. C., L. F. Lu, S. A. Quezada, S. Sakaguchi, and R. J. Noelle. 2005. analysis of T cell repertoires: spectratypes generated by multiplex polymerase Cutting edge: contact-mediated suppression by CD4ϩCD25ϩ regulatory cells chain reaction and evaluated by radioactivity or fluorescence. Hum. Immunol. 44: involves a granzyme B-dependent, perforin-independent mechanism. J. Immunol. 28–34. 174: 1783–1786.