Interleukin 12 Stimulates IFN-γ–Mediated Inhibition of Tumor-Induced Regulatory T-Cell Proliferation and Enhances Tumor Clearance

Clinical observations and animal models have established that the immune system responds to tumors, although they are derived from the host. Despite the observations that activated, tumor-specific immune cells exist in cancer patients and animal models, the immune system often fails to prevent tumor formation and metastasis (1). Thus, established tumors must have developed strategies to inhibit or evade the immune system.

Sakaguchi and colleagues (2, 3) first established the essential role of CD4⁺CD25⁺ regulatory T (Treg) cells in the maintenance of peripheral tolerance to self-antigens. More recent studies have implicated Treg cells in inducing tolerance to tumors (4, 5). Mounting evidence indicates that tumors can promote the expansion, recruitment, and activation of Treg cells (6). Increased numbers of CD4⁺CD25⁺Foxp3⁺ Treg cells have been found either in the circulation or in the tumors of patients with various cancers (lung, breast, colorectal, esophageal, gastric, ovarian, pancreatic, melanoma, hepatocellular, leukemia, and lymphoma; ref. 7). In addition, large numbers of Treg cells in tumors and malignant ascites were shown to be correlated with worse clinical outcomes in several types of cancer (8). Likewise, studies using in vivo mouse models have shown that Treg cell proliferation/activation induced by tumors or self-antigens can help tumors to escape immunosurveillance (9, 10). Furthermore, antigen-based tumor vaccines have also been found to activate Treg cells, which can considerably dampen antitumor immune responses (11).

Extensive investigations have improved our understanding of the molecular and cellular basis for the development, activation, and function of Treg cells. Antigen T-cell receptor signals are not only determinants for thymic selection and differentiation of Treg cells (12, 13), but they also control their peripheral activation and suppressive efficacy (14, 15). Although Treg cells predominantly recognize self-antigens and suppress self-reactive T cells (12, 13, 16), it has also been shown that they can respond to nonself-antigens, such as viral antigens, allo-antigens, and foreign antigens in transgenic mouse models (17–19). Cytokine signals also regulate Treg cell development, function, and homeostasis. Interleukin (IL)-2 has been shown to be indispensable for Treg expansion, activation, and maintenance (20, 21). Other γ-chain cytokines (IL-4, IL-7, and IL-15) can also support Treg cell development and function (22, 23). IL-10 has been reported to be a soluble mediator of Treg suppressive function (5, 7). Transforming growth factor-β (TGF-β) has been shown to promote the proliferation of natural Treg cells, and also the “conversion” of non-Treg cells into induced Treg cells (5, 7).

We recently showed that Treg cells in the tumor environment (but not Treg cells in the peripheral lymphoid tissues) use granzyme B to suppress antitumor immunity (24), which suggests that Treg cells are subject to regulation by local factors during antitumor immune responses. To identify such factors, we used the Foxp3-GFP reporter mice in this study, and developed an in vitro coculture system to study Treg cells in the tumor setting. We found that tumor cells indirectly drive Treg cell proliferation via accessory immune cells. We also discovered, inadvertently, that IL-12 profoundly inhibits tumor-induced Treg cell proliferation; this inhibition is partially dependent on IFN-γ. This study provides definitive evidence that an IL-12- IFN-γ axis can inhibit Treg cell proliferation. These findings show that Treg cells are subject to cytokine-mediated inhibition and support a notion that cytokines with such a property may have the potential to serve as an adjuvant for tumor vaccine-based immunotherapy.

Wild-type (WT) 129/SvJ, and WT and IFNγR1−/− mice in the C57Bl/6J strain were obtained from the Jackson Laboratory. WT 129/SvPas mice were obtained from the Charles River Laboratory. The IL12Rβ2−/− mice in the C57Bl/6J strain were maintained as described (25). The IFNγR1−/− mice in the 129/SvPas mice were obtained from Dr. Herbert Virgin (Washington University, St. Louis, MO). The RMAS lymphoma cells, originally derived from C57BL/6 mice (H-2^b, I-A^b), were obtained from Dr. Wayne Yokoyama (Washington University, St. Louis, MO) and maintained in complete K10 medium (24). The primary APL samples are cryopreserved cells harvested from the spleens of three independent leukemic mCG-PML-RARα transgenic mice in pure C57BL/6 strain (26, 27). All mice were bred and maintained in specific pathogen-free housing, and all experiments were conducted in accordance with Roswell Park Cancer Institute and Washington University School of Medicine animal care guidelines, using protocols approved by the animal studies committee.

The purified CD25 antibody (PC61) and rat IgG1 isotype control (HRPN) were purchased from Bio Express. The CD4⁺CD25⁺ Treg isolation and Pan T-cell isolation kits, and T-cell depletion (CD90) microbeads were purchased from Miltenyi Biotec. T-cell isolation or depletion was performed by following the manufacturer's instruction manuals. Antibodies used include anti-mouse CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD25 (7D4, PC61), CD16/32 (2.4G2; BD Biosciences), and Foxp3 (eBioscience, FJK-16s).

Unfractionated splenocytes (4 × 10⁶) were harvested from naïve mice and mixed with 1.2 × 10⁵ RMAS tumor cells (3,000 Rads irradiated) in 2 mL of K10 medium in tissue culture–treated 12-well plates for 4 to 5 d. To study exogenous cytokine effect, various doses of cytokines were added in the beginning of the coculture experiments. At the beginning and the end of cocultures, absolute numbers of various T-cell compartments were obtained by multiplying the percentages (determined by flow cytometry as described below, and represented in Supplementary Fig. S1) by the total numbers of live cells harvested from the cultures. T-cell expansion is presented as fold change by the equation below:

Cells (1 × 10⁶) were washed and resuspended in staining buffer (PBS, 0.5% bovine serum albumin, 0.5 mmol/L EDTA). Samples were labeled with primary-conjugated antibodies against cell surface markers (anti-mouse CD3, CD4, CD8, CD25; BD Biosciences). Samples were fixed, permeabilized (Foxp3 staining kit, eBioscience), and stained with primary-conjugated anti-Foxp3 antibody (FJK-16s; eBioscience) diluted at 1:400 in staining buffer. Sample data were acquired on a Cytek-modified FACScan (Becton Dickinson) flow cytometer, and analyzed using FlowJo (TreeStar) software.

CD4⁺CD25⁺ Treg cells were isolated from the resting spleens of WT mice, IL12Rβ2−/− mice, and IFNγR1−/− mice using the CD4⁺CD25⁺ Treg isolation kit, according to the manufacturer's instructions (Miltenyi Biotec). The purified Treg cells were labeled at 37°C in PBS with 150 nmol/L of carboxyfluorescein diacetate succinimidyl ester (CFSE) for 15 min and mixed into WT splenocytes, or IL12Rβ2−/− splenocytes, or IFNγR1−/− splenocytes cultured with irradiated RMAS cells for 4 to 5 d.

RMAS cells were transduced with a ΔU3 retroviral construct that drives the expression of a fusion cDNA encoding click beetle red luciferase and green fluorescence protein (GFP) as previously described (24). Click beetle red luciferase–expressing RMAS clones were injected i.p. into WT mice in 200 μL RPMI 1640 on day 0, and bioluminescence imaging was performed 2 h later and variously through 10 d after injection. For imaging, mice were injected i.p. with D-luciferin (150 μg/g body weight), anesthetized (2% isoflurane), and placed in an IVIS 50 imaging system (Caliper; exposure time, 1–30 s; binning, 2–8; no filter; f/stop, 1; FOV, 12 cm; 28). Regions of interest were defined manually over the whole body using Living-Image software (Igor Wavemetrics) for determining tumor burden signal intensities. Data were expressed as photon flux (photons/second) across all mice receiving the same treatment.

Naïve, unfractionated splenocytes from Foxp3-GFP knock-in mice were cocultured with irradiated RMAS cells for 4 d as described above. Lymphocytes were harvested from the cocultures and stained with primary-conjugated antibodies against cell surface markers (CD3, CD4, and CD8) and 7-AAD to assess cell death in Treg and other T-cell compartments by flow cytometry.

To identify factors that may modulate Treg cell function, we explored several approaches to study Treg cells in vitro in a tumor relevant setting. First, both purified Foxp3-GFP⁺ cells (from Foxp3-GFP reporter mice) and CD4⁺CD25⁺ cells failed to proliferate, and the majority of them died when cocultured with RMAS lymphoma cells (live or lethally irradiated) for 4 to 5 days (Fig. 1A). Second, when unfractionated splenocytes (which contain lymphocytes and accessory cells) were cocultured with live tumor cells, the tumor cells outgrew the lymphocytes (data not shown). Finally, we found that coculturing irradiated RMAS cells with unfractionated splenocytes induces significant expansion of CD4⁺Foxp3⁺ Treg cells, but not of CD4⁺Foxp3⁻ or CD8⁺ T cells (Figs. 1A and 2); this expansion occurs in K10 medium without any exogenous growth factors. To rule out an RMAS-specific effect, and to determine whether primary tumor cells can induce Treg expansion in the cultures, we used cryopreserved primary tumor cells derived from the splenic tumors of mice with acute promyelocytic leukemia (APL; refs. 26, 27). Similar amounts of Treg expansion occurred with three independent primary APL tumor samples in the cocultures (Fig. 1B).

Notably, RMAS cells and the primary APL cells used in this study (derived from B6 mice, and sharing MHC I and II with 129 mice, but expressing different minor histocompatibility antigens) induced greater Treg expansion when cocultured with 129/Sv splenocytes (up to 4-fold expansion), compared with B6 splenocytes (∼2-fold expansion; Fig. 1A). Both 129/SvJ and 129/SvPas splenocytes supported similar amounts of Treg expansion.

Because tumor-induced Treg expansion did not occur with purified Treg cells, we performed a series of experiments to define the cells and/or factors that promote tumor-induced Treg expansion. First, we used MACS separation techniques to fractionate total splenocytes into Pan T cells versus T cell–depleted accessory cells. Treg expansion occurred only when both non–T splenocytes and irradiated tumor cells were added to the cultures (Supplementary Fig. S2A and B). Second, because IL-2 and TGF-β have been shown to be relevant for Treg expansion, we used monoclonal antibodies to neutralize IL-2 and TGF-β in the tumor-splenocyte cocultures. IL-2 neutralization completely blocked Treg expansion (Supplementary Fig. S2C and D), whereas TGF-β neutralization had no effect. These results suggest that the accessory non–T splenocytes and IL-2 are both required for tumor-induced Treg expansion in vitro.

From the coculture experiments, we hypothesized that tumor cells may indirectly drive the expansion of Treg cells via antigen presentation and cytokine production by other splenocytes. To identify factors that may modulate Treg cell function in tumor-bearing mice in vivo, we used an array-based approach. We performed gene expression profiling on Treg cells using Affymetrix MOE430v2 arrays. Naïve Treg cells were purified from the resting spleens of Foxp3-GFP reporter mice using high-speed flow sorting on the GFP signal. Tumor-associated Treg cells were purified from the tumor ascites of Foxp3-GFP reporter mice injected 5 days earlier with RMAS cells, whereas Spleen^T Treg cells were purified from the spleens of the same tumor-bearing mice. RMA normalization was performed for all microarray data. Two-group comparison (t test) of Ascites versus Naïve samples, with multiple test corrections (adjusted P value/false discovery rate = q-value) was performed. Using these approaches, we have identified 419 probesets with q < 0.005 and linear fold change of ≥2 (Ascites/Naïve) and 114 probesets with q < 0.005 and linear fold change of ≤0.5 (Ascites/Naïve; Supplementary Table and Supplementary Fig. S8). By comparing cytokine receptor expressions of tumor-associated Treg cells to that of naïve or Spleen^T Treg cells, we identified several candidate cytokines, as shown in Supplementary Fig. S3. IL-2 has been shown to be essential for Treg homeostasis and function (15, 29, 30); as expected, IL-2 receptor α (CD25) is upregulated in tumor-associated Treg cells. IL-7 receptor is also upregulated, whereas IL-10 receptor mRNA abundance in Treg cells is unchanged. Neuropilin-1, a known vascular endothelial growth factor (VEGF) receptor that has been proposed as a Treg cell surface marker, is downregulated in the tumor-associated Treg cells (31). In addition, other pathways involved in Treg function are also differentially regulated. For example, granzyme B is the third most upregulated gene (39-fold), consistent with our previous report that tumor-associated Treg cells activate granzyme B to suppress antitumor immune response (24). Multiple cell division genes (Cdc6, Cdca5, Ccna2, Cdca8) are among the top 30 upregulated genes (Supplementary Table), suggesting that tumor-associated Treg cells may undergo active proliferation. These profiles suggest a complex network in which multiple pathways may influence the expansion and/or function of tumor-associated Tregs cells.

We tested the candidate cytokines for their activity in the tumor-splenocyte coculture system. IL-2 treatment, at doses between 0.1 to 20 ng/mL, mildly increases Treg expansion (data not shown). Although VEGF blockade was shown to reduce intratumoral Treg cells and enhance granulocyte macrophage colony-stimulating factor–stimulated tumor rejection (32), VEGF treatment revealed no effect on Treg expansion at doses ranging from 1 to 1,000 ng/mL. Likewise, most of the other candidates, such as IL-10 and tumor necrosis factor α, show no effect on Treg expansion, although multiple concentrations (directed by their levels in the tumor ascites) were tested individually and in combination with two to three candidates (data not shown).

Remarkably, addition of IL-12 to the culture, at doses between 0.1 and 2 ng/mL, significantly inhibited the expansion of CD4⁺Foxp3⁺ Treg cells in a dose-dependent fashion (Fig. 2A), but had no significant effect on the numbers of CD4⁺Foxp3⁻ or CD8⁺ T cells (Fig. 2). Similar IL-12 effects were observed when the primary APL tumor cells were used in the coculture assay (data not shown), ruling out a tumor cell type–specific effect. Notably, this finding is consistent with a previous report that IL-12 treatment reduced CD4⁺CD25⁺ T cell infiltration in a lung tumor model (33).

To further validate the IL-12 phenotype, we analyzed Treg expansion in IL-12 receptor β2–deficient (IL-12Rβ2−/−) mice. Treg cell numbers in the resting spleens of IL-12Rβ2−/− mice are comparable with that of strain-matched WT mice (Fig. 3A). Using IL-12Rβ2−/− splenocytes, we found that these cells are completely resistant to IL-12–mediated suppression (Fig. 3B). However, the lack of detectable IL-12 receptor mRNA in Treg cells suggested an indirect mechanism of IL-12 action (Supplementary Fig. S3). To test this hypothesis, we used CFSE to label CD4⁺CD25⁺ Treg cells purified from WT and IL-12Rβ2−/− mice, and measured the expansion of CFSE⁺ Treg cells mixed with WT splenocytes or IL-12Rβ2−/− splenocytes in coculture assays. In WT cocultures, WT Treg cells and IL-12Rβ2−/− Treg cells are equally susceptible to IL-12–mediated inhibition (Fig. 3C); in IL-12Rβ2−/− cocultures, both WT Treg cells and IL-12Rβ2−/− Treg cells are equally resistant to IL-12–mediated inhibition (Fig. 3D). These data suggest an indirect mechanism, in which certain IL-12–responsive cell types and/or cytokines mediate the inhibitory effect on Treg cells.

Interestingly, a subset of Treg cells were found to retain high levels of CFSE, whereas other subpopulations of Treg cells seemed to be more tumor reactive and diluted CFSE signal to lower levels by proliferating more extensively (Supplementary Fig. S4).

It has previously been shown that IFN-γ produced by CD8⁺ T cells may regulate the generation/activation of CD4⁺CD25⁺ Treg cells in vivo (34). IL-12 is known to induce IFN-γ production by other immune cells, including CD4⁺ and CD8⁺ T cells (35, 36). As expected, we found that Treg cells do express IFN-γ receptor 1 mRNA and protein (Supplementary Figs. S3 and S4A). Therefore, we used an IFN-γ–neutralizing antibody (H22) to determine whether IFN-γ is responsible for IL-12–mediated Treg inhibition. IFN-γ neutralization increased Treg cell expansion by >30% in WT cocultures in the absence of exogenous IL-12 (Fig. 4B), suggesting that a basal level of IFN-γ (produced by other immune cells in the cultures) inhibits Treg cell expansion. In addition, IFN-γ neutralization partially abolished IL-12–mediated Treg inhibition. Similarly, in the absence of exogenous IL-12, Treg cells in IFN-γ receptor 1–deficient (IFNγR1−/−) cocultures proliferated >60% more efficiently than Treg cells in strain-matched WT cocultures; Treg cells in the IFNγR1−/− cocultures are more resistant to IL-12–mediated inhibition (Fig. 4C). To determine the IFN-γ levels that inhibit Treg expansion, we performed ELISA using cell-free media harvested from the cultures. Splenocytes in the culture secreted a basal level of IFN-γ (0.2–0.9 ng/mL) in the absence or presence of tumor cells (Supplementary Fig. S5A). Exogenous IL-12 treatments enhanced IFN-γ secretion to 5 to 17 ng/mL. Furthermore, adding 1 to 100 ng/mL exogenous IFN-γ to WT cocultures inhibits Treg expansion by 30% to 50% (Fig. 4D). As expected, Treg cells in the IFNγR1−/− cocultures are completely resistant to IFN-γ–mediated inhibition (Fig. 4D), but partially resistant to IL-12 inhibition (Fig. 4C). Together, these data show that IFN-γ signaling partially accounts for IL-12–mediated inhibition of tumor-induced Treg expansion.

However, these data also suggest that there may be additional component(s) that mediate IL-12 inhibition of Treg expansion. Because neutralization of IL-2 completely blocks Treg expansion (Supplementary Fig. S2), we tested whether IL-12 treatment could influence IL-2 secretion. Interestingly, higher levels of IL-2 were detected in the splenocyte-only cultures (210–320 pg/mL) compared with the tumor-splenocyte cocultures (30–70 pg/mL), where the higher levels of IL-12, in the absence of tumor cells, did not correlate with significant Treg expansion. Nevertheless, IL-12 treatment significantly reduced IL-2 secretion in the cocultures (Supplementary Fig. S5B), which may account for the IFN-γ–independent inhibition of Treg cells.

To determine whether IFN-γ receptor–mediated signaling in Treg cells directly inhibits proliferation, we used CFSE to label CD4⁺CD25⁺ Treg cells purified from WT or IFNγR1−/− mice, and compared their proliferative activity in the WT splenocyte-based cocultures. IFNγR1−/− Treg cells proliferate significantly faster than WT Treg cells in the absence of exogenous IL-12 (Fig. 5A), and are significantly more resistant to IL-12 treatment. These data suggest that IFN-γ receptor–dependent signaling in Treg cells directly inhibits proliferation, and is required for IL-12–mediated maximal inhibition of Treg cell expansion. In addition, cell cycle analyses revealed that IL-12 treatment significantly reduces proliferating Treg cells in the S phase (Fig. 5B and C). In contrast, IFNγR1−/− Treg cells display significantly more cells in S phase in the absence of exogenous IL-12, and are more resistant to the cell cycle arrest caused by IL-12 treatment. Furthermore, IFN-γ treatment directly reduced proliferating Treg cells in the S phase (Fig. 5D), although to a lesser extent compared with IL-12 treatment that induced more intensive inhibition due to an additional mechanism, which may be caused by reduced secretion of IL-2 (Supplementary Fig. S5B). Interestingly, these treatments seem to “trap” more Treg cells in the G₂-M phase (Fig. 5B). It has been reported that disruption of T-cell receptor signaling (37), or IL-2 deprivation (38), can cause G₂-M arrest in CD4⁺ T cells, and that IFN-γ treatment can also induce G₂-M arrest in human cell lines (39). Our data suggest that lack of stimulation may account for G₂-M accumulation in the splenocyte-only cultures, whereas IL-12 treatment apparently resulted in an elevation of IFN-γ and a reduction of IL-2 (Supplementary Fig. S5) that may jointly lead to increased G₂-M arrest.

To study the impact of IL-12 on tumor growth in vivo, we first tested whether IL-12 could directly influence tumor cell proliferation or survival. We used various doses of IL-12 (0.1, 0.5, 1, 2, and 5 ng/mL) to treat RMAS cells for 2 weeks in culture. Fresh IL-12 was replenished at each passage of the culture. These treatments showed no effect on RMAS cell proliferation or survival, ruling out a direct effect of IL-12 on tumor growth.

To test whether IL-12–mediated inhibition of Treg cell proliferation has an impact on tumor clearance in vivo, we compared the effect of IL-12 treatment with that of Treg cell depletion. We used luciferase-expressing RMAS cells and performed bioluminescence imaging to monitor mice injected i.p. with these cells. Daily IL-12 treatment (100 ng per mouse) significantly enhances tumor clearance in (Supplementary Fig. S6; Fig. 6A). As expected, Treg cell depletion also increases tumor clearance. Notably, IL-12 treatment enhances tumor clearance more efficiently than Treg cell depletion, perhaps because it also activates immune effector cells (33). Finally, the combination of Treg depletion with IL-12 treatment seems to have a similar effect as IL-12 treatment alone (Fig. 6A), perhaps because IL-12 treatment alone has blocked Treg expansion in vivo. To test this hypothesis, we measured absolute T-cell numbers in the peritoneum of naïve mice versus mice i.p. injected with RMAS cells. CD4⁺Foxp3⁺ Treg cells are increased ∼20-fold in the tumor-bearing 129/SvJ mice, and IL-12 treatment almost completely blocks the increase (Fig. 6B). Similar tumor-induced increases in the populations of CD4⁺Foxp3⁻ and CD8⁺ effector T cells were also detected. IL-12 treatment also reduced the number of effector T cells, perhaps because it reduced tumor burden, which drove the increase of effector T cells in this model.

To define the effect of IL-12 and IFN-γ on tumor growth in vivo, we used RMAS cells to challenge IL-12Rβ2−/− and IFNγR1−/− mice that are in syngeneic C57BL/6J background. We injected mice with a tumor dose that caused ∼50% death of WT mice. As shown in Fig. 6C, all the WT mice treated by exogenous IL-12 were rescued from death caused by tumor growth; endogenous IL-12 was not sufficient to impact tumor growth because IL-12Rβ2−/− mice showed a survival rate similar to that of WT mice. However, all the IFNγR1−/− mice succumbed to tumor growth, indicating that endogenous IFN-γ is required for tumor immunity in this system. Furthermore, IL-12 treatment did not improve the survival of the IFNγR1−/− mice, suggesting that the antitumor effect of IL-12 is dependent on IFN-γ signaling. We also examined the effect of IL-12 and IFN-γ on Treg expansion in vivo. As shown in Fig. 6D, Treg expansion was observed in WT mice after tumor challenge, which was completely blocked by exogenous IL-12 treatment. Treg cells in IL-12Rβ2−/− mice expanded to levels comparable with that in WT animals, suggesting that endogenous IL-12 was not sufficient to control Treg expansion. As expected, significantly higher levels of Treg expansion were observed in IFNγR1−/− mice, which was partially inhibited by IL-12 treatment. However, this partial inhibition of Treg expansion by exogenous IL-12 did not improve the survival of IFNγR1−/− mice, perhaps because IFNγR1 deficiency disabled effector lymphocytes, which require normal IFN-γ function to clear the tumor cells (Fig. 6C).

In summary, these in vivo observations now confirm the in vitro data, and strengthen the role of the IL-12–IFN-γ axis in regulating Treg cells and tumor immunity.

In this study, we developed an in vitro system to study factors that influence Treg cell expansion in the tumor setting. Irradiated tumor cells preferentially induce Treg cells to expand, with little effect on T effector cells. Although we do not know the specific antigens that cause Treg expansion, our data suggest that both accessory cells and IL-2 are required for Treg expansion. We also tested the effects of candidate cytokines on Treg cell expansion; most cytokines had no effect, but IL-12 inhibited tumor-induced Treg expansion in an IFN-γ–dependent fashion. Mechanistic studies revealed that IFN-γ signaling directly inhibits Treg cell proliferation. Our findings support a model in which IL-12 stimulates immune effector cells to produce more IFN-γ, which then leads to the growth arrest of Treg cells. The inhibition of Treg cells may be critical in the early phase of an adaptive immune response, when accessory cells start to present tumor antigens, and also produce cytokines such as IL-12 and IFN-γ. These cytokines may work in a positive feedback loop to ensure a potent adaptive immune response, which could end prematurely should Treg cells also become activated.

A previous report suggested that IL-12 treatment induced the apoptotic cell death of intratumoral CD4⁺CD25⁺ T cells in vivo (33). In that study, Annexin V was used to measure apoptosis; local IL-12 treatment increased the percentage of CD4⁺CD25⁺ T cells that stained positive for Annexin V. However, CD25 is probably not a reliable Treg marker for in vivo mechanistic studies (40), especially when IL-12 is used to stimulate effector T cells during an antitumor response. It is possible that the observed death in CD4⁺CD25⁺ T cells was in fact the activation-induced death of effector T cells, rather than Treg cells. To clarify this issue, we used splenocytes from Foxp3-GFP knock-in mice in coculture assays and used 7-AAD to measure cell death. No significant cell death was observed in the CD4⁺GFP⁺ Treg cell compartment (Supplementary Fig. S7); however, IL-12 treatment did increase cell death in the CD4⁺GFP⁻ effector T-cell compartment. Because cells can lose GFP when membrane integrity is lost, these data alone cannot exclude Treg cell death that occurs in this setting. However, our cell cycle analyses suggest that the antiproliferative mechanism is important.

Our data show that IL-12 indirectly inhibits Treg cells, and that IFN-γ signaling accounts for at least part of this phenotype. Whereas an additional, probably IL-2–dependent mechanism may also account for part of IL-12–mediated inhibition of Treg proliferation, it is also possible that cellular components are involved. Purified Treg cells do not expand when cocultured with tumor cells; Non–T accessory cells in the cocultures are required to promote Treg expansion, presumably by presenting tumor antigens and/or by producing growth factors. Therefore, it is certainly possible that IL-12 can prevent Treg expansion by altering the function of an as yet unidentified population of accessory cells.

In summary, we show here that Treg inhibition by IL-12 may underlie some of its antitumor effects. Although the initial trials were suspended a decade ago because of dose- or regimen-related toxicities (41, 42), this cytokine has recently been revitalized for clinical studies in cancer patients (43–45). Since Treg cell proliferation (induced by tumors or even vaccines) can considerably dampen tumor immunity (9–11), this study provides mechanistic evidence to support the notion that IL-12, by inhibiting Treg cell proliferation, may serve as an adjuvant for tumor vaccine-based immunotherapy.

No potential conflicts of interest were disclosed.

Grant support: NIH grants DK49786 (T.J. Ley), P50 CA94056 (D.P. Worms) from the NIH, and a basic research fellow scholar award (X. Cao) from American Society of Hematology.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.