TGF-β Receptor II Loss Promotes Mammary Carcinoma Progression by Th17-Dependent Mechanisms

TGF-β plays a major role in the regulation of tumor initiation, progression, and metastasis and requires the type II TGF-β receptor (TβRII) for signaling. It has been shown that decreased expression of TβRII is correlated with an increased risk of developing invasive breast cancer (1) and that loss of TβRII is correlated with high-grade human carcinoma in situ and invasive breast cancer (2). In our laboratory, we have shown that conditional deletion of TβRII in mammary epithelial cells that also express the polyoma middle T (PyMT) oncogene under control of the mouse mammary tumor virus (MMTV) promoter resulted in shortened tumor latency and a 5-fold increase in lung metastases compared with PyMT tumors with intact TGF-β signaling (3, 4). We identified that TGF-β signaling mediates intrinsic, stromal–epithelial, and host–tumor interactions during breast cancer progression by regulating basal CXC chemokine ligand 1, CXC chemokine ligand 5 (CXCL1/5), and CC chemokine ligand 20 expression (5).

Interleukin (IL)-17 is a cytokine secreted by CD4 and CD8 cells (6–8). The differentiation and regulation of murine Th17 cells has been extensively studied in the past few years, and TGF-β, IL-6, and IL-23 have been implicated as critical regulators of the initiation of mouse Th17 cell differentiation (9, 10). Although the function of IL-17 is not fully understood, it is clear that IL-17 amplifies the immune response by inducing the expression of other chemokines, inflammatory cell-surface markers, and inflammatory mediators (11, 12). IL-17–producing cells are detected in patients with cancer and tumor-bearing mice (13, 14). Some investigators report that tumor growth is increased in IL-17^−/− mice (15, 16). However, Wang and colleagues (17) indicated that tumor growth is suppressed in IL-17^−/− mice. Recently, the authors of another study have shown that neutralization of IL-17 stunted tumor growth and that the systemic administration of IL-17 promoted tumor growth. Additional analysis indicated that IL-17 was required for the development and tumor-promoting activity of myeloid-derived suppressor cells (MDSC) in tumor-bearing mice (18).

In the current study, we examined the indirect role of impaired TGF-β signaling in carcinoma cells on tumor growth. In TβRII knockout tumor cells, we determined basal and IL-17–stimulated secretion of CXCL1/5 and expression of IL-17R. We analyzed the mechanisms that are involved in Th17 differentiation in mice and determined the role of IL-17 in the regulation of suppressive function of MDSCs and macrophages. By using anti–IL-17 Ab in vivo, we demonstrated a significant indirect role of impaired TGF-β signaling in carcinoma progression by enhanced Th17 response.

We have previously shown that conditional knockout of TβRII in mammary epithelial cells of MMTV-PyMT mice results in shortened tumor latency and an increased number of metastases in the lung (3, 4). In addition, we reported that 1 week after tumor palpation, the number of CD11b⁺Gr1⁺ cells (MDSCs) increased in tumor tissue in PyMT/Tgfbr2^KO mice. To examine the role of MDSCs in TGF-β–mediated tumor progression, we collected tumor tissue, lung, bone marrow, and spleen before tumor progression; on the day of tumor palpation; and 1, 2, 3, and 4 weeks after tumor palpation in PyMT/Tgfbr2^fl mice and mice without TβRII in the mammary epithelium. We observed a significant increase of CD11b⁺Gr1⁺ cells in spleen and in tumor tissue at each time point in the PyMT/Tgfbr2^KO mice (Fig. 1A, Supplementary Fig. S1) except at a late stage of tumor progression (day 28).

In the lungs of these tumor-bearing animals we found increased numbers of these cells on days 7, 14, and 21 of tumor progression in PyMT/Tgfbr2^KO mice. No differences were found in the bone marrow of these mice. In parallel with the increased number of CD11b⁺Gr1⁺ cells, we observed a significant increase of tumor-associated macrophages (TAM) in the PyMT/Tgfbr2^KO mice (Fig. 1A, Supplementary Fig. S1). Localization of CD11b⁺Gr1⁺ cells in the tumor tissue was similar to our previously published studies (data not shown) (19). Surprisingly, we found that CD11b⁺Gr1⁺ cells are negative for CXCR2 in lung and tumor tissues but positive in the spleen and bone marrow (Fig. 1B–C). We have previously shown that SDF-1/CXCR4 and CXCL5/CXCR2 are involved in the recruitment of MDSCs into tumors of PyMT/Tgfbr2^KO mice (19). Therefore, we concentrated our studies on mechanisms driving the migration of MDSCs to tumor tissue, which are recruited by CXCL1/5.

CXCL1/5 are the primary chemokines that signal through CXCR2 to induce myeloid cell chemotaxis. However, it is interesting that CXCR2 is not expressed in CD11b⁺Gr1⁺ cells found in the tumor tissue. To address this issue, we isolated CD11b⁺Gr1⁺ cells from the spleen of tumor-bearing mice and incubated these cells with varying concentrations of CXCL1/5. We observed a significant decrease in expression of CXCR2 starting at a concentration of 10 ng/mL. Incubation with greater concentrations of chemokines resulted in an almost complete loss of CXCR2 expression (Fig. 1D). As seen previously with CXCL5 (19), loss of TβRII in PyMT tumor cell lines resulted in an increased secretion of CXCL1 (Fig. 2A). Moreover, the significantly high level of CXCL1 secretion indicates a potentially important role in the induction of MDSCs chemotaxis. These data are further supported by the analysis of CXCL1/5 in tumor explants in which we also found an increased level of CXCL1/5 in PyMT/Tgfbr2^KO mice (Fig. 2B).

We observed a strong effect of IL-17 in promoting increased secretion of CXCL1/5 from carcinoma cells, and this effect depended on TβRII. Carcinoma cells (Ep-CAM⁺CD45⁻) were sorted from primary tumor tissue of PyMT/Tgfbr2^fl and PyMT/Tgfbr2^KO mice (Supplementary Fig. S2) and incubated in the presence of IL-17 with or without TGF-β. Reinforcing previous work performed with the use of cultured cell lines, we found that these sorted epithelial cells displayed the same profile of CXCL1/5 chemokine expression (Fig. 2C).

Incubation with IL-17 increased secretion of CXCL1/5 from both types of tumor cells with a more intense increase in CXCL5. Importantly, we observed that TGF-β completely inhibits IL-17—stimulated secretion of CXCL1/5 from PyMT/Tgfbr2^fl cells. The data indicate that loss of TGF-β signaling in tumor cells results in a subsequent increase in CXCL1/5 production and acts to exacerbate further induction of these chemokines by other factors by eliminating the regulatory role of TGF-β over their production. Thus, in vivo levels of CXCL1/5 could far exceed those measured ex vivo and lends credence to the increased infiltration of CD11b⁺Gr1⁺ cells into tumors with abrogated TGF-β signaling.

IL-17R is the primary receptor for IL-17 signaling (7) and because of the more robust response of PyMT/Tgfbr2^KO cells to IL-17 stimulation, alterations in IL-17R expression (CD217) caused by the loss of TβRII expression were examined. Increased expression of CD217 was observed on PyMT/Tgfbr2^KO cells (Fig. 2D). After incubating cells with varying concentrations of IL-17, we found that PyMT/Tgfbr2^KO carcinoma cells are more sensitive to IL-17, resulting in enhanced secretion of CXCL1/5 (Fig. 2E). Additional analysis showed that PyMT/Tgfbr2^KO cells are 4 times more sensitive to low concentrations of IL-17 (50 pg/mL) in stimulation of CXCL1/5 secretion (data not shown).

Deletion of TβRII in carcinoma cells displayed a double effect on CXCL1/5 secretion. First, deletion of TβRII increased expression of CD217 and as a consequence sensitized carcinoma cells to IL-17, and second, deletion of TβRII prevented inhibition by TGF-β of IL-17–stimulated secretion of chemokines.

The presence of IL-6, TGF-β, and IL-23 preferentially promotes differentiation of CD4⁺ T cells into CD4⁺Th17 cells (9). Tumor tissues from PyMT/Tgfbr2^KO mice were characterized by more myeloid cell infiltration, which acts as an additional source of IL-6, TGF-β, and IL-23. Thus, we hypothesized that the PyMT/Tgfbr2^KO tumors might be creating a cytokine milieu that promotes differentiation of CD4⁺T cells into Th17 cells.

To identify the primary source of TGF-β, IL-6, and IL-23, the presence of these cytokines in conditioned media of cultured tumor cells, tumor explants, isolated immune cells (CD45⁺), and nonimmune cells (CD45⁻) from tumor tissue was analyzed by ELISA (Fig. 3A). We observed that PyMT/Tgfbr2^KO carcinoma cells do not secrete significant levels of IL-6. However, tumor explants, CD45⁺, and CD45⁻ cells from tumor tissue have an increased level of IL-6 in PyMT/Tgfbr2^KO mice. Basal secretion of TGF-β by cultured carcinoma cells did not depend on TβRII expression, and the level of TGF-β in PyMT/Tgfbr2^KO cells was the same as in cells with intact TGF-β signaling. Levels of TGF-β in tumor explants and in conditioned media of CD45⁺ cells demonstrated the same pattern as IL-6, with increased expression in PyMT/Tgbfr2^KO mice but no differences in secretion by CD45⁻ cells. Neither carcinoma cell type secreted significant levels of IL-23. Also, there were no significant differences in the concentration of IL-23 in conditioned medium from CD45⁻ cells.

However, tumor explant and CD45⁺ cells from tumor tissue have increased levels of IL-23 in the PyMT/Tgfbr2^KO mice. Thus, we demonstrate that the primary source of the cytokines necessary for development of Th17 cells is immune cells (CD45⁺) that have migrated into the tumor tissue. Also, expression of IL-6, TGF-β, and IL-23 (p19) was examined by real-time PCR (RT-PCR) in PyMT/Tgbr2^fl vs PyMT/Tgfbr2^KO tumors with the use of freshly resected and PBS-perfused tissues (to avoid contamination from circulating cells), and these data correlated with our data shown in Figure 3E (Supplementary Fig. S3A).

In addition, we sorted MDSCs and TAMs from tumor tissue (Supplementary Fig. S3B) and analyzed the expression of the IL-6, IL-23 (p19), and TGF-β genes by RT-PCR. Previously, it has been shown that MDSCs are a major source of TGF-β in PyMT/Tgfbr2^KO tumors (19). Analysis of MDSCs from tumor tissue recapitulated these findings (Fig. 3B), but it was also observed that MDSCs have increased expression of IL-23(p19) and IL-6 in PyMT/Tgfbr2^KO.

TAMs from PyMT/Tgbr2^KO tumors have similar patterns of gene expression except for TGF-β. As a result, it was determined that 2 sources for increased expression of TGF-β, IL-6, and IL-23 exist. First, secretion of these cytokines is increased in PyMT/Tgfbr2^KO tumors mediated by increased immune cell infiltration. Migration of immune cells to tumor tissue with deleted TβRII is increased as a consequence of enhanced basal secretion of CXCL1/5. Second, the PyMT/Tgfbr2^KO tumor microenvironment up-regulates gene expression of TGF-β, IL-6, and IL-23 in myeloid cells.

Quantification of Th17 cells in the tumor microenvironment was performed by the use of both flow cytometry for cellular identification and ELISA analysis of Th17-specific cytokine production. A significant increase in the number of CD4⁺IL-17A⁺ cells in PyMT/Tgfbr2^KO mice was observed (Supplementary Fig. S3C). Similar results were obtained by the use of another tumor model, 4T1 mammary carcinoma cells, which produce more CXCL1/5 than PyMT/Tgfbr2^KO carcinoma cells (Supplementary Fig. S4). Quantitative analysis of the proportion of CD4⁺IL17A⁺ cells to the total number of CD4⁺ cells showed that PyMT/Tgfbr2^KO tumors contain on average 3.5 times more Th17 cells and also CD4⁺ T cells isolated from knockout tumors produced more IL-17s (Fig. 3C).

Because we found that the number of Th17 cells in tumor tissue is increased in PyMT/Tgfbr2^KO mice, we hypothesized that myeloid cells from tumor tissue of these mice have increased expression of protumorigenic genes as the result of the increased presence of IL-17. First, we checked the possible role of IL-17 in functions of myeloid cells. We incubated MDSCs with IL-17 and found an up-regulation in the expression of genes that correlated with the suppressive function of MDSCs, such as Arg, matrix metalloproteinase (MMP)-9, IDO, cyclooxygenase (COX)-2, and MMP-13 (Fig. 3D). An increased suppressive function of MDSCs on T-cell proliferation also was observed after incubation with IL-17 (Supplementary Fig. S3D). In monocytes we found that IL-17 up-regulates gene expression corresponding to the M2 type of TAMs, such as VEGF, IL-10, IL-6, Arg, and IDO (Fig. 3D). Second, we sorted MDSCs and TAMs from tumor tissue and analyzed gene expression. We found that MDSCs from PyMT/Tgfbr2^KO mice exhibited increased Arg, MMP-13, and TGF-β expression, whereas TAMs from these mice exhibited increased Arg, MMP-9, and IL-1β expression compared with PyMT/Tgfbr2^fl mice (Fig. 3E). The data suggest that increased infiltration of myeloid cells mediated by CXCL1/5 to the tumor tissue of PyMT/Tgfbr2^KO mice switches antitumorigenic properties of these cells to protumorigenic in the presence of IL-17.

Osmotic pumps were implanted on the day of tumor palpation with anti–IL-17 Ab or with Rat IgG for the control group. After 2 weeks, the osmotic pumps were removed and the tumors were allowed to develop for an additional 2 weeks. On the 28th day after tumor palpation, analysis of tumor size, immune cell infiltration, and metastasis was performed (Fig. 4). It was found that treatment with anti–IL-17 Ab decreased tumor growth (Fig. 4B). The treatment effect on PyMT/Tgfbr2^KO mice was more pronounced. Analysis of lung metastases demonstrated that the number of foci in the lungs isolated from PyMT/Tgfbr2^KO mice treated with IL-17 blocking Ab decreased dramatically. Flow cytometry analysis of immune cells showed a decreased number of MDSCs in the spleens of PyMT/Tgfbr2^KO mice (Fig. 4C).

One possible explanation for the significant decrease in the MDSC population could be decreased secretion of CXCL1/5 from carcinoma cells, which can be stimulated by IL-17. To evaluate this hypothesis, the level of CXCL1 was measured in tumor tissue lysates, and a significant decrease in CXCL1 expression in PyMT/Tgbr2^KO mice after treatment with anti–IL-17 Ab was observed (Fig. 4D). Flow cytometry analysis showed a decreased percentage of CD11b⁺Gr1⁺ cells in tumor tissues of PyMT/Tgfbr2^KO mice (Fig. 4E). In addition we sorted MDSCs from tumor tissues of PyMT/Tgfbr2^KO mice and found decreased expression of the Arg, MMP-13, and TGF-β genes, which are the primary mediators of the suppressive function of MDSCs (Fig. 4F). Taken together, these data indicate that IL-17 plays a major role in tumor progression, through the modulation of MDSC migration and function, and the effect of IL-17 is increased in mice with deleted TβRII in carcinoma cells.

To determine the relevance of these findings to human breast cancer, we analyzed microarray profiles of human breast cancer tissues with well-documented clinical data related to lymph node (LN) involvement, estrogen receptor (ER) status, and time of relapse detection during a 10-year period. In a mouse model, we found that deletion of TβRII in epithelial cells correlated with increased Th17 response. Therefore, we first conducted a correlation analysis between expression of genes from TGF-β signaling and genes that associated with Th17 response. We found that only 2 Th17 genes had a significant correlation with TGF-β superfamily genes IL-17A and IL-17F (data not shown). A large number of genes associated with TGF-β signaling had a negative correlation with IL-17A/F (data not shown).

Second, we analyzed the correlation between IL-17 and the CXCL1/5 genes and found a very strong positive correlation between IL-17A and CXCL1 and IL-17A and CXCL5 (r = 0.25, P = 0.0001 and r = 0.22, P = 0.0007, respectively; also, there is a marginal effect between IL-17RA/CXCL1 [r = 0.12, P = 0.07] and IL-17RC/CXCL5 [r = 0.12, P = 0.07]). These findings indicate that in human breast cancer, as well as in mouse models, decreased TGF-β signaling is correlated with increased CXCL1/5 genes and Th17 response.

Next, we examined the association of Th17 genes with tumor subtype and relapse-free survival (RFS) by testing whether LN status and ER status would modify the association (Fig. 5). Three genes from Th17 family had a significant interaction with RFS: IL-17RC, IL17RA, and IL-17F. IL17-RC was associated with reduced RFS in patients with LN⁺ breast cancer. IL17F had a significant association with reduced RFS in patients with luminal B subtype breast cancer. IL17RA had marginal association with increased RFS in ER⁺ patients, but this association was not significant in ER⁻ patients. RFS in ER⁻ patients with IL17RA is opposite to ER⁺ cancer patients, suggesting a potential role of IL17RA to decrease RFS in ER⁻ patients.

Our data suggest that some particular genes in human breast cancer may correlate with Th17 response and associate with reduced RFS in subsets of breast cancer patients. The nonsignificant correlations or marginal effects may be attributable to the lack of study power or the heterogeneity of the human data, and we conclude that further testing in larger human data sets may be necessary to confirm that our findings have a direct impact on the patient population. Collectively, our findings indicate a strategy for targeting tumor growth and tumor metastasis by disrupting Th17 cell development in selected breast cancer cases.

An important role of MDSCs in tumor-associated immune response has been established in a large number of studies. Here, we have shown the important role of TGF-β signaling in carcinoma cells in the stimulation of MDSC migration to the tumor site and enhanced protumorigenic properties of these cells through increased numbers of Th17 cells. We found significantly increased numbers of MDSCs in spleen, bone marrow, lung, and tumor tissue during the course of tumor progression. The role of MDSCs in metastasis initiation and formation has been well established (19–22). We suggest that the presence of increased MDSCs in both the tumor and lung of PyMT/Tgfbr2^KO mice is a basic mechanism of acceleration of tumor growth and increased lung metastasis.

Previously, we found that MDSCs in tumor tissue can express high levels of CXCR4 in PyMT/Tgfbr2^KO mice (19). In this study we found that CD11b⁺Gr1⁺ cells from tumor and lung did not express CXCR2. In vitro analysis of MDSCs indicates that CXCR2 is internalized as cells migrate to areas with a high concentration of CXCL1 and/or CXCL5. We suggest that the high level of CXCL1/5 secretion is the primary cause for CXCR2 internalization by tumor infiltrating MDSCs, and this finding suggests that CXCL1/5-CXCR2 axis plays a major role in the recruitment of MDSCs into tumor tissue.

Secretion of the CXCL1/5 chemokines can be induced by IL-1, LPS, TNF-α (23), prostaglandin E2 (24), adenosine (25), IL-17 (12), and other factors.

In parallel with increasing production of CXCL chemokines, IL-17 increases the local production of chemokines such as IL-8 (26) and MCP-1 (27), thereby promoting the recruitment of myeloid cells (28). In this study, we demonstrated that IL-17 stimulates secretion of CXCL1/5 from carcinoma cells, but TGF-β inhibits IL-17–stimulated secretion from carcinoma cells with an intact TGF-β signaling pathway. This finding indicates an important role of TβRII in regulation of IL-17–stimulated secretion of chemokines. Moreover, we suggest that signaling through TβRII regulates expression of CD217 and sensitivity to IL-17 in tumor cells.

Although IL-17 production is increased in inflammatory reactions and is considered to be an inflammatory cytokine-promoting factor in tumor development (29, 30), it was not known whether IL-17 has an effect on MDSCs in the tumor-bearing hosts. To address this question, He and colleagues (18) showed that IL-17 is required for the development of MDSCs in tumor-bearing mice. By using IL-17R^−/− mice and Ad-IL-17 mice, they showed that IL-17 can regulate expression of Arg-I, A100/A8/A9, and MMP9 molecules (18), which are known to be mediators of MDSC-mediated immunosuppression and tumor promotion (31, 32). However, this effect could be attributable to the systemic effects of loss or gain of IL-17 function.

In this study, by using a different approach, we showed that IL-17 increases expression of Arg, MMP-9, IDO, COX-2, and MMP-13 (Fig. 4). Furthermore, after incubating MDSCs with IL-17, these cells have a significantly increased immunosuppressive effect on T-cell proliferation. Also, we found that IL-17 up-regulates gene expression corresponding to M2 type TAMs. Recent work by DeNardo and colleagues (33) using the same mouse model as in our work, but with intact TGF-β signaling, showed an important role for T cells in tumor progression. IL-4 secreted by Th2 cells can switch the phenotype of TAMs from M1 to M2. Moreover, they showed that activation of TAMs by IL-4, in combination with factors derived from mammary epithelial cells (MEC), such as SDF1, regulates expression of EGF, which in turn stimulates EGFR-induced MEC invasive behavior in vitro and MEC entry into peripheral blood and pulmonary metastasis in vivo.

We found additional mechanisms in mammary tumor progression in mice with the loss of TGF-β signaling in carcinoma cells. The early stages of carcinoma progression are dependent on Th17 cells, which secrete IL-17 to switch the phenotype of TAMs to M2 and increase the number and suppressive activity of MDSCs. Late stages of tumor progression can be dependent on Th2 cells, which in parallel with increased numbers of MDSCs, increase protumorigenic and metastasis properties of M2 TAMs.

These data demonstrate that IL-17 signaling is required not only for increased migration of MDSCs by up-regulating CXCL1/5 but also for the tumor-promoting and immunosuppressive activity of myeloid cells. The administration of IL-17 Ab in vivo during tumor growth was used to support our hypothesis that deletion of TβRII in PyMT carcinoma cells increases tumor growth and metastasis by increasing the host's Th17 response. We demonstrated that administration of anti–IL-17 Ab significantly decreased tumor growth in both groups of mice with an even more significant reduction in PyMT/Tgfbr2^KO mice. The decreased number of metastases was more pronounced and significant in PyMT/Tgfbr2^KO mice (31.67 ± 6.01 with IgG vs. 7.33 ± 1.86 with anti–IL-17), which demonstrated an increased Th17 response in mice with abrogated TGF-β signaling.

It was recently shown that a gene expression signature associated with complete abrogation of TGF-β signaling correlated with reduced RFS in breast cancer patients; moreover, the strongest association was observed in patients with ER⁺, specifically within the luminal A subtype (5). It has also been reported that IL-17 associated with macrophages is increased in LN⁺grade 3 human breast cancer (34). Nam and colleagues (14) found increased IL-17 mRNA in stage II and stage III human breast cancer samples compared with normal breast. Langowski and colleagues (35) showed significant up-regulation of IL-23(p19) mRNA in the overwhelming majority of carcinoma samples from various organ types when compared with their adjacent normal tissue. The expression of IL-17 was also found to be significantly elevated in human tumors, consistent with activation of IL-23–induced processes.

On the basis of these findings, we hypothesized that human breast cancer associated with decreased TGF-β signaling would also correlate with increased Th17 response. We observed that increased Th17 response in human breast cancer patients correlated with reduced RFS in patients with ER⁻ tumors, LN⁺, specifically within the luminal B subtype, and decreased TGF-β signaling in human breast cancer also correlated with enhanced Th17 response similar to the mouse studies.

In summary, our results demonstrated that deletion of TβRII in carcinoma cells promotes mammary tumor growth by increasing the number of infiltrating Th17 cells. The mechanism behind this is increased tumor secretion of CXCL1/5. The result of increased cytokine secretion is an increase in the number of infiltrating MDSCs and TAMs in the tumor tissue. These infiltrated myeloid cells may enhance differentiation of Th17 cells through increased secretion of IL-6, TGF-β, and IL-23. Th17 cells in turn act to induce CXCL1/5 in mammary carcinoma cells through IL-17 signaling, which is regulated by TβRII.

IL-17 secreted by Th17 cells has a tumor-promoting effect by up-regulating CXCL1/5 secretion by epithelial cells and increasing the protumorigenic properties of myeloid cells (Fig. 6). Our studies provide insights into a novel mechanism by which epithelial TGF-β signaling modulates the tumor microenvironment and is involved in tumor progression. The importance of TGF-β signaling in tumor progression is irrefutable and has been implicated in a variety of cancers and at numerous stages of tumor progression. Because the effects of pharmacological targeting of the TGF-β pathway in vivo on tumor progression remain controversial as a result of the dual roles of TGF-β in tumor progression, the targeting of TGF-β effects is a viable option. Because IL-17 has protumorigenic effects on breast cancer, the targeting of this cytokine could be considered when it is associated with impaired TGF-β signaling in carcinoma cells.

4T1 breast cancer cell line (CRL-2539) was obtained from ATCC and was maintained via the manufacturer's protocols. Tgfbr2^fl and Tgfbr2^KO carcinoma cell lines were derived from primary tumors of MMTV-PyMT/Tgfbr2^flox/flox and MMTV-PyMT/Tgfbr2^KO mice, respectively, established and cultured in DMEM/F12 with 5% adult bovine serum as previously described (5, 19).

All studies were performed on MMTV-PyMT/Tgfbr2^flox/flox (PyMT/Tgfbr2^fl) and MMTV-PyMT/Tgfbr2^KO (PyMT/Tgfbr2^KO) mice, which were established and maintained as described (3). MMTV-Cre mice were used to delete Tgfbr2 in the mammary epithelium. These mice have pure FVB background and have spontaneous tumor formation of mammary gland on the 28th to 34th day of age. For additional experiments, 8- to 10-week-old female C57bl/6 and FVB mice were purchased from Harlan Inc. To examine the effect of neutralizing IL-17 on tumor growth, mice were implanted with osmotic pumps (DURECT Corporation) with normal rat IgG or rat anti-mouse IL-17 mAb (600 ng/hr for 2 weeks; R&D Systems). Pumps were implanted on day of tumor palpation and removed after 2 weeks. The studies were approved by IACUC at Vanderbilt University Medical Center.

Single-cell suspensions were made from spleens, bone marrow, and lungs of normal and tumor-bearing mice (25, 36), and tumor tissues (37). These cells were labeled with fluorescence-conjugated Abs (Biolegend; eBiosciense, BD) and isotype-matched IgG controls. The cells were analyzed on LSRII flow cytometry (Becton Dickinson) and the data were analyzed with FlowJo software. For intracellular staining, cells were fixed and permeabilized by the use of BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit following the manufacturer's protocol. T cells were stimulated with Dynabeads Mouse T-Activator CD3–CD28 (Invitrogen) or with Leukocyte Activation Cocktail, with BD GolgiPlug (BD) following the manufacturer's protocols.

Tumor-infiltrating host immune cells and T cells were magnetically separated from tumor tissue of mammary gland by use of CD45 or CD4 magnetic microbeads following application protocols of the manufacturer (Miltenyi Biotec).

Splenocytes from tumor-bearing mice and single-cell suspensions of tumor tissue were stained with fluorescence-labeled Abs and sorted with a FACSAria flow cytometer (Becton Dickinson). EpCAM⁺CD45⁻, CD11b⁺Gr1⁺, CD11b⁺Gr1⁻F4/80⁺, and CD4⁺CD45⁺CD11b⁻DAPI⁻ cells were collected for gene expression and cytokine secretion analysis.

Mice were sacrificed by anesthetic overdose. Lungs were processed as described (38). The tumor nodules in lung were then counted.

T-cell proliferation was measured using a mixed leukocyte reaction. Dendritic cells from C57BL/6 mice were differentiated from bone marrow cells for 6 days with granulocyte macrophage colony-stimulating factor (10 ng/mL) and IL-4 (10 ng/mL) in RPMI 1640 medium. T cells from naïve FVB mice were isolated by the use of R&D Systems Mouse T-cell Enrichment columns, according to the manufacturer's protocol. T cells were plated at 10⁵ T cells per well in 96-well plates. Each well contained 100,000 T cells, 25,000 MDSCs, and different numbers of dendritic cells. Cells were incubated for 72 hours. ³[H]-thymidine was then added at 1 μCi per 200 μL of cells per well for an additional 18 hours followed by cell harvesting and radioactivity count by the use of a liquid scintillation counter.

Tumor explants were prepared from freshly isolated tumors. Spontaneous tumors from Tgfbr2^fl and Tgfbr2^KO mice, without ulceration, approximately 1.5 cm in diameter were removed under sterile conditions after euthanizing the mouse. Tumors were minced into pieces <3 mm in diameter and digested in 1 mg/mL Collagenase Type I/Dispase Type II at 37°C for 2 hours. The digested tissue pieces were then pressed through a 70-μm mesh screen to create a single-cell suspension. Cells were washed with PBS and resuspended in DMEM supplemented with 200 U/mL penicillin plus 50 μg/mL streptomycin and 5% FBS. Cells were cultured overnight at 10⁷ cells/mL and the cell-free supernatant collected.

Total RNA was extracted from sorted CD11b⁺Gr1⁺, CD11b⁺Gr1⁻F4/80⁺, and CD4⁺ cells as described by the use of an RNeasy Mini Kit (QIAGEN). cDNA was synthesized with Invitrogen Superscript First-strand synthesis system for RT-PCR (Invitrogen). Primers specific for Arg, MMP-9, IDO, COX-2, IL-1β, TGF-β, MMP-13, IL-23, IL-6, IL-17a, IL-17f, IL-21, chemokine ligand 20, IL-10, VEGF, granulocyte macrophage colony-stimulating factor, and IFN-γ were used, and relative gene expression was determined with the use of the ABI PRISM 7900HT Sequence Detection System (PE Applied Biosystems). The comparative threshold cycle method was used to calculate gene expression normalized to β-actin as a gene reference. Primer sequences are available upon request.

Cytokine levels in CD4⁺, CD45⁺, CD45⁻, and Ep-CAM⁺CD45⁻ cell supernatants or tissue lysates were measured using the Mouse TGF-β1, IL-6, IL-23, IL-17A/F, and CXCL1/5 ELISA kits (R&D Systems) following the manufacturer's protocol.

Data were presented as mean ± SEM. Multiple comparisons between treatment groups and control untreated group were performed by the use of 1-way ANOVA followed by the Dunnett procedure for multiplicity adjustment. A 2-group comparison was performed with the use of 2-sample t tests. Correlation analyses for the gene expression between the signatures of interest were performed with data representing 1,319 patients from 4 independent, previously reported studies (Gene Expression Omnibus ID: GSE10886, GSE4922, GSE6532, and GSE2845) (5).

The gene expression was appropriately normalized across all arrays. Gene symbols were assigned with the manufacturer-provided annotation, but the analysis was performed at probe level. The association between RFS and IL17 family was analyzed with the univariate Cox proportional hazard regression model. We examined the modification effect of ER status and node status by testing the interaction between the IL-17 gene and ER or node status by using the multivariate Cox proportional hazard model. When a significant interaction was identified, a Kaplan-Meier survival curve was created for the “high versus low” of the IL-17 gene expression by dichotoming all the samples among the subtype (Fig. 5). All tests were 2-tailed. All statistical analyses used a P-value cutoff of 0.05 to determine significance. All data analysis for human breast cancer correlation was performed in R (39).

No potential conflicts of interest were disclosed.

Supported by NIH grants CA085492, A102162, and CA126505 and the T. J. Martell Foundation to H.L. Moses and NIH grant CA068485 for core laboratory support.