Transient regulatory T cell ablation deters oncogene-driven breast cancer and enhances radiotherapy

To determine the role of T reg cells in an oncogene-driven model of mammary tumorigenesis, we isolated carcinoma cells from C57BL/6 mice expressing a transgene encoding the PyMT oncogene under control of the murine mammary tumor virus (MMTV) promoter and implanted them in the knockin mice Foxp3^DTR, which express the human diphtheria toxin (DT) receptor (DTR) under the control of the Foxp3 locus (Kim et al., 2007). Orthotopic implantation of only 10⁵ tumor cells in the inguinal mammary gland of virgin female Foxp3^DTR mice on a C57BL/6 background results in uniformly growing mammary tumors that metastasize to the lungs with complete penetrance in ∼3–4 wk. Using this strategy, we were able to evaluate tumor growth in large cohorts of mice with synchronous, rapidly progressing metastatic mammary tumors. We ablated Foxp3⁺ T reg cells by injection of 50 µg/kg of DT on days 1, 2, 4, 6, and 13 after tumor cell implantation, and we observed significant reduction in tumor growth and incidence of lung metastasis (unpublished data). Flow cytometric analysis of lymphocyte populations isolated from enzymatically dissociated tumors demonstrated that the extent of T reg cell ablation was >99% (Fig. 1 B). We also found expansion and activation of CD4⁺ and CD8⁺ T cell subsets based on the increased expression of Ki67 and CD44 and decreased levels of CD62L (unpublished data). In addition, the proportion of CD4⁺ and CD8⁺ T cells expressing IFN-γ and TNF was markedly augmented (Fig. 1 B and not depicted). Furthermore, we also observed an increase in immature myeloid cells (unpublished data). Next, we sought to test the effect of T reg cell ablation on established tumors that reached exponential growth (∼250 mm³; Fig. 1 A). DT treatment of Foxp3^DTR mice with large tumors resulted in significant reduction of tumor burden (Fig. 1 C). In addition, the incidence and size of lung metastasis in mice bearing large tumors was reduced upon depletion of T reg cells (Fig. 1 D). The observed reduction in lung metastatic burden could have been secondary to reduced primary tumor volume in DT-treated animals because metastatic load is proportional to primary tumor size (Heimann and Hellman, 2000; Minn et al., 2007). To investigate whether the beneficial effect of T reg cell ablation on lung metastasis in these experiments was independent of the diminished primary tumor growth, we treated animals with DT after establishment of lung metastasis 2 wk after tail vein inoculation of 5 × 10⁵ PyMT cells (Fig. 1 A). Analysis performed 2 wk after DT injection showed that tumor burden in the lungs was markedly reduced as determined by histological quantification of tumor areas in lung sections, demonstrating a pronounced direct effect of T reg cell ablation on the disseminated tumors (Fig. 1 E). These studies demonstrate that T reg cell ablation is therapeutic not only for newly formed but also for large, rapidly growing primary mammary tumors and fully established lung metastasis.

The potent restraint of cancer progression and metastasis in the orthotopic transplantation model of breast carcinogenesis observed upon T reg cell ablation raised a question of whether it can be efficacious when applied to genetically induced oncogene-driven tumors. To address this issue, we introduced the Foxp3^DTR allele into mice coexpressing PyMT oncogene together with a luciferase reporter regulated by a doxycycline-inducible promoter and reverse tetracycline-controlled transactivator under the MMTV promoter (MMTV–reverse tetracycline transactivator [rtTA]; tet-O-MT:IRES:Luc or TOMT; Podsypanina et al., 2008). Upon doxycycline administration, these mice developed tumors in all mammary glands. Analysis of tumor-infiltrating lymphocytes showed that T reg cells were highly enriched within the CD4⁺ T cell subset (Fig. 2 A). We allowed mice to develop large invasive carcinomas that emitted a photon flux of 10¹⁰ photons per second and set the time point of analysis at 10 d after the initial dose of DT (Fig. 2 B). At that time, the mice are fully active and do not present any signs of morbidity despite sustained T reg cell ablation (Fig. 2 C). Because asynchronous and slow tumor growth in TOMT mice precludes the accurate evaluation of the effect of T reg cell ablation on growth kinetics, we analyzed expression of cleaved caspase-3 in cancer cells, a marker of apoptotic death, as a means to assess the consequence of T reg cell ablation on tumor viability. We observed a remarkable increase in apoptosis of tumor cells in mice treated with DT compared with control mice injected with PBS (Fig. 2 D). Concomitantly, we observed significant expansion and activation of intratumoral CD4⁺ and CD8⁺ T cells in T reg cell–depleted mice (Fig. 2 E). These results indicate that T reg cells represent a major cellular mechanism facilitating tumor progression by maintaining cancer cell viability in this experimental model of oncogene-driven breast cancer.

To minimize the potential side effects of T reg cell ablation and test whether continuous ablation was required to achieve the observed reduction in orthotopic tumor growth, we decided to limit the dose and frequency of the DT administration. We gave tumor-bearing animals only two doses of DT (25 µg/kg) once tumors reached a volume of ∼100 mm³. This treatment regimen allows for efficient (>99%) yet transient T reg cell ablation with minimal morbidity (slight short-term weight loss followed by a quick recovery; Fig. 3 C) and no gross organ immunopathology evaluated by histological examination 2 wk after DT (Fig. 3 D). Remarkably, despite lack of pronounced generalized immunopathology, this brief ablation of T reg cells significantly hindered primary tumor growth (Fig. 3 A) and resulted in the almost complete disappearance of metastatic tumor nodules in the lungs (Fig. 3 B). These experiments demonstrate that efficient ablation of T reg cells for a very short period of time provides therapeutic benefit similar to that of persistent ablation, while reducing the dangerous side effects to a bare minimum.

T reg cells could be beneficial to cancer cell growth and tumor progression in several ways. On one hand, they can suppress components of the adaptive immune system providing protection against antigen-specific tumor cell killing. On the other hand, they can modulate components of the tumor microenvironment that may directly or indirectly promote tumor progression. To better understand the early changes taking place in the tumor milieu upon T reg cell ablation, we performed a protein array of 66 cytokines and chemokines on tumor lysates prepared on day 5 after DT administration to evaluate early changes in these soluble mediators (Fig. 4 A). Comparison of control and DT-treated tumor lysates revealed significant increments in 12 cytokines, although only 5 of them increased above a twofold threshold (Fig. 4 B). The most prominent increase was observed in IFN-γ, a proinflammatory cytokine with potent anti-tumor effects, followed by CXCL9 and CXCL10 (Fig. 4 C). These two chemokines are produced by several cell types in response to IFN-γ and serve as chemoattractant for CXCR3-expressing leukocytes, most notably Th1 and NK cells, but also monocytes. To validate these observations and determine the source of each cytokine, we isolated T cells (CD45⁺CD3⁺CD11B⁻Gr1⁻) and myeloid cells (CD45⁺CD3⁻CD11B⁺Gr1⁻) from an independent group of control and DT-treated tumors by fluorescence activated cell sorting. Using primer-specific semiquantitative PCR, we determined their mRNA levels in these two populations. We observed that IFN-γ was produced in the T cell compartment and increased significantly upon T reg cell ablation, whereas CXCL9 and CXCL10 were significantly increased in the myeloid compartment upon DT treatment, likely as a response to IFN-γ (Fig. 4 D). Because IFN-γ is a potent classic activator of macrophages, we quantified the mRNA levels of iNOS, a prototypical IFN-γ–induced effector molecule of M1 polarized macrophages. We observed a marked up-regulation of iNOS expression in the myeloid cell compartment upon T reg cell ablation (Fig. 4 E). Altogether, our results suggest that T reg cell ablation leads to a strong IFN-γ–dominated tumor milieu that can promote the anti-tumor responses of Th1, NK, and CD8⁺ T cells or generation of M1 macrophages.

Given the predominance of IFN-γ in T reg cell–depleted tumors, we sought to evaluate its role in the observed delayed tumor progression. Thus, we injected tumor-bearing Foxp3^DTR mice with 1 mg IFN-γ neutralizing antibody alone or in combination with DT. Although IFN-γ antibody treatment alone did not have an impact on tumor growth in control mice, IFN-γ antibody almost completely abolished the effect of DT-mediated T reg cell ablation on tumor growth (Fig. 5 A). Next, we wanted to determine whether the anti-tumoral effect was mediated through cytotoxic T or NK cells. To test this, we ablated T reg cells in the presence of NK- or CD8-depleting antibodies. NK cell depletion using NK1.1 antibody did not have a detectable effect on growth of control or T reg cell–depleted tumors (Fig. 5 B). In addition, administration of CD8-depleting antibody during the course of T reg cell ablation did not affect the resulting tumor growth reduction (Fig. 5 C). To corroborate this finding, we made use of Foxp3^DTR mice lacking β2-microglobulin, which are required for MHC class I expression and proper maturation of CD8⁺ T cells (Gasteiger et al., 2013). In agreement with antibody-mediated depletion, T reg cell ablation in control or DT-treated Foxp3^DTR β2M^−/− mice resulted in comparable determent of tumor progression and indistinguishable tumor growth profiles (Fig. 5 D). In contrast, when CD4⁺ cells were depleted from the DT-treated tumors, there was a pronounced attenuation in the T reg cell ablation-mediated anti-tumor effect (Fig. 5 E). Together, these results indicate that NK and CD8⁺ T cells are not necessary for the anti-tumor effect of T reg cell ablation, which is dependent on CD4⁺ T cells and requires IFN-γ. In addition, these observations suggest that NK and CD8⁺ T cells are dispensable as sources of IFN-γ.

Encouraged by the potent anti-tumor effect achieved via T reg cell ablation, we looked for potentially cooperative therapeutic approaches. Highly expressed on activated and chronically stimulated (“exhausted”) effector cells, CTLA-4, PD-1, and its ligand PD-L1 are also present in high amounts on T reg cells (Pardoll, 2012), and their antibody-mediated inhibition has proven a viable immunotherapeutic strategy to treat solid tumors in recent preclinical studies and clinical trials. Therefore, immune checkpoint blockade jointly with T reg cell depletion could potentially further promote the effector response of newly recruited T cells in addition to reversing the exhausted state of preexisting tumor-infiltrating T cells.

In support of this concept, tumor-infiltrating lymphocytes in T reg cell–depleted tumors exhibited a marked increase in the expression of the PD-1 receptor on effector T cells. Additionally, expression of the PD-L1 ligand was increased on both T cells and myeloid cells (Fig. 6 B and not depicted). Thus, we sought to test whether combination with CTLA-4 or PD-1 checkpoint blockade could enhance the therapeutic effect of T reg cell ablation in our oncogene-driven breast cancer model. First, we examined the effects of targeting the CTLA-4 or PD-1/PD-L1 inhibitory pathways with blocking antibodies of corresponding specificity administered on days 0, 3, and 6 after tumors reached a volume of ∼100 mm³. Blockade of either one of these pathways by CTLA-4 or PD-1 or PD-L1 or a combination of PD-1 and PD-L1 antibodies had no significant effect on the growth of PyMT-driven orthotopic tumors (Fig. 6, A, C, and E; and not depicted). Of note, lung metastatic burden measured by enumerating tumor nodules on the lung surface was diminished by half upon the blockade of PD-1/PD-L1 but not CTLA-4 signaling (Fig. 6, D and F). A combination of DT with CTLA-4 antibody or with PD-1 and PD-L1 antibodies did not enhance the effect of T reg cell ablation alone on primary tumor progression (Fig. 6, C and E). DT treatment almost completely eliminated the appearance of metastatic nodules in the lungs (Fig. 6, D and F). Thus, we were unable to evaluate the potential synergistic effects of T reg cell ablation and PD-1/PD-L1 blockade in this experimental setup, although the latter did have a beneficial effect as monotherapy (Fig. 6 F). Our observations suggest that efficient targeting of T reg cells is necessary and sufficient to achieve an effective immunotherapeutic response to the growing tumor in this model of oncogene-dependent cancer.

Given that no advantage was offered by a combination with checkpoint blockade, the leading immune-based strategy in the treatment of metastatic tumors, we evaluated the ability of T reg cell ablation to increase the efficacy of ionizing radiation aimed at inhibiting proliferation and inducing cell death. Local radiotherapy, widely used in the management of breast cancer, has the potential to synergize with T reg cell ablation in several ways. First, T reg cells are markedly more resistant to radiation than conventional T cells, and increased T reg/T effector cell ratio upon radiotherapy may reduce its efficacy (Komatsu and Hori, 2007; not depicted). Second, radiation can modulate immune response through the release of tissue damage factors that attract immune cells, stimulate antigen presentation, increase tumor antigen pool, and sensitize cancer cells to immune-mediated killing. Lastly, the high rate of cancer cell death resulting in tumor debulking contributes to a decrease in persistent antigens that can induce tolerance (Drake, 2012; Coussens et al., 2013).

First, we compared the two most commonly used doses of experimental local tumor irradiation, 7.5Gy and 12Gy, and found that the latter reduced the size of the tumors by 50% in the 2 wk after treatment (unpublished data). Thus, we stereotactically delivered 12Gy radiation to mice bearing ∼100 mm³ (and 250 mm³; unpublished data) bilateral tumors, and we depleted T reg cells by administering DT on days 1 and 2 after radiation, before rise in T reg cell/T effector cell ratio (Fig. 7 A and not depicted). The combination of radiation with transient T reg cell ablation affected the tumor growth much more significantly than either treatment alone, with the most pronounced cooperative effects observed by the end of the experiment. By that time, control and single therapy groups of mice were euthanized due to heavy tumor burden before tumors in the combination treatment group reached exponential growth phase (Fig. 7 B). During the first 2 wk of the experiment, volumes of control tumors had increased 50-fold, irradiated tumors 10-fold, T reg cell–depleted tumors 7.5-fold, and tumors treated with the combination had only increased ∼2.5-fold (Fig. 7 C). When we measured the mean time tumors needed to reach a volume of ∼1,000 mm³, we observed that control tumors reached that size in ∼25 d, irradiated tumors in 28 d, T reg cell–depleted tumors in 32 d, and tumor treated with the combination needed a mean of 39 d (Fig. 7 D). Histological examination of tumors collected from the various groups 2 wk after treatment showed a significantly larger area of necrosis with the combination treatment than either of the single treatments and increased cleaved-caspase 3 staining in healthy areas of the tumor (Fig. 7 E and not depicted). In addition, tumors in the combination treatment group displayed a significant increase in the number of macrophages by double immunohistochemical staining with CD45 and Iba-1 markers (Fig. 7 F), as well as increase in CSF-R1⁺ tumor-infiltrating leukocytes, as determined by flow cytometry (not depicted). The differences observed in tumor growth translated to a very significant increase in mouse survival, with mice treated with the combination therapy living almost twice as long as the control, untreated mice (Fig. 7 G). Interestingly, lung metastatic burden analyzed in a time-matched manner was not affected by local radiation treatment, and it was not significantly improved by the combination therapy over the T reg cell ablation treatment alone, at least at the time of analysis (Fig. 7 H), suggesting that transient T reg cell ablation alone is effective at limiting distant metastasis.

Recent progress of immune-based cancer therapies is best exemplified by the clinical trials of CTLA4 blockade for the treatment of melanoma (Hodi et al., 2010) and by targeting PD-1 and PD-L1 in melanoma, renal cell carcinoma, and non–small cell lung and ovarian cancer (Brahmer et al., 2012; Topalian et al., 2012b). These interventions yield durable responses, but only in a limited fraction of recipients. Mechanistic studies in an oncogene-driven prostate cancer model showed that the therapeutic efficacy of monoclonal antibody-mediated checkpoint blockade depends on prior vaccination with cytokine-expressing tumor cells (Hurwitz et al., 2000). This finding, along with the observed clinical response rates of monotherapy checkpoint blockade in human trials, suggests that relieving the negative regulation exerted by these pathways alone is insufficient to attain potent anti-tumor immunity. Hence, combinatorial therapy of immunostimulatory antibodies has been proposed to increase response rates, with the potential risk of increasing dangerous side effects (Curran et al., 2010; Ascierto et al., 2013). In agreement with several studies in mouse models of experimental cancer, we found that as a single therapy, checkpoint blockade does not hinder primary tumor progression in a murine orthotopic model of oncogene-driven mammary carcinogenesis. However, efficient ablation of T reg cells alone achieved a significant reduction in tumor burden without the need for additional manipulation, revealing the essential role of T reg cells in oncogene-driven tumor growth. Consistent with these results, high tumor effector T cell to T reg cell ratio correlates with better tumor prognosis and clinical benefit of immunotherapy (Sato et al., 2005; Quezada et al., 2006; Balachandran et al., 2011). T reg cell ablation resulted in a sharply augmented expression of IFN-γ by tumor-infiltrating T cells, which was necessary for the observed determent of tumor progression. The role of IFN-γ in this model is consistent with its previously demonstrated role in immune surveillance of carcinogen-induced sarcomas and spontaneous epithelial carcinomas (Kaplan et al., 1998; Shankaran et al., 2001). Cytotoxic CD8⁺ T cells, as well as NK cells, were not necessary, whereas CD4⁺ T cells were required for anti-tumoral effect of T reg cell–mediated ablation. These results imply that IFN-γ production by NK or CD8⁺ T cells is dispensable for the anti-tumoral effect of T reg cell ablation, and they point to a potential role for CD4⁺ T cells as a nonredundant source of protective IFN-γ in PyMT breast carcinomas. CD4⁺ T cells can exert IFN-γ–dependent as well as direct cytotoxic effects on tumors cells (Shankaran et al., 2001; Quezada et al., 2010). Increased IFN-γ levels can contribute to the reduction of tumor growth by regulating the properties of tumor-infiltrating macrophages (O’Sullivan et al., 2012). In this regard, the observed sharp increase in expression of iNOS and proinflammatory chemokine expression by tumor-infiltrating myeloid cells raises the possibility that the therapeutic effect of T reg cell ablation is due to modulation of the accessory functions of tumor-infiltrating macrophages. The latter has been found to be essential for lung metastases in the MMTV-PyMT model (DeNardo et al., 2009).

In contrast to T reg cell ablation, anti-tumoral effect of both systemic and local administration of CTLA-4 neutralizing antibody is CD8⁺ T cell dependent but CD4⁺ T cell independent (van Elsas et al., 2001; Fransen et al., 2013). Likewise, the therapeutic effects of PD-1/PD-L1 blockade in chronic viral infection, and likely in cancer, are dependent on restoration of cytolytic responses and IFN-γ production by CD8⁺ T cells (Barber et al., 2006; Topalian et al., 2012a). The latter observations in combination with a dispensable role of CD8⁺ T cells for the therapeutic benefit of T reg cell targeting offer a likely explanation for the failure of PD-1/PD-L1 blockade to mount biological response in our experiments.

CTLA-4 is expressed by T reg cells and is required for their function (Wing et al., 2008). Elegant genetic studies demonstrated that targeting CTLA-4 in both effector and T reg cell subsets affords the maximal inhibition of tumor growth in a transplantable B16 melanoma model (Peggs et al., 2009). Considering these findings, our study raises the possibility that the success of the PD-1 and CTLA-4 checkpoint blockade may be primarily due to selective (or relative) depletion or functional impairment of T reg cells. This is consistent with the recent studies suggesting that anti–CTLA-4 therapy works primarily through macrophage-mediated T reg cell depletion (Selby et al., 2013; Simpson et al., 2013). Additionally, PD-1/PD-L1 pathway blockade has also been shown to diminish T reg cell suppressor function (Wang et al., 2009).

Although CTLA-4 blockade did not affect lung metastatic burden, PD-1/PD-L1 blockage significantly diminished the number of metastatic foci in the lungs. The reduction, albeit markedly less pronounced than the one achieved through T reg cell ablation, is suggestive of a specific role for PD-1/PD-L1 inhibitory pathway in the colonization of lungs by disseminated single cancer cells. The observed selective role for PD-1 in lung metastasis was consistent with its prominent role in blocking lung inflammation, i.e., pneumonitis resulting from PD-1/PD-L1 deficiency, and clinical responses of PD-1 blockade in non–small cell lung cancer patients (Lucas et al., 2008; Brahmer et al., 2012; Topalian et al., 2012b).

Immune therapeutic approaches, such as checkpoint blockade and T reg cell depletion, can lead to the breaking of immune self-tolerance, inducing a variety of side effects that include rash, colitis, hepatitis, and endocrinopathies (Postow et al., 2012). Moreover, sustained ablation of T reg cells leads to a fatal lymphoproliferative syndrome (Kim et al., 2007, 2009). We found that transient T reg cell ablation does not have a significant effect on the overall mouse health status, as indicated by normal mouse activity and weight, and minimal tissue immunopathology. These results bring hope to future translational studies, and they highlight the importance of developing methods of highly effective yet short-term T reg cell depletion that minimizes side effects.

Ionizing radiation therapy is widely used for the management of breast cancer due to its ability to induce cell death and proliferation arrest. In addition, radiotherapy can promote antigen presentation due to antigen and tissue damage signal release by dying cancer cells and reduce tolerance induction due to reduction of the tumor mass (Formenti and Demaria, 2009). Specific rationale for combined radiation and T reg cell ablation therapies is based on the known relative radioresistance of T reg cells (Komatsu and Hori, 2007). Our studies demonstrate that ablation of T reg cells during the course of radiotherapy significantly improves the outcome of treatment, not only by reducing tumor growth and metastasis but also by prolonging survival.

Altogether, our observations put forward the notion that targeting T reg cells is likely to result in pronounced clinical responses in breast cancer patients. Modulating this central mechanism of immune tolerance may expand the use of immunotherapy to tumor types which, like breast cancer, are not inherently immunogenic. Furthermore, current clinical outcomes might be significantly improved by combination of T reg cell depletion strategies with radiation, and possibly chemotherapy or targeted therapies against molecular drivers of oncogenesis.

Foxp3^DTR mice were generated in our laboratory and previously described (Kim et al., 2007). β2M^−/− mice were purchased from Taconic. Mice bearing the MMTV-rtTA and TetO-PyMT:IRES:Luc transgenes were provided by H. Varmus (National Cancer Institute, National Institutes of Health, Bethesda, MD). C57BL/6 MMTV-PyMT mice were a gift from M.O. Li. (Memorial Sloan-Kettering Cancer Center, New York, NY) All animal studies were performed in accordance with an IACUC-approved protocol at the Memorial Sloan-Kettering Cancer Center.

For oncogene induction, mice were placed on doxycycline-impregnated food pellets (625 ppm; Harlan-Teklad). For T reg cell ablation studies, DT (Sigma-Aldrich) was injected intravenously at 50 or 25 µg per kg of body weight at indicated times. Mammary tumorigenic cell lines were generated via enzymatic dissociation of invasive tumors from MMTV-PyMT mice, expanded in Dulbecco’s modified Eagle’s medium and high glucose medium supplemented with 10% FBS, and transduced with a Firefly Luciferase retroviral vector using standard techniques. For orthotopic implantation studies, 100,000 cells were resuspended in PBS and mixed in a 1:1 ratio with growth factor–reduced Matrigel (BD), and injected in the mammary fat pad of isoflurane-anesthetized mice. Primary tumor outgrowth was monitored daily by taking measurements of the tumor length (L) and width (W). Tumor volume was calculated as πLW²/6. For experimental lung metastasis assays, 500,000 cells were resuspended in PBS and inoculated via tail vein injection. Lung metastatic burden was quantified by counting the number of metastatic nodules under a dissection scope (Olympus), ex vivo bioluminescence using an IVIS200 imager (Xenogen), or calculating the ratio between the area covered by metastasis over the total area of the lung in histological sections. CTLA-4 (clone 9D9), PD-1 (clone RPM1-14), and PD-L1 (clone 10F.9G2) antibodies were administered i.p. at days 0, 3, and 6 at a dose of 100, 250, and 100 µg per mouse, respectively, as indicated in the text. IFN-γ (clone XMG1.2), NK cells (clone PK136), and CD4⁺ (clone GK1.5) and CD8⁺ (clone 2.43) T cell depletion was achieved through i.p. injection of 1 mg and 300, 100, and 250 µg, respectively, together with the second dose of DT (IFN-γ, CD4, and NK) or 4 d after DT injection (CD8). All antibodies for animal studies were obtained from Bio X Cell. Radiation was administered in a single fraction of 12Gy when tumors reached a volume of ∼100 or 250 mm³ using an X-RAD 225Cx micro-irradiator. In brief, individual mice were anesthetized using isoflurane and positioned on a platform where a cone-beam CT imaging of the animal was performed to allow targeting the radiation field to the tumor, avoiding normal structures.

Cytokines and chemokines were measured using a multiplex Luminex bead assay (Millipore). Tumors were lysed in buffer containing 50 mM Tris, 150 mM NaCl, 1% NP-40, 1 mM EDTA, and protease inhibitors. Cleared lysates were quantified and extracts bearing 20 µg of total protein were incubated with mouse cytokine/chemokine magnetic bead panels I, II, and III from Milliplex, according to the manufacturer’s instructions.

For histological analysis, tissues were fixed in 10% neutral buffered formalin and routinely processed for hematoxylin and eosin staining. Apoptosis (cleaved-caspase 3), proliferation (Ki67), leukocyte (CD45), and macrophage (IBA1) stainings were performed using automated IHC techniques by the Molecular Cytology Core Facility, and quantified using MetaMorph analysis.

Tumor-infiltrating lymphocytes were isolated by enzymatic dissociation of tumors using Liberase TL (Roche) and digested for 25 min, followed by Percoll (VWR) centrifugation to eliminate dead cells. Intracellular Foxp3 staining was performed using a Foxp3 mouse T reg cell staining kit (eBioscience). Cytokine staining was performed after stimulation of splenocytes or isolated TILs with 50 ng/ml PMA and 500 ng/ml ionomycin for 4–5 h in the presence of Golgi-Plug (BD). All antibodies used for flow cytometry staining were purchased from eBioscience, BioLegend, or BD. Stained cells were analyzed in a LSRII flow cytometer (BD). Data were analyzed using FlowJo software (Tree Star).

For qPCR analysis, tumors were processed by enzymatic digestion as previously described and myeloid or T cells were sorted based on their surface expression of CD45, TCR-β, CD11B, and Gr1 using a FACSAria II (BD). Sorted cells were lysed in TRIzol reagent (Invitrogen) and reverse-transcribed using SuperScript III Reverse transcription (Invitrogen). Semi-quantitative PCR was performed in an ABI Prism 7900HT instrument (Applied Biosystems) using SybrGreen PCR master mix (Applied Biosystems) and the following primers: β-actin forward, 5′-CTAAGGCCAACCGTGAAAAG-3′; β-actin reverse, 5′-ACCAGAGGCATACAGGGACA-3′; IFN-γ forward, 5′-ATCTGGAGGAACTGGCAAAA-3′; IFN-γ reverse, 5′-TTCAAGACTTCAAAGAGTCTGAGGTA-3′; CXCL9 forward, 5′-TTTTCCTTTTGGGCATCATCTT-3′; CXCL9 reverse, 5′-AGCATCGTGCATTCCTTATCACT-3′; CXCL10 forward, 5′-GAAATCATCCCTGCGAGCCT-3′; CXCL10 reverse, 5′-TTGATGGTCTTAGATTCCGGATTC-3′; iNOS forward, 5′-CTTTGCCACGGACGAGAC-3′; and iNOS reverse, 5′-TCATTGTACTCTGAGGGCTGAC-3′.

All statistical analysis was performed using Student’s t test or ANOVA analysis as indicated with Prism software (GraphPad Software).

We thank Drs. Harold Varmus and Ming Li for providing TOMT mice and C57BL/6 MMTV-PyMT mice, respectively.

This work was supported by a fellowship from the American Cancer Society (P.D. Bos), National Institutes of Health (NIH) Immunology Training grant T32CA009149-37 (P.D. Bos), a Geoffrey Beene Metastasis Center Grant Award (A.Y. Rudensky), and NIH grant R37 AI034206 and Ludwig Center at Memorial Sloan-Kettering Cancer Center (A.Y. Rudensky). A.Y. Rudensky is an investigator with the Howard Hughes Medical Institute.

The authors declare no competing financial interests.