Regulatory T cells Exhibit Decreased Proliferation but Enhanced Suppression After Pulsing with Sirolimus

Abstract

Although regulatory T cells (Tregs) suppress allo-immunity, difficulties in their large-scale production and in maintaining their suppressive function after expansion have thus far limited their clinical applicability. Here we have used our non-human primate model to demonstrate that significant ex vivo Treg expansion with potent suppressive capacity can be achieved and that Treg suppressive capacity can be further enhanced by their exposure to a short pulse of sirolimus. Both unpulsed and Sirolimus-Pulsed Tregs (SPTs) are capable of inhibiting proliferation of multiple T cell subpopulations, including CD4+ and CD8+T cells, as well as antigen-experienced CD28⁺CD95⁺ memory and CD28⁻CD95⁺ effector subpopulations. We further show that Tregs can be combined in vitro with CTLA4-Ig (belatacept) to lead to enhanced inhibition of allo-proliferation. SPTs undergo less proliferation in a mixed lymphocyte reaction (MLR) when compared with unpulsed Tregs, suggesting that Treg-mediated suppression may be inversely related to their proliferative capacity. SPTs also display increased expression of CD25 and CTLA4, implicating signaling through these molecules in their enhanced function. Our results suggest that the creation of SPTs may provide a novel avenue to enhance Treg-based suppression of allo-immunity, in a manner amenable to large-scale ex vivo expansion and combinatorial therapy with novel, costimulation-blockade-based immunosuppression strategies.

Introduction

While short-term success is now common after solid organ transplantation, long-term results are still inadequate and include chronic rejection, as well as off-target toxicities of life-long immunosuppression. There is a growing realization that effective immunomodulation will likely require the induction of active immune regulation as well as immunosuppression, and that CD4⁺ CD25⁺FoxP3⁺ regulatory T cells (Tregs) may be key participants in this process. In murine models there is growing evidence to support the role of Tregs in both autoimmunity and transplantation (1–3). These data provide the rationale for the development of strategies whereby Tregs are used in conjunction with pharmacologic immunosuppression to downregulate alloreactivity.

Tregs comprise between 5–10% of the peripheral CD4+ T cell pool (4) and develop either in the thymus (natural Tregs, nTregs) (5, 6) or are induced in the periphery (induced or iTregs) (7–9). Although both nTregs and iTregs have been implicated in regulating immune reponses, (6, 10–17) there are concerns with the in vivo stability of iTregs (18–20) due to the potential risk of their reversion toward an activated Teffector phenotype. Adoptive transfer of in vitro expanded nTregs has therefore moved the farthest clinically, with the first phase I trials of this strategy for GvHD prevention recently completed (21, 22).

While these studies have documented the feasibility of transfer of relatively low numbers of Tregs (~3–4×10⁶/kg) for GvHD prevention after BMT, many mechanistic and practical questions about their production and delivery remain. These include a determination of the dose-dependence of Treg therapy in vivo, and a determination of their potency when combined with pharmacologic immunosuppression. Of the standard immunosuppressive agents that have been combined with Tregs, perhaps the most detailed work has been accomplished with sirolimus. Studies have shown that Tregs are preferentially able to retain their suppressive function in the presence of sirolimus (23, 24). However, although Treg function persists, several studies have demonstrated that prolonged sirolimus exposure can substantially inhibit Treg expansion (25, 26).

To rigorously study the questions of Treg specificity and potentcy during adoptive transfer, we have established a non-human primate (NHP) model of Treg purification, expansion and function (27). In this study, we use an in vitro model of alloreactivity to provide the first evidence that NHP Tregs can effectively inhibit both naïve and memory T cell allo-proliferation, and that Tregs can combine with belatacept to induce CD8-predominant suppression of allo-proliferation. In addition, we show that the potency of ex vivo expanded Tregs can be significantly increased through a short pulse of sirolimus without compromising the ability to highly expand these cells ex vivo. These observations are expected to impact the development and implementation of strategies for combinatorial Treg-based therapy during both HSCT and solid organ transplantation.

Materials and Methods

Animals

Rhesus macaques from the Yerkes National Primate Research Center or the NIAID-sponsored Rhesus macaque colony in Yemassee, SC were used in this study. All animals were treated in accordance with Emory University IACUC regulations.

Treg Isolation

Peripheral blood lymphocytes (PBL) were purified from CPT tubes (BD Biosciences). RBCs were lysed (High-Yield Lyse, Invitrogen), the PBL washed with PBS, and then re-suspended at 10⁷ cells/40 µl in MACS buffer (Miltenyi Biotec). CD4+ T cells were purified aseptically by negative selection using the LD column platform (Miltenyi Biotec). Cells were stained for CD4 (clone SK3, BD), CD25 (clone 4E3, Miltenyi Biotec) and CD127 (clone eBioRDR5, eBioscience) and re-suspended in FACS sorting buffer (PBS, 2% fetal bovine serum, 25 mM HEPES buffer). CD4⁺CD25⁺⁺CD127^−/low (Tregs) and CD4⁺CD25^+/−CD127^high (non-Tregs) were then purified flow cytometrically using aseptic technique. Treg purity was assessed by staining for CD4, CD25, CD127 and FoxP3 (clone PCH101, eBioscience).

Flow cytometry

On day 0 or day 21, 0.5–1×10⁶ cells were stained for CD3 (clone SP34-2, BD), CD4, CD25, CD127, FoxP3, and CTLA-4 (clone BNI3, BD). Data were acquired on an LSR II flow cytometer (BD Biosciences) and analyzed using FlowJo cytometry analysis software (Treestar, Ashland, OR). Thresholds for identifying positively-staining cells were set with relevant isotype control antibodies. These controls were critical, as small differences in the binding of the isotype controls were noted between Treg and non-Treg cultures (Figure 1E), potentially due to cell-specific differences in non-specific antibody binding after bead-based stimulation.

Figure 1. Isolation, ex vivo expansion and analysis of CD4⁺CD25⁺⁺CD127^−/low “Tregs” and CD4⁺CD25^+/−CD127^high “Non-Tregs”.

(A): Purification strategy for Rhesus macaque Tregs: CD4+ T cells were first enriched from PBLs by depletion of non-CD4+ cells using a NHP-specific CD4+ T cell isolation kit. These ‘untouched’ CD4+ T cells were stained for CD4, CD25 and CD127 and flow-sorted into CD4⁺CD25⁺⁺CD127^−/low “Tregs” and CD4⁺CD25^+/−CD127^high“Non-Tregs” on a BD FacsAria cell sorter.

(B): The phenotype of the flow-sorted Tregs and Non-Tregs was confirmed by staining for CD3, CD4, CD25, CD127 and FoxP3. The data shown is representative of five independent experiments. Cells were first gated to identify CD3+/CD4+ cells, which were then queried for the level of expression of CD25, CD127 and FoxP3.

(C) The percent of CD4+ T cells expressing CD25 and FoxP3 in putative Tregs (blue) and non-Tregs (red) after flow-based cell sorting (n= 5 separate Treg donors). Shown is the mean +/− SEM.

(D): Ex vivo Expansion of Tregs and Non-Tregs: Flow-sorted Tregs and Non-Tregs were stimulated with microbeads coated with anti-rhesus-CD3 and anti-human CD28 antibodies at a cell: bead ratio of 1:2 and cultured in X-Vivo-15 medium supplemented with 5% human serum, antibiotics and 2000 IU/ml and 200 IU/ml of rhIL-2 for Tregs and non-Tregs respectively as described in Methods. Cells were re-stimulated on day 7 and day 14 and were harvested on day 21. Cell numbers were counted on day 7, 14 and 21. Shown are the average fold-increases over baseline ± SEM, n=5 separate Treg donors.

(E): The phenotype of the expanded Tregs and Non-Tregs was confirmed by staining for CD3, CD4, CD25, CD127 and FoxP3 at day 21 of culture. The data shown is representative of five independent experiments. Cells were first gated to identify CD3+/CD4+ cells, which were then queried for the level of expression of CD25, CD127 and FoxP3.

(F): The percent of CD4+ T cells expressing CD25 and FoxP3 in expanded Tregs (blue) and non-Tregs (red) after 21 days of culture (n= 5 separate Treg donors). Shown is the mean +/− SEM.

Ex vivo expansion of CD4⁺CD25⁺⁺CD127^−/low Tregs

Flow-sorted Tregs and non-Tregs were expanded by stimulating with anti-rhesus-CD3 and anti-human CD28 coated microbeads (Miltenyi Biotec) at a cell: bead ratio of 1:2 and culturing in X-Vivo-15 media (Lonza) supplemented with 5% human serum, 0.2% N-acetyl cysteine, 5 mM Hepes buffer, penicillin (100 IU/ml), streptomycin (100 µg/ml), gentamicin (20 µg/ml), and either 2000 or 200 IU/ml of rhIL-2 (R&D Systems) for Tregs or non-Tregs, respectively. Cultures were split and replenished with fresh media and rhIL-2 when the media became acidic (at a density of ~2–3×10⁶ cells/ml). At days 7 and 14, cell numbers were counted and cultures re-stimulated as on day 0. Cells were harvested on day 21, magnetic beads removed with a magnetic column (Miltenyi Biotec) and their phenotypic integrity assessed by staining for CD3, CD4, CD25, CD127 and FoxP3. In some cultures, 1–1000 nM of sirolimus was added at the time of each stimulation. To create Sirolimus Pulsed Tregs (SPTs), Tregs were expanded in the absence of sirolimus until day 19, and then pulsed with 100 nM of sirolimus (the standard dose used in human Treg cultures (24, 25, 28, 29), for the next 48 hours. The cultures were then harvested, washed free of sirolimus and cryopreserved.

Suppression assay to measure the inhibitory activity of ex vivo expanded Tregs

Treg-mediated suppression of allo-proliferation was assessed in an in vitro CFSE-MLR assay. 2×10⁵ ‘responder’ PBLs were labeled with CFSE as previously described (27), and then either cultured without stimulation, or in the presence of 4×10⁵ irradiated allogeneic ‘stimulator’ PBLs in the absence or presence of Tregs or non-Tregs. Treg cultures that were derived from the same animal from which the responder PBLs were collected were referred to as “responder-specific” Tregs. MLRs were cultured for 5 days at 37°C in OpTmizer T cell expansion media (Invitrogen) supplemented with 5% human serum, 2 mM glutamine, penicillin-streptomycin and gentamycin. On day 5, cells were stained for CD2, CD3, CD4, CD8, CD28, CD95 and FoxP3 and the proliferation of the responder T cells was assessed flow cytometrically by CFSE dilution. In some experiments, 200 µg/ml of belatacept (Bristol-Myers Squibb) was also added. The gating strategy used to assess proliferation is shown in Supplemental Figure S1 and was as follows: (1) Lymphocytes were identified with a forward-scatter (FSC) versus side-scatter (SSC) gate. (2) T cells were identified using a CD3 versus FSC gate. (3) The CD3 gate was further refined by gating on CD3+/CD2+ cells, which includes both memory and naïve T cell populations (30, 31). (4) The CFSE-labeled responder T cells were identified by applying a CFSE vs CD2 gate, which facilitated the elimination of non-CFSE-labeled cell populations that could otherwise confound the interpretation of the data. This gate distinguishes the allo-proliferating CFSE-labeled cells from the non-CFSE labeled cells based on the fact that CD2 expression increases during proliferation. Thus, the threshold between the highly divided cells (with the lowest CFSE fluorescence) and the non-CFSE-labeled cells was set based on their relative degree of CD2 fluorescence. (5) A CD4 versus CD8 gate was then applied, and the CFSE fluorescence of the CD4+ and CD8+ subpopulations determined.

Allo-proliferation of Tregs

The allo-proliferation of Tregs was assessed with an MLR assay, using Tregs that were first stained with 5 µM CellTrace Violet (CTV, Invitrogen). The gating strategy was identical to that described above except that to establish the threshold between unlabeled cells and highly divided CFSE- or CTV-labled cells, control cultures were used that contained only CFSE-labeled responder cells (no CTV-labeled Tregs), or only CTV-labeled Tregs (no CFSE-labeled responders). In these cultures, the proliferating cells were identified by comparing either CFSE or CTV vs CD2 fluorescence, which facilitated the establishement of a gate that distinguished the non-labeled cell populations from the allo-proliferating cells.

Statistical analysis

Data were analyzed by both paired and unpaired Student’s t test. p values of ≤ 0.05 were considered statistically significant. * = p <0.05, ** = p <0.01, *** = p <0.001.

Results

Purification and Expansion of Rhesus macaque CD4⁺CD25⁺⁺CD127^−/low Tregs

Rhesus macaque Tregs were purified by flow cytometry-based sorting of CD4⁺CD25⁺⁺CD127^−/low cells after initial column-based CD4 enrichment, using a FACSAria flow-based cell sorter (Figure 1A). This strategy resulted in sufficient yield (≥1×10⁶ Tregs) and purity (> 80%) of Tregs to proceed with expansion and functional characterization (Figure 1). As shown in Figure 1B, the vast majority of sorted Tregs were CD25-positive, CD127-negative/low and exhibited significant FoxP3 expression, unlike the sorted non-Tregs, which displayed significantly less CD25 and FoxP3 expression and significantly more CD127 expression. Figure 1C shows the quantification of FoxP3 expression between sorted Tregs and non-Tregs compared to the isotype control, which documents that, similar to what has been observed during the flow-based purification of human Tregs (21, 26, 28) prior to their expansion, 64.3 ± 6.9% of the Treg isolates were FoxP3+ compared to 2.8 ± 0.8% of the non-Treg isolates (p<0.001). Figure 1D shows the results of anti-CD3/CD28 bead-based expansion of both Treg (210–760-fold) and non-Treg (1200–2000 fold) cultures, and Figures 1E,F document that expanded Treg cultures maintained the CD25^high/CD127^low/FoxP3⁺ phenotype, with 90.2 ± 4.7% expressing FoxP3 after 21 days in culture, compared to 18.3 ± 5.6% FoxP3 expression in the expanded non-Treg cultures (p<0.001).

Only ex vivo expanded Tregs potently suppress allo-proliferation of T effector cells

The suppressive capacity of both ex vivo expanded Tregs and non-Tregs was assessed using a CFSE-MLR proliferation assay. As shown in Figure 2A, despite the fact that a small amount of FoxP3 expression was induced in the non-Treg cultures (Figure 1F), these cells were not suppressive: the addition of non-Tregs to MLR cultures was noted to slightly enhance the allo-proliferation of both CD4+ and CD8+ responder T cells, increasing the average percent proliferation of CD4+ cells from 9.4± 5% → 22.1± 15% (n=18, p= 0.004) and of CD8+ cells from 40.6 ± 14% → to 50.5± 20% (n=18, p= 0.01). In contrast, the addition of Tregs to the MLR strongly suppressed the allo-proliferation of responder T cells, resulting in 3.5 ± 1.2-fold inhibition of CD4+ proliferation and 3.0 ±1.3-fold inhibition of CD8+ proliferation (p <0.001 and p= 0.002, respectively, Figure 2B). Treg-mediated inhibition was calculated by comparing the percent of responder cells in Treg-containing MLRs that divided at least once with the percent of responder cells that had divided when non-Tregs were added to the MLR. Treg-mediated suppression of allo-proliferation occurred predominantly on highly proliferating cells. Thus, as shown in Table S1 and Supplemental Figure S2, Tregs modestly inhibited the first division cycle of responding CD4+ and CD8+ T cells (by 1.5- and 1.47-fold respectively), but showed more substantial inhibition of later stages of proliferation (inhibiting the second-sixth division cycle of responding CD4+ and CD8+ T cells by an average of 3.3 ± 0.5-fold and 3.3 ± 1.1-fold, respectively).

Figure 2. Ex vivo expanded Tregs potently suppress allo-proliferation of T effector cells in an MLR assay.

(A): CFSE-labeled responder T cells were allo-stimulated for 5 days with unlabeled antigen presenting cells from MHC disparate donors in the absence or presence of either ex vivo expanded Tregs or Non-Tregs at a Treg/Non-Treg:Responder T cell ratio of 1:1. Allo-proliferation of CD4+ and CD8+ T cells over the five-day period and its suppression by added Tregs was followed by CFSE dilution. The data shown is representative of nine independent experiments.

(B): Summary data showing paired analysis of MLR-based allo-proliferation (top panels) and the average fold inhibition of allo-proliferation (bottom panels) of CD4+ (left) and CD8+ T cells (right), in the presence of expanded non-Tregs and Tregs (n= 9 independent MLRs. These MLRs were performed using Treg cultures from 5 independent Treg donors. Treg cultures from these donors were assayed between 1–3 times in the MLR). Shown is the mean +/− SEM.

(C): CD28⁺CD95⁺ and CD28⁻CD95⁺ CD4+ and CD8+ T cell subpopulations were identified flow cytometrically after the five-day MLR and the suppression of their allo-proliferation by added expanded Tregs was determined by CFSE dilution of the respective responder populations. The data shown is representative of nine independent experiments.

(D): Summary data showing the relative allo-proliferation of CD4⁺CD28⁺CD95⁺ (upper left), CD4⁺CD28⁻CD95⁺ (lower left), CD8⁺CD28⁺CD95⁺ (upper right) and CD8⁺CD28⁻CD95⁺ (lower right) T cells in an MLR assay in the presence of expanded non-Tregs and Tregs (n= 9 independent MLRs. These MLRs were performed using Treg cultures from 5 independent Treg donors. Treg cultures from these donors were assayed between 1–3 times in the MLR. Shown is the mean +/− SEM). Shown is the mean +/− SEM.

(E): Tregs suppress allo-proliferation of sorted naïve and memory T cell populations. CD3+CD28+CD95− naïve T cells, CD3+CD28+CD95+ central memory T cells and CD3+CD28−CD95+ effector/effector memory T cells were sorted flow cytometrically and then placed into a CFSE-MLR in the absence or presence of ex vivo expanded Tregs or non-Tregs. Top row: CD28 and CD95 fluorescence before sorting (far left panel) and after sorting of CD3+CD28+CD95−, CD3+CD28+CD95+ and CD3+CD28−CD95+ cells. Bottom three rows: CFSE fluorescence (shown as dot plots on the left and the same data as histograms on the right) in the presence of allo-stimulation plus non-Tregs or Tregs.

(F): Combining Tregs with belatacept lead to enhanced inhibition of allo-proliferation. CFSE-labeled responder T cells were allo-stimulated as in Figure 2A in the absence or presence of either belatacept (200 µg/ml), ex vivo expanded Tregs, or both. The allo-proliferation of CD4+ (top two rows, shown as both dot plots and as histograms) and CD8+ T cells (bottom two rows, shown both as dot plots and as histograms) and its suppression by added belatacept and/or Tregs was determined by CFSE dilution. The data shown is representative of four independent experiments.

(G): Summary data showing relative allo-proliferation of effector CD4+ T cells (left) and CD8+ T cells (right) in an MLR assay in the presence of belatacept, Tregs and both (n= 4 independent MLRs from two individual Treg donors). Shown is the mean +/− SEM.

Tregs potently inhibited the accumulation of CD95⁺ T cells

Figure 2C–D documents Treg-mediated inhibition of accumulation of CD28⁺ and CD28⁻ CD95+ cells in the MLR. Thus, in a standard MLR, significant accumulation of CD95+ memory phenotype cells (32) occurred after five days in culture (Figure 2C). Whether this accumulation was due to the selective expansion of memory cells that were present in the CFSE-labeled responder T cell pool, or due to the conversion of naïve T cells toward a memory phenotype, could not be distinguished from these experiments. Nevertheless, when antigen-experienced memory CD28⁺CD95⁺ or effector CD28⁻ CD95⁺ cells were analyzed after 5 days of culture in the presence of ex vivo expanded Tregs, (Figure 2C,D) significant inhibition of their accumulation was observed. Thus, when normalized against cultures to which non-Tregs were added, MLRs treated with expanded Tregs showed 3.0 ±1.5-fold (p= 0.01) and 2.4 ± 1.3-fold (p=0.029) inhibition of CD28⁺CD95⁺ proliferation for CD4+ and CD8+ cells, respectively (Figure 2D, top row), and 2.9±1.7 (p= 0.021) and 2.7± 0.8 fold (p=0.007) inhibition for CD28⁻CD95⁺ proliferation for CD4+ and CD8+ cells, respectively (Figure 2D, bottom row). Tregs were also able to inhibit the proliferation of naïve and memory T cells which were purified flow cytometrically prior to placement in the MLR (Figure 2E). These results may have significant clinical relevance, given recent observations of costimulation blockade-resistant rejection mediated by CD28-negative memory populations. (30, 31) They led us to investigate whether Tregs could effectively combine with belatacept to more completely suppress alloreactivity.

Tregs effectively combine with belatacept to inhibit allo-proliferation

Figures 2F,G document that expanded Tregs could indeed combine with CD28-directed costimulation blockade to inhibit allo-proliferation. Thus, as shown in Figure 2F,G, when Tregs alone were added to effector cells (at a 1:1 ratio) in the MLR, they resulted in 3.4± 0.2-fold inhibition of CD8+ proliferation and a 3.9 ± 0.9-fold inhibition of CD4+ proliferation (p <0.001 for CD8+ proliferation and p<0.05 for CD4+ proliferation). While the amount of autologous T cell proliferation (occurring in the absence of allogeneic APCs) was too low to be able to determine an impact of belatacept (not shown), belatacept did clearly inhibit allo-proliferation, with a greater salutary effect on CD4+ proliferation (7.1 ± 2.7 fold inhibition, p <0.05) than on CD8+ proliferation (2.0 ± 0.2-fold inhibition, p <0.01). Importantly, the addition of ex vivo expanded nTregs to belatacept had an additive suppressive effect in the MLR, significantly increasing the belatacept-mediated inhibition of CD8+ proliferation, resulting in 4.7 ± 0.5-fold inhibition of proliferation (p <0.01 when compared to either Tregs or belatacept alone). While Tregs had a less substantial impact on belatacept-mediated inhibition of CD4+ proliferation (8.7± 3.5-fold with belatacept + Tregs compared to 7.1 ± 2.7 fold inhibition of proliferation with belatacept alone), in four independent experiments, the addition of Tregs to belatacept-containing cultures always resulted in an incremental inhibition of CD4+ accumulation, which resulted in a statistically significant effect in paired analysis (p <0.05). The CD8-directed additive effect of Tregs + belatacept was also evident when lower Treg:Teffector cell ratios were used (Supplemental Data, Figure S3, S4), implying that costimulation blockade with belatacept may be a clinically important partner for Treg-mediated cellular therapies. The inhibitory effect of belatacept on allo-proliferation was likely not due to the induction of Tregs in the MLR, since no increase in FoxP3 expression on the responder T cells was observed (Supplemental Figure S5). This is similar to what has been observed in transplant and autoimmunity patients (33–35) where no induction of Tregs has occurred in patients treated with CTLA4Ig.

Tregs proliferate in response to allogeneic stimulation while simultaneously suppressing the allo-proliferation of effector T cells

To quantify the impact that allo-stimulation made on the Tregs themselves, we labeled the expanded Treg cultures with CTV before adding them into MLR cultures containing either allogeneic APCs alone or APCs plus CFSE-labeled responder T cells. As shown in figure 3A–G, expanded Tregs were also capable of allo-specific proliferation (14.9 ± 5.2% proliferation in an MLR containing allogeneic APCs versus 1.9 ± 0.4% proliferation without allogeneic APCs, p = 0.003). Tregs proliferated similarly in response to allogeneic APCs, whether in the presence or absence of responder T cells (Supplemental Figure S6).

Figure 3. Tregs proliferate in response to allogeneic stimulation while simultaneously suppressing the proliferation of effector T cells.

Tregs were stained with the proliferation marker CellTrace Violet (CTV) as described in Methods, and then added to T effector cells stained with CFSE. Cultures were then allo-stimulated with APCs for five days and the ability of the labeled Tregs to suppress allo-proliferation of effector T cells and to undergo proliferation was studied by simultaneously determining the dilution of CTV and CFSE dyes.

(A): CFSE fluorescence from a “T-responders Alone” control in which CFSE-labeled responder T cells do not proliferate in the absence of allogeneic APCs.

(B) CTV fluorescence from a “Tregs Alone” control in which labeled Tregs do not proliferate in the absence of allogeneic APCs.

(C): CFSE and CTV fluorescence demonstrate the lack of proliferation of either responder T cells or Tregs in the absence of allogeneic APCs.

(D): CFSE fluorescence from an allo-stimulated MLR showing significant proliferation of the CFSE-labeled responder T cells.

(E): MLR consisting of CFSE-labeled responder T cells, unlabeled stimulator APCs and CTV-labeled Tregs. The dilution of both CFSE (responder cells) and CTV (Tregs) fluorescence demonstrates proliferation of both responder T cells and Tregs in the presence of allogeneic APCs. Note that the proliferation of the CFSE-labeled responder T cells is reduced in the presence of Tregs (compare panel E to panels D and F).

(F) MLR consisting of CFSE-labeled responder T cells, unlabeled stimulator APCs, and CTV-labeled non-Tregs (no Tregs were added). The dilution of both CFSE (responder cells) and CTV (non-Tregs) fluorescence demonstrates proliferation of both responder T cells and non-Tregs, but no inhibition of responder cell proliferation without added Tregs (compare Panel F to Panel E).

(G): A summary of the proliferation of Tregs after auto- or allo-stimulation in MLRs in four independent experiments is shown. Shown is the mean +/− SEM.

Significant inhibition of ex vivo Treg expansion by the continuous presence of sirolimus

While several studies have suggested that sirolimus may enrich bulk lymphocyte cultures for Tregs (26, 36) this effect appears to be due to preferential survival of Tregs compared to conventional T cells, rather than to Treg-specific expansion in sirolimus. (26, 37–39). In this study, we confirmed that, as we have previously shown for human umbilical cord blood (40) and peripheral blood (41) nTregs, the inclusion of continuous sirolimus in expanding NHP Treg and non-Treg cultures strongly inhibited their anti-CD3/CD28 mAb-mediated expansion (Supplemental Figure S7). Thus, tonic exposure to sirolimus was counter-productive to our Treg expansion strategy.

Sirolimus-pulsed Tregs (SPTs) demonstrate increased suppressive capacity

Given that previous studies have suggested that sirolimus may enhance Treg function (37, 42, 43), we determined whether short-term exposure to sirolimus could improve Treg potency while maintaining Treg yield. Thus, we first expanded nTregs for 19 days without sirolimus, and then pulsed them with 100 nM of sirolimus (24, 25, 28, 29) for 48-hours before Treg harvest and functional analysis. As shown in Figures 4A–C, the sirolimus pulse more than doubled the inhibitory capacity of expanded Treg cultures. Thus, sirolimus-pulsed Tregs (SPTs) exhibited a significantly enhanced suppressive capacity, which was predominantly directed against CD28⁺ CD4+ and CD8+ T cells. As shown in Figure 4C, SPTs demonstrated 2.7 ± 0.9 and 4.4 ± 2.2 – fold inhibition of CD28⁺ CD4+ and CD8+ T cells, respectively (p < 0.01 compared to un-pulsed Tregs). This effect was dose-dependent, as shown in the representative example in Figure 4D.

Figure 4. Sirolimus-pulsed Tregs (SPTs) demonstrate increased suppressive capacity, predominantly against the proliferation of CD28⁺CD95⁺ T cells.

Flow-sorted Tregs were stimulated with anti-rhesus-CD3 and anti-human CD28 coated microbeads and IL-2 as described in Methods. On day 19, 100 nM of sirolimus was added to the cultures. 48 hours later sirolimus-pulsed Tregs (SPTs) were harvested, washed free of sirolimus, beads removed and tested for their ability to suppress allo-proliferation in an MLR.

(A): Representative data at a 1:1 Treg: responder T cell ratio demonstrating the enhanced suppressive capacity of SPTs against CD4+ T cell proliferation. Row (i): autologous responder T cells alone control (no allogeneic APCs added to the MLR). Row (ii): MLR with responder T cells + allogeneic APCs plus expanded Non-Treg CD4+ T cells. Row (iii): MLR with responder T cells, + allogeneic APCs plus expanded Tregs. Row (iv): MLR with responder T cells, + allogeneic APCs plus sirolimus-pulsed Tregs (SPTs). Columns show CFSE fluorescence after gating on the following T cell subpopulations after the 5-day MLR: Column 1: total CD4+ T cells; Column 2: CD28⁺CD95⁺ CD4+ T cells; Column 3: CD28⁻CD95⁺ CD4+ T cells. The data shown is representative of eleven individual MLRs.

(B): Representative data at a 1:1 Treg:responder T cell ratio demonstrating the enhanced suppressive capacity of SPTs against CD8+ T cell proliferation. Row (i): autologous responder T cells alone control (no allogeneic APCs added to the MLR). Row (ii): MLR with responder T cells + allogeneic APCs plus expanded Non-Treg CD4+ T cells. Row (iii): MLR with responder T cells, + allogeneic APCs plus expanded Tregs. Row (iv): MLR with responder T cells, + allogeneic APCs plus sirolimus-pulsed Tregs (SPTs). Columns show CFSE fluorescence after gating on the following T cell subpopulations after the 5-day MLR: Column 1: total CD8+ T cells; Column 2: CD28⁺CD95⁺ CD8+ T cells; Column 3: CD28⁻CD95⁺ CD8+ T cells. The data shown is representative of eleven individual MLRs.

(C): Summary data showing the relative allo-proliferation of total CD4+ T cells (upper left), CD4⁺CD28⁺CD95⁺ cells (upper middle), CD4⁺CD28⁻CD95⁺ cells (upper right), total CD8+ cells (lower left) CD8⁺CD28⁺CD95⁺ cells (lower middle) and CD8⁺CD28⁻CD95⁺cells (lower right), in an MLR assay in the presence of Non-Tregs, unpulsed Tregs and SPTs. (n= 11 independent MLRs. These MLRs were performed using SPT cultures from 5 independent Treg donors. SPT cultures from these donors were assayed between 1–3 times in the MLR.) The mean fold-inhibition +/− SEM and p values for the comparison of Tregs vs SPTs were as follows. For total CD4+ T cells: Tregs: 3.1+/−0.28, SPTs: 4.5+/− 0.57, p= 0.006. For CD4+CD28+CD95+: Tregs: 1.9+/−0.16, SPTs: 2.7+/− 0.3, p= 0.008. For CD4+CD28−CD95+: Tregs: 2.2+/−0.21, SPTs: 2.9+/−0.35, p= 0.02. For Total CD8+ T cells: Tregs: 2.9+/−0.34, SPTs: 6.3+/− 0.79, p= 0.0001. For CD8+CD28+CD95+: Tregs: 1.6+/−0.12, SPTs: 4.4+/− 0.73, p= 0.003. For CD8+CD28−CD95+: Tregs: 9.9+/−2.4, SPTs: 10.5+/− 2.4, p= 0.673.

(D): Dose-response curve showing the degree of suppression when Tregs and SPTs were included in an MLR at the following ratio compared to the responder T cells: 1:1; 1:2; 1:4; 1:8; 1:16; 1:32. Data is representative of two individual experiments.

The enhanced function of SPTs was accompanied by alterations in their proliferative capacity and phenotype

To determine the mechanisms by which SPTs exhibited their increased inhibitory capacity, we determined their phenotype and proliferative capacity using flow cytometric analysis. As shown in Figure 5A, B, SPTs exhibited significantly less allo-proliferation than control Tregs, (p = 0.015), consistent with previous studies, which have documented an inverse relationship between Treg proliferation and function (44, 45). In addition, as shown in Figure 5C, all of the proliferating cells in the SPT cultures were FoxP3+, reinforcing the fact that the allo-proliferation that we measured likely emanated from Tregs and not from the small number of FoxP3-negative cells present in the ex vivo expanded cultures. In addition, flow cytometric analysis demonstrated increased expression of both CD25 (Figure 5D, p = 0.04) and CTLA4 (Figure 5E, p = 0.009) on SPTs compared to Tregs, consistent with the previously documented role of CD25 and CTLA4 expression in Treg function (46, 47). In contrast, expression levels of CD3, CD4, CD27, CD45RA, CD62L, CD127, CD179b, CD223, CD279, GITR, MHC Class II and phospho-Stat-5 did not show any significant change in SPTs (data not shown), although FoxP3 expression was slightly decreased on SPTs (MFI = 5308 ± 1243 for SPTs, 6464 ± 1270 for Tregs, p = 0.03, data not shown).

Figure 5. The enhanced function of SPTs is accompanied by alterations in their proliferative capacity and phenotype.

(A): SPTs are less proliferative than unpulsed Tregs in an allo-MLR. SPTs and control Tregs were stained with CTV dye and were added to an MLR in which effector T cells were stained with CFSE. Suppression of effector T cell allo-proliferation (not shown) and proliferation of SPTs and control Tregs were determined. Shown is a representative tracing of proliferation as measured by the dilution of CTV fluorescence, comparing unpulsed Tregs to SPTs.

(B): Summary of the relative proliferation of SPTs compared to control Tregs from three independent experiments. Shown is the mean +/− SEM.

(C): Both dividing and non-dividing cells from the ex vivo expanded Treg cultures express FoxP3: SPTs were labeled with CTV and placed into the MLR. FoxP3 fluorescence was measured on the proliferating SPTs (CTV^low, green traces) and non-proliferating SPTs (CTV^high, purple traces) as well as on the non-Tregs (black traces).

(D). SPTs exhibit higher cell-surface expression of CD25 compared to unpulsed Tregs. SPTs and control, unpulsed Tregs were stained for CD3, CD4 CD25 and intracellular FoxP3 and flow cytometry data was acquired on BD LSR II and analyzed by Flowjo. A representative histogram (left) and summary expression data (displayed as the mean fluorescence intensity (MFI) from five independent experiments (right) for CD25 fluorescence intensity is shown. Shown is the mean +/− SEM.

(E) SPT cultures have more CTLA-4^High Tregs compared to control, unpulsed Treg cultures. SPTs and control unpulsed Tregs were stained for CD3, CD4, total CTLA-4 and intracellular FoxP3, and flow cytometry data was acquired on BD LSR II and analyzed by Flowjo. Left: Representative density plots showing the number of Tregs in each culture that was CTLA-4^High Right: Summary data of CTLA-4^High expression from four independent experiments is shown. Shown is the mean +/− SEM.

Discussion

While murine studies have documented the ability of both nTregs and iTregs to downregulate allo-immunity (1–3, 10, 43) the broad translation of these observations to large animal models and to the clinic, especially for solid organ transplantation, has yet to occur. However, the recent publication of the first phase I clinical trials using Tregs (which both employed Tregs during BMT as part of post-transplant GvHD immunoprophylaxis) (21, 22) suggest the feasibility of Treg-based approaches.

There have, historically, been significant barriers to the wide spread use of adoptive Treg immunotherapy. These have included both the difficulty in producing these cells in sufficient quantities for in vivo use and with the maintenance of adequate suppressive function in massively expanded Treg cultures. While Hippen et al (12) have recently reported a significant breakthrough in Treg expansion, their production remains technically challenging and extremely costly. Thus, designing strategies to increase the potency of Tregs and identifying immunosuppressive agents with which they can effectively combine (especially at lower nTreg doses) remains a key challenge for the field.

In this manuscript, we use an in vitro model of primate allo-stimulation to document several mechanistic observations that may be critical for the translation of nTreg strategies to the clinic. This model is strengthened by its ability to measure the impact of Tregs on responder T cell proliferation after allogeneic stimulation, rather than after mitogenic (anti-CD3) antibody stimulation. Given that the mechanisms governing CD3-stimulation and allo-stimulation are likely distinct, and that allo-stimulation is expected to be the main driver of the immune response after transplant, the ability to interrogate mechanisms controlling allo-stimulation is expected to add to the predictive power of this model with respect to in vivo Treg function and potency.

Using this model we show that NHP Tregs are potent inhibitors of allo-proliferation of both CD28⁺ and CD28⁻ memory T cells. This observation suggests that nTregs may be good candidates for combination with costimulation blockade-based immunosuppression, given recent data suggesting that costimulation blockade-resistant rejection is often mediated by CD28-negative effector/memory T cells (30, 31, 48, 49). The observations made in this study with respect to Treg-mediated inhibition of memory T cell allo-proliferation are quantitatively different than those documented by Yang et al, (50) which showed that in mice, while Tregs could inhibit in vitro allo-proliferation of both naive and memory T cells at high Treg:Teff ratios, Tregs were less potent at inhibiting memory T cell-driven allo-proliferation when present at lower ratios, and were unable to inhibit memory-driven allograft rejection in vivo. While not yet proven experimentally, these differences may reflect species-specific mechanistic differences in Treg function. Indeed, in humans, Tregs have been demonstrated to inhibit both αβ and γδ memory T cell function, (51, 52) suggesting that the breadth and strength of the Treg response may differ between mice and primates.

Our results also identify a one-step strategy that significantly increased the in vitro potency of Tregs. The production of SPTs succeeded in improving the suppressive capacity of these cells while maintaining their ability to be highly expanded in vitro prior to sirolimus exposure. Thus, while there is now substantial data from this study and others (12, 26, 53) demonstrating that continuous exposure to sirolimus significantly impairs Treg expansion, here we demonstrated that a short pulse of sirolimus, which can be delivered after expansion, is sufficient to alter nTreg phenotype and function, resulting in improved inhibitory capacity. Our results with SPTs do not yet distinguish whether exposure to sirolimus potentiates the functional competence of all of the Tregs in culture, or whether it selects for a highly-active Treg subpopulation. This question remains an important area for future investigation.

Our observation that nTreg-mediated suppression may be inversely related to their allo-proliferative capacity suggests that on a per-cell level, anergy may be mechanistically tied to suppression, and mTOR inhibition, in its ability to potentiate anergy, may therefore potentiate the nTreg suppressive phenotype. These data are distinct from those recently described by Procaccini et al (54), in which transient exposure of mouse and human Tregs to sirolimus enhanced their subsequent proliferation in response to CD3/CD28-mediated polyclonal stimulation, as part of an oscillatory response to mTOR signaling. However, our observations using allo-stimulation rather than anti-CD3/CD28 mAb-mediated stimulation are similar to those recently described by Wang et al (55) who showed that in the setting of non-lymphopenic homeostatic and allo-stimulated proliferation, sirolimus significantly inhibited Treg replication. These results suggest that a delicate balance may be required for the design of transplantation strategies based on chronic sirolimus exposure, given that continuous exposure to sirolimus may actually result in the inhibition of Treg expansion, despite potentially increasing their suppressive potency.

These results also underscore the complexities associated with Treg function, expansion, and in vivo use, and support testing in large animal models such as the rhesus macaque in order to determine the optimal strategies to move forward to the clinic. These studies will allow us to rigorously determine the ability of Tregs to proliferate in vivo, and the impact that exposure of these cells to sirolimus (prior to transfer or during ongoing immunosuppression), and to other conventional and novel immunosuppressive agents (including belatacept) may make on their expansion, survival and function.

Non-standard Abbreviations

Tregs

regulatory T cells

SPTs

sirolimus pulsed Tregs

BMT

bone marrow transplant

GvHD

graft-versus-host disease

CTV

CellTrace Violet

MLR

mixed lymphocyte reaction