Interleukin-35-mediated induction of a novel regulatory T cell population

Abstract

Regulatory T cells (T_regs) play a critical role in the maintenance of immunological self-tolerance. Naïve human or murine T cell treatment with the inhibitory cytokine IL-35 induces a regulatory population, termed iT_R35, that mediates suppression via IL-35, but not IL-10 or TGFβ, neither express nor require Foxp3, are strongly suppressive in five in vivo models, and exhibit in vivo stability. T_reg-mediated suppression induces iT_R35 generation in an IL-35- and IL-10-dependent manner in vitro, and in inflammatory conditions in vivo in Trichuris-infected intestines and within the tumor microenvironment, where they appear to contribute to the regulatory milieu. iT_R35 may constitute a key mediator of infectious tolerance, may contribute to T_reg-mediated tumor progression, and ex vivo generated iT_R35 may possess therapeutic utility.

Regulatory T cells (T_regs) are a unique subset of CD4⁺ T cells that are essential for maintaining peripheral tolerance, preventing autoimmunity, and limiting chronic inflammatory diseases. However, they also prevent beneficial anti-tumor responses and sterilizing immunity against certain chronic infections. Consequently, the modulation of T_reg activity or generation of T_regs ex vivo are important goals of immunotherapy. Naturally-occurring, thymus-derived CD4⁺ T_regs (nT_regs) express the lineage specific transcription factor Foxp3 (forkhead box P3), which is required for their development, homeostasis and function^1–4. Despite their limited numbers (5–10% of CD4⁺ T cells), T_regs play a pivotal role in immune homeostasis. Indeed, it has been suggested that the suppressive milieu is potentiated by in vivo conversion of non-T_regs into suppressive cells, a process termed ‘infectious tolerance’. This contagious spread of suppression is thought to be a primary mechanism underlying transplantation tolerance⁵ and modulating autoimmune and inflammatory diseases, such as experimental allergic encephalomyelitis (EAE)⁶ and asthma⁷. While the mechanisms that mediate infectious tolerance remain obscure, both TGFβ and IL-10 have been implicated.

Induced regulatory T cell populations (iT_R) can be generated in the periphery, or in vitro, from conventional CD4⁺Foxp3⁻ T cells (T_conv)^8–10. There is substantial interest in the therapeutic potential of iT_R as it has been shown that antigen-specific regulatory populations can be generated that are potently inhibitory in vivo^{11, 12}. Two types of iT_R have been described based on the cytokines that induce them; TGFβ- and IL-10-iT_R. TGFβ-iT_R are generated following T cell activation in the presence of TGFβ with or without retinoic acid and IL-2. Both types of iT_R are potently suppressive both in vitro and in vivo^{11, 13, 14}, but possess distinct molecular signatures. While TGFβ iT_R express Foxp3 and primarily secrete TGFβ, IL-10 iT_R cells remain Foxp3⁻ following conversion and are defined by high IL-10 secretion.

T_reg-based approaches to treating inflammatory conditions such as allergy, autoimmune diseases, and graft-versus-host responses have great potential, but also have limitations [reviewed in ¹²]. The therapeutic potential of human T_regs is limited by their polyclonal specificity, poorly defined markers for enrichment, and reduced proliferative capacity which limits ex vivo expansion. Antigen-specific iT_R (IL-10 iT_R or TGFβ iT_R) can be generated ex vivo but their utility is restricted by technical complexities in their generation, limited potency and/or ambiguity regarding stability and longevity in vivo. Thus, the identification of a well-defined population of T_regs which can be readily generated ex vivo, and are stable and potently inhibitory in vivo is a critical goal for effective cell-based immunotherapy.

We have recently described a novel T_reg-specific cytokine, IL-35, that is required for maximal regulatory activity of murine T_regs in vitro and in vivo¹⁵. In this study, we show that IL-35, like IL-10 and TGFβ, can generate human and murine iT_R and address four questions: (1) what is their in vivo efficacy and stability, (2) can they be generated by nT_regs, (3) are they generated at inflammation sites, and (4) what is their physiological contribution to the regulatory milieu established by nT_regs?

RESULTS

Human IL-35 treated T_conv acquire a regulatory phenotype

Human IL-35 can suppress the proliferation of umbilical cord-derived human CD4⁺ T_conv cells to a degree similar to that seen by activated T_regs (see Supplementary Information and Supplementary Figs. 1 and 2). T_conv cells activated with anti-CD3- + anti-CD28-coated latex beads (αCD3/CD28) in the presence of IL-35 dramatically upregulated EBI3 and IL12A mRNA, the two constituents of IL-35 (Ebi3 and p35, respectively) (Fig. 1a), but not IL-10 or TGFβ (Supplementary Fig. 3). Single cell analysis by both intracellular cytokine staining (Fig. 1b) and confocal microscopy (Fig. 1c) suggests that IL-35, but not control protein, treatment induces homogeneous expression of IL-35 in human CD4⁺ T_conv cells. Similarly, CD4⁺CD45RA⁺CD25⁻ T_conv cells from adult peripheral blood expressed EBI3 and IL12A, but not TGFB or IL10 mRNA following activation in the presence of IL-35 (Supplementary Fig. 3i,j and data not shown).

Figure 1. Human IL-35 treatment of T_conv induces autocrine IL-35 expression and confers a regulatory phenotype.

T_conv purified by FACS from cord blood were treated with IL-35 or control at 25% of total culture volume for 9 days during activation (αCD3/CD28, and IL-2). (a) RNA was extracted, cDNA generated and qPCR performed. Relative Ebi3 (left panel) and Il12a (right panel) mRNA expression. (b) Cytokine treated cells were re-purified and stained with an isotype or a p35-specific antibody following 4h activation with PMA and ionomycin. Intracellular staining of IL-35 was determined by FACS. (c) Microscopic analysis of p35 expression was similarly determined following 4h activation with PMA and ionomycin. Anti- p35 or isotype control antibodies (shown in red), phalloidin (shown in green) and DAPI (shown in blue). (d) Proliferation of cytokine treated cells was determined by [³H]-thymidine incorporation (e) T_conv cells were mixed at indicated ratios (T_conv: suppressor) with control or IL-35 treated T_conv, hIL-2 and anti-CD3- + anti-CD28-coated latex beads for 9 days. Proliferation was determined by [³H]-thymidine incorporation. The mean of 4 representative experiments with similar cpm is shown. (f) Control or IL-35 treated T_conv were cultured in the top chambers of a Transwell^™ culture plate as indicated. Freshly purified responder T_conv were cultured in the bottom chamber of the 96-well flat bottom plates in medium containing hIL-2 and anti-CD3- + anti-CD28-coated latex beads. Top chambers were removed and [³H]-thymidine was added directly to the responder T_conv cells in the bottom chambers of the original Transwell^™ plate for the final 8 h of the 9 day assay. (g) T_conv cells were activated in the presence of IL-35 at 25% of total culture volume. Following conversion with cytokines, suppression assays were supplemented with neutralizing IL-10, TGFβ, or IL-35 to assess their requirement for indicated cytokines to mediate suppression. Counts per minute of T_conv cells activated alone, in the absence of control or IL-35, were 20,000–125,000 (f) and 25,000 – 570,000 (g). Data represent the mean ± SEM of (a)10 (b) 4, (c-g) independent experiments [* p < 0.05, ** p < 0.005, *** p < 0.001, NS = not significant].

We next assessed if IL-35-treated cells assumed the functional phenotype of iT_R. T_conv cells activated in the presence of IL-35 but not control were hyporesponsive to secondary restimulation (Fig. 1d). To determine whether IL-35-pretreated T_conv cells had acquired regulatory capacity, they were co-cultured as potential suppressors with freshly purified responder T_conv. While control-treated cells lacked any suppressive capacity, IL-35 treated cells were strongly suppressive (Fig. 1e). Human IL-35, but not control-treated, T_conv cells also suppressed responder T_conv cell proliferation across a permeable membrane, in the absence of direct cell contact, supporting a role for cytokine-mediated suppression (Fig. 1f). Moreover, neutralizing mAbs to IL-35, but not IL-10 or TGFβ, blocked their suppressive capacity (Fig. 1g, Supplementary Information and Supplementary Fig. 3). Taken together, these data suggest that IL-35 can convert human T_conv into a homogeneous population of iT_R cells that suppress via IL-35.

IL-35 treated murine T_conv acquire a regulatory phenotype in vitro

Given that human IL-35 can mediate iT_R generation, we then asked if murine IL-35 possessed a similar capacity (see Supplementary Information and Supplementary Fig. 4). Analysis of T_conv cells activated in the presence of murine IL-35 upregulated both Ebi3 and Il12a, but not Il10 or Tgfb mRNA (Fig. 2a, Supplementary Information and Supplementary Fig. 5). Immunoprecipitation and western blot analysis demonstrated that only IL-35 treated cells secrete IL-35, which was equivalent to the amount of IL-35 produced by natural T_regs. Both control-treated T_conv cells and IL-35-treated Ebi3^−/− T_conv cells did not secrete IL-35 (Fig. 2b). We next assessed if IL-35-treated murine cells, like their human counterparts, assumed an iT_R phenotype. Consistent with earlier reports¹⁶, previously activated T_conv cells proliferated well in response to secondary re-stimulation (Fig. 2c). IL-10 and IL-27 pre-treated T_conv also proliferated strongly in response to re-stimulation (note that short-term IL-10 treatment alone, in the absence of DCs, is insufficient to mediate IL-10 iT_R conversion¹⁴) (Supplementary Fig. 6). However, both IL-35 and TGFβ pretreated T_conv cells were hyporesponsive to re-stimulation, albeit to a lesser degree than freshly purified nT_regs. To determine whether these cytokine-pretreated T_conv cells had acquired regulatory capacity, they were co-cultured as potential suppressors with freshly purified responder T_conv cells (Fig. 2d and Supplementary Fig. 6). Whereas the control, IL-10- and IL-27-treated T_conv cells had no effect on responder proliferation, TGFβ-treated T_conv cells suppressed responder T cell proliferation¹³. As seen with human T cells, murine T_conv cells pretreated with IL-35 were also capable of suppressing responder T cell proliferation. Furthermore, IL-35- but not control-treated T_conv could suppress T cell proliferation in a contact-independent manner, across a permeable membrane, implicating soluble suppressive mediators (Fig. 2e). Using an IL-35 sandwich ELISA we also showed that approximately 500–700pg/ml IL-35 is required to mediate the induction of Ebi3 and Il12a expression and of the suppressive phenotype (Supplementary Fig. 4).

Figure 2. Murine IL-35 treatment of T_conv converts cells to an IL-35 producing suppressive population.

T_conv purified by FACS from C57BL/6, Ebi3^−/− or Il10^−/− mice were treated with indicated cytokines for 72 h during activation (αCD3/CD28). (a) RNA was extracted and cDNA generated from T_conv following control or IL-35 treatment. Relative Ebi3 (left panel) and Il12a (right panel) mRNA expression. (b) T_conv cells were cultured with control protein or IL-35 for 72 h. Cells were re-purified and cultured for an additional 24 h to facilitate IL-35 secretion. Culture supernatants from indicated cultures, or T_conv and T_reg as control, were immunoprecipitated with anti-p35 mAb, resolved by SDS-PAGE and probed with anti-Ebi3 mAb to identify IL-35 secretion. (c) Proliferative capacity, determined by [³H]-thymidine incorporation, of T_conv treated with indicated cytokines for 72 h. (d) T_conv cells were mixed at indicated ratios (T_conv: suppressor) with control or IL-35 treated T_conv and anti-CD3- + anti-CD28-coated latex beads for 72 h. Proliferation was determined by [³H]-thymidine incorporation (e) Control or IL-35 treated T_conv were cultured in the top chambers of a Transwell^™ culture plate as indicated. Freshly purified wild-type responder T_conv were cultured in the bottom chamber of the 96-well flat bottom plates in medium containing anti-CD3- + anti-CD28-coated latex beads. After 60 h in culture, top chambers were removed and [³H]-thymidine was added directly to the responder T_conv cells in the bottom chambers of the original Transwell^™ plate for the final 8 h of the 72 h assay. (f) T_conv from C57BL/6, Ebi3^−/− or Il10^−/− mice were activated in the presence of IL-35 at 25% of total culture volume, for 72 h to generate suppressive cells. Cells were re-purified and mixed at 4:1 ratio (T_conv: suppressor) and proliferation was determined. (g) Wild-type T_conv cells were activated in the presence of IL-35 at 25% of total culture volume. Following conversion with cytokines, suppression assays were supplemented with neutralizing IL-10, TGFβ, or IL-35 to assess their requirement for indicated cytokines to mediate suppression. Cells were cultured at a 4:1 ratio in suppression assays as described in e. (h) Suppressive capacity of IL-35 treated T_conv cells supplemented with titrations of isotype control or neutralizing IL-35 mAbs. Counts per minute of T_conv cells activated alone were 21,000–64,000 (e-h). Data represent the mean ± SEM of 4–8 independent experiments [* p < 0.05, ** p < 0.005, *** p < 0.001, NS = not significant].

To determine the mechanism of suppression, we first showed that IL-35-pretreated Il10^−/− (which cannot make IL-10), but not Ebi3^−/− (which cannot make IL-35), T_conv could suppress responder T cells (Fig. 2f). In addition, TGFβR.DN T_conv that are unable to respond to TGFβ were fully suppressed by IL-35-treated T_conv (see Supplementary Information and Supplementary Fig. 7). Using cytokine neutralizing mAbs, we then showed that IL-35-pretreated T_conv mediated suppression via IL-35 but not IL-10 or TGFβ (Fig. 2g,h [note that the anti-Ebi3 mAb used neutralizes IL-35 but not IL-27; Supplementary Fig. 8]). Collectively, these data suggest that murine IL-35 converts T_conv into an iT_R population that appears to mediate suppression exclusively via IL-35.

IL-35-treated murine T_conv exhibit a highly restricted genetic signature

Given that IL-35 can convert proliferative, Foxp3⁻ T_conv cells into hypo-responsive, strongly suppressive iT_R, we next sought to define their genetic signature. Interestingly, Foxp3 is neither induced nor required for the generation of IL-35-iT_R. While natural T_regs and TGFβ-iT_R express Foxp3, neither control nor IL-35 treated T_conv cells express Foxp3 (Supplementary Fig. 9). Moreover, Foxp3^−/− T_conv cells could be converted to IL-35 iT_R which expressed IL-35 and mediated suppression in a manner indistinguishable from their wild type counterparts (Supplementary Fig. 9c). In addition, iT_R35 cells do not express Foxp3 following in vivo inoculation as demonstrated by utilizing iT_R35 generated from Foxp3^gfp mice in an in vivo model of homeostatic expansion. Seven days post-transfer into Rag1^−/− mice, iT_Rcontrol, wild-type iT_R35 and Ebi3^−/− iT_R35 were purified based on congenic Thy1 markers and Foxp3 expression assessed by flow cytometric analysis of GFP expression. No induction of Foxp3 was seen in any of the transferred iT_R (Supplementary Fig. 9d).

We next compared the global gene expression of IL-35- and control-treated T_conv cells using Affymetrix GeneChip microarrays. Prior to analyses, we verified that the IL-35-iT_R used for analysis expressed Ebi3 and p35 mRNA, secreted IL-35 and suppressed responder T_conv cells (Supplementary Fig. 10). While clear differences were observed between nT_reg and T_conv, no genes appear to be significantly up- or down-regulated between IL-35- and control-treated T_conv (>3-fold; Supplementary Information and Supplementary Fig. 11). Nevertheless, the IL-35-iT_R cells generated from five independent experiments expressed Ebi3 and Il12a mRNA, secreted IL-35 and mediated potent in vitro suppression. The minimalistic genetic alteration observed in IL-35- compared to control-treated T_conv was further supported by analysis of T cell activation and co-stimulatory molecule expression and cytokine production. While the expression of most of the proteins/cytokines examined was indistinguishable between IL-35- and control-treated T_conv, reductions in the secretion of GM-CSF, IFNy, and IL-4 were observed, albeit not statistically significant (Supplementary Fig. 12). In addition, surface molecules such as CLTA-4 and CD25, which have been previously described as mediators of nT_reg suppression, were similarly upregulated in both control and IL-35-treated T_conv, cells, arguing against an exclusive role in the latter. Furthermore, the percentage of CTLA4⁺ IL-35-iT_R was relatively small (<15%). Taken together, these results suggest that IL-35 treatment mediates surgical rather than global gene expression changes.

Given that IL-35 is central to both the generation and suppressive function of IL-35-iT_R, we refer to this novel iT_R population as iT_R35 (for induced T regulatory population making IL-35) [Control-treated T_conv, which do not acquire a suppressive phenotype, are referred to as iT_Rcontrol]. While our data suggest that iT_R35 have a highly restricted Cd4⁺/Foxp3⁻/Ebi3⁺/Il12a⁺/Il10⁻/Tgfb⁻ genetic signature, we cannot rule out the possibility that there are some molecular changes that might distinguish IL-35- from control-treated T_conv that were not revealed by this analysis, or that this signature may apply to an as yet undefined non-regulatory population (See Supplementary Information).

iT_R35 are potently suppressive in vivo

The regulatory capacity of iT_R35 was tested in five different in vivo models. We first assessed whether iT_R35 could restore immune homeostasis and prevent lethal autoimmunity in Foxp3^−/− mice^{17, 18}. iT_R35 and various control populations were transferred into newborn (2–3 day old) Foxp3^−/− mice. Approximately 25 days later, an external clinical score, splenic and lymph node CD4⁺ T cell numbers, and a histological score (lungs, liver and skin) were determined (see Supplementary Information)¹⁹. As expected, nT_regs and TGFβ-iT_R, but not control-treated T_conv (iT_Rcontrol) were able to restore immune homeostasis and prevent autoimmunity (Fig. 3a-d and Supplementary Fig. 13). Interestingly, iT_R35 were as effective as nT_regs at restoring immune homeostasis and preventing autoimmunity in Foxp3^−/− mice. Importantly, neither Ebi3^−/− nor p35^−/− iT_R35 could restore immune homeostasis demonstrating the necessity for IL-35 production in vivo by iT_R35.

Figure 3. iT_R35 are suppressive in vivo.

Control treated (iT_Rcontrol), IL-35 treated (iT_R35) or TGFβ treated (TGFβ iT_R) cells were generated from FACS purified T_conv from wild type C57BL/6 or Ebi3^−/− mice. (a-c) Cells were injected i.p. into 2–3 day old Foxp3^−/− mice. Recovery from disease was monitored and reported by reduction in (a) clinical score (b) splenic CD4⁺ T cell number as determined by flow cytometric analysis and (c) combined histological score from lungs, liver and skin (d) Representative histopathology images of ear pinna following T cell transfer into Foxp3^−/− mice. Panel 1: Wild type mouse receiving no T_regs is normal and has no inflammatory cells in the dermis. Panel 2: Foxp3^−/− mouse receiving no T_regs is thickened with a dense inflammatory cell infiltrate causing distortion of the cartilage. Panel 3: Foxp3^−/− mouse receiving nT_regs is mildly thickened with mild edema and a sparse inflammatory cell infiltrate. Panel 4: Foxp3^−/− mouse receiving TGFβ iT_R cells is moderately thickened with inflammatory cells separated by edema fluid. Panel 5: Foxp3^−/− mouse receiving iT_Rcontrol cells is markedly thickened with edema dispersed inflammatory cells causing distortion of the cartilage. Panel 6: Foxp3^−/− mouse receiving iT_R35 cells is mildly thickened with edema and a mild inflammatory cell infiltrate. (e-j) iT_Rcontrol or iT_R35 were generated from FACS purified T_conv from C57BL/6 or Ebi3^−/− (Thy1.2) or B6.PL (Thy1.1) mice. (e) Homeostatic expansion was monitored by i.v. injection of Thy1.1⁺ T_conv cells alone or with Thy1.2⁺ iT_Rcontrol or iT_R35 cells (as regulatory cells) into Rag1^−/− mice. Seven days after transfer, splenic T cell numbers were determined by flow cytometry. Thy1.2⁺ regulatory T cell numbers (e). Thy1.1⁺ target T_conv cell numbers (f). (g) EAE was induced by immunizing mice with MOG35–55 peptide in complete Freund’s adjuvant followed by pertussis toxin administration. iT_Rcontrol, iT_R35 or nT_reg (10⁶) were transferred i.v. into C57BL/6 mice 12–18 hours prior to disease induction. Clinical disease was monitored daily. (h) Rag1^−/− mice received indicated cells via the tail vein on day –1 of experiment. On day 0, all were injected with 120,000 B16 cells i.d. in the right flank. Tumor diameter was measured daily for 15 days and is reported as mm³. (i-j) IBD was induced by injecting Rag1^−/− mice with T_conv cells via the tail vein. After 3–4 weeks, mice developed clinical symptoms of IBD and were given iT_Rcontrol or iT_R35 cells. Percentage weight change after iT_Rcontrol or iT_R35 cell transfer (i). (j) Colonic histology scores of experimental mice. Data represent the mean ± SEM of 8–12 mice per group from at least 2 independent experiments [* p < 0.05, ** p < 0.005, *** p < 0.001, NS = not significant].

Second, T_regs can control the homeostatic expansion of T_conv cells in lymphopenic recombination activating gene 1 (Rag1)^−/− mice^{15, 20, 21}. Purified wild-type Thy1.1⁺ T_conv cells, either alone or in the presence of control- or IL-35-treated Thy1.2⁺ T cells were adoptively transferred into Rag1^−/− mice, and splenic responder (Thy1.1⁺) and suppressor (Thy1.2⁺) T cell numbers determined 7 days later¹⁹. Thy1.2⁺ iT_Rcontrol expanded significantly and failed to block the expansion of Thy 1.1⁺ responder T_conv cells (Fig. 3e, f). In contrast, Thy1.2⁺ T_conv iT_R35, but not Ebi3^−/− iT_R35, had low proliferative capacity and significantly limited proliferation of Thy 1.1⁺ responder T_conv cells.

Third, experimental autoimmune encephalomyelitis (EAE) is a model for the human autoimmune disease multiple sclerosis and can be induced in C57BL/6 mice following immunization with MOG_35–55 peptide. Adoptively transferred nT_regs have been shown to reduce EAE disease severity^{6, 23, 24}. To determine whether iT_R35 could slow or prevent EAE, 10⁶ nT_regs, iT_Rcontrol or iT_R35 cells were transferred into mice prior to EAE induction¹⁹. Consistent with previous reports, clinical scores were reduced in nT_regs recipients, while disease course in mice receiving the iT_Rcontrol cells or saline alone was unaffected (Fig. 3g). Strikingly, the iT_R35-treated mice were completely protected from EAE, while mice receiving Ebi3^−/− iT_R35 were indistinguishable from saline control mice, suggesting that IL-35 production by iT_R35 in vivo is required for protection.

Fourth, T_regs can prevent anti-tumor CD8⁺ T cell responses against the poorly-immunogenic B16 melanoma^{25, 26}. Wild type naïve CD4⁺CD25⁻ and CD8⁺ T cells alone or in combination with nT_regs or iT_R35 cells were adoptively transferred into Rag1^−/− mice followed by i.d. injection of B16 melanoma cells and tumor size monitored daily¹⁹. As expected, tumor size was reduced in CD4⁺/CD8⁺ T cell recipients lacking T_regs compared with the untreated Rag1^−/− mice (Fig. 3h). In contrast, transfer of either nT_regs or iT_R35 cells completely blocked the anti-tumor response resulting in more aggressive tumor growth that was comparable to untreated Rag1^−/− mice. Surgical excision of the primary tumor and subsequent secondary tumor challenge at a distal site demonstrated that concomitant tumor immunity was also prevented by both nT_regs and iT_R35 (Supplementary Fig. 13).

Fifth, inflammatory bowel disease (IBD) is initiated by the adoptive transfer of naïve CD4⁺CD45RB^hiCD25⁻ T cells into Rag1^−/− recipient mice and disease assessed by weight loss and colonic histopathology²². After mice developed clinical symptoms of IBD (^~4 weeks post T cell transfer), they received iT_Rcontrol or iT_R35 and were monitored daily19. Recovery from disease, marked by weight gain (Fig. 3i) and decreased histopathology (Fig. 3j and Supplementary Fig. 13), was observed in mice that received iT_R35 but not the iT_Rcontrol cells. We also used this model to further demonstrate that TGFβ is not required for the in vivo suppressive capacity of iT_R35 (see Supplementary Information and Supplementary Fig. 14).

Finally, we further assessed the importance of IL-35 production by iT_R35 in vivo using a unique Ebi3-specific mAb that neutralized IL-35 but not IL-27 (Supplementary Fig. 8). Administration of this mAb, but not an isotype control, blocked the suppressive capacity of iT_R35 in an in vivo model of homeostatic expansion (Supplementary Fig. 13g), consistent with our observations using Ebi3^−/− or p35^−/− iT_R35 in three in vivo models (Fig. 3). Taken together, these data clearly demonstrate that iT_R35 have potent suppressive capacity in a wide variety of in vivo models and that this activity is dependent on IL-35 production in vivo.

iT_R35 are stable in vivo

Some have suggested that ex vivo generated iT_R cells are unstable in vivo. Although our five in vivo transfer experiments suggest that iT_R35 have some degree of stability, we used two approaches to assess this question directly. First, CD45.2⁺ iT_R35 or TGFβ-iT_R were generated in vitro and adoptively transferred into CD45.1⁺ C57BL/6 mice to monitor cell recovery and function over time (see Supplementary Information). Both iT_R35 and TGFβ-iT_R can be recovered from the spleen post-transfer, and retain expression of their signature genes, Ebi3/Il12a and Foxp3/Tgfb, respectively (Supplementary Fig. 15d). As in vitro, iT_R35 inoculated in vivo fail to induce Foxp3 expression suggesting that this critical nT_reg transcription factor is not required for iT_R35 maintenance and function. While 33% of the initial iT_R35 inoculum was recovered 3 weeks post-transfer, only 12% of TGFβ iT_R cells were recovered (Fig. 4a). In addition, purified iT_R35 cells still retained strong suppressive capacity, whereas the function of TGFβ-iT_R cells was reduced by approximately 50% (Fig. 4b). While this suggests that iT_R35 may be more stable in vivo, it does not exclude the possibility that iT_R35 and TGFβ-iT_R may home to different anatomical locations in the mouse, which could affect their recovery from the spleen. Second, we transferred nTregs, iT_R35 or TGFβ-iT_R into 2–3 day old Foxp3^−/− mice and determined how long they could prevent the onset of a moribund state (clinical score ≥4). By 5 weeks post-transfer, all the TGFβ-iT_R recipients were moribund compared with 40% and 33% of nT_regs and iT_R35 recipients, respectively (Fig. 4c). Furthermore, survival of the remaining nT_reg and iT_R35 recipients was longer, with 100% moribund not being reached until 6.5 and 8 weeks, respectively. Although additional experiments will be required to fully evaluate the long-term stability of iT_R35 in homeostatic and inflammatory environments, these data suggest that they may be functionally stable in vivo.

Figure 4. Comparative stability of iT_R35 and TGFβ iT_R in vivo.

iT_R35 or TGFβ iT_R were generated in vitro with CD45.2⁺ T_conv cells and adoptively transferred into CD45.1⁺ C57BL/6 mice to assess stability. (a) Splenic iT_R cell number was determined by flow cytometric analysis of CD45.2⁺ cells. Percentage of total cells injected that were recovered 25 days post-transfer. (b) T_conv cells were mixed at indicated ratios (T_conv: suppressor) with either freshly generated iT_R cells (pre-transfer) or iT_R cells recovered after in vivo resting (post-transfer) and anti-CD3- + anti-CD28-coated latex beads for 72 h. Proliferation was determined by [³H]-thymidine incorporation. (c) Natural T_regs or iT_R were injected into 2–3 day old Foxp3^−/− mice. Mice were monitored for clinical signs of sickness and scored accordingly. Mice obtaining a clinical score of 4 were considered moribund. Survival, based on longevity of mice, was significantly less in recipients of TGFβ iT_R when compared to nT_regs. Conversely, mice that received iT_R35 survived significantly longer than those that received nT_regs (p < 0.005). Counts per minute of T_conv cells activated alone were 18,000–44,000 (b). Data represent the mean ± SEM of 5–12 mice per group from at least 2 independent experiments [* p < 0.05, *** p < 0.005].

T_reg-mediated suppression generates iT_R35

It has been suggested that T_regs can amplify their suppressive capacity by converting non-regulatory populations into suppressive cells, consistent with the concept of infectious tolerance, and that this process might be cytokine-mediated^27–29. We have previously shown that nT_regs are a natural source of IL-35, which increases 5–10-fold upon contact with the target T_conv cells^{15, 30}. Thus, we asked whether nT_reg-derived IL-35 could mediate iT_R35 conversion. We first purified T_conv cells that had been cultured with, and suppressed by, nT_regs for 3 days (which we refer to as suppressed T_conv) and found that expression of both Ebi3 and Il12a (p35) mRNA was significantly up-regulated following co-culture, to a level comparable with nT_regs (Fig. 5a,b). Furthermore, suppressed T_conv generated by co-culture with wild type, but not Ebi3^−/− nT_regs, secrete a significant amount of IL-35 (Fig. 5c). This demonstrates that IL-35 secretion by nT_regs is required to induce IL-35 secretion by co-cultured, suppressed T_conv cells. To determine whether suppressed T_conv express Foxp3, a prerequisite for mediating the regulatory activity of nT_regs and TGFβ-iT_R, we activated Thy1.2⁺ Foxp3^gfp T_conv cells alone or in combination with Thy1.1⁺ nT_regs. Unlike TGFβ-iT_R, but similar to iT_R35, suppressed T_conv remain Foxp3⁻ following activation in the presence of T_regs suggesting that TGFβ may not mediate this conversion (Supplementary Fig. 16). These data raise the possibility that iT_R35 are generated within the suppressed T_conv population.

Figure 5. T_regs generate iT_R35 in an IL-35- and IL-10-dependent manner.

T_conv were activated in the presence of T_reg at a 4:1 ratio (responder: suppressor) for 72 h. (a) RNA was extracted and cDNA generated from resting or activated T_conv cells or from T_conv:T_reg co-cultures (resorted based on differential Thy1 markers). Ebi3 (a) and Il12a (b) expression of the populations indicated. (c) Following co-culture, suppressed T_conv were re-purified and cultured for an additional 24 h. Secretion of IL-35 was determined by IP/WB of culture supernatants and compared to that of freshly cultured T_conv and T_reg. (d) Following co-culture, suppressed T_conv were re-purified and activated (αCD3/CD28). Proliferative capacity was assayed by [³H]-thymidine incorporation. (e) The suppressive capacity of suppressed T_conv upon fresh responder T_conv cells was determined by [³H]-thymidine incorporation. (f) Anti-IL-10, anti-TGFβ, or anti-IL-35 neutralizing antibodies were added to co-cultures to inhibit cytokine driven “conversion” into suppressed T_conv (top panel) or added in secondary proliferation assays to inhibit cytokine driven suppression of “function” (bottom panel). (g) T_conv cells alone or with C57BL/6, Ebi3^−/− suppressed T_conv (as regulatory cells) were injected into Rag1^−/− mice. Seven days after transfer, splenic T cell numbers were determined by flow cytometry. (h) EAE was induced by immunizing mice with MOG35–55 peptide in complete Freund’s adjuvant followed by pertussis toxin administration. 1 × 10⁶ suppressed T_conv or natural T_reg were transferred i.v. into C57BL/6 mice 12–18 hours prior to disease induction. Clinical disease was monitored daily. Clinical score was statistically significant between mice receiving saline control and T_reg or suppressed T_conv at days 16–18. Counts per minute of T_conv cells activated alone were 14,000–34,000 (e, f). Data represent the mean ± SEM of 8–12 mice per group from at least 2 independent experiments [* p < 0.05, ** p < 0.005, *** p < 0.001, NS = not significant].

We next assessed whether suppressed T_conv gained the phenotypic characteristics of a regulatory population. Interestingly, suppressed T_conv were profoundly unresponsive to anti-CD3 stimulation and were potently suppressive in vitro (Fig. 5d, e). T_regs can secrete IL-10, TGFβ and IL -35 which may influence their ability to convert T_conv into suppressed T_conv. Likewise, the same cytokines could be secreted by suppressed T_conv and contribute in an autocrine fashion to their conversion and/or their suppressive activity. To address these questions we first co-cultured T_conv and T_regs that were wild type or lacked the capacity to produce IL-35 (Ebi3^−/− or Il12a^−/−) or IL-10 (Il10^−/−), or were unable to respond to TGFβ (TGFβR.DN). While the generation of suppressed T_conv that were hyporesponsive and possessed regulatory capacity did not require TGFβ-mediated signaling, the absence of both IL-35 and IL-10 in the nT_reg:T_conv co-culture blocked their development and/or function (Fig. 5d, e). Further analysis using nT_reg:T_conv co-cultures in which only one population was mutant revealed that both nT_reg-derived IL-10 and IL-35 was required for the generation of the regulatory suppressed T_conv population. Interestingly, suppressed T_conv-derived IL-35 was required for conversion as suggested by qPCR analysis of signature genes (see Supplementary Information and Fig. 15b).

Neutralizing antibodies were included during the conversion process or in the secondary suppression assay to further assess the role of IL-35, IL-10 and TGFβ (Fig. 5f). While anti-TGFβ had no effect at either stage, IL-10 neutralization partially blocked conversion but not the regulatory capacity of suppressed T_conv, suggesting that IL-10 is required for optimal conversion of suppressed T_conv into a regulatory population. In contrast, IL-35 neutralization prevented both the conversion and regulatory function of suppressed T_conv. Collectively, these data suggest that some or all of the suppressed T_conv are iT_R35. The precise contribution of IL-10 to T_reg–mediated conversion of iT_R35 remains to be fully elucidated, as IL-10 alone does not induce Ebi3 and Il12a mRNA expression (Supplementary Fig. 5). However, the addition of low dose IL-10 appears to augment IL-35-mediated conversion, which may help offset the delayed production of IL-35 by nT_regs (see Supplementary Information and Supplementary Fig. 16c). Taken together, these data suggest that IL-35, either from a natural source (nT_regs) or supplemented exogenously, mediates iT_R35 conversion.

Next, the regulatory capacity of nT_regs-suppressed T_conv (nT_regs-induced iT_R35) was assessed in vivo. First, they were able to significantly suppress the homeostatic expansion of co-transferred naïve T_conv in Rag1^−/− mice in a manner comparable to nT_regs and iT_R35 (Fig. 5g). However, suppressed T_conv generated from Ebi3^−/− T_conv cultured with wild-type nT_reg, failed to suppress the expansion of co-transferred T_conv. Second, in the EAE model, peak clinical disease scores were decreased by suppressed to a level T_conv comparable with nT_regs (Fig. 5h). However, suppressed T_conv could not ameliorate EAE as effectively as iT_R35 suggesting either that only a proportion of this suppressed T_conv population are iT_R35 or that conversion in vitro is suboptimal due the time required for potentiation of IL-35 production by nT_regs15, 30. Nevertheless, these data support the notion that iT_R35 are generated from T_conv, to some degree, by nT_regs during suppression. In contrast, there is no evidence for the generation of IL-10-iT_R or TGFβ-iT_R in this setting.

T_reg-mediated induction of iT_R35 in vivo

We reasoned that iT_R35 generation in vivo would occur predominantly in inflammatory or disease environments where optimally stimulated nT_regs might be secreting high amounts of IL-35. Infection with Trichuris muris, an intestinal nematode, is known to promote T_regs responses at the site of infection, the large intestine³¹. Thus, we assessed whether iT_R35 could be detected following Trichuris muris infection, using the CD4⁺/Foxp3⁻/Ebi3⁺/Il12a⁺ iT_R35 signature. Foxp^gfp mice were infected with Trichuris muris and Foxp3⁺ and Foxp3⁻ T cells were purified from spleens, small intestines and large intestines 14 days post-infection. Both Ebi3 and Il12a (p35) expression were dramatically increased in Foxp3⁺ T_regs, in both the small and large intestines, compared with splenic T_regs, consistent with our previous observations that nT_regs increase IL-35 expression ^~10-fold in the presence of T_conv cells¹⁵ (Fig. 6a,b). While essentially no Ebi3/Il12a expression was observed in splenic Foxp3⁻CD4⁺ T cells, expression was substantial in comparable isolates from the small and, especially, the large intestines (the primary site of infection). Indeed, Ebi3/Il12a expression in Foxp3⁺ T_regs and Foxp3⁻CD4⁺ T cells from the large intestines was statistically indistinguishable. It should be emphasized that we have never observed IL-35 expression by naïve, activated or memory CD4⁺ T cells¹⁵, and Foxp3⁻CD4⁺ T cells from the intestines or MLN of uninfected mice do not express Ebi3/Il12a (data not shown) raising the possibility that iT_R35 are being generated by T_regs within this inflammatory microenvironment.

Figure 6. IL-35-producing Foxp3⁻ iT_R35 develop in vivo.

Foxp3^gfp mice were infected with Trichuris muris. CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ cells were purified from spleen, small intestine IEL and LPL, or large intestine IEL and LPL by FACS, RNA extracted and cDNA generated. Ebi3 (a) and Il12a (b) expression of the populations indicated. (c, d, g, h) Foxp3^gfp mice or Ebi3^−/− Foxp3^gfp were injected with 120,000 B16 cells i.d. on the right flank. Tumors and spleens were excised after 15–17 days, CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ cells, purified by FACS, RNA extracted and cDNA generated. Ebi3 (c) and Il12a (d) expression of the populations indicated. (e, f) Foxp3^gfp mice or Ebi3^−/− Foxp3^gfp were injected with 2×10⁶ MC38 cells subcutaneously on the right flank. Tumors and spleens were excised after 12 days, CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ cells purified by FACS, RNA extracted and cDNA generated. Ebi3 (e) and Il12a (f) expression of the populations indicated. (g) Following B16 cell inoculation, purified T cells from the spleen or tumor were cultured for 24 h to allow secretion of IL-35. Culture supernatants from indicated cultures were immunoprecipitated with anti-p35 mAb, resolved by SDS-PAGE and probed with anti-Ebi3 mAb to identify IL-35 secretion. No IL-35 secretion was seen in either the splenic or tumor-infiltrating lymphocytes from Ebi3^−/− mice (g). Purified cells were assayed for regulatory capacity by mixing populations indicated at a 4:1 ratio with fresh responder T_conv cells for 72 h. Proliferation was determined by [³H]-thymidine incorporation. Counts per minute of T_conv cells activated alone were 16,000–33,000 (h). Data represent the mean ± SEM of 8–10 mice per group from 2–3 independent experiments (B16) and 1 experiment (MC38) [* p < 0.05, ** p < 0.005, *** p < 0.001, NS = not significant].

The inflammation induced by solid tumors is known to attract T_regs^32–37. Using B16 melanoma and MC38 colorectal adenocarcinoma as model systems^{25, 38, 39}, tumor cells were inoculated into Foxp^gfp mice, solid tumors resected 15–17 days (for B16) or 12 days (for MC38) post-transfer and Foxp3⁺ and Foxp3⁻ T cells purified from spleens and tumors. Interestingly, tumor-infiltrating Foxp3⁺ T_regs had substantially increased expression of both Ebi3 and Il12a (Fig. 6c-f). Surprisingly, tumor-infiltrating Foxp3⁻ T cells also dramatically upregulated Ebi3 and Il12a expression. It is noteworthy that comparable observations were made in two distinct tumor types. We further analyzed IL-35 secretion and its physiological relevance using the B16 melanoma system. While a moderate amount of IL-35 secretion by splenic Foxp3⁺ T_regs was observed ex vivo, a substantial and comparable amount of IL-35 was secreted by both Foxp3⁺ T_regs and Foxp3⁻CD4⁺ tumor-infiltrating T cells (Fig. 6e). Lastly, we assessed whether tumor-infiltrating CD4⁺/Foxp3⁻/Ebi3⁺/Il12a⁺ T cells were able to suppress the proliferation of fresh responder T_conv in vitro. Although their suppressive capacity was not as potent as that of tumor-infiltrating Foxp3⁺ T cells, our results clearly demonstrate that tumor-derived Foxp3⁻CD4⁺ T cells can mediate effective suppression in vitro in an IL-35-dependent manner (Fig. 6f).

iT_R35 contribute to the regulatory milieu in vivo

Finally, we assessed the physiological contribution of iT_R35 to the T_reg-induced regulatory milieu. We reasoned that if iT_R35 development within the tumor played a significant role in blocking anti-tumor immunity, then mice reconstituted with T_conv cells that lacked the ability to be converted into iT_R35 would develop smaller tumors. Therefore, Rag1^−/− mice were reconstituted with wild type CD8 cells, with or without wild type T_regs plus either wild type or Ebi3^−/− CD4⁺ T_conv cells. As expected, tumor size was reduced in CD4⁺/CD8⁺ T cell recipients lacking T_regs (50–90mm³) regardless of whether wild type or Ebi3^−/− CD4 T_conv cells were transferred. Co-transfer of nT_regs with wild type CD4⁺/CD8⁺ T cells blocked anti-tumor immunity resulting in aggressive tumor growth (470mm³) (Fig. 7). Analysis of congenically-marked tumor-infiltrating CD4⁺ T cells confirmed high Ebi3/Il12a expression that was comparable to the T_regs (see Supplementary Information and Supplementary Fig. 17). Thus it should be noted that in this instance both IL-35-producing nT_regs and iT_R35 contribute to the suppressive milieu (Fig. 6, 7 and Supplementary Fig. 17). Strikingly, co-transfer of nT_regs and CD8⁺ T cells with Ebi3^−/− CD4⁺ T_conv (which are unable to be converted to IL-35-producing iT_R35) only partially blocked anti-tumor immunity resulting in intermediate tumor growth (220mm³). These results suggest that T_reg-mediated induction of iT_R35 development has a significant impact on tumor burden and is responsible for approximately half the regulatory milieu within the tumor microenvironment (as the tumor burden is reduced from 470mm³ to 220mm³). Furthermore, these data suggest that nT_regs mediate their suppressive activity, in part, by inducing iT_R35.

Figure 7. The suppressive T cell milieu in the tumor microenvironment is largely due to iT_R35.

Rag1^−/− mice were reconstituted with wild type C57BL/6 CD8⁺ T cells and wild type or Ebi3^−/− T_conv cells with or without wild type T_regs. The following day, all mice were injected with 120,000 B16 cells i.d. on the right flank. Tumors were excised on day 15 post-inoculation to facilitate analysis of tumor infiltrating lymphocytes and tumor diameter reported as mm³. Data represent the mean ± SEM of 6–12 mice per group from at least 2 independent experiments [* p < 0.05, *** p < 0.005, NS = not significant].

DISCUSSION

iT_R35 cells represent a new and unique member of the regulatory T cell family that are generated by, and mediate their suppression via, IL-35. They do not express Foxp3 and are quite distinct from the currently known induced regulatory populations, TGFβ iT_R and IL-10 iT_R (see Supplementary Information and Supplementary Fig. 18). The concept of infectious tolerance, whereby T_reg confer a suppressive phenotype upon T_conv cells, has been previously described in both murine and human systems^{27, 29}. It has been suggested that TGFβ may play a critical role in mediating infectious tolerance⁴⁰. Since IL-35-secreting T_regs can convert T_conv cells into iT_R35, a suppressed T_conv population with regulatory potential, this raises the possibility that IL-35 and iT_R35 may represent additional, important mediators of infectious tolerance. Indeed, our data suggest that up to half the regulatory microenvironment within the tumor is mediated by T_reg-induced iT_R35. This also suggests that iT_R35 may contribute to tumor progression. While iT_R35 may play a significant physiological contribution to the regulatory milieu established by T_reg, further studies will be required to fully elucidate their contribution in diverse disease settings.

Whether a similar system operates in humans remains to be determined, but may be controversial given recent studies suggesting that human T_regs do not make IL-35^{41, 42}, findings that await further confirmation. Nevertheless, human iT_R35 can be generated and can suppress primary human T cell proliferation in an IL-35-dependent manner. The potential therapeutic application of ex vivo generated IL-10 iT_R and TGFβ iT_R is complicated by complexities in their generation, their short half-life and reversal of their suppressive capacity in time or by IL-2 (see Supplementary Information)^{13, 43, 44}. Although additional experiments are needed to fully assess the clinical potential of iT_R35, our data suggest that they represent a new, stable iT_R population that may have significant therapeutic utility.

METHODS

Methods and any associated references are available in the online version of the paper at http://www.nature.com/natureimmunology/.