Recent data have demonstrated that treatment with αβ-TCR+CD3+CD4−CD8−NK1.1− double negative (DN) regulatory T cells (Tregs) inhibits autoimmune diabetes and enhances allotransplant and xenotransplant survival in an Ag-specific fashion. However, the mechanisms whereby DN Tregs suppress Ag-specific immune responses remain largely unknown. In this study, we demonstrate that murine DN Tregs acquire alloantigen in vivo via trogocytosis and express it on their cell surface. Trogocytosis requires specific interaction of MHC-peptide on APCs and Ag-specific TCR on DN Tregs, as blocking this interaction prevents DN Treg-mediated trogocytosis. Acquisition of alloantigen by DN Tregs was required for their ability to kill syngeneic CD8+ T cells. Importantly, DN Tregs that had acquired alloantigen were cytotoxic toward Ag-specific, but not Ag-nonspecific, syngeneic CD8+ T cells. These data provide new insight into how Tregs mediate Ag-specific T cell suppression and may enhance our ability to use DN Tregs as a therapy for transplant rejection and autoimmune diseases.
Regulatory T cells (Tregs)4 play an important role in controlling the development of various immune pathologies, including transplant rejection and autoimmune disease development (1, 2). Many subsets of Tregs have been identified, including CD4+CD25+ Tregs, Tr1 cells, Th3 cells, CD8+ T cells, γδ-TCR+ cells, NK T cells, and NK−αβ-TCR+CD4−CD8− double negative (DN) Tregs (1, 2, 3, 4, 5). Both murine and human DN Tregs have been shown to suppress allogeneic immune responses in an Ag-specific fashion (6, 7, 8). DN Tregs have been demonstrated to enhance donor skin, islet, and heart graft survival (6, 8, 9) and play a role in preventing graft-vs-host disease (10, 11). DN Treg-mediated suppression requires cell-cell contact and occurs via direct cytotoxicity toward T cells (6, 7, 8, 9). However, much remains unknown regarding the mechanisms whereby DN Tregs can interact with and kill Ag-specific syngeneic CD8+ T cells.
Studies originating in the early 1970s have shown that many lymphocytes, including T cells, are able to acquire foreign proteins, including membrane-bound proteins, from APCs (12, 13). Membrane acquisition, recently termed trogocytosis, can occur when T or B cells contact Ag-expressing cells in vitro (13, 14, 15, 16, 17, 18, 19). T cells have been shown to acquire MHC class I and class II, CD80, CD86, and ICAM-1 via trogocytosis from APCs (14, 15, 16, 17). Acquisition occurs in a rapid fashion and, depending on the cell type and the activation status, may require cell-cell contact (20, 21) or interaction with membrane exosomes that are shed from the APC (22). The mechanisms leading to trogocytosis are largely unknown, and the functional consequences of trogocytosis remain controversial. No previous experiments have investigated whether acquisition of alloantigens by a Treg population can augment its ability to suppress syngeneic immune responses in an Ag-specific fashion.
In this study, we investigated the Ag specificity of trogocytosis mediated by DN Tregs and whether trogocytosis enables DN Tregs to recognize and suppress T cell responses in an Ag-specific fashion. We demonstrate that DN Tregs can acquire allo-MHC-peptides in vivo and maintain expression for a longer period of time than CD8+ T cells. DN Tregs acquire alloantigen in a TCR-specific fashion. Importantly, we establish for the first time that alloantigen acquisition by DN Tregs in vivo enables them to induce apoptosis of Ag-specific, but not Ag-nonspecific, syngeneic CD8+ T cells. These findings suggest that alloantigen acquisition allows DN Tregs to specifically recognize target CTLs to inhibit immune responses. Understanding the functions and mechanisms of DN Tregs may move us closer to realizing their potential as an Ag-specific immune therapy.
Materials and Methods
2C (H-2b, expressing the 1B2+ anti-Ld transgenic TCR) breeders on a C57BL/6 background were kindly provided by Dr. D. H. Loh (Nippon Roche Research Center). BALB/c-H2dm2/KhEgJ mice, a BALB/c Ld-loss mutant, (H-2d, Ld0, The Jackson Laboratory) were bred with 2C mice to create 2CF1 mice (1B2+TCR, H-2b/d, Ld−) or with C57BL/6 mice to create B6 × Dm2 (H-2b/d, Ld−) mice. CB6F1/J mice (H-2b/d), referred to as CBy and CBy-SCID mice, were created by crossing C57BL/6 (H-2b) with BALB/c (H-2d) mice (Charles River) and B6.CB17-Prkdcscid/SzJ (H-2b) to CBySmn.CB17-Prkdcscid/J (H-2d) mice (The Jackson Laboratory), respectively. All mice were housed in specific pathogen-free conditions at the University Health Network (Toronto, Canada). All experiments have been approved by the University Health Network animal care committee.
Abs and peptides and PKH26 labeling
Monoclonal Abs used to identify DN Tregs include 1B2, specific for the 2C TCR (hybridoma provided by Dr. D. H. Loh, Nippon Roche Research Center), anti-CD4, anti-CD8 (eBioscience) and anti-Ld (HB-31). QL9 (QLSPFPFDL), mouse CMV (MCMV; YPHFMPTNL), p2Ca (LSPFPFDL), and P1A (LPYLGWLVF) peptides were synthesized by Sigma-Genosys. Ld+ or H-2s+ cells were labeled with the membrane-specific dye PKH26 (Sigma-Aldrich) by incubation for 10 min at room temperature then washed five times in cold PBS with 5% BSA.
Cell isolation, cloning, and culture
DN Treg clones were generated from Ld skin graft-tolerant 2CF1 mice as previously described (6). Primary Ld-specific CD8+ T cells, DN Tregs, and DN Treg clones were stimulated with irradiated CBy splenocytes in vitro in α-MEM medium supplemented with 10% FCS and 0.2% 2-ME with 50 U/ml IL-2 and 30 U/ml IL-4. Ld-GFP transfected Drosophila Schneider cells were generated and cultured as previously described (15). Expression of the transfected Ld-GFP was induced by the addition of 1 mM CuSO4 24 h before coculture with DN Tregs.
Ld-high- and Ld-low-expressing 2CF1 DN Tregs were sorted using a FACSAria cell sorter (BD Biosciences). DN Tregs were coincubated at the indicated ratios for 2–6 h with syngeneic 2CF1 or B6 × Dm2 CD8+ T cells that had been preactivated by in vitro stimulation with irradiated CBy (Ld+) or C3H (H-2k+) splenocytes, respectively, for 4 days. Cells were then stained with anti-CD8 mAb followed by labeling with annexin V and 7-aminoactinomycin D (7-AAD; BD Biosciences) according to the manufacturer’s directions.
Statistics were done using a paired Student’s t test. p < 0.05 was considered significant.
Results and Discussion
DN Tregs acquire alloantigens in vivo
Most reports that describe the acquisition of foreign proteins by T cells rely on data from in vitro studies. In this study we investigated the ability of DN Tregs to acquire MHC class I alloantigen in vivo. 2CF1 (transgenic anti-Ld TCR+, Db/d+, Kb/d+, Ld−) mice were infused with splenocytes from MHC class I alloantigen-expressing CBy (Db/d+, Kb/d+, Ld+) mice. Groups of mice were sacrificed at various time points and the spleen cells were stained to analyze their ability to acquire Ld alloantigen. As shown in Fig. 1,a, 1 day following infusion a similar proportion of DN Tregs and CD8+ T cells acquired and expressed Ld. However, the expression of Ld on CD8+ T cells diminished after 2 days and was at a very low level after 4 days (Fig. 1,a). In contrast, DN Tregs retained higher levels of Ld alloantigen throughout the course of the experiment (Fig. 1 a). These data demonstrate for the first time that DN Tregs can acquire MHC class I-Ld alloantigen in vivo and retain expression of alloantigen for at least 7 days. The ability of DN Tregs to express acquired membrane fragments for an extended period of time may allow the small number of DN Tregs to interact with and kill many effector T cells during an immune response.
Although DN Tregs were shown to acquire alloantigen in vivo, the percentage of cells that acquired Ld was very low. This may be due to either a limited overall alloantigen exposure in this model or because only a small subset of DN Tregs is able to acquire alloantigen in vivo. To distinguish these two possibilities, we increased the chance of alloantigen exposure by adoptively transferring Ld− 2CF1 DN Tregs into Ld+ CBy-SCID mice. We found that up to 70% of DN Tregs could acquire Ld alloantigen (Fig. 1,b), which was markedly increased compared with cells exposed to Ld+ alloantigen as a result of allogeneic lymphocyte infusion (Fig. 1,a). These data demonstrate that a high percentage of the overall DN Treg population is able to acquire alloantigens in vivo when unlimited Ag is present. Furthermore, alloantigen expression was higher on the surface of Ld-specific TCR-expressing (1B2+) DN Tregs when compared with DN T cells that do not express the Ld-specific TCR (1B2−) at all time points assessed (Fig. 1, b and c). These data suggest that alloantigen acquisition mediated by DN Tregs may occur in a TCR-alloantigen-specific fashion in vivo.
Trogocytosis of alloantigens requires specific TCR-MHC interaction
To confirm that DN Tregs mediate trogocytosis in an Ag-specific manner, APCs from either H-2d+ or H-2s+ mice were labeled with the membrane-specific dye PKH26, coincubated with 1B2+DN Tregs, and alloantigen acquisition was measured. As shown in Fig. 2 a, 1B2+DN Tregs that were incubated together with H-2d+ but not H-2s+ APCs acquired membrane fragments, suggesting that the acquisition of membrane fragments by DN Tregs requires TCR-specific alloantigen expression by APCs.
To further determine whether TCR-specific interaction with MHC expressed on an APC is required for DN Treg acquisition we used MHC tetramers and mAbs to block the Ld-specific TCR on the DN Tregs. Ld tetramers loaded with QL9, but not those loaded with MCMV peptides, can interact specifically with the 2CF1 1B2+ TCR. 1B2+DN Tregs were preincubated with either 1B2 Fab′ or Ld tetramers that were loaded with QL9 or MCMV peptides. As shown in Fig. 2 b, blocking either the TCR on DN Tregs with 1B2 Fab′ or Ld-QL9 tetramers abrogated the acquisition of membrane fragments from H-2d+ APCs, whereas preincubation with Ld MCMV tetramers could not block DN Tregs from acquiring Ld peptides from APC. These data further demonstrate the importance of the TCR on DN Tregs for the acquisition of MHC-peptide complexes from APCs.
We next determined whether the affinity of the peptide-MHC complex for the TCR on DN Tregs would affect the acquisition of alloantigen. Primary 1B2+DN Tregs were cocultured with CBy Ld+ APC in the presence of peptides that bind to Ld molecules but have differing affinities for the anti-Ld TCR. Two hours later, the percentage of DN Tregs that had acquired Ld molecules was assessed by flow cytometry. As some endogenous peptide-MHC complexes can bind to the anti-Ld-TCR (23), DN Tregs acquire Ld when cocultured with Ld+ APCs in the absence of additional peptides. Addition of P1A, which has no detectable affinity to the anti-Ld TCR (23), reduced the acquisition of Ld (Fig. 2,c), suggesting that this peptide prevents the acquisition of endogenous peptide-MHC by DN Tregs. Importantly, addition of either the high affinity peptide QL9 (binding affinity: Ka = 2 × 107 M−1) or the medium affinity peptide p2Ca (Ka = 2 × 106 M−1) (23), resulted in significantly higher levels of Ld acquisition when compared with those cocultured in the presence of P1A or without additional peptides (Fig. 2 c). These data suggest that DN Tregs have an increased ability to acquire Ld peptide complexes that bind with high affinity to their TCR.
One limitation of using flow cytometry to assess alloantigen acquisition is the inability to differentiate whether it is the T cell that is expressing the APC-associated molecules or the allogeneic APC that is expressing the T cell-associated markers. Moreover, whether DN Tregs directly interact with APCs to acquire alloantigens is not known. To overcome this limitation, we used confocal microscopy to directly visualize the acquisition of alloantigens mediated by 1B2+DN Tregs. 1B2+DN Treg clones were cocultured with APCs that had been transfected with the fluorescent marker GFP linked to MHC class I-Ld (15) in the presence of MCMV (Fig. 2,di) or QL9 (Fig. 2,d, ii--vi). Cells were then stained with the anti-Ld TCR-specific mAb 1B2-PE and visualized by fluorescence confocal microscopy. After 30 min, APCs could be seen interacting directly with DN Tregs (Fig. 2,d, i and ii). However, only DN Tregs that had been cocultured with QL9 (Fig. 2,diii), but not with MCMV (Fig. 2,di), expressed GFP on their cell surfaces. Furthermore, when the two fluorescent channels for green (Fig. 2,div) and red (Fig. 2,dv) are overlaid, the 1B2+ TCR on DN Tregs is seen to colocalize with GFP-Ld molecules (Fig. 2 dvi). These results further confirmed that it is the DN Treg that acquires alloantigens as opposed to the APCs acquiring TCR molecules. Collectively, our data demonstrate that acquisition of MHC-peptide complexes is mediated in an Ag-specific fashion through the interactions between the TCR on DN Tregs and specific peptide-MHC on APCs.
DN Treg trogocytosis of alloantigen is necessary for their ability to induce apoptosis in an Ag-specific fashion
Previous work has shown that human DN Tregs that had acquired alloantigen-peptide complexes in vitro are able to induce apoptosis of allogeneic target cells in an Ag-specific fashion (7). In this study we determined whether in vivo acquisition and expression of alloantigen is critical for DN Tregs to kill activated syngeneic CD8+ T cells. 2CF1 DN Tregs that had been i.v. injected into allogeneic Ld+ CBy-SCID mice and had acquired and expressed Ld on their surfaces were sort purified into either Ld-high- or Ld-low-expressing populations and used as putative effector cells to kill syngeneic CD8+ T cells. As shown in Fig. 3 a, the Ld-high DN Tregs, but not the Ld-low DN T cells, were able to induce CD8+ T cell apoptosis in a dose-dependent manner. These data demonstrate that only those DN Tregs that are able to acquire Ld in vivo and express the acquired alloantigens on their surfaces at a high level can kill syngeneic CD8+ T cells. Furthermore, it suggests that acquisition and expression of alloantigen by DN Tregs is required for their suppressive function, perhaps by facilitating Ag-specific target cell recognition.
We next wanted to investigate whether DN Tregs that have acquired alloantigen can induce apoptosis of target CD8+ T cells in an Ag-specific fashion. To do this, 2CF1 DN Tregs that had acquired Ld alloantigen in vivo were cocultured with syngeneic CD8+ T cell targets that had been activated by either the same alloantigen that DN Tregs had acquired (Ld) or a third party alloantigen (H-2k) and the expressions of annexin V (apoptosis marker) and 7-AAD (cell viability marker) were assessed. As shown in Fig. 3, b and c, annexin V and 7-AAD increased in a dose-dependent fashion on CD8+ T cells that had been activated with Ld alloantigen before coculture with DN Tregs that had acquired Ld in vivo. However, CD8+ T cells that had been activated with H-2k alloantigen did not increase the expression of annexin V or 7-AAD when cocultured with DN Tregs that had acquired Ld alloantigen. This suggests that DN Tregs that have trogocytosed alloantigen are only able to induce apoptosis of syngeneic CD8+ T cells that had been activated by the same alloantigen that has been acquired by the DN Treg and not those activated by a third party alloantigen. These data demonstrate for the first time that trogocytosis of alloantigen by peripheral DN Tregs is required for their ability to kill syngeneic CD8+ T cells and that DN Tregs that have acquired alloantigen induce cytotoxicity in an Ag-specific fashion. DN Tregs have been shown to increase in tolerant allografts and xenografts and to suppress allograft and xenograft rejection by suppressing CD4+ T cells (9, 24, 25). Further studies are required to determine whether trogocytosis occurs in transplanted organs and tissues and whether it plays a role in the ability of DN Tregs to suppress CD4+ T cell proliferation and survival.
Previous studies have shown that acquisition of Ag by CD8+ T cells sensitizes them to apoptosis by neighboring activated CD8+ T cells (6, 26). Tsang et al. show that CD4+ T cells that had acquired alloantigen were able to induce either the proliferation of naive T cells or the apoptosis of activated T cells (26). Interestingly, a recent study suggests that the acquisition of HLA-G by human CD4+ and CD8+ T cells confers suppressive function (21). However, it is not yet known whether FoxP3+CD4+ Tregs acquire alloantigens as a mechanism of suppression of CD4+ T cell proliferation. In our study we demonstrated that Ag-specific acquisition in vivo and the expression of alloantigen by DN Tregs allows them to specifically trap and kill syngeneic CD8+ T cells that can interact with the acquired alloantigen on the DN Tregs. These studies suggest that trogocytosis of Ag allows cells to either prime or inhibit immune responses, perhaps depending on the type of cell that has acquired the alloantigen as well as the types of Ag acquired and/or the type of target cell.
Taken together, our data presented here suggest the following model to explain how DN Tregs are able to specifically suppress syngeneic CD8+ T cells that carry the same TCR specificity. DN Tregs acquire foreign MHC-peptides from APC in a TCR-specific fashion and express them on their cell surfaces. CD8+ T cells that have the same TCR specificity as DN Tregs will recognize the molecules that have been acquired and expressed on the DN Treg surface. This Ag-specific recognition will bring DN Tregs into close contact with the CD8+ T cells and lead to the death of Ag-specific CD8+ T cells. The prolonged expression of acquired MHC molecules by DN Tregs could provide a window of opportunity for the low numbers of DN Tregs to kill many Ag-specific CD8+ T cells, making them highly potent Ag-specific Tregs. These findings unravel some of the critical mechanism whereby DN Tregs can suppress graft rejection in an Ag-specific fashion (6, 8, 9) and may facilitate the use of DN Tregs as an Ag-specific immune therapy for graft rejection, autoimmune type 1 diabetes, and graft-vs-host disease (10, 11, 27).
We thank Dr. Zeling Cai for generously providing the Drosophila cell line, Dr. Mark Peterson for assistance with confocal microscopy, and Joyce Pun and Michelle Tsang for cell sorting.
The authors have no financial conflict of interest.
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 with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Canadian Institute of Health Research and the Canadian Cancer Society.
Abbreviations used in this paper: Treg, regulatory T cell; 7-AAD, 7-aminoactinomyin D; DN, double negative; MCMV, mouse CMV.