Low-Dose Interleukin-2 Combined With Rapamycin Led to an Expansion of CD4⁺CD25⁺FOXP3⁺ Regulatory T Cells and Prolonged Human Islet Allograft Survival in Humanized Mice

Clinical islet transplantation is a cutting-edge therapy for the treatment of type 1 diabetes (T1D) and hypoglycemic unawareness (1–3). However, its success is limited by the need for long-term multiagent immunosuppression and the ensuing side effects, coupled with the progressive loss of graft function over time, most likely from chronic rejection (1,2,4). Clinically, there are competing objectives of reducing immunosuppressive burden while maintaining long-term graft function.

Altering the immunological microenvironment can potentially prolong islet allograft survival while simultaneously reducing the requirement for long-term immunosuppression. This requires the removal of potential and existing alloreactive T cells and, at the same time, expanding regulatory T cells (Tregs). Two such agents with the capacity to achieve these dual objectives are recombinant human interleukin-2 (IL-2) and rapamycin, both of which are used clinically. In the clinic, low-dose IL-2 therapy has been used successfully to treat chronic graft-versus-host disease (GVHD) (5), hepatitis C virus–induced vasculitis (6), systemic lupus erythematosus (SLE) (7), and T1D (8). Rapamycin (sirolimus) is approved for the prevention of rejection in kidney transplantation (9), and animal studies have shown that low-dose rapamycin allows selective expansion of Tregs in vivo (10,11). Furthermore, rapamycin has been used successfully in clinical islet transplantation as part of the Edmonton protocol (12). In animal models, IL-2 treatment prolonged corneal allograft survival (13), and the combination of IL-2 and rapamycin treatment prolonged skin graft survival in minor antigen mismatch and semiallogeneic mouse recipients (14); in many of these studies, selective expansion of FOXP3⁺ Tregs was found (5–7,13,14).

Rapamycin through its inhibition of the mammalian target of rapamycin pathway affects Treg homeostasis (9,12) and suppresses the activation and proliferation of immune cells, including T cells (15). It inhibits IL-2–induced phosphorylation and subsequent T-cell proliferation (16) and impairs CD8⁺ T-cell antitumor immune responses (17). Rapamycin also causes expansion of functional CD4⁺CD25⁺FOXP3⁺ Tregs in both healthy individuals and patients with T1D in vitro (18) and increases Treg-specific suppressive ability in patients with T1D (19). In experimental studies, rapamycin has been shown to be protolerogenic by expanding naturally occurring Tregs while also depleting CD4⁺ T effector cells (18) by promoting their apoptosis (20). The combination of IL-2/rapamycin can also induce FOXP3⁺ Tregs from CD4⁺CD25⁻ conventional T cells (21).

Because of the limited success of translating the findings obtained in naïve mouse models to the clinic, we established a humanized mouse model that would allow us to study the human T-cell response to islet allografts in vivo. Previously, in this model, we showed that infusion of in vitro–expanded human FOXP3⁺ Tregs prevented porcine neonatal islet cell cluster xenograft rejection by suppressing the T-cell–mediated antigraft response (22). In this study, we aimed to determine whether low-dose IL-2 and/or rapamycin could promote the expansion of CD4⁺FOXP3⁺ Tregs and limit effector T cells in vivo to promote islet allograft survival.

Breeding pairs of NOD-scidIL2Rγ^null (NSG) mice were obtained from The Jackson Laboratory and housed under specific pathogen–free conditions. NSG mice also were obtained from the Animal Resource Centre (Murdoch, Western Australia, Australia). Experiments were conducted using established guidelines for animal care and approved by the animal ethics committee of Western Sydney Local Health District.

Human islets were isolated from donor pancreata and cultured for 24 h (2). Human islets that did not reach release criteria for clinical transplantation were used in this study. Human spleen mononuclear cells (hSPMCs) were isolated by Ficoll density gradient centrifugation (Sigma-Aldrich, Castle Hill, New South Wales, Australia). The use of human tissue for research was approved by the human research ethics committee of the Western Sydney Local Health District.

NSG mice were administered hSPMCs (fresh and frozen) (hNSG mice) intraperitoneally. Recombinant human IL-2 (Aldesleuki; Chiron and Novartis) and/or rapamycin (sirolimus, oral solution; Wyeth) treatments were administered intraperitoneally into transplanted NSG mice with hSPMCs (hNSG recipient mice). Treatment groups are outlined in Supplementary Table 1 and included 1) IL-2 at 0.3 × 10⁶ IU/m² of body surface area/day (IL-2^0.3) (equivalent to 7.5 IU/g) and 2) 1 × 10⁶ IU/m² of body surface area/day (IL-2¹) (equivalent to 24.9 IU/g) for 6 days/week for 3 weeks, 3) IL-2^0.3 or 4) IL-2¹ plus rapamycin 1 mg/kg for week 1 and 0.5 mg/kg for weeks 2 and 3 (rapamycin 0.5/1 mg/kg), and 5) rapamycin (0.5/1 mg/kg) for 3 weeks. The doses of rapamycin were chosen on the basis of studies that showed that a dose of 1 mg/kg/day corresponded to therapeutic doses for solid organ transplantation (23).

NSG mice were injected with streptozotocin at doses of 170–180 mg/kg to induce diabetes, which was defined as blood glucose level (BSL) ≥20 mmol/L (24). Diabetic mice were transplanted with 2,000 human islets/mouse under their left renal capsule. After successful transplantation, mice received one of two doses of IL-2 as above for 1 day (day −1) intraperitoneally. The following day, mice were transfused with allogenic 1 × 10⁷ hSPMCs (day 0). At the same time, untreated mice were divided into rapamycin treatment or no treatment as above, and IL-2–treated mice were continually treated with IL-2 or IL-2/rapamycin as described above. Successful islet function was defined as random BSL of ≤12 mmol/L after islet transplantation before transfusion and ≤10 mmol/L after IL-2 and/or rapamycin treatment. Nephrectomy was performed on control NSG mice after transplantation and before reconstitution.

Flow cytometry was performed on an LSR II cytometer, and data were analyzed using FACSDiva (BD Biosciences) and FlowJo version 10 (25) software. Peripheral blood (PB) and/or spleen samples of the hNSG mice were collected and stained at weeks 3, 4, 5, 7, and/or 9 after transfusion with hSPMCs. Antibodies comprised allophycocyanin-conjugated anti-human CD3, CD4, CD25; fluorescein isothiocyanate–conjugated anti-human CD4, CD19; Pacific blue–conjugated anti-human CD3, CD4; V450-conjugated anti-mouse CD45; phycoerythrin-conjugated anti-human CD3, CD8, CD56; peridinin chlorophyll–conjugated and BUV395-conjugated anti-human CD45; BV421-conjugated anti-human perforin; and PE-CF594–conjugated anti-T-bet (BD Biosciences) and phycoerythrin-conjugated anti-human FOXP3 (eBioscience). LIVE/DEAD Fixable Near-IR stain (Thermo Fisher Scientific) was used before surface and intracellular staining (FOXP3, perforin, and T-bet).

The left kidney containing the human islet grafts, right kidney, skin, intestine, liver, and spleen of the NSG mice, hNSG mice, and hNSG recipient mice were fixed in 10% formalin or frozen in optimal cutting temperature compound (Miles Scientific) for hematoxylin-eosin (H-E) staining and immunohistochemistry (IHC). Polyclonal guinea pig anti-insulin (Dako) with the R.T.U. VECTASTAIN Universal ABC Kit (Vector Laboratories) were used for detection of insulin secretion by human islet allografts as described previously (26). Primary anti-human CD4 (monoclonal, RPA-T4) and anti-human CD8 (monoclonal, RPA-T8) (eBioscience) antibodies were used for detection of infiltrating T cells within islet allograft (IHC). Immunofluorescence staining of human CD4⁺FOXP3⁺ (hCD4⁺FOXP3⁺) Tregs by primary rabbit anti-human CD4 (clone: EPR6855) and mouse anti-human FOXP3 (clone: 236A/E7) (Abcam) with secondary goat anti-mouse IgG (Alexa Fluor 488) and donkey anti-rabbit IgG (TRITC; Jackson ImmunoResearch) antibodies was performed on kidney human islet allograft paraffin sections that were counterstained with VECTASHIELD Antifade Mounting Medium with DAPI (Jackson ImmunoResearch) (26).

IHC images stained for human CD4⁺ (hCD4⁺) and CD8⁺ (hCD8⁺) T cells and insulin were blindly analyzed using Aperio ImageScope version 12.4.0.7018 software (Leica Biosystems). Fields used for statistical analysis were restricted to islet allograft sites, and staining was quantified using Aperio Positive Pixel Count Algorithm version 9 software (Leica Biosystems) according to the manufacturer’s instructions and as reported by previous studies (27,28). Quantitative analysis of immunofluorescence-positive cells was reported in a similar way to previous studies (29).

The histology of skin, intestine, liver, and right kidney from experimental and control NSG mice was used for xeno-GVHD evaluation and for evaluating the alloimmune response in the islet allografts of left kidney. H-E images of skin, intestine, liver, and right kidney (Aperio ImageScope version 12.4.0.7018 software) were blindly scored as grade 0–4 to assess degree of organ and tissue damage. The scoring criteria are listed in Supplementary Table 2.

The hCD4⁺ and hCD8⁺ T cells were sorted from spleens of hNSG recipient mice and donor hSPMCs on a FACSAria III Cell Sorter (BD Biosciences). Allophycocyanin-conjugated anti-human CD3, phycoerythrin-conjugated anti-human CD4, and fluorescein isothiocyanate–conjugated anti-human CD8 (BD Biosciences) and DAPI (Invitrogen) were used for staining.

Real-time RT-PCR was used for analysis of human IL-2, IL-4, IL-6, IL-10, interferon-γ (IFN-γ), and transforming growth factor-β (TGF-β) in sorted hCD4⁺ and hCD8⁺ T cells; perforin-1 on sorted hCD8⁺ T cells; and IL-17A and B-lymphocyte–induced maturation protein-1 (Blimp-1), T-bet, GATA3, and RORγt in sorted hCD4⁺ T cells from the spleens of hNSG recipient mice and donor hSPMCs. TaqMan gene expression assays were used with GAPDH (Thermo Fisher Scientific) for normalized gene expression.

Human IFN-γ, tumor necrosis factor (TNF), IL-2, IL-4, IL-6, IL-10, and IL-17A in mouse serum were detected on BD FACSCanto II (BD Biosciences) by BD cytometric bead array human T helper 1 (Th1)/Th2/Th17 cytokine kit. Data were analyzed by FCAP Array software (BD Biosciences) (26).

GraphPad Prism 7.0 (GraphPad Software, San Diego, CA) was used for statistical analysis. The log-rank (Mantel-Cox) test was used for comparison of survival data between groups. Differences between two groups were evaluated using unpaired t test or Mann-Whitney test (nonparametric test), while three or more groups were compared using a one-way ANOVA followed by Tukey multiple comparison test and Kruskal-Wallis test (nonparametric) followed by Dunn multiple comparison test. P < 0.05 was considered statistically significant. Data are expressed as mean ± SEM.

The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To establish a humanized mouse model for studies of human islet transplantation, the relative importance of cell number and time from transfusion in NSG mice was evaluated. Initially, 5 × 10⁶ and 1 × 10⁷ of hSPMCs were both assessed. The proportion of human CD45⁺ (hCD45⁺) cells after transfusion in NSG mice was found to be time dependent (Supplementary Fig. 1A and B). A greater proportion of hCD45⁺ cells was found at week 9 compared with weeks 4 and 5, regardless of whether mice were transfused with 5 × 10⁶ fresh hSPMCs or 1 × 10⁷ fresh or frozen hSPMCs (Supplementary Fig. 1B). In the doses used, the number of cells used (5 × 10⁶ and 1 × 10⁷) had no significant impact on the degree of hCD45⁺ cell reconstitution in NSG mice (Supplementary Fig. 1B). hCD45⁺ cells in PB and spleen of hNSG mice were predominantly CD3⁺ T cells (Fig. 1A and B), comprising hCD4⁺ and hCD8⁺ subsets (CD4⁺ T cells > CD8⁺ T cells) (Fig. 1C and D) with low proportions of B cells or natural killer cells (Supplementary Fig. 1C–E).

To determine whether hNSG mice were capable of rejecting human islet allografts, human islets were transplanted under the kidney capsule of diabetic NSG mice, and graft function was confirmed by BSL monitoring, histology, and/or nephrectomy (Fig. 2, Supplementary Fig. 2, and data not shown). NSG mice with functioning islet grafts were transfused with 1 × 10⁷ hSPMCs. Functioning islet allografts were rejected with a mean survival time (MST) of 27 days (Fig. 2A and B). Histological evaluation of rejecting human islet allografts showed marked cellular infiltration, predominantly composed of hCD4⁺ and hCD8⁺ T cells with no significant difference in the numbers of each subset and no intact islets (detected by insulin stain), while control islet grafts from NSG mice without transfusion had intact islets with significantly stronger insulin staining by IHC (Fig. 2C–F).

Graft loss was considered to be due to the islet alloimmune T-cell response and not as a result of xeno-GVHD because there was no clinical evidence of xeno-GVHD at the time of graft failure. Only one mouse had weight loss of >10% (Fig. 2A). The skin, intestine, liver, and contralateral kidney without the islet graft were examined histologically, and no xeno-GVHD was seen at the time of rejection (Fig. 3). The only abnormality in the nongrafted tissues was the presence of a scattered inflammatory infiltrate surrounding the branches of the portal vein of the liver at day 35 after transfusion (Fig. 3). In untransplanted NSG mice transfused with hSPMCs, evidence of xeno-GVHD was seen at week 9 after transfusion, with histology demonstrating loss of hair follicles in the skin and massive cellular infiltration in the liver with significant increases in histopathological GVHD grades compared with normal controls. Structural changes were also present in the spleen and minimal cellular infiltrate within the kidneys (Fig. 3 and Supplementary Fig. 3).

hCD4⁺FOXP3⁺ Tregs, including CD25⁺Foxp3⁺ and CD25⁻Foxp3⁺ Tregs, were assessed in PB and spleen of hNSG mice. The proportion of CD4⁺FOXP3⁺ Tregs was reduced in the PB and spleens of hNSG mice following transfusion alone (Fig. 4A and B) and subsequent to transplantation (Fig. 4C). At week 5, there was no significant difference in the proportion of CD4⁺FOXP3⁺ Tregs in the PB of hNSG mice with and without transplantation (Fig. 4C). These CD4⁺FOXP3⁺ Tregs were predominantly CD25⁺FOXP3⁺ (Fig. 4A). However, by week 9 after transfusion, the proportion of CD4⁺FOXP3⁺ Tregs had fallen in both PB and spleen in hNSG mice compared with the proportion found at week 5 (Fig. 4B).

To determine the importance of IL-2 dose for expansion of hCD4⁺FOXP3⁺ Tregs and whether expansion of hCD4⁺FOXP3⁺ Tregs could prolong human islet allograft survival, recipient mice of human islet allografts were treated with one of two doses of IL-2 alone (IL-2^0.3 or IL-2¹) or combination of IL-2 with rapamycin for 3 weeks. Treatment with either dose of IL-2 alone and rapamycin alone did not prolong islet allograft survival (MST 35, 26, and 32 days, respectively) (Fig. 5A). However, the combination of either dose of IL-2 with rapamycin significantly prolonged human islet allograft survival (MST 60 and 42 days, respectively) (Fig. 5A). hNSG recipient mice that did not receive treatment had an MST of 27 days (Figs. 2 and 5A).

Rejection and survival were monitored by BSL and confirmed by histology of islet allografts. At day 35, islet allografts from hNSG recipient mice receiving either IL-2 alone or rapamycin alone showed a heavy infiltrate of mononuclear cells and no intact islets (Fig. 5B–E). Histological findings were similar to those observed in rejected allografts in the hNSG recipient mice without treatment (Fig. 2C). In contrast, hNSG recipient mice treated with combined IL-2/rapamycin had intact islets with positive insulin staining and were surrounded by a smaller cellular infiltrate at the same time point (Fig. 5B–E). By day 44, intact islets remained with very few infiltrating lymphocytes (Fig. 5B–E). However, this increased to a larger mononuclear cell infiltrate on day 62, containing hCD4⁺ and hCD8⁺ T cells (Supplementary Fig. 4). There was no histological evidence of xeno-GVHD in skin, intestine, or right kidney at days 35–44 after reconstitution, and less liver damage was observed (Fig. 6A–C). However, signs of chronic xeno-GVHD with limited skin and liver involvement was observed at day 62 despite functional islet allografts (Fig. 6A–D), with no significant histological evidence of GVHD in the contralateral right kidney (Supplementary Fig. 4).

To determine whether prolonged islet allograft survival following IL-2/rapamycin treatment was associated with the expansion of human Tregs (hTregs), the proportion of hCD45⁺ cells and hCD4⁺FOXP3⁺ Tregs was assessed. Compared with the no-treatment group, reconstituted hCD45 cells at week 3 of transfusion was decreased in all treatment groups (except IL-2 only) (Fig. 7A). However, by week 5, a similar proportion of hCD45 cells was seen in all groups (Fig. 7A). In the PB of mice treated with either IL-2^0.3/rapamycin or the higher dose of IL-2¹/rapamycin, the proportion of CD4⁺FOXP3⁺ Tregs was significantly increased (Fig. 7B). Interestingly, when the IL-2¹/rapamycin–treated mice were divided into those with a functioning graft and those with a rejected graft, the proportion of splenic CD25⁺FOXP3⁺ Tregs was markedly increased in the subset with functioning islet allografts (Fig. 7C). Furthermore, these findings, in PB and spleen, were replicated in the islet allograft, where a reduction in infiltrating CD4⁺ T cells and an increased proportion of CD4⁺FOXP3⁺ Tregs were seen in hNSG recipient mice following IL-2/rapamycin treatment (Fig. 8A and B). Taken together, these data support the proposal that IL-2/rapamycin therapy led to prolonged human islet allograft survival that was closely associated with the expansion of CD4⁺CD25⁺FOXP3⁺ Tregs in the graft, spleen, and PB.

Serum from hNSG recipient mice 35 days after transfusion was examined for human IL-2, IL-4, IL-6, IFN-γ, IL-10, TNF, and IL-17A by cytometric bead array. Other than human IFN-γ, the remainder of the cytokines were absent or low in the serum, with no significant difference observed between groups (Fig. 8C and data not shown). While hIFN-γ was detectable in hNSG recipient mice receiving no treatment or receiving rapamycin alone, in those receiving IL-2^0.3/rapamycin, the serum concentration of hIFN-γ was significantly reduced compared with NSG mice transfused with hSPMCs only (Fig. 8C).

Spleens were removed from hNSG recipient mice with and without IL-2 treatment alone at the time of rejection or following IL-2/rapamycin treatment (day 35). Splenocytes were analyzed by flow cytometry for T-bet and perforin, respectively (Fig. 8D). Intracellular staining of splenocytes from hNSG recipient mice confirmed a significant decrease in perforin expression in reconstituted hCD8⁺ T cells following IL-2/rapamycin treatment compared with no treatment or IL-2 only (Fig. 8D). These findings were supported by gene expression analysis by RT-PCR on sorted hCD4⁺ and hCD8⁺ T cells where there was a significant reduction in IFN-γ gene expression and a trend to lower perforin-1 expression in hCD8⁺ T cells in hNSG recipient mice receiving IL-2/rapamycin treatment, with no difference in IL-10, TGF-β, or IL-2 gene expression (Supplementary Fig. 5A).

In splenic hCD4⁺ T cells, there was a significant increase in the expression of T-bet, a key regulator of Th1 responses, in hNSG recipient mice receiving IL-2 alone or IL-2/rapamycin treatment compared with the no-treatment group (Fig. 8D). This finding was supported by gene expression data that showed a trend toward increased expression of Blimp-1 (30), T-bet, and TGF-β and no differences in expression of GATA3 and RORγt, which are transcriptional factors for Th2 and Th17 T-cell subsets, respectively (Supplementary Fig. 5B).

Our study shows that low-dose IL-2 in combination with low-dose rapamycin leads to an increased proportion of CD4⁺FOXP3⁺ Tregs, which was associated with prolonged (human) islet allograft survival in hNSG recipient mice reconstituted with hSPMCs. In this model, hSPMC transfusion led to hCD45⁺ cell reconstitution predominantly with CD4⁺ and CD8⁺ T cells and at early time points after reconstitution. CD4⁺FOXP3⁺Tregs were present in similar proportions to that found in human blood or spleen. As shown previously, this is an excellent model of human T-cell–driven allograft rejection (31). While several studies have shown that human T cells expand rapidly, express an activated immune phenotype, and cause xeno-GVHD after transfer to NSG and other immunodeficient mouse models, the time frame for GVHD exceeds that of allograft rejection, allowing the rejection response to be studied (31,32). It was hypothesized that in vivo expansion of hTregs in the presence of antigen would be more efficient at preventing rejection than infusion of hTregs that were expanded in vitro. To achieve in vivo hTreg expansion, a combination of IL-2 and rapamycin was used. IL-2 is an essential growth factor for Tregs, which constitutively express the high-affinity IL-2R (33). Clinically, low-dose IL-2 has been used in several auto- and alloimmune diseases, including SLE and chronic GVHD (5,7). Low-dose IL-2 treatment leads to a selective expansion of Tregs and clinical improvement in patients with GVHD (5,34). Similarly, in patients with SLE, low-dose IL-2 selectively modulates CD4⁺ T-cell subsets, with an increase in Treg number and function, and decreases follicular helper T and Th17 cells but not Th1 and Th2 cells (7). The combination of IL-2 and rapamycin has been used in a phase 1 trial of early-onset T1D leading to increased Tregs but a transient impairment of β-cell function (35). This highlights the complexities of clinical translation. In T1D, an autoimmune memory T-cell response is well established that is not able to be reversed, whereas islet transplantation has the advantage of initiating therapy at the time of transplantation when effector memory cells have not yet developed and where their development can be suppressed. Rapamycin has also been reported in some studies to be toxic to islets; however, the extent of this remains controversial, and rapamycin remains the main immunosuppressive for preventing clinical islet allograft rejection (36,37).

On the basis of the complementary role of IL-2 and rapamycin on Treg homeostasis, we proposed that these agents could be used to selectively expand Tregs in vivo and limit effector T-cell expansion. In the humanized mouse model of human islet allograft rejection and at the doses used in this study, the addition of IL-2 alone did not prolong islet allograft survival, although it expanded Tregs in vivo. Likewise, the use of rapamycin in the doses used here led to neither Treg expansion in vivo nor prolongation of graft survival. However, the combination of low-dose IL-2 and rapamycin for 3 weeks resulted in the expansion of CD25⁺FOXP3⁺ Tregs and prolonged islet allograft survival. In hNSG recipient mice with long-term surviving islet allografts, the proportion of splenic CD25⁺FOXP3⁺ Tregs was enhanced, suggesting that CD25⁺FOXP3⁺ Tregs were responsible for promoting graft survival. Interestingly, the IL-2/rapamycin treatment groups showed a significant decrease in the proportion of hCD45⁺ cells within the PB of the transplanted mice, suggesting that rapamycin either delayed hCD45⁺ cell reconstitution or suppressed effector T-cell proliferation after hSPMC transfusion. Taken together, these results suggest that IL-2 was an important driver of Treg expansion and that rapamycin played an important role in selectively suppressing effector T-cell expansion and T-cell–mediated islet alloimmunity.

IFN-γ is associated with rejection of solid organ and islet allografts in both experimental mouse models (38,39) and human biopsy specimens (40). IFN-γ was essential for rejection in a CD8⁺ T-cell–mediated rejection model of islet allotransplantation (38) as well as a CD4⁺ T-cell–mediated model of cardiac allograft rejection (39). Although IFN-γ is crucial for Th1 responses, it is secreted by natural killer T cells (41), CD8⁺ T cells (38), and CD4⁺CD25⁺ Tregs (42). It is also associated with maintenance of transplant tolerance (42,43) and plays a presumed role in rejection (38–40). In the findings presented here, serum and gene expression levels of IFN-γ were associated with increased T-bet gene expression, a Th1-specific transcription factor. When evaluated in the context of the study’s graft survival data, the findings were consistent with Treg-mediated suppression of the Th1 alloimmune response. Increased T-bet expression may reflect expression in Th1-specific Tregs. Graft survival was associated with reduced perforin-1 in CD8⁺T cells, suggesting an associated reduction in CD8⁺ T-cell–mediated effector responses (44).

CD4⁺FOXP3⁺ Tregs have been shown to be potent suppressors in experimental models of allo- and autoimmunity (45,46). Infusions of FOXP3⁺ Tregs expanded in vitro have been evaluated in phase 1/2 clinical trials of renal transplantation and GVHD as well as in autoimmune conditions such as T1D (47–50) (www.onestudy.org). Major limitations of this approach include the large numbers of Tregs that lack antigen specificity, the requirement for multiple infusions, and the cost of regulatory requirements and production. The ability to expand autologous Tregs in the presence of antigen is an attractive alternative to in vitro manipulation and expansion. We have reported here that the combination of low-dose IL-2 and rapamycin can successfully lead to in vivo expansion of Tregs, suppression of effector T-cell proliferation, and prolongation of islet allograft survival without the requirement for in vitro cell manipulation. It has the added advantage that both IL-2 and rapamycin are approved for clinical use in humans with an acceptable safety profile.

These findings need to be interpreted within the limitations of the model. NSG mice infused with hSPMCs did not have a completely reconstituted human immune system. The reconstituted cells were predominantly T cells, with limited B-cell repopulation. Previous studies have demonstrated that B cells contribute to allograft tolerance in human and mouse studies, making prolonged islet graft survival in the hNSG mice more remarkable (51). While in vivo–expanded Tregs were associated with prolonged graft survival and an immunoregulatory phenotype, their antigen-specific regulatory function was not confirmed. However, in previous studies, we have shown that in models of islet tolerance, expansion of Foxp3⁺ Tregs in the presence of antigen was a potent regulator of the graft immune response (26). This model was focused on the strategies that could inhibit the alloimmune response to human islets while limiting the burden of immunosuppression, which remains a major limitation to clinical islet transplantation. It does not address the issue of autoimmune disease recurrence in the host, which may not be amenable to this approach. Despite these limitations, this study demonstrates the potential of low-dose IL-2/rapamycin to limit human islet allograft rejection by expanding Tregs and limiting effector T-cell proliferation. These findings expand on successful results in murine experiments using IL-2 or IL-2/anti–IL-2 antibody complexes and rapamycin as a combination, leading to Treg expansion and effector T-cell suppression (14,52). These studies for the first time show the efficacy of IL-2 and rapamycin as a combination in a human immune model of allorejection of human islet transplants. Many strategies to expand naïve Tregs, ex vivo or in vivo, have failed in primate studies or human trials because of preexisting memory responses or a lack of antigen specificity, despite being successful in naïve mouse models. Our results suggest that it may be possible, using agents already approved for clinical use, to develop a clinical regimen that can minimize the burden of long-term immunosuppression or encouraging operational tolerance through the expansion of CD4⁺FOXP3⁺ Tregs in vivo, leading to long-term islet allograft survival in humans. Such a strategy, if successful, would have the potential to expand the indications for islet transplantation in T1D.

Acknowledgments. The authors thank the members of the Department of Animal Care at Westmead Hospital. Cell sorting and histological sample processing were performed at the Westmead Scientific Platforms, which are supported by the Westmead Research Hub, the Westmead Institute for Medical Research, the Cancer Institute New South Wales, the National Health and Medical Research Council of Australia (NHMRC), and the Ian Potter Foundation.

Funding. This work was supported by the NHMRC, JDRF/Australian Research Council, and Diabetes Australia and University of Sydney. M.H. was awarded an early career fellowship from the NHMRC (GNT1013185), a Deputy Vice Chancellor University of Sydney research fellowship (IRMA178768) and Sydney Medical School Early Career Researcher Scheme, and Diabetes Australia research program from Diabetes Australia and University of Sydney (Y16G-HUMI). N.R. is a recipient of an NHMRC career development fellowship (GNT1158597). P.J.O. was a recipient of a senior practitioner fellowship from the NMHRC (GNT1125456) and the principal investigator of JDRF/Australian Research Council grant 4-SRA-2016-265-M-B.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.H. designed and performed experiments, analyzed research data, wrote the manuscript, and finalized the manuscript for publication. W.J.H. performed human islet isolation and islet transplantation. L.N. performed double immunofluorescence staining and histological image analysis. H.B., Y.W.Q., D.L., E.J.V., Y.V.C., and L.W. performed human islet and spleen isolation, histological examination, and/or animal monitoring. S.Y. contributed to the discussion of the research data. K.K. and D.W. contributed to the discussion of the research data and histological examination. N.R. and S.I.A. contributed to the discussion and experimental design and edited the manuscript. P.J.O. proposed the study and designed the experiments, edited and reviewed the manuscript, and gave final approval of the final manuscript for publication. M.H. and P.J.O. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in oral abstract form at the 27th International Congress of the Transplantation Society, Madrid, Spain, 30 June–5 July 2018.