T Cells Suppressive Function of Human Regulatory CD86 And

CD86 and CD80 Differentially Modulate the Suppressive Function of Human Regulatory T Cells

This information is current as Yong Zheng, Claire N. Manzotti, Michael Liu, Fiona Burke, of September 23, 2021. Karen I. Mead and David M. Sansom J Immunol 2004; 172:2778-2784; ; doi: 10.4049/jimmunol.172.5.2778 http://www.jimmunol.org/content/172/5/2778 Downloaded from

References This article cites 45 articles, 25 of which you can access for free at: http://www.jimmunol.org/content/172/5/2778.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

by guest on September 23, 2021 *average

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 © 2004 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

CD86 and CD80 Differentially Modulate the Suppressive Function of Human Regulatory T Cells1

Yong Zheng, Claire N. Manzotti, Michael Liu, Fiona Burke, Karen I. Mead, and David M. Sansom2

Regulatory T cells (Treg) are important in maintaining tolerance to self tissues. As both CD28 and CTLA-4 molecules are implicated in the function of Treg, we investigated the ability of their two natural ligands, CD80 and CD86, to inﬂuence the Treg-suppressive capacity. During T cell responses to alloantigens expressed on dendritic cells, we observed that Abs against CD86 potently enhanced suppression by CD4؉CD25؉ Treg. In contrast, blocking CD80 enhanced proliferative responses by impairing Treg suppression. Intriguingly, the relative expression levels of CD80 and CD86 on dendritic cells are modulated during pro- gression from an immature to a mature state, and this correlates with the ability of Treg to suppress responses. Our data show that CD80 and CD86 have opposing functions through CD28 and CTLA-4 on Treg, an observation that has signiﬁcant implications Downloaded from for manipulation of immune responses and tolerance in vivo. The Journal of Immunology, 2004, 172: 2778Ð2784.

he controlled manipulation of immune responses is an T cell activation (11, 12). However, this view does not incorporate important therapeutic goal. The ability to prevent T cell the role of CTLA-4 as a negative regulator, and an alternative T activation and establish long term tolerance is desirable in possibility is that the absence (or decreased levels) of CD86 might both transplantation and autoimmune disease settings. In contrast, affect the function of CTLA-4 via its alternative ligand CD80 (1). http://www.jimmunol.org/ heightened T cell activation is seen as a potential tumor therapy. Thus, the relative levels of CD80 and CD86 expressed on APCs The CD28-CTLA-4 pathway holds considerable promise for im- might affect the balance between CD28- and CTLA-4-dependent munotherapy because it is clear that CD28 is a key activator, outcomes. whereas CTLA-4 is a crucial attenuator, of T cell responses. CD28 Recently, it has become clear that, uniquely, both CD28 and is known to provide important stimulatory signals for T cells re- CTLA-4 are expressed constitutively on regulatory T cells (Treg)3 sulting in enhanced proliferation, cytokine production, and anti- and can have substantial influence on their function (13–16). As apoptotic signals. However, CD28 functions are opposed by the Treg are thought to suppress a variety of autoimmune diseases (17) related receptor CTLA-4 (CD152) (see Refs. 1–4 for review). The and be involved in the tolerance to allografts (18, 19), understand- importance of CTLA-4 in limiting T cell activation is demon- ing how these cells are regulated is of considerable significance. by guest on September 23, 2021 strated by the fatal hyperactivation of T cells seen in CTLA-4- Intriguingly, there have been suggestions that the state of DC mat- deficient mice (5), resulting in the destruction of a variety of tis- uration can influence the development and function of some types sues. This suggests that under conditions of normal homeostasis, of regulatory T cells (20) and promote T cell tolerance (21, 22). engagement of CTLA-4 is important in preventing T cell activa- With these issues in mind, we investigated the function of CD80 tion to self-Ags. It is therefore somewhat paradoxical that CTLA-4 and CD86 in modulating the function of Treg via CD28 and shares its only known ligands, CD80 and CD86, with CD28. CTLA-4. In response to allogeneic dendritic cells (DCs) we ob- A major gap in our understanding of the CD28/CTLA-4 system served that blockade of CD86 potently inhibited CD4ϩ T cell re- has been the role played by CD80 and CD86. Both ligands are sponses, whereas anti-CD80 enhanced the responses. Surprisingly, found on APCs and are known to provide efficient costimulation the inhibitory effect of anti-CD86 required the presence of via CD28; however, they have distinct patterns of expression. In CD4ϩCD25ϩ T cells, and the removal of these cells prevented general, CD86 is thought to be more widely expressed and at suppression. This suggested that stimulation of Treg via CD86 higher levels than CD80 (6–10). In addition, the more immuno- inhibits their suppressive function. Consistent with this view, al- deficient phenotype of the CD86 knockout mouse has generated a loresponses to cultured mature DCs (mDCs) that expressed high strong perception that CD86 is the more important costimulator of levels of CD86 were resistant to Treg suppression, whereas responses to immature DCs (iDCs) that expressed lower levels of CD86 were inhibited. Taken together, our data show that CD80 Medical Research Council Center for Immune Regulation, University of Birmingham and CD86 have opposite roles in the functioning of Treg via CD28 Medical School, Birmingham, United Kingdom and CTLA-4. Received for publication August 25, 2003. Accepted for publication December 18, 2003. Materials and Methods The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance Cell isolation and generation of DC with 18 U.S.C. Section 1734 solely to indicate this fact. Human monocytes were purified from PBMC by negative selection using 1 This work was supported by the Wellcome Trust (to Y.Z.) and the Arthritis Research human monocyte enrichment mixture and magnetic colloid according to Campaign (to D.M.S., C.N.M., and M.L.). D.M.S. is an Arthritis Research Campaign the instructions of the manufacturer (StemCell, Meylan, France). Briefly, Senior Fellow. K.M. is an Medical Research Council Ph.D. student. PBMC were isolated from fresh buffy coats (provided by the National 2 Address correspondence and reprint requests to Dr. David M. Sansom, Medical Research Council Center for Immune Regulation, University of Birmingham Medical School, Vincent Drive, Birmingham, U.K. B15 2TT. E-mail address: d.m.sansom@ 3 Abbreviations used in this paper: Treg, regulatory T cell; CHO, Chinese hamster bham.ac.uk ovary; DC, dendritic cell; mDC, mature DC.

Blood Transfusion Service (Birmingham, U.K.)) using a Ficoll-Paque gra- min. The tube was filled with PBS, and the cells were washed by centrif- dient. Cells were washed twice with PBS, then resuspended at 1 ϫ 108 ugation three times. The final cell pellet was made up to 2 ϫ 106 cell/ml. cells/ml in isolation buffer, and incubated with the monocyte enrichment Ab mixture at 4°C for 30 min. The cells were washed and subsequently Transfectants incubated with magnetic colloid at 4°C for 30 min. Unlabeled monocytes passed through the MACS column and were collected. To generate iDCs, Chinese hamster ovary (CHO) cells transfected with CD80 and CD86 were monocytes were cultured in RPMI 1640 medium containing 10% FCS and generated and used as previously described (15). Before use, cells were antibiotics with GM-CSF (PeproTech, Rocky Hill, NJ; 800 U/ml) and IL-4 fixed with 0.025% glutaraldehyde in PBS for 2–3 min, washed extensively with medium containing 10% FCS, and recounted. (PeproTech; 500 U/ml) at concentration of 2 ϫ 106 cells/ml. Half the medium was replaced every other day with GM-CSF- and IL-4-containing medium. Mature DCs were generated by stimulating iDCs with LPS (026: Results B6; Sigma-Aldrich, St. Louis, MO; 1 ␮g/ml) on day 5 for an additional Anti-CD86 inhibits alloresponses by affecting Treg suppression 24 h. To investigate the functional effects of CD80 and CD86 expression ϩ ϩ ϩ Ϫ Purification of human CD4 CD25 and CD4 CD25 T cells on DCs, we used allostimulation assays in which CD4ϩ T cells CD4ϩCD25ϩ and CD4ϩCD25Ϫ T cells were separated using specific anti- were stimulated with cultured DCs in the presence of blocking CD25 microbeads (Miltenyi Biotec, Auburn, CA) and positive or negative Abs. As shown in Fig. 1a, although CD86 blockade was highly selection, respectively. Initially, CD4ϩ T cells were purified by negative ϩ effective at inhibiting stimulation by mDCs, CD80 blockade was selection by incubating PBMC with human CD4 T cell enrichment mix- ineffective and even enhanced responses in a number of experi- ture and magnetic colloid according to the manufacturer’s instructions (Stemsep). CD4ϩ T cells were then resuspended in MACS buffer, incu- ments. These effects were observed at two different DC:T cell ra- bated with CD25ϩ microbeads on ice for 30 min, washed, and loaded on tios. Given the expression levels of CD80 and CD86 (Fig. 1a, Downloaded from the column. CD4ϩCD25Ϫ T cells, which did not bind to the column, were ϩ ϩ inset), these data seemed inconsistent with a simple model in collected from the flow-through and washed before use. CD4 CD25 T which anti-CD86 blocked CD28 costimulation, because in the ab- cells were subsequently retrieved from the column and washed before use. sence of CD86 interactions, CD80 should be able to compensate Flow cytometry and provide costimulation through CD28. In control experiments For analysis of DC phenotype, DCs were collected in cold PBS and pre- both anti-CD80 and anti-CD86 Abs completely and specifically incubated in 100 ␮l of rabbit serum at 37°C for 30 min to block FcRs. abolished costimulation by CD80 or CD86 transfectants, demon- mAbs directly conjugated to FITC or PE were subsequently used against strating their blocking ability over a range of Ab concentrations http://www.jimmunol.org/ CD80, CD86, HLA-DR CD14, CD40, and CD83 (BD PharMingen, San (Fig. 1b). Furthermore, despite using doses of anti-CD80 that were Diego, CA). In time-course experiments, cells were collected at different time points during the culture of monocytes in GM-CSF and IL-4 and stained with FITC-labeled CD80 and CD86. Stained cells were analyzed on a FACScan flow cytometer using CellQuest software (BD Biosciences, Mountain View, CA). In cell sorting experiments CD4ϩCD25ϩ T cells were stimulated with PMA (5 ng/ml) for 2 h to induce CTLA-4 recycling. Cells were stained for CTLA-4 expression at 37°C using CTLA-4-PE (BN13; BD PharMingen) and sorted into positive and negative populations

using a MoFlow cell sorter. Sorted cells were added to alloresponses as by guest on September 23, 2021 detailed below. Allostimulation Primary DC-stimulated MLR was conducted in 96-well, U-bottom tissue culture plates in 200 ␮l of RPMI 1640 containing 10% FCS and antibiotics. DCs were mixed with 1 ϫ 105 allogeneic total CD4ϩ T cells or CD4ϩCD25Ϫ T cells at a ratio between 1:10 and 1:100 DC:T cells. Cul- tures were also conducted in the presence or the absence of neutralizing mAbs: anti-human CD80 and CD86 (R&D Systems, Minneapolis, MN), anti-CD28 FabЈ (9.3; a gift from C. June (University of Pennsylvania, Ј ␮ Philadelphia, PA)) or anti-CTLA-4 F(ab )2 (Alexis) and used at 10 g/ml. Assays were incubated for 5 days, and during the last 16 h [3H]thymidine was added at 1 ␮Ci/well. [3H]thymidine incorporation was measured by scintillation counting, and proliferative responses were expressed as the mean [3H]thymidine incorporation (counts per minute) of triplicate wells Ϯ SD. Counts due to DCs alone were routinely Ͻ1000 cpm. Results shown are representative examples of a minimum of five experiments performed. ϩ ϩ FIGURE 1. Anti-CD86 inhibits alloresponses to cultured DCs by af- CD4 CD25 T cell assays fecting CD4ϩ CD25ϩ T cells. Total CD4ϩ (a)orCD4ϩ CD25Ϫ (d) T cell CD4ϩCD25ϩ T cells (1 ϫ 105) were preincubated with DCs (1 ϫ 104) for proliferation to allogeneic mDCs was measured at 6 days by [3H]thymidine 18 h in the presence of anti-CD80, anti-CD86, anti-CD28, anti-CTLA-4, or incorporation. Cultures were conducted at two different DC:T cell ratios control mouse IgG (10 ␮g/ml) as shown. Cells were then washed and and treated with control Ig (Ⅺ) or Abs to CD80 (f) or CD86 (o)at10 ϩ Ϫ transferred to CD4 CD25 T cells (1 ϫ 105), stimulated with allogeneic ␮ Ϯ 3 ϩ g/ml. Bars represent the mean of triplicate wells SD. [ H]thymidine DCs (at a ratio of either 1:10 to 1:100), and incubated for 5 days. CD25 Ͻ Ϫ incorporation by DCs alone was routinely 1000 cpm. Analysis of CD80 and CD25 cells were always from the same donor. As controls, and CD86 expression on mDC by FACS is shown in the inset to a. Filled CD4ϩCD25Ϫ T cells were preincubated with DCs and then transferred to ϩ Ϫ histograms represent specific Ab staining compared with isotype controls into CD4 CD25 alloresponses for 5 days. Proliferative responses were assessed by 3H incorporation) as detailed above. Results shown are repre- (open histograms). b and c, Titrations of anti-CD80 and -CD86 Abs were sentative examples of a minimum of six experiments performed. performed to confirm costimulation blockade. Purified CD4 T cells were stimulated with anti-CD3 (1 ␮g/ml) and transfectants expressing either CFSE labeling CD80 or CD86 in the presence of blocking Abs (b). c, Alloresponses stim- To determine proliferation, Treg cells were washed twice with PBS and ulated by mDCs were incubated in the presence of the blocking Abs shown. 3 incubated with 2.5 ␮M CFSE for 10 min at room temperature, agitating Responses in b and c are shown as a percentage of the maximal [ H]thy- gently every 2–3 min. The reaction was quenched by the addition of an midine incorporation without Ab blockade. The experiments shown are equal volume of RPMI 1640 containing 10% FCS and incubation for 1 representative of three to five independent experiments. 2780 REGULATION OF CD25ϩ T CELLS

20-fold in excess of that required to block costimulation by trans- inhibitory effects of Treg. Both these effects required the presence fectants, we did not observe inhibition using DCs (Fig. 1c). We of DCs, ruling out the possibility that the CD80 and CD86 Abs therefore investigated whether the differing effects of CD86 and acted directly on the Treg. Pretreatment of DCs with Abs in the CD80 blockade on DCs were due to effects other than blockade of absence of Treg did not modulate suppression (data not shown), CD28 costimulation. CD4ϩCD25ϩ cells were therefore depleted indicating that the effects were not due to Ab carried over into the from the total CD4ϩ population. Strikingly, the inhibitory effect of second culture or to alterations in DC phenotype due to Ab-in- CD86 blockade (Fig. 1d) was abrogated in the absence of Treg. duced signaling. We therefore concluded that the interaction of Thus, rather than simply inhibiting CD28 costimulation of CD25Ϫ Treg with CD86 on DCs inhibited their function (as CD86 block- T cells, CD86 blockade appeared to enhance the inhibitory func- ade enhanced suppression) and that CD80 interaction with Treg tion of Treg. promoted suppression (as CD80 blockade reversed suppression). To directly assess the effect of CD80 and CD86 blockade on As the Abs were only present during the initial 18-h Treg contact Treg function, we performed experiments on puriﬁed CD25ϩ T with DCs, this suggested that the ability of Treg to suppress was cells. CD4ϩCD25ϩ T cells were therefore preincubated in the acquired during this period. presence (Fig. 2a) or the absence (Fig. 2b) of DCs and blocking CD80 or CD86 Abs. These pretreated Treg were then washed to CD28 and CTLA-4 have opposite effects on Treg function remove the Ab and tested for their capacity to suppress allore- Given that CD80 and CD86 appeared to be differentially control- sponses. The results of this experiment (Fig. 2a) revealed that the ling Treg function, we investigated how this might be mediated by addition of stimulated Treg that had not been treated with blocking their receptors, CD28 and CTLA-4. Treg were therefore primed Ϫ Ab (control Ig) modestly inhibited the responses of CD25 cells. with DCs in the presence of anti-CTLA-4 and anti-CD28 Abs (ei- Downloaded from Ј However, pretreatment of Treg with anti-CD86 Ab enhanced sup- ther whole or F(ab )2), washed, and used to suppress alloresponses Ј pression, whereas pretreatment with anti-CD80 Ab diminished the as before. This demonstrated that blocking CTLA-4 using F(ab )2, during Treg contact with DCs, reversed their suppressive activity, whereas whole anti-CTLA-4 Ab had no blocking activity (Fig. 3a). In contrast, blocking anti-CD28 FabЈ Ab potentiated the suppres-

sive activity of Treg in a manner similar to that of anti-CD86 (Fig. http://www.jimmunol.org/ 3b). As the expression of CTLA-4 on puriﬁed CD25ϩ cells is not homogeneous, we hypothesized that CTLA-4 expression should correlate with suppressive activity. We therefore performed cell- sorting experiments in which CD4ϩ CD25ϩ T cells were stimulated to induce surface CTLA-4 expression and then sorted into CTLA-4-positive and -negative populations (Fig. 3c). The suppressive functions of these cells were then assessed in alloresponses as before. This experiment clearly demonstrated that CTLA-4ϩ Treg were substantially more suppressive than by guest on September 23, 2021 CTLA-4Ϫ Treg, consistent with a role for CTLA-4 in Treg function. Together, our data demonstrated that blocking CD86 or CD28 enhanced Treg function and, conversely, that blocking CD80 or CTLA-4 limited the suppressive activity of Treg. We therefore concluded that CD80 and CD86 have distinct functional interactions with CTLA-4 and CD28 on Treg Immature DC support Treg inhibition more effectively than mDC The above data suggested a model in which the relative expression levels of CD80 and CD86 on DCs could potentially modulate the potency of Treg via differential interactions with CD28 and CTLA-4. We therefore analyzed the expression patterns of CD80 and CD86 during the culture and maturation of monocyte-derived DCs to determine whether this might inﬂuence Treg behavior. This revealed clear changes in the expression patterns of these molecules (Fig. 4). Firstly, we observed that the initial PBMC population was CD86 positive and CD80 negative. However, in culture with GM-CSF and IL-4, we noted that by 24 h CD80 began to be expressed, whereas CD86 expression began to diminish. By 96 h the cells resembled iDCs and expressed substantial levels of CD80, but lower levels of CD86. To establish that these cells could progress further into mature DCs, the cells were stimulated with

ϩ LPS. This clearly showed substantial up-regulation of CD86 by FIGURE 2. CD80 and CD86 Abs directly affect CD25 T cell:DC in- 24 h (Fig. 4a) and that the mature cells up-regulated CD83 as well teractions. CD25ϩ T cells were incubated in the presence (a) or the absence as CD40 and MHC class II (Fig. 4b). (b) of stimulating mDCs in the presence of the Abs shown for 18 h. Cells were then washed and added to a CD25Ϫ alloresponse at a ratio of 1:1 Based on this expression pattern, we reasoned that the suppres- CD25ϩ:CD25Ϫ T cells. The CD25Ϫ alloresponse with control CD25Ϫ cells sive capacity of Treg should be greater when stimulated by iDCs added is shown (CD25Ϫ). Proliferation was measured by [3H]thymidine than by mDCs, because the levels of CD86 are lower. To test this Ϫ incorporation on day 6. Bars represent the mean of triplicate wells Ϯ SD. hypothesis, alloresponses were established using CD25 T cells The experiment is representative of more than ﬁve experiments performed. and either iDC or mDC. Treg were then added to these cultures at The Journal of Immunology 2781 Downloaded from http://www.jimmunol.org/ by guest on September 23, 2021

FIGURE 3. CTLA-4 and CD28 modulate suppression by CD25ϩ cells. CD25ϩ T cells were pretreated for 18 h with DCs and anti-CTLA-4 Abs (a)or anti-CD28 Abs (b) and compared with anti-CD80 or anti-CD86 blockade. All Abs were used at at10 ␮g/ml. Pretreated cells were washed and tested for suppression of an alloresponses between CD25Ϫ T cells and DCs (using a ratio of 1:1 CD25ϩ:CD25Ϫ T cells). c, For CTLA-4 sorting CD4ϩCD25ϩ T cells were stimulated with PMA (5 ng/ml) for 2 h, labeled with anti-CTLA-4-PE, and sorted into positive and negative populations as shown. The upper panel shows CTLA-4 expression before sorting (bold line, stimulated cells; dashes, unstimulated; thin line, isotype control), and the lower panel shows the positive and negative populations after sorting. Sorted cells were added to a CD25Ϫ alloresponse at the above ratio. DCs were present at 1:10 or 1:100 T cells. Proliferation was measured by [3H]thymidine incorporation. Bars represent the mean of triplicate wells Ϯ SD. Data are representative of three experiments performed. different ratios of Treg:T cells. The results of this experiment re- mained undivided, consistent with the nature of allostimulation. vealed that adding Treg to iDC (Fig. 5a) significantly suppressed However, when mDC were used in the presence of blocking CD80 their proliferation at a ratio of 1:1 Treg:T cells. In contrast, using and CD86 Abs, we observed that blocking CD86 inhibited divi- mDCs (Fig. 5b), little suppression was observed. To confirm that sion, whereas blocking CD80 enhanced Treg division. These find- CD86 was involved in abrogating suppression by mDC, blocking ings suggested that CD80 and CD86 engagement had opposite Abs were added. This revealed (Fig. 5c) that blockade of CD86 on effects on Treg division as well as suppressive function. mDC could indeed restore Treg suppression. Overall, this indicated that Treg suppression is significantly influenced by DC maturation Discussion and reflects the balance between CD80 and CD86 expression. It is clear that T cell tolerance to self-Ags is not purely a conse- Recent reports suggest that CD25ϩ Treg can proliferate in re- quence of T cell deletion in the thymus, but that other mechanisms sponse to DCs (23, 24). We therefore investigated whether CD80 exist to allow discrimination between pathogen-associated TCR or CD86 blockade might also influence this response. Treg were signaling events and those generated by our own tissues. Consid- labeled with CFSE, and their division was measured in response to erable evidence suggests that this relates to the recognition of DCs. This revealed (Fig. 6) that mDC induced more T cells to pathogen-associated molecules, or so-called danger signals, which divide than iDC. However, substantial numbers of T cells re- are recognized by cells of the innate immune system such as DCs. 2782 REGULATION OF CD25ϩ T CELLS

FIGURE 4. Expression patterns of CD80 and CD86 on iDCs and mDCs. a, DCs were derived from PBMC, and CD80 and CD86 expression was measured at various times in culture in GM- CSF and IL-4. DCs were matured by the addition of LPS for 24 h (mDC). Cells were stained using specific Abs shown (filled histograms), compared with a flu- orochrome-matched isotype control Ab (open histograms), and analyzed by FACS analysis. b, iDC and mDC were assessed for CD83, CD40, and MHC class II molecule expression. Data are representative of Ͼ10 independent experiments performed. Downloaded from

DC maturation occurs as a consequence of recognition of these CTLA-4 in preventing rejection (29, 30). Furthermore, other evi- http://www.jimmunol.org/ pathogen-associated molecules via a variety of Toll-like receptors dence, such as the ability of Abs to CD80 to exacerbate disease in (25, 26), but exactly how this information is subsequently com- NOD mice (31) and the tolerogenic potential of CD80-expressing, municated to T cells is not entirely clear. It is generally thought but not CD86-expressing, tumors (32), is consistent with the view that increased expression of MHC class II and costimulatory mol- that CD80 may be the more effective ligand for CTLA-4. ecules such as CD86 is important in this process. However, CD86 Our data also revealed that both CD28 and CD86 blockade had and its relative, CD80, bind to both stimulatory (CD28) and in- similar effects on Treg function, indicating that CD28 is primarily hibitory (CTLA-4) receptors on T cells, raising the significant involved in sensing the signals from CD86. Consistent with this question of how a stimulatory CD28 signal, as opposed to an in- observation Sakaguchi et al. (33) showed that CD28 stimulation hibitory CTLA-4 signal, is ensured. The present study demon- could abolish the inhibitory capacity of CD25ϩ regulatory T cells. by guest on September 23, 2021 strates that CD86 expression regulates this decision not as a result It therefore seems likely that CD86 represents a CD28-biased li- of enhanced costimulation, but by diminishing the capacity of Treg gand. In contrast, although CD80 is certainly an effective CD28 to suppress responses. ligand, its stimulatory effects are substantially opposed by its in- Our data support a model in which the relative expression levels teractions with CTLA-4, which is expressed on Treg after contact of CD80 and CD86 on DCs could dictate the balance between with DCs. This interpretation also receives strong support from stimulatory and inhibitory outcomes by modifying the potency of recent biophysical data (34) showing that CD80-CTLA-4 interac- Treg. This scenario is consistent with the fact that CD86 is highly tions are likely to be highly favored compared with CD86. Further responsive to danger signals such as LPS, CFA, and many other data consistent with a role for CD28 on Treg have been obtained inflammatory stimuli (12) that enhance T cell responses. Recently, from NODϫCD28Ϫ/Ϫ mice (16), which have exacerbated diabetes others have also suggested that DC maturation can influence the due to a lack of Treg. Thus, CD28 signals are probably important inhibitory effects of Treg (27). However, in contrast to our find- for the expansion and/or survival of Treg, a role that we would ings, these studies showed that cytokine signals, possibly mediated suggest requires CD86 engagement. This concept gains further by IL-6, could limit the ability of the responder T cells to be reg- support from our observation that mDCs, which express higher ulated by Treg. Combined with our data this suggests that the levels of CD86 than iDCs, are more effective at driving Treg pro- ability of Treg to suppress T cell responses may be dictated by liferation. Here again, our experiments revealed that CD80 and both signals received by the Treg themselves and the status of the CD86 have opposing roles in influencing Treg proliferation. responder T cells. In contrast to the roles of CD86 and CD28, we observed that Our experiments clearly indicate distinct functions for CD80 both CD80- and CTLA-4-blocking Abs enhanced T cell responses and CD86. The differential functions of these molecules have been to DCs, indicating that CD80 acts as a ligand for CTLA-4 on Treg the subject of considerable study, with most data suggesting that and is involved in their suppressive function. The role of CTLA-4 the two ligands share substantially overlapping functions (4, 12, in Treg function is somewhat controversial, because it appears to 28). However, it seems likely that such studies measured predom- play a role in some models (13, 14, 18, 19), but not in others (35, inantly CD28-dependent activation of CD25Ϫ T cells, in which 36). In our experiments both CD80 and CTLA-4 Abs inhibited both CD80 and CD86 appear to perform similarly. There is, how- suppression by Treg. However, these effects were less pronounced ever, increasing evidence to support the view that CD80 may be a than the enhancement of suppression observed with CD28 and more effective ligand for CTLA-4 than CD86. Our own studies, CD86 Abs. We believe that this is due to the difficulty of disrupt- directly comparing the ability of CD80 and CD86 transfectants to ing CD80-CTLA-4 interactions. Firstly the affinity of this interac- stimulate T cell responses, indicate that CTLA-4 inhibition is only tion is 100-fold greater than that of CD86-CD28 and possibly as observed with CD80 as the ligand (15). In addition, in transplan- great as 10,000-fold if the bivalent nature of CTLA-4 binding is tation models it is clear that CD80 is the effective ligand for taken into account (34). Secondly, delivery of CTLA-4 directly The Journal of Immunology 2783 Downloaded from http://www.jimmunol.org/

FIGURE 5. Alloresponses stimulated by iDCs are more effectively suppressed by CD25ϩ T cells than those of mDCs. Six-day alloresponses were established using CD25Ϫ T cells iDCs (a) or mDCs (b). CD25ϩ T cells were titrated into these cultures at the ratios of CD25ϩ T cells:CD25Ϫ T cells shown by guest on September 23, 2021 to assess potency. Experiments were conducted at a DC:T cell ratio of either 1:10 or 1:100. Responses were measured by [3H]thymidine incorporation. Bars represent the mean of triplicate wells Ϯ SD. c, mDC were used to stimulate responses as described in b at a ratio of 1:1 CD25Ϫ:CD25ϩ in the presence of Abs to CD80 (f), CD86 (^), or control Ig (Ⅺ). Data are representative of between three and five experiments. from an intracellular compartment to the immune synapse (37) at blockade, because these have sufficient overall avidity combined makes it very inaccessible to blockade by Ab. Accordingly, in our with the small physical size required for access to the immune Ј ϩ experience only F(ab )2, but not whole CTLA-4 Abs, are effective synapse. Our demonstration that CTLA-4 Treg are more suppressive than CTLA-4Ϫ Treg provides further strong support for a role for CTLA-4 in the function of Treg. Finally, consistent with a role for CD80 and CTLA-4 in Treg function, it is interesting to note that CD80-deficient NOD mice have exacerbated diabetes, where it is known that Treg play a role in protection (2). The CD28-CTLA-4 pathway has been extensively targeted in immunotherapy and is therefore of considerable clinical relevance (38–40). The most well-developed reagent, CTLA-4-Ig, a soluble antagonist of both CD80 and CD86, has now been used in human clinical trials (41, 42). However, a potential limitation of this reagent is that it has the capacity to interfere with the natural inhibitory functions of CTLA-4 as well as inhibit the activating function of CD28. Thus, CTLA-4-Ig may be less effective than reagents that leave the natural inhibitory function of CTLA-4 intact. In other strategies, CD80 and CD86 have been targeted to tumors in an attempt to stimulate T cell responses. However, as there is evidence that CTLA-4 is involved in suppressing responses to tumors (43–46), a knowledge of how CD80 and CD86 ligands interact ϩ with CD28 and CTLA-4 is essential to avoid immune suppression FIGURE 6. CD80 and CD86 differentially modulate CD25 T cell division. CFSE-labeled Treg cells were incubated with mDC or iDC in an rather than stimulation. Thus, there is a need to clearly define the allostimulation assay for 6 days (a). The effects of anti-CD80 and anti- roles of CD80 and CD86 and their contributions to both stimulat- CD86 Abs (10 ␮g/ml) on division stimulated by mDC were studied (b). ing and inhibiting T cell responses. Our data now present clear Cell division was monitored by flow cytometry after 6 days. evidence that for human T cells, CD86-CD28 interactions represent a 2784 REGULATION OF CD25ϩ T CELLS potent signal that interferes with the inhibitory function of Treg, 23. Yamazaki, S., T. Iyoda, K. Tarbell, K. Olson, K. Velinzon, K. Inaba, and ϩ ϩ whereas CD80 ligation promotes regulatory function by interacting R. M. Steinman. 2003. Direct expansion of functional CD25 CD4 regulatory T cells by antigen-processing dendritic cells. J. Exp. Med. 198:235. with CTLA-4. These observations provide further evidence of differ- 24. Walker, L. S., A. Chodos, M. Eggena, H. Dooms, and A. K. Abbas. 2003. An- ential functions of CD80 and CD86 and suggest approaches for the tigen-dependent proliferation of CD4ϩ CD25ϩ regulatory T cells in vivo. J. Exp. development of more rational immunotherapies. Med. 198:249. 25. Janeway, C. A., Jr., and R. Medzhitov. 2002. Innate immune recognition. Annu. Rev. Immunol. 20:197. Acknowledgments 26. Medzhitov, R., and C. Janeway, Jr. 2000. The Toll receptor family and microbial recognition. Trends Microbiol. 8:452. We thank Drs. Peter Lane and Lucy Walker for helpful comments. 27. Pasare, C., and R. Medzhitov. 2003. Toll pathway-dependent blockade of CD4ϩCD25ϩ T cell-mediated suppression by dendritic cells. Science 299:1033. References 28. Lanier, L., S. O’Fallon, C. Somoza, J. H. Phillips, P. S. Linsley, K. Okumura, D. Ito, and M. Azuma. 1995. CD80(B7) and CD86(B70) provide similar co- 1. Sansom, D. M., C. N. Manzotti, and Y. Zheng. 2003. What’s the difference stimulatory signals for T cell proliferation, cytokine production and generation of between CD80 and CD86? Trends Immunol. 24:313. CTL. J. Immunol. 154:97. 2. Salomon, B., and J. A. Bluestone. 2001. Complexities of CD28/B7: CTLA-4 costimulatory pathways in autoimmunity and transplantation. Annu. Rev. Immu- 29. Yamada, A., K. Kishimoto, V. M. Dong, M. Sho, A. D. Salama, N. G. Anosova, nol. 19:225. G. Benichou, D. A. Mandelbrot, A. H. Sharpe, L. A. Turka, et al. 2001. CD28- 3. Sansom, D. M. 2000. CD28, CTLA-4 and their ligands: who does what and to independent costimulation of T cells in alloimmune responses. J. Immunol. 167: whom?. Immunology 101:169. 140. 4. Sharpe, A., and G. J. Freeman. 2002. The B7-CD28 superfamily. Nat. Rev. Im- 30. Judge, T. A., Z. Wu, X. G. Zheng, A. H. Sharpe, M. H. Sayegh, and L. A. Turka. munol. 2:116. 1999. The role of CD80, CD86, and CTLA4 in alloimmune responses and the 5. Waterhouse, P., J. M. Penninger, E. Timms, A. Wakeham, A. Shahinian, induction of long-term allograft survival. J. Immunol. 162:1947. K. P. Lee, C. B. Thompson, H. Griesser, and T. W. Mak. 1995. Lymphoprolif- 31. Lenschow, D. J., S. C. Ho, H. Sattar, L. Rhee, G. Gray, N. Nabavi, K. C. Herold, erative disorders with early lethality in mice deficient in CTLA-4. Science and J. A. Bluestone. 1995. Differential effects of anti-B7-1 and B7-2 monoclonal Downloaded from 270:985. antibody treatment on the development of diabetes in the nonobese diabetic 6. Inaba, K., M. Inaba, M. Witmer-Pack, K. Hatchcock, R. Hodes, and mouse. J. Exp. Med. 181:1145. R. M. Steinman. 1995. Expression of B7 costimulator molecules on mouse den- 32. LaBelle, J. L., C. A. Hanke, B. R. Blazar, and R. L. Truitt. 2002. Negative effect ϩ ϩ dritic cells. Adv. Exp. Med. Biol. 378:65. of CTLA-4 on induction of T-cell immunity in vivo to B7-1 , but not B7-2 , 7. Inaba, K., M. Witmer-Pack, M. Inaba, K. S. Hathcock, H. Sakuta, M. Azuma, murine myelogenous leukemia. Blood 99:2146. H. Yagita, K. Okumura, P. S. Linsley, S. Ikehara, et al. 1994. The tissue distri- 33. Takahashi, T., Y. Kuniyasu, M. Toda, N. Sakaguchi, M. Itoh, M. Iwata, bution of the B7-2 costimulator in mice: abundant expression on dendritic cells J. Shimizu, and S. Sakaguchi. 1998. Immunologic self-tolerance maintained by CD25ϩCD4ϩ naturally anergic and suppressive T cells: induction of autoimmune in situ and during maturation in vitro. J. Exp. Med. 180:1849. http://www.jimmunol.org/ 8. Azuma, M., D. Ito, H. Yagita, K. Okumura, J. Phillips, L. Lanier, and C. Somoza. disease by breaking their anergic/suppressive state. Int. Immunol. 10:1969. 1993. B70 antigen is a second ligand for CTLA-4 and CD28. Nature. 366:76. 34. Collins, A., D. Brodie, R. Gilbert, A. Iaboni, R. Manso-Sancho, B. Walse, 9. Freeman, G. J., G. S. Gray, C. D. Gimmi, D. B. Lombard, L. Zhou, M. White, D. Stuart, P. van der Merwe, and S. Davis. 2002. The interaction properties of J. D. Fingeroth, J. B. Gribben, and L. M. Nadler. 1991. Structure, expression, and costimulatory molecules revisited. Immunity 17:201. T cell costimulatory activity of the murine homologue of the human B lympho- 35. Levings, M. K., R. Sangregorio, and M. G. Roncarolo. 2001. Human cyte activation antigen B7. J. Exp. Med. 174:625. CD25ϩCD4ϩ T regulatory cells suppress naive and memory t cell proliferation 10. Freeman, G. J., J. G. Gribben, V. A. Boussiotis, J. W. Ng, V. A. Restivo, and can be expanded in vitro without loss of function. J. Exp. Med. 193:1295. L. A. Lombard, G. S. Gray, and L. M. Nadler. 1993. Cloning of B7-2: a CTLA-4 36. Shevach, E. M. 2002. CD4ϩ CD25ϩ suppressor T cells: more questions than counter-receptor that costimulates human T cell proliferation. Science 262:909. answers. Nat. Rev. Immunol. 2:389. 11. Sharpe, A. H. 1995. Analysis of lymphocyte costimulation in-vivo using trans- 37. Egen, J., and J. Allison. 2002. Cytotoxic T lymphocyte antigen-4 accumulation in genic and knockout mice. Curr. Opin. Immunol. 7:389. the immunological synapse is regulated by TCR signal strength. Immunity 16:23.

12. Borriello, F., M. P. Sethna, S. D. Boyd, A. N. Schweitzer, E. A. Tivol, D. Jacoby, 38. Lenschow, D., Y. Zeng, J. Thistlethwaite, A. Monty, W. Brady, N. Gibson, by guest on September 23, 2021 T. B. Strom, E. M. Simpson, G. J. Freeman, and A. H. Sharpe. 1997. B7-1 and P. Linsley, and J. Bluestone. 1992. Longterm survival of xenogeneic pancreatic B7-2 have overlapping, critical roles in immunoglobulin class switching and islet grafts induced by CTLA4Ig. Science 257:789. germinal center formation. Immunity 6:303. 39. Linsley, P. S., P. M. Wallace, J. Johnson, M. G. Gibson, J. L. Greene, 13. Read, S., V. Malmstrom, and F. Powrie. 2000. Cytotoxic T lymphocyte-associ- ϩ ϩ J. A. Ledbetter, C. Singh, and M. A. Tepper. 1992. Immunosuppression in vivo ated antigen 4 plays an essential role in the function of CD25 CD4 regulatory by a soluble form of the CTLA-4 T cell activation molecule. Science 257:792. cells that control intestinal inﬂammation. J. Exp. Med. 192:295. 40. Larsen, C. P., E. T. Elwood, D. Z. Alexander, S. C. Ritchie, R. Hendrix, 14. Takahashi, T., T. Tagami, S. Yamazaki, T. Uede, J. Shimizu, N. Sakaguchi, C. Tucker-Burden, H. R. Cho, A. Aruffo, D. Hollenbaugh, P. S. Linsley, et al. T. W. Mak, and S. Sakaguchi. 2000. Immunologic self-tolerance maintained by ϩ ϩ 1996. Long-term acceptance of skin and cardiac allografts after blocking CD40 CD25 CD4 regulatory T cells constitutively expressing cytotoxic T lympho- and CD28 pathways. Nature 381:434. cyte-associated antigen 4. J. Exp. Med. 192:303. 15. Manzotti, C. N., H. Tipping, L. C. A. Perry, K. I. Mead, P. J. Blair, Y. Zheng, and 41. Guinan, E. C., V. A. Boussiotis, D. Neuberg, L. L. Brennan, N. Hirano, D. M. Sansom. 2002. Inhibition of human T cell proliferation by CTLA-4 utilizes L. M. Nadler, and J. G. Gribben. 1999. Transplantation of anergic histoincom- CD80 and requires CD25ϩ regulatory T cells. Eur. J. Immunol. 32:2888. patible bone marrow allografts. N. Engl. J. Med. 340:1704. 16. Salomon, B., D. J. Lenschow, L. Rhee, N. Ashourian, B. Singh, A. Sharpe, and 42. Abrams, J. R., M. G. Lebwohl, C. A. Guzzo, B. V. Jegasothy, M. T. Goldfarb, J. A. Bluestone. 2000. B7/CD28 costimulation is essential for the homeostasis of B. S. Goffe, A. Menter, N. J. Lowe, G. Krueger, M. J. Brown, et al. 1999. the CD4ϩCD25ϩ immunoregulatory T cells that control autoimmune diabetes. CTLA4Ig-mediated blockade of T-cell costimulation in patients with psoriasis Immunity 12:431. vulgaris. J. Clin. Invest. 103:1243. 17. Sakaguchi, S. 2000. Animal models of autoimmunity and their relevance to hu- 43. Sutmuller, R. P., L. M. van Duivenvoorde, A. van Elsas, T. N. Schumacher, man diseases. Curr. Opin. Immunol. 12:684. M. E. Wildenberg, J. P. Allison, R. E. Toes, R. Offringa, and C. J. Melief. 2001. ϩ ϩ Synergism of cytotoxic T lymphocyte-associated antigen 4 blockade and deple- 18. Kingsley, C. I., M. Karim, A. R. Bushell, and K. J. Wood. 2002. CD25 CD4 ϩ regulatory T cells prevent graft rejection: CTLA-4- and IL-10-dependent immu- tion of CD25 regulatory T cells in antitumor therapy reveals alternative path- noregulation of alloresponses. J. Immunol. 168:1080. ways for suppression of autoreactive cytotoxic T lymphocyte responses. J. Exp. 19. Taylor, P. A., R. J. Noelle, and B. R. Blazar. 2001. CD4ϩCD25ϩ immune reg- Med. 194:823. ulatory cells are required for induction of tolerance to alloantigen via costimu- 44. Hodi, F. S., M. C. Mihm, R. J. Soiffer, F. G. Haluska, M. Butler, M. V. Seiden, latory blockade. J. Exp. Med. 193:1311. T. Davis, R. Henry-Spires, S. MacRae, A. Willman, et al. 2003. Biologic activity 20. Jonuleit, H., E. Schmitt, G. Schuler, J. Knop, and A. H. Enk. 2000. Induction of of cytotoxic T lymphocyte-associated antigen 4 antibody blockade in previously interleukin 10-producing, nonproliferating CD4ϩ T cells with regulatory prop- vaccinated metastatic melanoma and ovarian carcinoma patients. Proc. Natl. erties by repetitive stimulation with allogeneic immature human dendritic cells. Acad. Sci. USA 100:4712. J. Exp. Med. 192:1213. 45. Phan, G. Q., J. C. Yang, R. M. Sherry, P. Hwu, S. L. Topalian, 21. Dhodapkar, M. V., R. M. Steinman, J. Krasovsky, C. Munz, and N. Bhardwaj. D. J. Schwartzentruber, N. P. Restifo, L. R. Haworth, C. A. Seipp, L. J. Freezer, 2001. Antigen-speciﬁc inhibition of effector T cell function in humans after in- et al. 2003. Cancer regression and autoimmunity induced by cytotoxic T lym- jection of immature dendritic cells. J. Exp. Med. 193:233. phocyte-associated antigen 4 blockade in patients with metastatic melanoma. 22. Steinman, R. M., and M. C. Nussenzweig. 2002. Inaugural Article: Avoiding Proc. Natl. Acad. Sci. USA horror autotoxicus: the importance of dendritic cells in peripheral T cell toler- 46. Leach, D. R., M. F. Krummel, and J. P. Allison. 1996. Enhancement of antitu- ance. Proc. Natl. Acad. Sci. USA 99:351. mour immunity by CTLA-4 blockade. Science 271:1734.