Instability of Helios-deficient Tregs is associated with conversion to a T-effector phenotype and enhanced antitumor immunity

Significance

Depletion or inhibition of regulatory T cells (Tregs) has been associated with increased effector T-cell activation that may enhance antitumor responses. A potentially more effective strategy depends on induction of lineage instability that allows conversion of intratumoral but not systemic Tregs into effector T cells (Teffs). We show that targeted deletion of the Helios transcription factor within CD4 Tregs promotes instability and effector cell conversion of Tregs in tumors and increased antitumor immunity. Ab-dependent ligation of Treg surface receptors that diminishes Helios expression can also induce intratumoral Treg conversion. These findings indicate that targeting of signaling pathways that reduce Helios expression by intratumoral Tregs may represent a potentially robust approach to cancer immunotherapy.

Abstract

Expression of the transcription factor Helios by Tregs ensures stable expression of a suppressive and anergic phenotype in the face of intense inflammatory responses, whereas Helios-deficient Tregs display diminished lineage stability, reduced FoxP3 expression, and production of proinflammatory cytokines. Here we report that selective Helios deficiency within CD4 Tregs leads to enhanced antitumor immunity through induction of an unstable phenotype and conversion of intratumoral Tregs into T effector cells within the tumor microenvironment. Induction of an unstable Treg phenotype is associated with enhanced production of proinflammatory cytokines by tumor-infiltrating but not systemic Tregs and significantly delayed tumor growth. Ab-dependent engagement of Treg surface receptors that result in Helios down-regulation also promotes conversion of intratumoral but not systemic Tregs into T effector cells and leads to enhanced antitumor immunity. These findings suggest that selective instability and conversion of intratumoral CD4 Tregs through genetic or Ab-based targeting of Helios may represent an effective approach to immunotherapy.

Although CD4 regulatory T cells (Tregs) are essential for the maintenance of self-tolerance, accumulation of Tregs within the tumor microenvironment (TME) can inhibit antitumor immunity. The increased numbers of Tregs within tumors compared with peripheral lymphoid tissues may reflect, in part, efficient Treg recruitment and expansion. Enrichment of intratumoral Tregs may also reflect increased autoreactivity of naturally occurring FoxP3⁺ Treg cells (nTregs), including recognition of tumor-associated antigens (TAAs) (1, 2). These findings suggest that interventions that target Tregs may effectively complement current approaches aimed at checkpoints that control effector T-cell activation. Indeed, administration of Abs targeted at Tregs alone (e.g., TIGIT, CCR4, FR4) can strengthen antitumor immune responses (3–5).

The transcription factor (TF) Helios is expressed by two regulatory T-cell lineages—FoxP3⁺CD4⁺ and Ly49⁺CD8⁺ Tregs—which are essential to maintain self-tolerance (6, 7). Recent studies have revealed that Helios expression by Tregs ensures stable expression of a suppressive and anergic phenotype in the face of intense inflammatory responses, in part by ensuring Treg survival and stability through activation of the IL-2Rα–STAT5 pathway (6). These observations have also suggested that loss of Treg lineage stability in inflammatory environments can result in conversion of Tregs into cells that express an effector T-cell phenotype and produce proinflammatory cytokines (6).

The essential contribution of Helios to the maintenance of Treg size and functional stability in the face of diverse immunological perturbations may be relevant to the strategies that underpin current tumor immunotherapy. Current approaches rely mainly on depletion or blockade of CD4 Tregs to shift the systemic balance toward Teff cells. However, alterations in this balance may provoke severe autoimmune disorders. A more effective strategy may depend on approaches that selectively convert intratumoral Tregs into Teff cells without affecting the systemic Treg population.

Here we report that selective Helios deficiency within CD4 Tregs leads to enhanced antitumor immunity through induction of an unstable phenotype by intratumoral but not systemic Tregs and conversion of these Tregs into Teff cells within the TME. Induction of an unstable Treg phenotype is associated with enhanced activity of tumor-infiltrating CD4⁺ and CD8⁺ Teff lymphocytes and significantly delayed tumor growth. Moreover, Ab-dependent engagement of Treg surface receptors that down-regulate Helios expression also promote effector cell conversion of intratumoral but not systemic Tregs and enhanced antitumor immunity. These findings suggest that selective intratumoral inactivation and conversion of CD4 Tregs through targeting Helios may represent an effective approach to cancer immunotherapy.

Results

Intratumoral CD4 Tregs Express an Enhanced Suppressive Phenotype.

The contribution of Helios to stabilization of the Treg phenotype in the face of chronic inflammatory conditions may include Treg stability within progressively growing tumors. A comparison of the phenotype of intratumoral Tregs with Tregs in peripheral lymphoid tissues of B16 tumor-bearing mice indicated that intratumoral CD4 Tregs expressed significantly higher levels of Helios compared with splenic or lymph node (LN) Tregs (Fig. 1A and SI Materials and Methods). Increased Helios expression by intratumoral Tregs may reflect preferential migration of this Treg subpopulation into tumors and/or preferential expansion and survival within the TME. In either case, increased expression of Helios by intratumoral CD4 Tregs may signal enhanced suppressive activity, as judged by expression of a gene profile associated with effector CD4 Tregs. A comparison of gene expression by Helios⁺ and Helios^– FoxP3⁺ CD4 Tregs after separation by surrogate markers (GITR^hiICOS^hi vs. GITR^loICOS^lo) revealed that Helios⁺ FoxP3⁺ CD4 Tregs showed increased expression of KLRG1, GZMB, IL-10, and ICOS, i.e., molecules that are associated with robust suppressive activity (Fig. 1 B and C) (8). These observations together with our recent findings that Helios can stabilize the suppressive CD4 Treg phenotype under inflammatory but not steady-state conditions, suggest that disruption of Helios expression by intratumoral CD4 Tregs might enhance antitumor immunity.

Fig. 1.

Intratumoral CD4 Tregs display increased suppressive phenotype. (A) Analysis of Helios expression by FoxP3⁺ CD4 Tregs in spleen, LNs, and tumor of B16/F10 melanoma-bearing mice. Percent of Helios⁺ FoxP3⁺ CD4 Tregs is shown (n = 3). (B) Comparison of gene expression between Helios⁺ and Helios^–/lo FoxP3⁺ CD4 Tregs. Helios⁺ and Helios^–/lo FoxP3⁺ CD4 Tregs were sorted from spleen of WT B6 mice after staining with Abs for CD3, CD4, CD25, ICOS, and GITR. Expression of Helios, ICOS, and GITR by FoxP3⁺ CD4 cells: ICOS^hiGITR^hi and ICOS^loGITR^lo cells represent Helios⁺ and Helios^–/lo cells, respectively. (C) DNA microarray was performed using the Affymetrix Gene-Chip and gene expression was compared.

Selective Deletion of Helios in FoxP3⁺ CD4 Tregs Enhances Antitumor Immunity.

Previous studies have demonstrated that intratumoral FoxP3⁺ CD4 Tregs express surface molecules, including TIM3 and TIGIT, that are associated with robust immunosuppressive activity as well as dysfunctional tumor-infiltrating lymphocytes (TILs) (3, 9). However, the impact of the Treg-specific Helios TF has not been investigated. Although Helios deficiency promotes an unstable FoxP3⁺ CD4 Treg phenotype under inflammatory conditions (6), whether defective Helios expression by intratumoral Tregs impairs Treg function within this local inflammatory environment is unclear. We therefore analyzed tumor growth in Helios-deficient (Helios^fl/fl.FoxP3-Cre; Helios KO) mice after s.c. inoculation of transplantable melanoma (B16/F10) or colon adenocarcinoma (MC38). Helios KO mice showed substantial delay of tumor growth compared with Helios WT mice, resulting in prolonged survival (Fig. 2 A and B), indicating that Helios expression by FoxP3⁺ CD4 Tregs may normally contribute to Treg-mediated repression of antitumor immunity. Indeed, delayed tumor growth was associated with increased IFNγ production by CD8⁺ TILs in Helios KO mice compared with TILs from Helios WT mice (Fig. 2C).

Fig. 2.

Mice with Helios deficiency in CD4 Tregs show enhanced antitumor immunity. (A and B) Helios^fl/fl and Helios KO mice were injected s.c. with 2 × 10⁵ B16/F10 (A) or MC38 (B) and tumor growth and survival of mice were monitored. (C) Enhanced IFNγ production by CD8 Teff cells in B16/F10 or MC38 tumors from Helios KO mice. Two weeks after tumor inoculation, lymphocytes were enriched from tumor cell samples and stimulated with PMA and ionomycin in vitro followed by FACS analysis of IFNγ expression by CD8 cells. (D) Helios WT (FoxP3-Cre) and KO (Helios^fl/fl.FoxP3-Cre) mice were inoculated with 2 × 10⁵ B16-Ova, vaccinated with GVAX on days 3, 7, and 9, and tumor growth was monitored. (E) At day 21, effector CD4 and CD8 T cells were analyzed for TNFα production.

Vaccination with GM-CSF–secreting irradiated tumor cells (GVAX) represents a prototypic paracrine cytokine adjuvant that induces differentiation of dendritic cells (DCs) leading to tumor antigen uptake, trafficking to tumor-draining lymph nodes, and enhanced inflammation. Because the contribution of the Helios TF to stable FoxP3⁺ Treg suppressive activity is critically important in inflammatory settings (6), we examined Helios WT and Helios KO mice after s.c. inoculation with B16-Ova melanoma cells and treatment with GVAX at days 3, 7, and 9. This regimen resulted in a substantial decrease in tumor growth in Helios KO mice compared with WT mice, indicating that Helios-deficient CD4 Tregs display reduced immunosuppressive activity (Fig. 2D). Indeed, intratumoral CD4 and CD8 Teff cells in Helios KO mice expressed high levels of TNFα compared with modest levels by TILs from Helios WT mice (Fig. 2E).

Enhanced Antitumor Immunity by Helios KO Mice Is Associated with an Unstable Treg Phenotype.

Helios-deficient CD4 Tregs develop an unstable phenotype during inflammatory responses characterized by reduced FoxP3 expression and increased effector cytokine expression secondary to diminished activation of the STAT5 pathway (6). Growing tumors attract a wide variety of cytokine/chemokine-producing leukocytes that shape the inflammatory microenvironment during tumor progression (10) and promote increased proportions of FoxP3⁺ CD4 cells compared with their frequency in lymphoid tissues. Thus, ∼40% of CD4⁺ cells within B16 melanoma are FoxP3⁺, whereas ∼10% of splenic CD4 cells in tumor-bearing mice are FoxP3⁺ (Fig. 3A). However, the frequency of FoxP3⁺ CD4 Tregs within CD4⁺ TILs in Helios KO mice is not increased (∼10–12%) compared with spleen (Fig. 3A). Moreover, FoxP3⁺ CD4 Tregs isolated from tumors grown in Helios KO displayed a nonanergic phenotype, as judged from decreased ratio between FR4^hiCD73^hi (anergic) and FR4^loCD73^lo (nonanergic) FoxP3⁺ cells (Fig. 3B). This difference was confined to intratumoral CD4 Tregs; splenic FoxP3⁺ CD4 Tregs showed a similar anergic phenotype in tumor-bearing Helios WT and KO mice (Fig. 3B). Consistent with their diminished anergic surface phenotype, Helios-deficient intratumoral CD4 Tregs expressed relatively high levels of the IFNγ effector cytokine, in contrast to WT intratumoral CD4 Tregs, which displayed a stable phenotype and minimal IFNγ production (Fig. 3C). Conversion of Helios-deficient FoxP3⁺ CD4 Tregs to Teff cells was also noted in B16-bearing mice that had been treated with GVAX. Helios-deficient intratumoral Tregs expressed high levels of TNFα, whereas Helios WT intratumoral CD4 Tregs produced marginal levels (Fig. 3D).

Fig. 3.

Unstable phenotype of Helios-deficient CD4 Tregs within tumors. Helios^fl/fl (WT) and Helios KO mice were injected s.c. with 2 × 10⁵ MC38. Two weeks later, splenocytes and intratumoral lymphocytes were analyzed. (A) Percent of FoxP3⁺ cells within TCR⁺CD4⁺ in spleen and tumor is shown. (B and C) Intratumoral but not splenic Helios-deficient CD4 Tregs display a nonanergic phenotype. FoxP3⁺ CD4 cells from spleen and tumor were analyzed for expression of FR4 and CD73 (B) and IFNγ expression after in vitro restimulation with PMA and ionomycin (C). (D) Effector cytokine TNFα expression by CD4 Tregs isolated from tumors in Helios KO mice that were treated as described in Fig. 2D.

The Impact of Helios Deficiency in CD4 Tregs on Antitumor Immunity is Cell Intrinsic.

To determine whether enhanced antitumor immunity displayed by Helios KO mice is Treg intrinsic, we transferred purified Helios WT or Helios KO CD4 Tregs (CD45.2) along with CD4 and CD8 Teff cells (CD45.1) into Rag2^−/− hosts and monitored for tumor growth. Rapid tumor growth was observed in hosts that had received WT CD4 Tregs, whereas tumor development was delayed in hosts that had received CD4 Tregs from Helios KO mice (Fig. 4A). Analysis of CD4 Tregs recovered from adoptive hosts also revealed that Helios-deficient CD4 Tregs displayed reduced FoxP3 expression compared with Helios WT CD4 Tregs (Fig. 4B). FoxP3 down-regulation by Helios-deficient CD4 Tregs within tumors was more pronounced than in spleen, again suggesting that Helios expression by Tregs might be particularly important for stable FoxP3 expression within the chronic inflammatory environment of growing tumors. In accord with FoxP3 down-regulation, intratumoral Helios-deficient CD4 Tregs expressed high levels of IFNγ, suggesting a Treg→Teff cell conversion within the tumor (Fig. 4C). Moreover, intratumoral CD4 and CD8 Teff cells in adoptive hosts transferred with Helios KO CD4 Tregs displayed increased IFNγ expression (Fig. 4D).

Fig. 4.

Isolated function of Helios-deficient CD4 Tregs in antitumor immunity. (A) Rag2^−/− hosts were transferred with purified CD4 and CD8 T cells (CD4 and CD8 Tregs depleted) along with CD4 Tregs isolated from Helios WT (FoxP3-Cre) or KO (Helios^fl/fl.FoxP3-Cre) mice. Two days later, Rag2^−/− hosts were inoculated s.c. with MC38 and tumor growth was monitored. (B) Twenty-one days after cell transfer, cells from spleen and tumor were analyzed for their origin (CD45.1 vs. CD45.2) and FoxP3 expression. (C) IFNγ expression by Helios WT (CD45.2⁺) or KO (CD45.2⁺) CD4 Tregs recovered from spleens of Rag2^−/− hosts was analyzed after in vitro restimulation. (D) IFNγ expression by intratumoral effector CD4 and CD8 T cells from Rag2^−/− hosts that were transferred with Helios WT or KO CD4 Tregs was analyzed after in vitro restimulation.

Helios Deficiency Promotes in Vitro Conversion of Tregs into Teff Cells.

To further investigate the basis for the intrinsic instability of Helios-deficient CD4 Tregs, we used an in vitro system that allows analysis of Treg stability in the presence of inflammatory cytokines IL-2/IL-4 (6). Here we analyzed the responses of Helios^+/+ and Helios^−/− CD4 Tregs to graded concentrations of IL-2 and the proinflammatory cytokine IL-4 (11) (Fig. 5A). Helios-deficient Tregs displayed profoundly reduced expression of both FoxP3 and CD25 and enhanced expression of IFNγ in an IL-2 dose-dependent manner (Fig. 5A); Helios WT CD4 Tregs expressed high levels of FoxP3 and CD25 that were not affected by IL-2 concentrations. Although Helios-deficient CD4 Tregs showed a significant increase in CD25 expression in the presence of higher concentrations of IL-2, it is unlikely that increased CD25 expression accounted for cytokine conversion, because FoxP3^lo cells marked by low CD25 expression produced the highest levels of IFNγ (Fig. 5B). Taken together, these findings indicate that Helios makes a critical contribution to the stability and survival of FoxP3⁺ CD4 Tregs in the presence of inflammatory cytokines in this in vitro assay.

Fig. 5.

Helios deficiency and Treg-to-Teff conversion. (A) Reduced CD25 and FoxP3 expression by Helios-deficient CD4 Tregs upon exposure to IL-4 in vitro. Sorted CD4 Tregs from Helios^+/+ and Helios^−/− mice were stimulated for 4–5 d with coated anti-CD3/CD28 Abs in the presence of IL-2 (0–50 ng/mL) and IL-4 (20 ng/mL) before levels of CD25, FoxP3, and IFNγ expression were measured. Representative data from three independent experiments are shown. (B) Helios-deficient CD4 Tregs from the above cultures (A) were divided into FoxP3^hi and FoxP3^lo and IFNγ expression levels analyzed. (C) CD4 Tregs from WT B6 mice were treated in vitro with DMSO or AG-490 in the presence of IL-4 (20 ng/mL) and increasing concentrations of IL-2 (0–50 ng/mL) for 48 h. Cell numbers and FoxP3 expression were measured. IFNγ production by recovered cells was analyzed by intracellular cytokine staining. (D) FoxP3⁺ CD4 Tregs were isolated from spleens of WT B6 and cultured in anti-CD3/CD28-coated wells in the presence of IL-4, increasing concentrations of IL-2 and anti-GITR (DTA-1) or isotype Abs. After 5 d, cells were analyzed for FoxP3 expression and IFNγ production.

The dependence of CD4 Tregs on IL-2 for survival and lineage stability reflects, in part, an interaction between STAT5 and the FoxP3 intronic CNS2 region that promotes stable FoxP3 expression (12). We previously noted that Helios may control STAT5-dependent stabilization of FoxP3 expression (6). This hypothesis is strengthened by the finding that the unstable phenotype of Helios-deficient Tregs can be induced by blockade of STAT5 activation. The AG-490 STAT5 inhibitor reduced Treg survival, diminished FoxP3 expression, and enhanced IFNγ effector cytokine expression (Fig. 5C). These data taken together suggest that approaches that down-regulate Helios in vivo may enhance tumor immunity via reduction of CD4 Treg numbers and conversion of a portion of the surviving Tregs into an effector cell phenotype.

Engagement of Glucocorticoid-Induced TNF Receptor Induces Helios Down-Regulation by CD4 Treg and Enhanced Antitumor Immunity.

The observation that Helios-deficient CD4 Treg convert to Teff cells within the TME and enhance antitumor responses opened the possibility that an immunotherapeutic regimen that induces Helios down-regulation might be exploited to enhance antitumor immunity. To this end, we used an in vitro Treg conversion assay (Fig. 5D) to screen for Abs that induced conversion by Helios WT Tregs. One Ab identified was specific for glucocorticoid-induced TNF receptor (GITR), a member of the TNF receptor gene family that is prominently expressed by CD4 Tregs compared with other T cells that normally display low expression levels (13). Engagement of GITR on CD4 Tregs by Abs in vitro in the presence of proinflammatory cytokine IL-4 resulted in induction of an unstable Treg phenotype characterized by reduced FoxP3 expression and IFNγ production (Fig. 5D). Because the agonistic Ab DTA-1 has been shown to diminish CD4 Treg activity and enhance antitumor immunity, which has been attributed in part to decreased Treg lineage stability within tumors (14), we tested whether engagement of GITR results in Helios down-regulation and enhanced antitumor immunity. Although repeated administration of a relatively low dose of DTA-1 (200 μg) did not induce obvious side effects (15), both prophylactic and therapeutic regimens significantly delayed tumor growth (Fig. 6A) and were associated with diminished expression of Helios and increased IFNγ production by intratumoral CD4 Tregs (Fig. 6B). Moreover, CD8 Teff cells isolated from tumors in DTA-1–treated mice displayed increased IFNγ and TNFα production compared with CD8 Teff cells from isotype-control–treated mice (Fig. 6C). The phenotypic changes in CD4 Tregs after DTA-1 treatment can potentially reflect a contribution of Teff cells that express GITR after activation. To more directly analyze the isolated effects of DTA-1 on CD4 Tregs, we transferred purified CD4 Tregs (YFP⁺ from FoxP3^YFP-Cre mice) into Rag2^−/− hosts before treatment with DTA-1 or control Ig. Tregs in DTA-1–treated adoptive hosts showed a profound decrease in Helios expression, acquisition of a nonanergic phenotype, and reduced survival (Fig. 6 D and E). Moreover, ∼20% of the (Helios^lo) FoxP3⁺ CD4 Tregs from DTA-1–treated mice produced IFNγ and TNFα effector cytokines, consistent with Treg→Teff conversion (Fig. 6F). However, the dramatic reduction in the anergic phenotype by DTA-1–treated CD4 Tregs indicates the impact of DTA-1 on the reactivity of Tregs is not limited to up-regulation of TNF-α/IFNγ. Together, these data suggest that the enhanced antitumor immunity after administration of DTA-1 reflects, at least in part, induction of an unstable CD4 Treg phenotype and Treg conversion. These data also suggest that Helios may represent a previously unrecognized target for cancer immunotherapy in light of its impact on intratumoral Treg stability.

Fig. 6.

Engagement of GITR induces Helios down-regulation by CD4 Tregs. (A) WT B6 mice inoculated s.c. with B16/F10 and anti-GITR or isotype Abs were injected on days 3, 6, and 9 (prophylactic treatment, Left) or on days 10, 12, 14, and 16 (therapeutic treatment, Right). Tumor growth was monitored. (B) Fourteen days later, FoxP3⁺ CD4 cells from spleens and tumors of tumor-bearing mice were analyzed for IFNγ production after in vitro restimulation with PMA and ionomycin. (C) Intratumoral CD8 Teff cells were analyzed for IFNγ and TNFα production after in vitro restimulation with PMA and ionomycin. (D and E) Rag2^−/− hosts were transferred with Helios WT (FoxP3-Cre) or Helios KO (Helios^fl/fl.FoxP3-Cre) CD4 Tregs. Rag2^−/− hosts were injected with anti-GITR (DTA-1) or isotype control Ab on days 0, 7, 14, and 20. On day 21, Tregs from spleen cells were analyzed for recovery, Helios expression, and anergic phenotype (FR4 and CD73). (F) Spleen cells were analyzed for expression of effector cytokines (TNFα and IFNγ) after in vitro stimulation with PMA and ionomycin.

SI Materials and Methods

Mice.

C57BL/6J (B6) and B6.129(Cg)-Foxp3^{tm4(YFP/icre)Ayr}/J (FoxP3^YFP-Cre) mice were purchased from The Jackson laboratory. B6.129S6-Rag2^tm1Fwa N12 (B6 Rag2^−/−) and B6.SJL-Ptprc^a/BoyAiTac (B6 CD45.1⁺) were from Taconic Farms. Ikzf2^fl/fl (Helios^fl/fl) mice, which bear a loxP-flanked Helios allele, were kindly provided by Ethan Shevach (25). Helios^fl/fl.FoxP3^YFP-Cre mice were generated by crossing Helios^fl/fl mice to FoxP3^YFP-Cre mice in our animal facility.

Cell Lines.

B16/F10 melanoma cells were purchased from American Type Culture Collection. B16-OVA and B16-GVAX (B16-GM-CSF) were maintained in complete Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific) containing 10% (vol/vol) FCS (Sigma-Aldrich) and 250 μg/mL of G418 (Thermo Fisher Scientific). MC38 colon cancer cells were cultured in complete RPMI-1640 (Sigma-Aldrich) containing 10% (vol/vol) FCS. All tumor cell lines were maintained at 37 °C with 5% CO₂.

Isolation of Tumor-Infiltrating Lymphocytes.

Mice were killed before tumor sizes reached 2,000 mm³ and analyzed. Harvested tumors were mechanically chopped and dispersed into small pieces followed by collagenase digestion for 1 h with 50 units/mL collagenase type I (Thermo Fisher Scientific) and 20 units/mL DNase I (Roche). Digested samples were filtered and enriched for tumor-infiltrating lymphocytes by centrifugation through a Ficoll-Paque 1.084 density gradient (GE Healthcare).

Flow Cytometry and Cell Sorting.

Fluorescence dye conjugated monoclonal antibodies specific for CD4 (RM4-5), CD8 (53-6.7), CD25 (PC61), TCR V_β (H57-597), FoxP3 (FJK-16s), Helios (22F6), GITR (DTA-1), ICOS (7E.17G9), IFNγ (XMG1.2), TNFα (MP6-XT22), FR4 (12A5), CD73 (TY/11.8), and CD45.1 (A20) were purchased from BD Bioscience, eBioscience, or BioLegend. IFNγ and TNFα were detected after restimulation of cells in vitro with leukocyte activation mixture with BD GolgiPlug (BD Bioscience) for 5 h. Stimulated cells were stained for surface markers first, then fixed, permeabilized using the FoxP3 staining buffer set (eBioscience) and stained with antibodies for cytokines. Samples were measured by BD LSRFortessa X-20 (BD Bioscience) and data were analyzed using FlowJo v10 (FlowJo). For CD4 Treg isolation, cells were enriched for CD4⁺CD25⁺ cells using a CD4 Treg enrichement kit (Miltenyi) followed by sorting for CD4 Tregs using BD FACSAria IIIu (BD Bioscience).

Antibody Treatment.

Anti-GITR monoclonal antibody (clone: DTA-1) and isotype control (Rat IgG2b clone: LTF-2) were purchased from Bioxcell. For prophylactic treatment, 200 μg of antibody was i.v. injected into the tail vein of mice at day 0, 3, 6, and 9 after tumor cell injection.

Cell Purification and Adoptive Transfer.

CD4⁺CD25⁻ effector cells were negatively isolated from spleens of CD45.1 mice using a Mouse CD4 T Lymphocyte Enrichment Set supplemented with biotinylated anti-CD25 antibodies (BD Bioscience). CD8⁺Ly49⁻ effector cells were negatively isolated from spleens of CD45.1 mice using a Mouse CD8 T Lymphocyte Enrichment Set (BD Bioscience) supplemented with biotinylated anti-Ly49C/I/F/H antibodies (14B11). Purity of CD4 and CD8 cells was >90%. CD4 Tregs were obtained from spleens of Helios^fl/fl.FoxP3^YFP-Cre and FoxP3^YFP-Cre (Helios WT) mice by sorting TCR⁺CD4⁺YFP⁺ cells after enrichment using a CD4⁺CD25⁺ Regulatory T Cell Isolation Kit (Miltenyi Biotec). CD4 Treg purity was >95%. The 5 × 10⁵ CD4 Tregs were transferred i.v. into Rag2^−/− hosts along with 2 × 10⁶ CD4 and 1 × 10⁶ CD8 T effector cells on day 0. To establish tumors, 2 × 10⁵ MC38 tumor cells were inoculated s.c. on day 2, and tumor growth was monitored.

In Vitro Stimulation of CD4 Tregs.

CD4 Tregs were isolated from Helios^+/+ and Helios^−/− mice (11) using a CD4⁺CD25⁺ Regulatory T Cell Isolation Kit (Miltenyi Biotec) followed by sorting for CD4⁺CD25⁺ cells. CD4 Treg purity was > 95%. Sorted CD4 Treg were cultured on a 96-well flat bottom plate coated with anti-CD3 (17A2, eBioscience) and anti-CD28 antibody (37.51, eBioscience) in the presence of IL-4 (20 ng/mL) and IL-2 (0–50 ng/mL) (eBioscience) for 4–5 d before flow cytometry analysis. For the in vitro STAT5 inhibition assay, isolated CD4 Tregs were cultured with DMSO or AG490 (50 μM) (Sigma-Aldrich).

CD4 Treg Transfer into Rag2^−/− Mice and Antibody Treatment.

CD4 Tregs were isolated from FoxP3^YFP-Cre (Helios WT) mice as described above and transferred into Rag2^−/− hosts. DTA-1 or isotype Abs were injected i.v. via tail vein on days 0, 7, 14, and 20. Spleens were harvested on day 21 and analyzed by flow cytometry.

Discussion

The definition of immunoinhibitory pathways that are up-regulated after T-cell activation has led to significant insight into tumor escape mechanisms. Many of these findings have been incorporated into approaches that combine checkpoint blockade with immunostimulatory agents that can promote sustained antitumor immune responses (16). In some cases, these approaches also may affect Treg function. For example, the antitumor activity of some anti–CTLA-4 Abs may reflect FcγR-dependent depletion of intratumoral Tregs in addition to targeting of Teff cells (17). Likewise, PD-1–based approaches may affect both Treg suppressive activity as well as effector T-cell responses (18). Because depletion of Treg activity may also produce adverse autoimmune side effects (19), approaches that preferentially target intratumoral Tregs without affecting the Treg systemic phenotype potentially represent a more effective strategy for cancer immunotherapy.

Here we show that a selective deficiency of Helios in FoxP3⁺ CD4 Tregs results in increased Treg instability and conversion of intratumoral CD4 Treg to Teff cells and enhanced antitumor immunity. Instability of intratumoral Tregs may increase the numbers of Teff cells within tumors as a combined result of Treg conversion and reduced Treg suppressive activity. In addition, defective IL-2 responses of Helios-deficient intratumoral Tregs resulting in decreased numbers of activated Tregs may also contribute to increased intratumoral effector T-cell activity. Interaction between tumor cells and infiltrating immune cells results in secretion of inflammatory mediators, including TNF-α, IL-6, IL-17, IL-1, and TGF-β, and the formation of a local inflammatory environment. Although the signaling pathway(s) that leads to effector cell conversion of Helios-deficient Tregs has not been identified, cytokine-mediated inflammation, competition for limited amounts of IL-2, and hypoxic conditions within the TME may promote conversion within the TME but not peripheral tissues, perhaps by skewing the pSTAT5/pSTAT3 ratio bound to Treg-specific demethylated regions (6, 12). Because conversion of Helios-deficient Tregs occurs within the local inflammatory environment of the tumor (e.g., Figs. 3 and 4), this approach may not provoke the autoimmune side effects associated with systemic reduction of Tregs (12, 16).

Although thymic-derived Tregs that recognize tissue-specific antigens expressed by tumors and their parent tissues may be highly represented within tumors (20), under normal conditions, this autoreactive TCR repertoire does not translate into robust antitumor responses. Our data suggest that this tumor recognition bias of Treg may be exploited by approaches that induce Treg conversion into MHC class II/peptide-reactive effector cells that directly kill tumor cells (21–23). These considerations suggest that protocols for transfer of TAA-specific CD4 T cells may benefit from approaches that down-regulate Helios expression by CD4 Tregs to obtain increased antitumor reactivity from both conventional CD4 cells and Helios-deficient converted Tregs.

This study also suggests that Treg→Teff conversion of Helios-deficient Tregs within the local inflammatory setting of growing tumors can be mimicked by Ab-dependent engagement of surface receptors that down-regulate Helios expression. An in vitro screen of Abs that might reduce Helios expression and enhance Treg conversion suggested the potential contribution of GITR to this process. The impact of GITR Abs on tumor immunity has been described and may depend in part on engagement of GITR⁺ Tregs (14). We note that this TNFR costimulatory molecule, which is constitutively and highly expressed on Tregs and induced after activation of Teff cells, induces Helios down-regulation and Treg conversion that is restricted to tumor sites. Our CD4 Treg transfer experiments also suggest that potentiation of antitumor immunity by anti-GITR Ab administration can be attributed largely to induction of an unstable Treg phenotype.

Administration of anti-GITR Ab may lead to untoward side effects in mice (15); however, relatively low doses of anti-GITR Ab (200 μg) suffice to induce relatively selective Treg conversion, in view of the relatively low GITR levels expressed by activated Teff cells (13). Although the mechanism that translates GITR ligation into an unstable nTreg phenotype is unknown, some recent studies may be relevant. Ligation of GITR inhibits induced-Treg (iTreg) generation and diverts these cells to the Th9 effector cell lineage, in part by inducing chromatin remodeling at the Foxp3 and Il9 loci via regulation of histone acetylation (24). Induction of STAT6 after GITR ligation of iTregs also suggests that a similar GITR-dependent pathway in nTregs may enhance STAT6 competition with STAT5 for binding to CNS2, leading to diminished FoxP3 expression (12). These findings suggest that administration of relatively low doses of anti-GITR Ab to impair Treg activity, perhaps in combination with T-cell–activating immunotherapy, may yield strong antitumor immunity. In sum, we anticipate that strategies that specifically harness Helios-dependent control of the intratumoral Treg phenotype hold significant promise for improving cancer immunotherapy.

Materials and Methods

Mice and Treatment.

All mice were purchased from The Jackson Laboratory or Taconic Farms and maintained in specific pathogen-free conditions in the Dana-Farber Cancer Institute (DFCI) Animal Resource Facility. For tumor induction in B16/F10 and MC38 transplantable tumor models, mice were injected with 2 × 10⁵ cells s.c. in the right flank. In the B16-GVAX model, mice were injected s.c. with 2 × 10⁵ B16-OVA cells in the right flank followed by vaccination with 1 × 10⁶ irradiated (150 Gy) B16-GMCSF cells on the contralateral flank at day 3, 6, and 9. Tumor growth was monitored every 2 d and tumor volume was calculated by: tumor volume (mm³) = longest diameter (mm) × shortest diameter (mm) × width (mm)/2. All animal protocols in this study were approved by DFCI’s Animal Care and Use Committee, and all animal experiments were performed in compliance with federal laws and institutional guidelines.

DNA Microarray.

Helios⁺ and Helios^– CD4⁺CD25⁺ T cells were separated by sorting cells after staining with surrogate markers ICOS and GITR (ICOS^hiGITR^hi: Helios⁺; ICOS^loGITR^lo: Helios^–/lo). RNA was prepared with the RNeasy mini kit (Qiagen). RNA amplification, labeling, and hybridization to MOA430 2.0 chips (Affymetrix) were performed at the DFCI Molecular Biology Core Facility. Relative gene expression by ICOS^hiGITR^hi and ICOS^loGITR^lo cells was analyzed by the Multiplot program.

Statistical Analysis.

Statistical significance was calculated according to the Wilcoxon–Mann–Whitney rank sum test. A P value of <0.05 was considered to be statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001).