IL-21 Is an Antitolerogenic Cytokine of the Late-Phase Alloimmune Response

The following mice were used: C57BL/6, DBA/1J, BALB/c, Rag1^tm1Mom, B6-FoxP3-GFP knock-in, IL-21R^−/−, and hIL-21-Tg (15), all in a C57BL/6 background.

Transgenic C57BL/6 mice expressing mIL-21 cDNA under the control of a tetracycline-dependent promoter were crossed to ROSA-rtTA mice (see Supplementary Data). Mice were administered streptozotocin (STZ), transplanted with BALB/c islets, and treated the same day with doxycycline (2 mg/mL in drinking water until rejection).

Anti-CD3/anti-CD28 (anti-CD3/CD28) stimulation assay and mixed-lymphocyte reaction (MLR) assay were performed as described previously (19). In vitro assays to study generation, survival, and function of CD4⁺CD25⁺ Tregs were performed as previously described (17,20,21).

Islet-transplanted mice were treated with 400 μg mIL-21R.Fc (>99% purity; Pfizer, Cambridge, MA) or with 400 μg i.p. of a negative-control IgG2a antibody (anti–Eimeria tenella) every day for 10 days and then three times per week for a total of 4 weeks (22). A group of mice was treated with mIL-21R.Fc in combination with CTLA4-Ig (Abatacept, Orencia; Bristol-Myers Squibb, Princeton, NJ) (500 μg at day 0 and 250 μg at days 2, 4, 6, 8, and 10).

Data are expressed as mean ± standard error. Kaplan-Meier analysis (log-rank test) was performed for survival studies. A P value of <0.05 (by two-tailed testing) was considered an indicator of statistical significance. Analyses of data were performed using an SPSS statistical package for Windows (SPSS Inc., Chicago, IL).

IL-21R is highly expressed on naïve unstimulated CD4⁺, CD8⁺, and B220⁺ cells at baseline, and in contrast with previously published data (23), the percentage of IL-21R⁺ cells is unchanged on stimulated cells at day 1 and at day 3 (Fig. 1A and B) (some data not shown). It is interesting that after 5 days of anti-CD3/CD28 stimulation, a reduction in the percentage of IL-21R⁺ T cells and B cells was evident compared with baseline (IL-21R⁺CD4⁺ T cells = 8.5 ± 1.7%; IL-21R⁺CD8⁺ T cells = 14.9 ± 5.6%; IL-21R⁺B220⁺ cells = 21.5 ± 14.5%; baseline vs. all, P < 0.05) (Fig. 1C and D). IL-21 was undetectable in the supernatant at day 1 (Fig. 1E) and at day 3 (data not shown). However, increased levels of IL-21 in the supernatant were found at day 5 compared with day 1 (P < 0.001) in the anti-CD3/CD28 stimulation assay (Fig. 1E and F). When 150 ng/mL of recombinant mouse (rm)IL-21 was added in vitro, IL-21 supernatant levels were higher at day 1 compared with day 5 (P < 0.001) (Fig. 1E and F), possibly for rmIL-21 degradation.

At day 5 (but not at days 1 or 3), an overall reduction in the percentage of IL-21R⁺CD4⁺/CD8⁺/B220⁺ cells was evident (Fig. 1A–D) (some data not shown) regardless of the presence of rmIL-21 or mIL-21R.Fc. The addition of mIL-21R.Fc reduced IL-21 levels in the supernatant at day 5, when IL-21 first became detectable (Fig. 1F). Data obtained from anti-CD3/CD28 stimulation assay data were confirmed in the MLR assay as well (data not shown). To confirm that mIL-21R.Fc binds IL-21 cytokine (thus reducing IL-21 detectable levels in vitro), we added serial concentrations of mIL-21R.Fc (5, 50, and 150 ng/mL) to 150 ng/mL of rmIL-21 and evaluated IL-21 levels by Luminex assay (Millipore, Billerica, MA) at days 1, 3, and 5 after seeding. Reduced IL-21 levels were detected when rmIL-21 and mIL-21R.Fc were cocultured (i.e., IL-21 levels at day 1: no drug = 3290 ± 337 vs. mIL-21R.Fc 150 ng/mL = 1135 ± 284; P < 0.05).

We first challenged CD4⁺ and CD8⁺ T cells extracted from the spleens of 10-week-old C57BL/6 mice in an anti-CD3/CD28 stimulation assay with the addition of rmIL-21 or mIL-21R.Fc. Addition of rmIL-21 led to a dose-dependent increase of IFN-γ–producing CD4⁺ T cells (no drug = 70.5 ± 6.5; 5 ng/mL rmIL-21 = 91.7 ± 15.9; 50 ng/mL rmIL-21 = 104.8 ± 7.7; 150 ng/mL rmIL-21 = 110.8 ± 8.3, counted as number of IFN-γ–producing cells per 0.2 × 10⁶ total CD4⁺ T cells; no drug vs. 50 ng/mL rmIL-21 and 150 ng/mL rmIL-21, P < 0.01) (Fig. 1G) and CD8⁺ T cells (Fig. 1H). IL-4–producing CD4⁺ T cells increased in frequency as well when rmIL-21 was added to the anti-CD3/CD28 assay, while no IL-4–producing CD8⁺ T cells were detected in this assay (data not shown). These data were confirmed in an allostimulation assay (MLR) (data not shown). Addition of mIL-21R.Fc to the anti-CD3/CD28 assay induced a decrease in IFN-γ levels and left IL-4 unchanged (Fig. 1I), while rmIL-21 increased IFN-γ and IL-4 levels (Fig. 1J) 5 days after stimulation.

We then evaluated the effect of genetically overexpressing or knocking down the IL-21/IL-21R pathway. We used CD4⁺ T cells obtained from IL-21 plasmid–treated mice (injected with IL-21 pOrf plasmid) and from IL-21R^−/− mice (with genetic deletion of the IL-21R gene) for in vitro studies, confirming the data obtained with the rmIL-21 and mIL-21R.Fc (Supplementary Fig. 1A–D).

We performed a Treg generation assay (20) to investigate whether conversion from CD4⁺CD25⁻ cells (and, thus, FoxP3⁻) into CD4⁺FoxP3⁺ cells was affected by rmIL-21 or by the overexpression of IL-21 (through the use of hIL-21-Tg mice or IL-21 plasmid–treated mice). In a complementary experiment, we evaluated the effect of targeting IL-21 chemically (mIL-21R.Fc) or genetically (IL-21R^−/− mice) in the same assays. We found a dose-dependent decrease in Treg frequency when rmIL-21 was added compared with the percentage obtained without supplementation (no drug = 33.0 ± 0.4%; 150 ng/mL rmIL-21 = 11.5 ± 0.4%; P < 0.0001) (Fig. 2E and F). An increase in Tregs was evident when mIL-21R.Fc was added at high concentrations only (no drug = 33.0 ± 0.4%; 150 ng/mL mIL-21R.Fc = 42.3 ± 2.8%; P = 0.03) (Fig. 2E and F). When CD4⁺CD25⁻ cells obtained from hIL-21-Tg or IL-21 plasmid–treated mice were used, fewer Tregs were generated as compared with when using CD4⁺CD25⁻ cells from wild type (WT) and IL-21R^−/− mice (Fig. 2G–L).

To explore whether Tregs generated in the presence of IL-21 are dysfunctional and unable to suppress Teff proliferation in vitro, CD4⁺CD25⁻ 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled T cells obtained from WT C57BL/6.Thy1.2 mice were plated with the same number of C57BL/6.Thy1.1 Tregs generated in the presence of high concentrations (150 ng/mL) of rmIL-21 or mIL-21R.Fc or with medium alone. After 72 h of stimulation, cells were harvested, and CFSE dilution was assessed on Thy1.1-negative cells. It is interesting that an increased percentage of Teff proliferation was evident when Tregs generated in the presence of rmIL-21 were added compared with Tregs generated in the absence of drugs or with mIL-21R.Fc (no drug = 19.1 ± 0.5%; rmIL-21 = 27.7 ± 0.9%; mIL-21R.Fc = 16.1 ± 0.3%; rmIL-21 vs. all, P < 0.05) (Fig. 3A and B). This effect may be the result of both a reduced suppressive ability of Tregs and increased Treg apoptosis. Control CFSE-labeled CD4⁺CD25⁻ T cells stimulated in the absence of Tregs were shown to undergo robust proliferation (Fig. 3A and B). To assess the mechanism through which IL-21 mediates a decrease in in vitro Treg generation, we performed a Treg maintenance assay in which GFP⁺ cells (sorted by FoxP3-GFP mice) or CD4⁺CD25^hi cells (obtained through cell sorting of splenocytes extracted by IL-21 plasmid–treated or IL-21R^−/− mice) were plated with anti-CD3/CD28 monoclonal antibodies as well as IL-2 as a survival stimulus (17). A greater percentage of apoptotic Tregs (AnnexinV⁺/7-AAD⁻) was observed in the presence of 150 ng/mL rmIL-21 compared with Tregs cultured in the absence of any drug (no drug = 36.5 ± 0.9%; rmIL-21 = 53.4 ± 3.4%; P = 0.04) (Fig. 3C). In a similar manner, IL-21 overexpressing Tregs (from IL-21 plasmid–treated mice), but not IL-21R^−/− Tregs, displayed an increased apoptotic rate in the maintenance assay (Fig. 3D). No differences in mRNA levels of Bcl2, Bcl6, Bax, Caspase 3, Caspase 8, cMyc, Fas, and FasL were detected in Tregs cultured in the presence of rmIL-21 compared with control Tregs (data not shown).

To better understand the interplay of IL-21/IL-21R signaling with Tregs, we performed RT-PCR studies to evaluate mRNA transcript levels of FoxP3 and FoxP3-dependent genes in Tregs cultured with 150 ng/mL rmIL-21 or mIL-21R.Fc (17). RT-PCR expression studies confirmed the reduction of FoxP3 mRNA levels (rmIL-21 = 0.2 ± 0.0 fold decrease compared with baseline, P = 0.01) (Fig. 3E) and of FoxP3-targeted genes (CTLA4, Nrp-1, and FasL) in the presence of rmIL-21 (Fig. 3F–H). mIL-21R.Fc did not induce any modifications in FoxP3 and FoxP3-targeted gene expression. To assess whether IL-21 inhibition of TGF-β–driven differentiation of Tregs is mediated by FoxP3 epigenetic reprogramming, we evaluated the DNA methylation profile, through pyrosequencing (see Supplementary Data), of two regions of the FoxP3 gene (Intron 1 and Distal Promoter) known to be highly and poorly methylated, respectively, at basal conditions (16) in Tregs generated in the presence of rmIL-21 or mIL-21R.Fc or without any supplementation (Fig. 3I). No differences were evident in the CpG base pair percentage in the Intron 1 region (mainly hypermethylated) or the Distal Promoter region (mainly hypomethylated) regardless of rmIL-21 or mIL-21R.Fc presence (Fig. 3I).

Further confirmation of a non-FoxP3 methylation–mediated mechanism of IL-21 action was sought by examining the ability of decitabine or 5-Aza-2′-deoxycytidine (a DNA hypomethylating agent), to promote Treg generation in the presence of rmIL-21 (24). As previously shown (24), treatment with decitabine alone was able to induce Treg generation (Δ-percentage compared with no drug: no drug = 100.0 ± 17.7% vs. decitabine = 180.3 ± 17.3%; P = 0.01) (Fig. 3J). Decitabine was, however, unable to increase CD4⁺FoxP3⁺ cells in the presence of rmIL-21 (Δ-percentage compared with no drug: rmIL-21 = 29 ± 3.6% vs. rmIL-21 + decitabine = 26 ± 2.9%; NS) (Fig. 3J).

We then evaluated the percentage of IL-21R⁺ T cells and B cells as well as levels of IL-21 during the alloimmune response in a murine model of islet transplantation in vivo. Islets harvested from BALB/c mice were transplanted under the kidney capsule of hyperglycemic C57BL/6 (STZ) mice and treated with mIL-21-R.Fc, treated with control antibody (anti–E. tenella), or left untreated. No changes in CD4⁺IL-21R⁺, CD8⁺IL-21R⁺, or B220⁺IL-21R⁺ cell percentage were evident in the spleen at days 7 and 14 after transplantation (Fig. 4A–C). IL-21 serum levels analyzed at the same time points as above were not different at day 7 (Fig. 4D) but increased at day 14 after islet transplantation in the three groups of mice (Fig. 4D). The expression of the IL-21/IL-21R pathway was further evaluated in the graft infiltrate of untreated and mIL-21R.Fc–treated islet-transplanted mice at 7 and 14 days after islet transplantation. We found that IL-21R mRNA significantly increased at day 14 compared with baseline (in freshly isolated islets) and at day 7 (day 14 = 11.7 ± 2.5 fold increase compared with baseline; baseline vs. day 14, P = 0.04) (Fig. 4E) in untreated mice, while a lower but not significant intraislet expression of IL-21R mRNA was evident 14 days after islet transplantation in mIL-21R.Fc–treated compared with untreated mice (Fig. 4E).

We then tested the impact of targeting the IL-21/IL-21R axis (using the mIL-21R.Fc compound) on intragraft FoxP3⁺ cell expansion and ultimately on allograft survival and tolerance induction in a model of the alloimmune response in vivo (i.e., islet transplantation). Untreated C57BL/6 mice rapidly rejected islet allografts (mean survival time [MST] of 14 days, n = 5), and treatment with the negative control antibody anti–E. tenella led to similar graft survival (MST of 13 days, n = 5) (Fig. 4F). mIL-21R.Fc treatment prolonged graft survival, leading to long-term tolerance, defined as MST >100 days, in 20% of treated mice (MST: untreated = 14 days; anti–E. tenella treated = 13 days; mIL-21R.Fc treated = 19 days; mIL-21R.Fc treated vs. all, P < 0.05) (Fig. 4F). We therefore combined mIL-21R.Fc with CTLA4-Ig treatment (CTLA4-Ig: 500 μg at day 0; 250 μg at days 2, 4, 6, 8, and 10), with the aim of suppressing early T-cell activation (which is likely IL-21 independent) through CTLA4-Ig (25). CTLA4-Ig synergized with mIL-21R.Fc to further increase allograft survival compared with treatment with either mIL-21R.Fc or CTLA4-Ig alone, resulting in tolerance induction in 100% of mice (percentage of tolerant mice: CTLA4-Ig + mIL-21R.Fc treated = 100%; CTLA4-Ig alone treated = 55%; mIL-21R.Fc treated = 20%; CTLA4-Ig + mIL-21R.Fc treated vs. all, P ≤ 0.01) (Fig. 4F). Histological sections of the grafts were analyzed at day 14 and at day 100 (Fig. 4G1–J5). Mice treated with anti–E. tenella at day 14 after transplantation displayed a similar severe infiltration of the graft, primarily consisting of CD3⁺ and B220⁺ cells (Fig. 4G1–3 and H1–3), with absent insulin staining (Fig. 4G5 and H5) and very few islet-nested FoxP3⁺ cells (Fig. 4G4 and H4). Conversely, in the mIL-21R.Fc–treated group at day 14, islet morphology was still well preserved with several detectable insulin-producing cells and a mild CD3⁺ cell infiltrate, primarily made up of islet-nested FoxP3⁺ cells and very few B cells (Fig. 4I1–5). Islet morphology in tolerant CTLA4-Ig + mIL-21R.Fc–treated mice at 100 days after transplantation showed no T cells or B cells infiltrating the graft and a clear insulin staining (Fig. 4J1–5). An increase in the FoxP3⁺-to-CD3⁺ cell ratio, quantified with Aperio Technology (Vista, CA) (19), was observed at day 14 in mIL-21R.Fc–treated mice compared with both untreated mice and mice treated with anti–E. tenella (Fig. 4K). We confirmed our mIL-21R.Fc data and recent data (13) by transplanting BALB/c islets into chemically induced (STZ) hyperglycemic IL-21R^−/− mice, and a prolongation of islet graft survival compared with WT control mice was indeed observed (MST: WT = 14 days; IL-21R^−/− = 17 days; P = 0.01; data not shown).

A reduction in the percentage of peripheral CD4⁺CD44^hiCD62L^low Teffs was evident in mIL-21R.Fc–treated mice at 14 days after transplant compared with mice treated with anti–E. tenella (controls) and untreated mice and CD8⁺CD44^hiCD62L^low cells (Fig. 5A–C). The percentage of peripheral CD4⁺CD25⁺FoxP3⁺ cells (Tregs) increased in mIL-21R.Fc–treated mice compared with the other two groups (Fig. 5D and F), and a higher Treg-to-Teff ratio was evident in mIL-21R.Fc–treated mice as well (Fig. 5E). Ex vivo rechallenge of splenocytes harvested at day 14 from islet-transplanted mice with donor alloantigen (i.e., BALB/c splenocytes) displayed a profound regulation of the antidonor-specific immune response in mIL-21R.Fc–treated mice, as shown by the reduction of IFN-γ–producing cells in the ELISpot assay (BD Bioscience, San Jose, CA) compared with mice treated with anti–E. tenella and untreated mice (IFN-γ–producing cells per 0.5 × 10⁶: untreated = 125.0 ± 11.3; anti–E. tenella treated = 163.8 ± 30.6; mIL-21R.Fc treated = 25.3 ± 4.9; mIL-21R.Fc treated vs. all, P ≤ 0.01) (Fig. 5G). Of interest, 14 days after islet transplantation, a reshaping of the peripheral cytokine profile was evident in mIL-21R.Fc–treated mice as well, with an overall suppression of proinflammatory cytokines, as shown by the reduction of peripheral serum levels of IFN-γ, IL-2, IL-6, and IL-15 (Fig. 5H) compared with controls and untreated mice. IL-4, IL-10, and IL-17 peripheral levels remained unchanged in response to all treatments (Fig. 5H).

We next asked whether targeting the IL-21/IL-21R axis would be capable of promoting Treg neogenesis in vivo. Thus, we tested whether mIL-21R.Fc treatment could promote the generation of FoxP3 Tregs after depletion of CD25 using an anti–CD25-Ig injection (500 μg at day −6 and at day −1 before islet transplantation). Treatment with anti–CD25-Ig completely depleted CD25⁺ cells and FoxP3⁺ cells (data not shown). Targeting of the IL-21/IL-21R axis (with mIL-21R.Fc) during anti–CD25-Ig treatment slightly prolonged islet graft survival compared with anti–CD25-Ig treatment alone (MST: anti–CD25-Ig + mIL-21R.Fc treated = 25; anti–CD25-Ig treated = 20 days; P = 0.04) (Fig. 6A). Moreover, mIL-21R.Fc showed the ability to newly generate splenic CD4⁺FoxP3⁺ cells (Fig. 6B). To verify the robustness of the tolerance induced by mIL-21R.Fc + CTLA4-Ig treatment, we recovered 1 × 10⁷ splenocytes from long-term tolerant mice at day 50 after islet transplantation and adoptively transferred them into immunodeficient C57BL/6-RAG mice that had been islet transplanted the day before with BALB/c islets. Splenocytes from rejecting islet-transplanted C57BL/6 mice were used as controls. Of interest, 100% of C57BL/6-RAG mice that were adoptively transferred with splenocytes obtained from long-term tolerant mice did not reject BALB/c islets compared with the control group, which promptly rejected BALB/c islets (MST: long-term tolerant mice = 100 days; rejecting mice = 14 days; P < 0.0001) (Fig. 6C). To evaluate whether mice were immunocompetent, we transplanted skin grafts from third-party DBA/1 J (H2^q) donors into mIL-21R.Fc + CTLA4-Ig–treated long-term tolerant mice and into the control group of rejecting mice. Both groups of mice invariably rejected third-party skin grafts at ∼14 days after transplantation, thus confirming that mice treated with mIL-21R.Fc + CTLA4-Ig are immunocompetent (Fig. 6D). Finally, we tested mIL-21R.Fc + CTLA4-Ig treatment in a stringent model of tolerance resistance and in which BALB/c islets were transplanted into NOD mice, confirming a prolongation of graft survival in mice treated with mIL-21R.Fc + CTLA4-Ig compared with both untreated mice and CTLA4-Ig alone–treated mice (Fig. 6E).

To confirm the paramount effect of IL-21 on FoxP3 and Treg function/survival in vivo, we generated the doxycycline-inducible ROSA-rtTA-IL-21-Tg mouse (Doxa-inducible IL-21-Tg). This transgenic C57BL/6 mouse expresses mIL-21 cDNA in a tetracycline-responsive element composed of the tet operator sequence under the control of a tetracycline-controlled transcriptional activator M2 promoter (Fig. 6F–H). We then transplanted BALB/c islets into the Doxa-inducible IL-21-Tg mice, followed by daily doxycycline administration beginning from day 0 until rejection. Doxycycline-treated Doxa-inducible IL-21-Tg mice rejected the grafts with the same timing of Doxa-inducible IL-21-Tg untreated mice (MST: WT doxycycline treated = 14 days vs. Doxa-inducible IL-21-Tg untreated = 17 days, NS) (data not shown). Doxa-inducible IL-21-Tg mice treated with doxycycline showed similar percentage of CD4⁺CD25⁺FoxP3⁺ cells (data not shown), with a clear expansion of CD4⁺CD44^hiCD62L^low cells (Fig. 6I), CD8⁺CD44^hiCD62L^low cells (Fig. 6J), and CD4⁺CD25⁺FoxP3⁻ cells (Doxa-inducible IL-21-Tg untreated = 9.7 ± 1.2%; Doxa-inducible IL-21-Tg doxycycline treated = 30.8 ± 5.0%; P = 0.002) (Fig. 6K).

Finally, CD4⁺CD25⁺Fox3⁺-to-CD4⁺CD25⁺FoxP3⁻ cell ratio is reduced in Doxa-treated mice compared with controls (Fig. 6L). The absence of CD4⁺CD25⁺FoxP3⁺ cell reduction in Doxa-inducible IL-21-Tg mice observed in the acute model may be related to the fast tempo of acute rejection, which may not fit with the IL-21 role. To further clarify this issue, we have obtained long-term islet function in C57BL/6 islet-transplanted mice with CTLA4-Ig + mIL-21R.Fc, and we have then induced IL-21 by doxycycline administration 40 days after transplantation. Very few Tregs (3.7 ± 0.5%) were detected in mice 20 days after doxycycline administration, when mice started to be sick.

FoxP3⁺ Tregs play an essential role in maintaining immunological unresponsiveness to alloantigens and in suppressing the immune response (26). Herein, we propose IL-21 as an anti-FoxP3/anti-Treg cytokine of the late-phase alloimmune response that disrupts FoxP3 and alters Treg function, thereby acting as an antitolerogenic factor, thus eliciting an enhanced alloimmune response, and creating a barrier to transplant tolerance (1). In our study, IL-21 was increased in vitro after 5 days in anti-CD3/CD28 and MLR assays and in vivo 14 days after islet transplantation. IL-21R expression decreased in vitro after 5 days of stimulation and was found elevated in the graft infiltrate of rejecting untreated mice at later time points as well. Furthermore, mIL-21R.Fc (a fusion protein that prevents the binding of murine IL-21 to IL-21R) (22) is effective in vitro only in the late phase of allostimulation. These data, combined with the head-to-head comparison of IL-2 and IL-21 levels in vitro and in vivo (Fig. 7), revealed that in a “physiological” alloimmune setting, IL-21 is not responsible for the early response to alloantigen (unlike IL-2), but it does exert its function in the late-phase immune response (Fig. 7) (27). It is possible that IL-21 ensures the redundancy of the system, thus maintaining activation of the alloimmune response and avoiding Treg counterregulation (Fig. 7). IL-21 is capable of enhancing the alloimmune response, as shown by our in vitro experiments with rmIL-21 and with IL-21 plasmid–treated or hIL-21-Tg mice. However, it is likely that IL-21 may have a different role in the physiological alloimmune setting as a mediator of FoxP3 and of Treg dysfunction. The IL-21 interplay with Tregs was tested in vitro, using assays for Treg generation, survival, and function, during the challenge with rmIL-21 or mIL-21R.Fc, or by using hIL-21-Tg, IL-21 plasmid–treated (which overexpress IL-21), or IL-21R^−/− mice. We show that TGF-β–driven conversion of CD4⁺CD25⁻ T cells (FoxP3⁻), Treg survival, and Treg suppressive function are negatively affected by rmIL-21 but positively affected by mIL-21R.Fc. Indeed, we demonstrate an inhibition of FoxP3 and FoxP3-dependent gene transcripts by IL-21 (28), without evidence of IL-21–mediated epigenetic reprogramming of the FoxP3 gene, at least at the methylation level. These data were further confirmed by the inability of the demethylating agent decitabine to rescue Tregs in the presence of rmIL-21, thus indicating that despite the fact that decitabine is a robust inducer of Tregs in vitro (29), IL-21 is more potent in inhibiting Treg generation. We therefore demonstrate that blockade of the IL-21/IL-21R axis (both chemically and genetically) during the alloimmune response in vivo promotes the expansion of intragraft and peripheral Tregs and induces tolerance when combined with CTLA4-Ig, which targets early T-cell activation and is likely IL-21 independent. We show that mIL-21R.Fc promotes FoxP3 Treg neogenesis in vivo and that the mechanism of tolerance induction is robust and transferable to nude mice. The strong effect of the combination of mIL-21R.Fc and CTLA4-Ig on alloimmunity does not apply to NOD mice, which show the coexistence of strong allo- and autoimmune anti-islet responses (30). Autoimmune response is mediated by several different cell types (e.g., B cells, NK cells, macrophages, and Th17 cells), which may not be targeted by mIL-21R.Fc, but this aspect will be further investigated. Finally, we provide evidence that IL-21 promotes in vivo FoxP3⁻ cell expansion. In fact, in our islet-transplanted Doxa-inducible ROSA-rtTA-IL-21-Tg mice (31,32), in which IL-21 is induced by doxycycline administration, both CD4⁺ and CD8⁺ activated/effector cells were found increased compared with controls, while CD25⁺FoxP3⁺-to-CD25⁺FoxP3⁻ cell ratio was reduced. Furthermore, long-term tolerant mice overexpressing IL-21 (by doxycycline administration) showed very few Tregs. Taken together, our data suggest that IL-21 is a potent inhibitor of the FoxP3 Treg chain and an inducer of Teffs and that the blockade of the IL-21/IL-21R pathway may induce a more tolerogenic environment.

We posit that IL-21, together with other recently identified cytokines (33–35), is part of a network of antitolerogenic cytokines that through induction of Treg apoptosis or dysfunction, could represent a barrier to allograft tolerance. We thus posit that blockade of the IL-21/IL-21R pathway should be a precondition for many tolerogenic protocols in transplantation. Our strategy may be relevant in controlling acute as well as chronic rejection, which causes delayed islet graft loss (36–38) and represents a barrier for long-term allograft survival (39,40), since it abrogates T-cell activation in the early phase of posttransplant alloimmune response (with CTLA4-Ig) and expands/induces Tregs in the late phase (with IL-21 antagonist). We thus propose to start both drugs at the day of transplantation, discontinuing CTLA4-Ig treatment later on and prolonging the administration of IL-21 antagonist.

A.J.Z. has received National Institutes of Health Grant U01-AI-082966. P.F. has received a Juvenile Diabetes Research Foundation Career Development Award and the American Society of Nephrology Career Development Award. W.J.L. is supported by the Intramural Research Program, National Heart, Lung, and Blood Institute, National Institutes of Health.

D.Y. is a paid employee of Pfizer. P.F. has received support from an Independent Investigation Research Proposal from Novartis. W.J.L. is an inventor on patents and patent applications related to IL-21. Pfizer provided the mIL-21R.Fc compound. No other potential conflicts of interest relevant to this article were reported.

A.P. performed research, analyzed data, and wrote the manuscript. M.C. performed research and analyzed data. A.V., K.M.L., S.T., M.D., S.K., L.C., and B.G.M. performed research. P.H., A.S., W.J.L., J.F.M., and A.J.Z. designed research. D.Y. contributed vital new reagents. M.H.S. designed research. P.F. designed research and edited the manuscript.

Parts of this study were presented in abstract form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010.

The authors are grateful to Pfizer for providing the mIL-21R.Fc compound.