|Home | About | Journals | Submit | Contact Us | Français|
Graft rejection by the immune system is a major cause of transplant failure. Lifelong immunosuppression decreases the incidence of graft rejection; however, nonspecific immunosuppression results in increased susceptibly to infection and cancer. Regulatory T cells (Tregs), which suppress the activation of the immune system and induce tolerance, are currently under evaluation for use in clinical transplantation. Ex vivo expanded polyclonal Tregs that are introduced into transplant recipients alter the balance of T effector cells to Tregs; however, experimental data suggest that alloantigen-specific Tregs would be more effective at preventing graft rejection. We have developed a method to enrich alloantigen-specific human Tregs based on the coexpression of activation markers, CD69 and CD71. These Tregs could be readily expanded in vitro and demonstrated potent antigen-specific suppression. In a humanized mouse model of alloimmune-mediated injury of human skin grafts, alloantigen-specific Tregs resulted in a significant reduction in clinically relevant indicators of dermal tissue injury when compared with polyclonal Tregs, restoring a histology comparable to healthy skin. This method of human allospecific Treg selection should be scalable to the clinic. The improved in vivo efficacy of alloantigen-specific Tregs over polyclonal Tregs shown here suggests that generating “customized” Tregs with defined anti-donor allospecificities may improve current practice in clinical immunotherapy.
Patients who receive organ transplants undergo lifelong immunosuppressive therapy to maintain their allograft. Immunosuppression is required to control the vigorous immune response triggered by allogeneic major histocompatibility complex (MHC) antigens from donor tissue, which pose the main obstacle to achieving transplantation tolerance. Two main pathways of antigen presentation, direct and indirect, are responsible for donor alloantigen recognition: Each pathway makes a distinct contribution to the development of the alloimmune response. Although immunosuppression can effectively manage acute rejection, it is unable to prevent generation of anti-donor alloreactivity over time. As a consequence, many patients succumb to immune-mediated chronic allograft rejection (1, 2) or the associated toxicities and side effects of long-term immunosuppression. Treatment of chronic rejection often requires retransplantation, further compounding the shortage of organs available for transplantation. Development of regulatory T cells (Tregs) as novel therapeutics is therefore based on the need for an alternative to the current clinical management of transplant recipients.
Naturally occurring Tregs are a functionally and phenotypically distinct lymphocyte subset constituting 1 to 5% of total CD4+ T cells. In humans, Tregs are defined by constitutive expression of the transcription factor FoxP3 (Forkhead box P3), the surface profile CD25hiCD127lo/−, and mechanistically broad immunosuppressive effects (3, 4). Although their important immunomodulatory role has been recognized for many years (5), Tregs have only recently made the transition from bench to bedside; they are currently being evaluated as cellular therapeutics in several clinical transplantation settings (6, 7).
One of the main approaches to using Tregs in clinical transplantation (8) is to enrich Tregs from a prospective transplant recipient, expand them ex vivo, and then reintroduce them into the patient after transplantation. This strategy aims to alter the in vivo balance of T effector cells (Teffs) to Tregs by exploiting their natural suppressive functions. However, this approach would not negate all the caveats associated with standard immunosuppression, because polyclonal Tregs would also deliver pan-immunosuppressive effects. Experimental rodent models of transplantation (9-11) and immune monitoring of transplant patients (12, 13) indicate that these side effects could be avoided by specific suppression of the expansion of donor-alloreactive Teffs after transplantation, which can be achieved using donor allospecific Tregs. Collectively, these studies demonstrate that Tregs with defined specificities for donor alloantigens are more effective at preventing allograft rejection and improving clinical outcome than polyclonal Tregs. The advantage conferred by donor-specific Tregs may be attributed to their immunomodulatory functions being concentrated at the site of alloantigen source and immune activation, whereas polyclonal Tregs would be more systemically distributed (14, 15). Moreover, fewer donor-specific Tregs than polyclonal Tregs would be required to suppress the expansion of donor-reactive Teffs in vivo. However, challenges remain in the ability to test the relative efficacy of human polyclonal and antigen-specific Tregs, including the ability to generate large numbers of highly pure human alloantigen-specific Tregs and also the need of an appropriate in vivo model to test human Treg function.
Current methods for ex vivo expansion of human donor-specific Tregs require repetitive stimulation of polyclonal Tregs with donor antigen-presenting cells (APCs) to increase the proportion of Tregs with (direct pathway) allospecificity (16). This approach is strictly limited by the logistical challenge of expanding such a small population, and persistence of Tregs with other specificities may hinder the expansion of alloantigen-specific Tregs, ultimately masking their true functional potency. To test human Treg therapy, humanized mouse models have emerged as an alternative to nonhuman primates to study human immunity, particularly in the context of allotransplantation (17-19). Although these models are crucial in validating the in vivo function of Tregs, much work remains to optimize their translation to clinical application, allowing us to evaluate the contribution of antigen specificity in determining Treg function.
Here, we demonstrate a method that enriches human Tregs with direct pathway allospecificity, allowing them to be expanded to clinically useful numbers in vitro without the need of further restimulation. This protocol has permitted the in vivo demonstration of the superior efficacy of antigen-specific Tregs, compared to polyclonal Tregs, in protecting against alloimmune-mediated injury of human skin grafts using a humanized mouse model of xenotransplantation.
Cell surface markers indicating Teff cellular activation have been studied comprehensively; however, few studies have reported on expression of activation markers by Tregs. To select alloantigen-specific Tregs, we hypothesized that by exposing a polyclonal pool of Tregs to an allogeneic stimulator, Tregs with specificity for the nominal alloantigen could be detected by display of activation markers. Freshly isolated human Tregs and autologous Teffs (Fig. 1A) were screened by flow cytometry for the basal expression of potential markers of cellular activation, in addition to markers associated with Treg phenotype (Fig. 1, B and C) (20-22). As a positive control for cellular activation, T cell subsets were stimulated polyclonally using CD3CD28 beads for 1 to 5 days (Fig. 1D). Tregs responded to polyclonal stimulation with similar alterations in activation marker expression to Teffs, namely, steady up-regulation of all markers studied over 5 days, with the exception of CD62L, which was transiently down-regulated by a proportion of the Teff cell subset. The most consistent and useful markers of activation were CD69 and CD71, because these had low basal expression in unstimulated cultures and were up-regulated by most of the T cell population after polyclonal stimulation.
The kinetics of activation marker expression in response to direct pathway alloantigen presentation was next examined in mixed lymphocyte reaction (MLR) using HLA-DRβ1–mismatched, blood-derived myeloid CD1c+ dendritic cells (DCs) as stimulators (Fig. 2A). Alloantigen-activated Tregs demonstrated similar modulation of CD39, CD62L, CD38, CD24, CD69, and CD71 expression as observed on polyclonal stimulation, although the proportion of activation marker expressing cells was considerably higher than expected for markers such as CD39 and CD38, suggesting that they were nonspecific markers of activation. Alloantigen-activated Tregs up-regulated expression of CD69 and CD71 with similar kinetics and in a similar proportion, with expression peaking at day 3 and sustained until day 5. When examined together, CD69 and CD71 were found to be discretely coexpressed by activated Tregs after either MLR or polyclonal stimulation (Fig. 2B). The use of CD69 and CD71 as dual parameters enabled more accurate identification of recently activated T cells, excluding any possible preactivated (CD69−CD71+) or bystander activated T cells (CD69+CD71−) (23), and was therefore used for analysis of MLR cultures after 4 days. Addition of HLA-DRβ1–blocking antibody to cultures inhibited expression of CD69 and CD71 in MLR, but not polyclonally stimulated cultures, confirming that activation marker up-regulation by Tregs was specifically in response to the recognition of intact allogeneic HLA molecules (Fig. 2C). More detailed phenotypic analysis of enriched CD69+CD71+ activated subpopulations showed that alloantigen responder Tregs were CD27hi, CTLA-4+, GITR+, TGFβLAP+, HLA-DR+, CD73−, CD38+/−, CD39+ and predominantly of a naïve CD62L+CD45RA+ phenotype (Fig. 2D).
Using sorted CD4+CD25+CD127lo/− Tregs, we determined the mean percentage of direct pathway alloreactive Tregs identified by CD69+CD71+ coexpression as 2.59 ± 1.19% (Table 1). Combining activation marker expression with cell proliferation analysis showed that after 4 days of MLR, most alloactivated CD69+CD71+ Tregs had undergone several divisions, whereas nonactivated CD69−CD71− cells had remained largely undivided, although some CFSEdim cells were detectable (fig. S1). This latter observation suggests that CFSE (carboxyfluorescein diacetate succinimidyl ester) dilution alone would not result in the specific enrichment of T cell receptor (TCR)–stimulated Tregs.
To validate whether the activation profile of CD69 and CD71 coexpression could be used to enrich Tregs with antigen specificity for the primary stimulator, we set up MLRs between Tregs and allogeneic myeloid DCs, after which CD69+CD71+ Tregs were sorted by flow cytometry. As expected, only a small number of cells were selected using this method (mean, ~2.8 × 104 cells; range, 1 × 104 to 4.5 × 104, n = 5). In parallel, Tregs from the same donor were stimulated polyclonally and CD69+CD71+ activated Tregs were sorted. Cells were rested for 48 hours before their suppressive function and antigen specificity was tested in vitro by CFSE dilution assays (Fig. 3A and fig. S2). Autologous naïve CD4+CD25− Teffs were used as responders using allogeneic DCs from the primary Treg MLR, and an additional allogeneic DC HLA-disparate to both the responder and the primary allostimulator, as Donor and 3rdParty stimulators, respectively. CD69+CD71+ Tregs enriched after activation with Donor DCs were able to mediate potent and specific suppression of autologous Teff proliferation in response to Donor DC stimulation at Treg/Teff ratios as low as 1:100, which was not observed after stimulation with 3rdParty DCs (Fig. 3A). In contrast, polyclonally activated Tregs (Pc-Tregs) were able to suppress Teff proliferation in response to both Donor and 3rdParty stimulation to similar levels. The CD69+CD71+ Treg activation profile could be reproducibly used as a protocol to identify and enrich alloantigen-specific Tregs (AgS-Tregs) with potent antigen-specific suppressive properties (Fig. 3B).
Relevant for cell therapy, sorted alloantigen-activated Tregs could be readily expanded in vitro in the presence of high doses of exogenous interleukin-2 (IL-2) (250 U/ml), without further restimulation; these cells expanded several log fold (mean, ~1000-fold expansion; range, 340- to 5130-fold, n = 4) over 4 to 6 weeks to numbers sufficient for in vivo use in patients (mean numbers expanded, ~2.8 × 107; range, 1.5 × 107 to 5.1 × 107) (Fig. 3C). Both antigen-specific and polyclonal Treg lines maintained expression of Treg phenotypic markers after 4 to 6 weeks of in vitro culture and did not demonstrate a significant outgrowth of interferon-γ (IFN-γ)– or IL-17–producing populations, suggesting that the purity of Tregs was maintained throughout culture (Fig. 3D). In vitro expanded Tregs were rested in media supplemented with low IL-2 (10 U/ml) before being tested for maintenance of their antigen specificity (Fig. 3E) and subsequent in vivo use.
We developed our previously described model of human skin xenotransplantation (24) into a model of allogeneic CD4-mediated allograft injury to compare the relative protective function of allospecific and polyclonal Tregs (Fig. 4A). Skin donor alloantigen–specific Tregs were enriched using the activation profile protocol described above with dermal CD1c+ DCs extracted from donor skin as allostimulators (fig. S3) (25), which in all experiments were equally effective as blood-derived DCs.
Mice transplanted with human skin grafts received 5 × 106 human allogeneic Teffs (HLA-DRβ1 mismatched to skin), which resulted in engraftment of human CD3+CD4+ populations (table S1 and fig. S4). Skin allografts were harvested after 4 to 6 weeks and analyzed by histological immunostaining (Fig. 4B). Histopathological analysis of dermal tissue provided a more clinically relevant assessment of tissue damage, compared to topical assessment of allografts, which was not considered. Compared to phosphate-buffered saline (PBS) control animals (Nil), animals receiving allogeneic Teffs showed intense human CD45+ mononuclear cellular infiltrates within allografted tissue. The presence of band-like human CD45+ infiltrates localized to the dermo-epidermal junction was associated with allograft pathology, clinically resembling the immunohistological appearance of human cutaneous graft-versus-host disease (GVHD). Allogeneic Teffs also induced an increase in proliferating keratinocytes across epidermal stratum basale, loss of involucrin expression in upper stratum spinosum and granulosum, and loss of the stratum corneum (Fig. 4B). Combined, these parameters were indicative of active skin inflammation and loss of viable dermo-epidermal integrity and concur with previous studies examining T cell–mediated rejection of skin allografts in xenograft models (26, 27).
Using this model of alloimmune-mediated injury of human skin allografts, we compared the protective function of donor alloantigen-specific Tregs (AgS-Tregs) and polyclonal Tregs (Pc-Tregs) by transfer of Teffs alone or in combination with each Treg (ratio of 5:1) (Fig. 5A). This ratio was selected because in vitro both AgS-Tregs and Pc-Tregs mediated a comparable suppression of Teff proliferation in response to donor allostimulation at Treg/Teff ratios between 1:1 and 1:10. High numbers of human CD45+ cell infiltrates were detected in allografts, which was equivalent between all treatment groups and reflective of human CD45+ splenic engraftment (fig. S4). However, fewer CD45+ cells were localized to the dermo-epidermal junction in both Treg-treated groups compared to the group treated with Teffs alone. Donor skin from both Treg-treated groups also demonstrated reduced inflammatory-associated keratinocyte proliferation, which was largely undetectable in AgS-Treg–treated animals. In addition, although both Treg-treated groups displayed preservation of the stratum corneum and involucrin expression, skin histology of AgS-Treg–treated animals was more comparable to that of animals receiving no allogeneic Teffs (Fig. 5A).
More comprehensive quantitative analysis of clinically relevant parameters of dermal pathology revealed that AgS-Tregs provided significantly improved protection against allogeneic Teff-mediated damage compared to Pc-Tregs (Fig. 5, B to I). Keratinocyte proliferation in allografted tissue showed an epidermal hyperproliferative compartment in Teff-treated animals with strong suprabasal keratinocyte expression of Ki67, whereas Ki67 expression in grafts from untreated animals was more discrete and restricted to fewer basal epidermal keratinocytes (Teffs versus Nil, P < 0.0001) (Fig. 5B). Histological quantification showed that Teff-induced keratinocyte hyperproliferation was completely inhibited by cotransfer of AgS-Tregs, delivering a histology similar to that of untreated animals (Teffs versus AgS, P < 0.0001; Nil versus AgS, P = 1.0), whereas Pc-Treg–treated animals displayed moderately reduced levels of keratinocyte proliferation (Teffs versus Pc, P = 0.289; Nil versus Pc, P = 0.006; Pc versus AgS, P = 0.021) (Fig. 5F). Similarly, cotransfer of AgS-Tregs completely inhibited inflammatory-associated tissue damage (Fig. 5C), because fewer apoptotic cells were detected in allografted dermal tissue from this group, which was comparable to that of untreated animals (Nil versus AgS, P = 1.0; Teffs versus AgS, P < 0.0001) (Fig. 5G). In contrast, similar numbers of apoptotic cells were detected in Teffs alone and Pc-Treg–treated animals (Teffs versus Pc, P = 0.084), suggesting that Pc-Tregs were unable to regulate allogeneic inflammatory processes as efficiently as AgS-Tregs (Pc versus AgS, P = 0.012). Consistent with these observations, analysis of human CD31+ endothelial vessels in the superficial dermis of allografts (Fig. 5D) showed that microvessel injury induced by allogeneic Teffs (Nil versus Teffs, P < 0.0001) could be significantly reduced by cotransfer of AgS-Tregs compared to Pc-Tregs (Teffs versus AgS, P < 0.0001; Teffs versus Pc, P = 0.58) (Fig. 5H). Combined, these data show that allografted tissue is more fully protected from immune-mediated injury by donor-allospecific Tregs, preserving skin pathology similar to that of untreated animals, despite the presence of an allogeneic cellular infiltrate. These data suggest that active regulation of allogeneic effectors had occurred, rather than deletion or inhibition of homeostatic proliferation. These findings were also reproducible using the BALB/cRag2γc−/− immunodeficient mouse strain (fig. S5). A significantly higher proportion of CD3+FoxP3+ cells detectable within allografts of AgS-Treg–treated animals (Teffs versus AgS, P = 0.001; Pc versus AgS, P = 0.006) correlated well with reduced indicators of inflammation and allograft damage detected within this treatment group (Fig. 5, E and I). Finally, investigation of early trafficking of adoptively transferred Tregs showed that although similar numbers of AgS-Tregs and Pc-Tregs were detected in skin allografts, a higher proportion of AgS-Tregs interacting with skin-resident HLA-DR+ cells was observed compared to Pc-Tregs (P = 0.036) (fig. S6). These data support the speculation that AgS-Treg–mediated suppression occurs primarily at the site of alloantigen source, potentially through modulation of APCs, resulting in the active regulation or earlier, more rapid control of infiltrating donor-reactive T effector responses in vivo.
Here, we addressed several key questions relating to the use and translation of antigen-specific Tregs as cellular therapeutics in clinical transplantation. We have identified a clinically transferable approach to enrich human Tregs with direct pathway alloantigen specificity based on the detection of an activation profile of CD69+CD71+ coexpression. Alloreactive Tregs demonstrated potent alloantigen-specific suppression at Treg/Teff ratios as low as 1:100 and could be readily expanded ex vivo (up to ~5000-fold). Enriched CD69+CD71+ Tregs maintained their antigen-specific suppressive function after in vitro expansion and, when tested in an in vivo model of immune-mediated allograft injury, provided significantly improved protection of human skin allografts compared to polyclonal Tregs. Previous experimental studies using MHC-mismatched combinations in murine skin and heart allotransplantation (10, 28, 29) have shown that alloantigen-specific Tregs are more effective at inducing indefinite allograft survival than polyclonal Tregs. In the study reported here, we have translated these observations into a human in vivo model system.
Modulating a transplant recipient’s immune response toward allograft acceptance using polyclonal Treg preparations is now technically achievable using Good Manufacturing Practice (GMP) facilities. Compared to the relative ease of expanding polyclonal Tregs, allospecific Tregs are difficult to isolate and expand. Clinical-grade cell sorting and associated reagents have recently been established, and therefore, enrichment of CD69+CD71+ allospecific Tregs represents a feasible approach for clinical-grade cellular preparation. CD69 is one of the most widely used markers of cellular activation in T cell biology (30, 31), although few studies have examined its expression by Tregs. CD71, also a well-defined marker of Teff activation (32), has only recently been reported as an activation marker of human Tregs, where elevated numbers of CD71+ Tregs could be detected in peripheral blood of patients with juvenile idiopathic arthritis, compared to healthy individuals (33). In addition, a recent study has applied the use of CD71 to clinical hematopoietic stem cell transplantation (HSCT), using it to deplete alloreactive Teffs from donor lymphocyte infusions prepared for patients, as a strategy to prevent GVHD (34). Previous approaches to generate allospecific Tregs have used allogeneic PBMCs or B cells to expand Tregs, which in vitro were shown to mediate antigen-specific suppression (16, 22, 35). Here, we selected irradiated mature myeloid DCs as allostimulators, first, to provide a potent allostimulatory signal and, second, for its compatibility with clinical-grade Treg preparation, eliminating the potential persistence of allostimulators within cultures. Although these cells are effective, a parallel comparison of other cells of donor origin would be useful in determining which allostimulator delivers optimal Treg activation and expansion.
The main limitation of current protocols for allospecific Treg expansion is that they rely on repetitive allostimulation to maintain the efficient in vitro expansion and function of Treg lines. Selection of CD69+CD71+ alloactivated Tregs eliminates the need for restimulation, because selected Tregs readily expand in the presence of exogenous IL-2 to numbers permitting in vivo use. This proliferative capacity may be associated with their initial enrichment from nonactivated Tregs, which could otherwise compete for IL-2 consumption and capacity within cultures. Detailed phenotypic analysis also shows that CD69+CD71+ Tregs fall within subpopulations defined by markers associated with potent suppressive function and stability, namely, CD27hiCD62L+ (22, 36), HLA-DR+ (37), and CD45RA+FoxP3+ (38, 39).
On adoptive transfer, allospecific Tregs were found to persist in vivo and could be detected up to 4 weeks after transfer. Because higher numbers of allospecific Tregs were found to be localized to donor-allografted tissue after 4 weeks, compared to polyclonal Tregs, it could be postulated that either antigen-driven expansion or preferential survival of allospecific Tregs occurred in vivo, contributing toward the protective advantage conferred by their use. These findings also reflect previous studies of Treg-mediated transplantation tolerance induction in a mouse model (28). In addition, the accumulation of Tregs in graft draining lymph nodes (DLNs) or sites of alloantigen sequestration supports the notion that the superior protection mediated by allospecific Tregs observed in vivo may also be associated with their concomitant expression of CD62L+ (40).
Development of the humanized mouse has permitted preclinical experimental models to bridge a long-standing gap in the translation of basic research observations to clinical use (18), although much work remains to study the stability, kinetics, and protective mechanisms of human Treg use in vivo. One limitation of this study is that the in vivo model used has focused on direct pathway CD4-mediated damage of allografted skin, whereas other studies have demonstrated a clear effector role for both CD4+ and CD8+ T cell–mediated human allograft rejection (26, 27). Because a recent study has shown that protection of human skin allografts using polyclonal Tregs (19) is associated with a modest decrease in graft infiltrating CD8+ T cells, it is possible that the mechanistic basis and protective potential of Treg therapy may differ depending on the immune component targeted. In this respect, the challenge of generating more clinically comparable models of transplantation in the humanized mouse needs to be addressed, allowing us to assess the ability of Tregs to inhibit, for example, alloantibody-mediated immunity or regulation of the indirect pathway during graft rejection.
The data we present here have provided in vivo evidence of the relative functional efficacy of human alloantigen and polyclonal Tregs in ameliorating immune-mediated allograft pathology. The findings hold important implications for the methods of Treg generation and selection for in vivo clinical use and suggest that efforts should be refocused on the goal of generating allospecific Tregs for the induction of antigen-specific immune regulation.
All cell cultures and assays were performed in complete media [RPMI 1640 supplemented with 2 mM l-glutamine, penicillin (100 IU/ml), and streptomycin, all from Invitrogen], containing 10% human AB serum (BioSera), and performed in humidified incubators at 37°C and 5% CO2. Isolated cells were HLA-typed with standard techniques (41).
Human Tregs were enriched from peripheral blood TRIMA cones from healthy volunteers, obtained with full informed consent and ethical approval (National UK Blood Service). PBMCs were separated by Lymphoprep density gradient centrifugation (PAA), and erythrocytes were lysed with ACK buffer (Sigma). Total CD4+ T cells were isolated by staining PBMCs with a cocktail of mouse anti-human monoclonal antibodies against CD8 (clone FK-18), CD14 (B-A8), CD16 (B-E16), CD56 (B-A19), CD19 (B-C3), γδTCR (B1), and CD33 (4D3), followed by negative bead selection with pan-mouse immunoglobulin G Dynabeads (from Invitrogen and Tepnel). CD4+CD25+ Tregs were then selected by positive isolation with Dynal CD25-conjugated beads (Invitrogen) or cell-sorted based on expression of CD4+CD25hiCD127lo (FACSAriaII Cell Sorter, Becton Dickinson). Tregs were rested overnight in complete media supplemented with IL-2 (10 ng/ml) (Proleukin, Novartis) before further use. Autologous PBMCs depleted of CD25+ cells or CD4+CD25− Teffs were cryopreserved in human serum containing 10% dimethyl sulfoxide for later use.
Human myeloid CD1c+ (BDCA-1) DCs were enriched from PBMCs with a microbead positive selection kit according to the manufacturer’s instructions (Miltenyi Biotech). DCs were cultured overnight in complete media with lipopolysaccharide (LPS) (100 ng/ml) (Serotype 055.B5, Sigma-Aldrich) to induce maturation. Cells were then washed thoroughly and then cryopreserved for later use, or irradiated [3000 gray (Gy)] and used directly as allogeneic stimulators. CD1c+ dermal DCs were extracted from human skin explants (described below) with a previously described protocol (25). In brief, skin was cut into 1- to 2-mm slices and immersed in DispaseII solution (5 mg/ml) (Roche) for 5 hours at 37°C. After removing the epithelium, the remaining dermis was cultured in six-well plates in complete media for 5 days, allowing dermal DCs to migrate out into the well. Dermal CD1c+ and CD14− DCs were further selected by cell sorting. Because skin-derived DCs showed a mature phenotype, they were not further matured by LPS but were irradiated before use.
Tregs were screened for expression of markers of T cell activation after polyclonal or allogeneic stimulation for 1 to 5 days by flow cytometry. Activation markers analyzed included CD24 (clone ML5), CD27 (O323), CD38 (HIT2), CD39 (ebioA1), CD62L (DREG56), CD69 (FN50), CD71 (OKT-9), CD73 (AD2), HLA-DR (T36), and TGFβ1LAP (FAB2463A) (from eBioscience, BD Biosciences, and R&D Systems). For intra-cellular cytokine staining of IFN-γ (B.27) and IL-17 (ebio64), Tregs were stimulated with phorbol 12-myristate 13-acetate (100 ng/ml) and ionomycin (1 μmg/ml) (both Sigma) for 5 hours with 1× Brefeldin A (eBioscience). CTLA-4 (BN13) and FoxP3 (PCH101) staining was performed with eBioscience Fix/Perm kit under the manufacturer’s directions. Dead cells were excluded from analysis by Live/Dead counterstaining (Invitrogen). Flow cytometric data were acquired with an LSRII flow cytometer (Becton Dickinson), analyzed with FlowJo 7.5 software (TreeStar Inc.), and included doublet discrimination.
Tregs were polyclonally stimulated with CD3CD28-coated T cell expander beads (25 μml/106 cells) (Invitrogen). Allogeneic MLRs were set up with irradiated peripheral blood– or skin-derived myeloid DCs and Tregs at a 1:3 ratio in complete media containing IL-2 (10 ng/ml) for 1 to 5 days. Markers where cellular expression was found to be modulated compared to unstimulated cells were then further validated in MLR cultures in the presence of an anti–HLA-DR–blocking antibody (clone L243, 10 μmg/ml, eBioscience).
After polyclonal or allogeneic stimulation, Treg cultures were stained and gated on CD3 (OKT3, eBioscience) and subjected to doublet discrimination to exclude any residual DC stimulator. CD3CD28 beads were removed from polyclonally activated cells before cell sorting. Cells were then sorted into two populations, CD69+CD71+ and CD69−CD71−, and either expanded in vitro in the presence of IL-2 (250 U/ml) or rested (48 hours) in complete media with IL-2 (10 U/ml) before performing suppression assays. Antigen specificity of Tregs was tested by in vitro cell proliferation assays. Tregs were labeled with 2.5 μmM eFluor670 (eBioscience) and added at ratios of 1:1 to 1:500 relative to autologous Teffs, which were labeled with 5 μmM carboxyfluorescein succinimidyl ester (CFSE) (Invitrogen). Allogeneic and polyclonal stimulators were used as described above. Assays were performed over 5 days, after which T cell proliferation was analyzed for CFSE expression by flow cytometry. Percentage suppression was calculated using the percentage of divided cells on stimulation, with and without Tregs. Where sufficient allostimulators were not available to perform sequential MLRs to generate alloantigen-specific Tregs and test their suppressive function in vitro, as with some skin-derived DCs, known single HLA-DR–expressing Epstein-Barr virus (EBV)–transformed B lymphoblastoid cell lines were used as surrogate allostimulators (10th International Histocompatibility Workshop, WS3, WS45, WS29, WS36) selected to match the donor HLA type or specific mismatch, along with an HLA-disparate 3rdParty stimulator.
NOD/scid/IL-2Rγ−/− (NOD.cg-PrkdcscidIl2rgtm1Wjl/SzJ) (NSG; The JacksonLaboratory) miceandBALB/cRag2−/−γc−/−mice (from A.Hayday, King’s College London) were used between 4 and 14 weeks at the time of the first experimental procedure. Animals were bred and maintained in the Biological Services Unit of King’s College London. All mice were kept under pathogen-specific sterile conditions, and procedures were conducted in accordance with institutional guidelines and the Home Office Animals Scientific Procedures Act (1986).
Human skin was obtained from routine abdominoplasty and reduction mammaplasty surgery, with informed consent and ethical approval (Guy’s & St Thomas’ NHS Foundation Trust and King’s College London). In brief, split-thickness (500 to 700 μmm) skin explants were harvested with a dermatome (Zimmer) and 1-cm2 pieces were transplanted onto NSG and Rag2−/−γc−/− mice, as previously described (24). Human skin grafts were fixed epidermal side upward onto similarly sized defects on recipient backs with an absorbable tissue seal (Vet-Bond, Braun Medical). Grafts were covered with Fucidin microbicide (Leo Laboratories Ltd.) and secured with Tegaderm film dressing (3M) for 7 days. Skin was allowed to engraft for 4 to 6 weeks before each experiment was initiated. Graft integrity and healing was assessed by histological analysis 4 weeks after transplantation to ensure that epidermal thickness was comparable to the pre-graft specimen (preservation of rete peg pattern and minimal evidence of inflammation). Allogeneic (one to two HLA-DRβ1 mismatches to skin) PBMCs depleted of CD25+ cells or enriched CD4+CD25− cells (5 × 106) were then adoptively transferred by intravenous injection to induce allograft damage, with or without in vitro–expanded Tregs (1 × 106). Higher doses of PBMC inoculates (10 × 106 to 20 × 106 cells per mouse) routinely resulted in engraftment of both CD4 and CD8 T cells, whereas lower doses resulted in predominant engraftment of CD4 T cells, with no detectable APC compartments (table S1). No animals developed xenogeneicGVHD symptoms, confirmed by maintenance of stable body weight. Weekly peripheral tail bleeds were performed to monitor human CD45+ cell engraftment. Animals were considered as successfully engrafted when human CD45+ cells constituted >0.5% of total splenic lymphocytes. Only these animals were selected for further analysis of allografted skin. Skin grafts were subjected to visual and tactile inspection two times weekly for evidence of graft injury, and experiments ended 4 to 6 weeks after Teff transfer, depending on human CD45+ engraftment detected in peripheral blood. Splenocytes and DLNs were harvested and analyzed for cellular composition.
Frozen sections (6 to 8 μmm) of human skin allografts were prepared after snap freezing in optimum cutting temperature reagent over precooled isopentane. Tissue sections were air-dried and fixed in acetone (immunohistochemistry) or histological fixative saline (immunofluorescence). Nonspecific Fc receptor binding was measured with isotype nonbinding control antibodies. Immunohistochemical staining was performed for human CD45 (clone HI30; eBioscience), human involucrin (SY5; Sigma), and human Ki67 (MM1; Leica) with avidin-biotinylated antibody pairs and developed with the ABC Vectastain Elite kit (Vector Laboratories). Double immunofluorescence stains were performed in the following combinations: CD45, Ki67; FoxP3 (259D/C7), CD3; involucrin (SY5), CD31 (ab28364). Sections were incubated with fluorescence-conjugated secondary antibodies and mounted with Prolong Gold AntiFade Reagent with 4′,6-diamidino-2-phenylindole (DAPI) (all from Invitrogen). Samples were analyzed by fluorescence microscopy, and quantitative analysis was performed by counting a minimum of three nonoverlapping visual fields per animal and blinded to treatment protocols. Quantification of CD31+ vessels was determined by counting only perfused intact vessels and not single CD31+ cells. Apoptotic cells were detected by TUNEL (terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling) staining according to the manufacturer’s directions (Roche). Sections were counterstained with DAPI, and a minimum of six visual fields was quantified for double-positive nuclei.
Data shown are means ± SD where indicated. Statistical comparisons of Treg suppression assays were made with two-tailed paired Student’s t tests. Statistical comparisons of immunohistological parameters CD31, TUNEL, and Ki67 quantitative analysis between treatment groups were made with nonparametric Kruskal-Wallis tests (α significance level, 99.9%). Statistical comparisons of FoxP3/CD3 ratio between treatment groups were made with one-way analysis of variance (ANOVA) (α significance level, 99.9%), and all significance is denoted as follows: *P < 0.05, **P < 0.01, and ***P < 0.001.
We thank H. Fraser and C. Scotta for advice on Treg expansion; I. Rebollo-Mesa (Medical Research Council Centre for Transplantation, King’s College London, London, UK) for statistical advice; H. Sreeneebus, A. Clifford, and I. Tosi for access to skin; G. Perera and D. Kassen for developing human skin xenograft procedures; C. Chu (St John’s Institute of Dermatology, King’s College London, London, UK) for advice on dermal DC extraction; S. Heck and P. J. Chana (Guy’s & St Thomas’ NHS Foundation Trust and King’s College London, cBRC Flow Core Facility) for cell sorting; and R. Vaughan and E. Kondeatis (Guy’s & St Thomas’ NHS Foundation Trust and King’s College London, Clinical Transplantation Laboratory) for HLA typing.
Funding: P.S. is a Biomedical Research Centre Research Fellow and acknowledges financial support from the Department of Health via the National Institute for Health Research (NIHR) comprehensive Biomedical Research Centre award to Guy’s & St Thomas’ NHS Foundation Trust in partnership with King’s College London and King’s College Hospital NHS Foundation Trust. This work was also supported by MRC Centre for Transplantation, British Heart Foundation, NIH (RO1AR040065), Wellcome Trust Programme (GR078173MA), and Guy’s & St Thomas’s Charity.
Competing interests: The authors declare that they have no competing interests.
Fig. S1. Cell proliferation and activation marker expression analysis of alloactivated Tregs.
Fig. S2. In vitro assessment of Treg-mediated suppression of autologous Teff proliferation.
Fig. S3. Use of dermal DCs to generate donor skin alloantigen-specific Tregs.
Fig. S4. Immune reconstitution characteristics in the humanized mouse.
Fig. S5. Improved allograft protection by antigen-specific Tregs is also observed in a BALB/cRag2−/−γ−/− model of human skin alloimmune injury.
Fig. S6. Antigen-specific Tregs traffic to skin allografts in vivo and interact with HLA-DR+ cells.
Table S1. Human lymphocyte engraftment characteristics in NSG mice.