Expression of IDO in melanoma is associated with aggressive tumors

In contrast to the high levels of IDO expression observed in human melanoma tissue (Fig. 1A), we did not detect IDO expression in the well-described murine B16 melanoma cell line (Fig. S2A). This observation may reflect the heterogeneity of IDO expression in human tumors. Thus, in order to model melanoma cells expressing IDO, we generated a B16 melanoma cell line overexpressing IDO-GFP fusion protein (B16-IDO) and a control cell line overexpressing GFP alone (B16-WT). Expression of IDO was confirmed by flow cytometry and Western blot analysis (Fig. S2A and S2B). When injected into naive mice, the B16-IDO tumors grew significantly faster and produced larger tumors (Fig. 2A). IDO expression did not influence the growth of B16 cells in vitro (Fig. S2C), or in irradiated naive mice (Fig. 2A), suggesting that the enhanced growth of the B16-IDO tumors was not related to a higher intrinsic growth rate of these cells. The levels of tryptophan and its IDO-generated degraded product, L-kynurenine, were evaluated by ELISA in blood and tumor supernatants, demonstrating decreased level of tryptophan and increased level of L-kynurenine in the B16-IDO tumors, but not in peripheral blood (Fig. 2B). To determine whether constitutive expression of IDO by B16 cells carried any implications for efficacy of cancer immunotherapy, animals bearing B16-IDO or B16-WT tumors were treated with antibodies to CTLA-4, PD-1, or combination of both. While combination therapy led to 60% long-term animal survival in the B16-WT tumor-bearing mice, B16-IDO tumors were completely resistant to therapy (Fig. 2C). Altogether, these results suggest that IDO expression by B16 tumor cells promotes tumor growth and resistance to immune checkpoint blockade immunotherapy.

Open in a separate window

Figure 2

Expression of IDO promotes tumor growth and resistance to immune checkpoint blockade

(A) Mean weight and average growth rate of B16-WT and B16-IDO tumors in naive and irradiated mice. (B) Levels of tryptophan and L-kynurenine in tumor and blood evaluated by ELISA. (C) Mean size of αCTLA-4 + αPD-1 treated B16-IDO and B16-WT tumors and long-term survival. Data are represented as mean ± SEM.

IDO inhibits proliferation and accumulation of tumor-specific T cells at the tumor site

To assess tumor-specific immune responses in the B16-IDO and B16-WT tumors, mice received a labeled cohort of activated transgenic CD8⁺ T cells, recognizing a peptide from the melanosomal antigen gp100 (pmel). In mice with B16-WT tumors, we observed a robust proliferation and increase in number of pmels in TDLNs as well as in tumors (Fig. S2D). In contrast, pmels had significantly lower accumulation in the B16-IDO tumors and TDLNs (Fig. S2D). These data indicate that tumor-expressed IDO leads to inhibition of proliferation of tumor-specific T cells, and prevents their accumulation at the tumor site, which may be responsible for the observed enhanced tumor progression.

Suppressive myeloid cells are expanded in IDO-expressing tumors

To establish the mechanism responsible for the inhibition of tumor-specific T cells in the IDO-expressing tumors, we analyzed the tumor immune composition. While the absolute numbers of CD45⁺ cells in the B16-IDO and B16-WT tumors were comparable (Fig. S3A), the relative frequency of CD3⁺ cells was significantly lower in the B16-IDO tumors (Fig. 3A). This decrease in CD3⁺ T cells corresponded to a diminished infiltration of both CD8⁺ and CD4⁺ Teff cells (Fig. 3B), but a notable increase in Foxp3⁺ Tregs (Fig. 3B). Activation of Tregs in tumor models in vivo has previously been linked to IDO so this finding was not surprising (Sharma et al., 2007). There were also decreased percentages of CD19⁺ B cells, NK cells, and natural killer T cells (NKT) noted at the tumor site, but these differences did not reach statistical significance (Fig. 3A and B). Importantly, there was a significant increase in myeloid CD11b⁺ cells in the B16-IDO tumors compared to B16-WT tumors (Fig. 3A), resulting in decreased intratumoral ratio of CD3⁺ T cells to CD11b⁺ cells (Fig. 3C). Of note, the CD11b⁺ cells isolated from the B16-IDO tumors, but not from the B16-WT were capable of suppressing CD8⁺ T-cell proliferation in vitro (Fig. 3D), suggesting that these cells may play a role in the observed aggressiveness of the B16-IDO tumors.

Open in a separate window

Figure 3

IDO promotes expansion of myeloid CD11b⁺ cells in tumors

(A) B16-IDO and B16-WT tumors were analyzed for infiltrating immune cells. Results for CD3⁺ T cells, CD19⁺ B cells, CD3⁻NK1.1⁺ NK cells, and CD11b⁺ myeloid cells as frequencies of CD45⁺ cells and representative dot plots. (B) Results for T cell subsets: NK1.1⁺ T cells (NKT), CD4⁺Foxp3⁺ Tregs, CD4⁺Foxp3⁻ Teff cells, and CD8⁺ T cells as frequencies of CD45⁺ cells. (C) CD3⁺/CD11b⁺ ratios. (D) In vitro suppressive activity of CD11b⁺ cells from B16-IDO or B16-WT tumors. Representative histograms of CD8⁺ T-cell proliferation are shown in CD11b⁺/CD8⁺ ratio of 1:1. (E) Average growth and mean weight of IDOi and vehicle treated B16-IDO tumors. (F) Relative percentages of CD11b⁺ and CD3⁺ cells of CD45⁺ cells in IDOi and vehicle treated B16-IDO tumors and representative flow plots. (G) Calculated CD3⁺/CD11b⁺ ratios. (H) Percent of tumor-infiltrating CD4⁺ and CD8⁺ T cells gated on CD45⁺ population and representative plots. (I) Example of CXCR3 staining within the CD8⁺ gated population. (J) Representative plots and graphs showing percentage pmels of total CD8⁺ T cells in LN, TDLN, spleen, and tumor, and proliferation of pmels in tumor and TDLNs, of IDOi and vehicle treated B16-IDO tumor-bearing mice. Data are shown as mean ± SEM.

Inhibition of IDO blocks expansion of myeloid CD11b⁺ cells in tumors

To determine whether the expansion of CD11b⁺ cells in B16-IDO tumors was indeed caused by IDO, mice with B16-IDO tumors were treated with the IDO inhibitor indoximod (IDOi). Therapy with IDOi reduced the growth rate and size of B16-IDO tumors significantly (Fig. 3E), which was associated with decrease in the number of CD11b⁺ cells and increase in CD3⁺ T cells in tumors (Fig. 3F), resulting in increased CD3⁺/CD11b⁺ ratios (Fig. 3G). Furthermore, treatment with IDOi led to an increase in CD4⁺ and CD8⁺ T cells in the tumor (Fig. 3H) with upregulation of the Teff cell marker CXCR3 on the tumor-infiltrating CD8⁺ T cells (Fig. 3I). To assess for tumor-specific responses, we adoptively-transferred activated pmels. IDOi treatment resulted in increase and significant proliferation of pmels within the TDLNs and tumors of B16-IDO tumor-bearing mice (Fig. 3J), and led to significantly reduced tumor growth (Fig. S3B). These results indicate that tumor IDO expression not only favors expansion of CD11b⁺ but also blocks trafficking and proliferation of tumor-specific T cells at the tumor site.

CD11b⁺ cells infiltrating IDO-expressing tumors exhibit MDSC properties

In order to evaluate the distinct suppressive abilities of the CD11b⁺ cells found in the B16-IDO and B16-WT tumors, we initially analyzed expression of the myeloid differentiation antigen Gr1 (Ly6G). It has previously been demonstrated that the intensity of Gr1 staining identifies CD11b⁺ cells with functional heterogeneity (Gabrilovich and Nagaraj, 2009), and consisted of two major subsets with different Gr1 expression levels, termed Gr1^high and Gr1^int cells, but not Gr1^high cells being suppressive (Dolcetti et al., 2010). The majority of the tumor-infiltrating CD11b⁺ cells of both B16-IDO and B16-WT tumors expressed Gr1 (Fig. 4A). Interestingly, the increase in myeloid CD11b⁺ cells observed in the B16-IDO tumors corresponded to a predominant increase in the Gr1^int population (Fig. 4A). We found that CD11b⁺Gr1^int but not CD11b⁺Gr1^high cells purified from B16-IDO tumors were capable of suppressing T-cell proliferation (Fig. 4B). Interestingly, CD11b⁺Gr1^int cells isolated from B16-WT tumors were incapable of suppressing CD8⁺ T-cell proliferation (Fig. 4B). The CD11b⁺Gr1^int subsets from B16-IDO and B16-WT tumors were then tested for differential expression of various surface markers, which have previously been described in tumor-derived CD11b⁺ cells with immunosuppressive properties (Dolcetti et al., 2010). Phenotypically, CD11b⁺Gr1^int cells from both B16-IDO and B16-WT tumors showed low levels of MHC class II and CD11c and intermediate expression of F4/80 (Fig. 4C), confirming their status as immature myeloid cells and distinguishing them from mature myeloid cells such as dendritic cells and macrophages. Furthermore, this population showed expression of the typical monocyte marker Ly6C (Fig. 4C). Thus, while we were not able to distinguish the two CD11b⁺Gr1^int populations based on the majority of phenotypic markers, the cells demonstrated clear functional differences with regards to immunosuppression. To this end, we observed that expression of IL-4Rα, which has been suggested as a functional marker of immunoregulatory CD11b⁺ cells (Gallina et al., 2006), was higher in the CD11b⁺Gr1^int population from B16-IDO tumors (Fig. 4D), and thus, IL-4Ra was the only predictive marker of the suppressive activity. Taken together, these data indicate that tumor IDO expression results in expansion of CD11b⁺Gr1^int cells, which possess immunosuppressive properties, consistent with their functional definition as MDSCs. These results support our initial findings in the patient tumor samples and thus highlight an important link between IDO expression and MDSCs in cancer.

Open in a separate window

Figure 4

Phenotype and suppressive capabilities of tumor-infiltrating myeloid CD11b⁺ cells

(A) CD11b⁺ subsets obtained from B16-IDO and B16-WT tumors were evaluated for CD11b versus Gr1 expression and graphed as percent Gr1^int and Gr1^high cells. (B) Suppressive properties of CD11b⁺Gr1^high and CD11b⁺Gr1^int cells from B16-IDO or B16-WT tumors. Representative histograms of CD8⁺ T-cell proliferation are shown in CD11b⁺ to CD8⁺ T cell ratio of 1:1. (C) CD11b⁺Gr1^int subsets from B16-IDO and B16-WT tumors were evaluated for expression of F4/80, CD11c, MHC class II and Ly6C markers (open histograms) against their matched isotype controls (filled histograms). (D) IL-4Rα expression (open histograms) within the CD11b⁺Gr1^int gated populations against matched isotype control (filled histograms). (E) Arg1 activity in CD11b⁺Gr1^high and CD11b⁺Gr1^int subsets from B16-IDO and B16-WT tumors. (F) Nitrite/NO concentration in supernatants of LPS stimulated CD11b⁺Gr1^high and CD11b⁺Gr1^int cells. (G) Suppressive activity of CD11b⁺Gr1^int cells from B16-IDO tumors in the presence of inhibitors of Arg1 (nor-NOHA), iNOS (LNMMA), TGFβ (anti-TGFβ) and/or PD-L1 (anti-PD-L1). Data are shown for CD8⁺ to CD11b⁺Gr1^int ratios of 1:1. (H) Suppressive activity CD11b⁺ cells from B16-IDO tumors measured with CD11b⁺ cells in contact with CD8⁺ T cells or placed in 0.4 μm cell culture inserts. Data are shown for CD8⁺ to CD11b⁺ ratios of 1:1. Data are represented as mean ± SEM.

B16-IDO tumor-infiltrating MDSCs upregulate arginase, iNOS and production of immunosuppressive cytokines

To gain further insight into the different suppressive character of CD11b⁺Gr1⁺ cells found in B16-IDO and B16-WT tumors, we measured the production and expression of factors that have been linked to the immunosuppressive activity of MDSCs (Gabrilovich et al., 2012). We observed that CD11b⁺Gr1^int cells sorted from B16-IDO tumors displayed significantly higher arginase-1 (Arg1) and NO production as compared to CD11b⁺Gr1^int cells from B16-WT tumors or to their granulocytic CD11b⁺Gr1^high counterparts (Fig. 4E and F). In contrast, CD11b⁺Gr1^high subsets produced highly elevated levels of ROS compared with CD11b⁺Gr1^int cells, as evidenced by increased DCFDA incorporation (Fig. S4A). To verify that these molecules were indeed important for suppression, we measured T-cell proliferation with CD11b⁺Gr1^int cells obtained from B16-IDO tumors in presence of inhibitors of Arg1 and inducible NO synthase (iNOS). Indeed, combination of Arg1 inhibitor with iNOS inhibitor completely blocked the suppressive activity (Fig. 4G). It has also been suggested that activated MDSCs produce high levels of transforming growth factor-β (TGFβ) and mediate their suppressive activity through PD-L1 expression (Gabrilovich et al., 2012). To this end, the addition of either αTGFβ or αPD-L1 antibody to Arg1 and iNOS inhibitors further augmented T-cell proliferation (Fig. 4G). Using a transwell system, we further found that inhibition of CD8⁺ T-cell proliferation was contact-dependent (Fig. 3H). Importantly, increased expression of Arg1, NO, TGFβ, PD-L1, and IL-4Rα was confirmed on mRNA level by real time PCR (RT-PCR) (Fig. S4B). The upregulation of these molecules in CD11b⁺Gr1^int cells from B16-IDO tumors compared to CD11b⁺Gr1^int cells from B16-WT tumors could explain their distinct suppressive capacity.

Activation of MDSCs is dependent on tumor-expressed IDO

We next proceeded to evaluate whether IDO was also required for tumor-infiltrating CD11b⁺Gr1^int cells to be activated and become functional MDSCs. Treatment of B16-IDO tumor-bearing animals with IDOi revealed a significant reduction in intratumoral CD11b⁺Gr1^int cells (Fig. 5A), with a concomitant decrease of IL-4Rα expression (Fig. 5B) and loss of suppressive activity in this subset (Fig. 5C). Thus, inhibition of IDO did not only reduce the frequency of tumor-infiltrating MDSCs, but also inhibited their capability to suppress T-cell proliferation, indicating that activation of MDSCs in tumors requires IDO. Addition of IDOi in vitro on the other hand, did not restore T-cell proliferation (Fig. 5C), indicating that MDSCs do not exert their suppressive capacity through expression of IDO. In support of the latter, B16-IDO tumors grown in IDO^−/− mice exhibited a similar aggressive behavior and inhibition of IDO in these animals also reduced tumor growth (Fig. 5D) and increased ratios of intratumoral CD3⁺ to CD11b⁺Gr1⁺ cells (Fig. 5E), indicating that the contribution of host-derived IDO is absent or minor in this model.

Open in a separate window

Figure 5

Activation of MDSCs is dependent on IDO expression

(A) Frequencies of Gr1^high and Gr1^int subsets of total CD11b⁺ cells from B16-IDO tumors of IDOi or vehicle treated mice. (B) MFI for IL-4Rα expression within the CD11b⁺Gr1^int population. (C) T cell suppression assay with CD11b⁺Gr1^int cells purified from B16-IDO tumors in the presence of an IDOi added in vitro or administered in vivo. (D) Growth curve for B16-IDO tumors in IDOi treated IDO^−/− mice. (E) Ratio of CD3⁺ to CD11b⁺Gr1⁺ cells in B16-IDO tumors in IDOi treated IDO^−/− mice. (F) Average growth rate and mean weight of IDOi and vehicle treated 4T1 tumors. (G) Relative percentages of Gr1⁺ of total CD11b⁺ cells in 4T1 tumors after IDOi therapy. (H) IL-4Rα expression within the CD11b⁺Gr1^int gated population of TILs from untreated 4T1 tumor-bearing mice. Data are represented as mean ± SEM.

MDSCs are expanded in a murine model expressing endogenous IDO

We next sought to determine whether IDO also drives the expansion of MDSCs in a tumor model naturally expressing high levels of IDO. Analysis of several murine cancer cell lines revealed high levels of IDO expression in the 4T1 breast cancer cell line when implanted in vivo (Fig. S5). Interestingly, this model has previously been shown to expand immunosuppressive CD11b⁺Gr1^int cells (Dolcetti et al., 2010). Treatment of 4T1 tumor-bearing mice with IDOi reduced the tumor growth rate (Fig. 5F). This was accompanied by a significant decrease in the intratumoral CD11b⁺Gr1^int cells, but not CD11b⁺Gr1^high cells (Fig. 5G). Notably, CD11b⁺Gr1^int cells purified from the 4T1 tumors were characterized by expression of IL-4Rα (Fig. 5H), suggesting that this cell population is indeed immunosuppressive.

Tumor IDO-induced expansion of MDSCs is systemic

Expansion of MDSCs has been previously described in the spleens of EL-4, CT26 and 4T1 mouse tumor models (Dolcetti et al., 2010; Youn et al., 2008), suggesting that tumors drive systemic MDSC expansion. Similar to these findings, analysis of spleens from B16-IDO tumor-bearing mice revealed expansion of CD11b⁺ cells (Fig. 6A), with predominance of Gr1⁺ cells (Fig. 6B). Surprisingly, while the CD11b⁺Gr1^int splenocytes from the 4T1 and EL-4 tumor-bearing mice were shown to possess immunosuppressive activity (Dolcetti et al., 2010; Youn et al., 2008), the CD11b⁺Gr1^int cells from spleens of B16-IDO tumor-bearing mice did not suppress CD8⁺ T-cell proliferation (Fig. 6C). While these splenic CD11b⁺Gr1^int cells were phenotypically similar to the corresponding CD11b⁺Gr1^int cells isolated from B16-IDO tumors (Fig. S6A), they produced significantly lower levels of Arg1 (Fig. S6B) and NO (Fig. S6C) and showed lower expression of IL-4Rα (Fig. S6D). The splenic expansion of CD11b⁺Gr1^int cells could be significantly reduced when mice were treated with IDOi (Fig. S6E), further confirming the role of IDO in controlling this process.

Open in a separate window

Figure 6

Local and systemic effects of tumor IDO expression

(A) Frequencies of CD11b⁺ myeloid cells and CD3⁺ T cells of total CD45⁺ cells in splenocytes from B16-IDO and B16-WT tumor-bearing mice and representative dot plots. (B) Expression of Gr1 within the CD11b⁺ gated population and representative plots. (C) Suppressive properties of CD11b⁺Gr1^int splenocytes from B16-IDO tumor-bearing mice. (D) Bilateral flank tumor models: (1) B16-WT tumors on both flanks, (2) B16-IDO tumors on both flanks, and (3) B16-IDO tumor on one flank and B16-WT on the contralateral flank. (E) Mean growth and weight of bilateral flank tumors. (F) Infiltration of CD11b⁺Gr1^int and CD11b⁺Gr1^high myeloid cells in bilateral flank B16-IDO and B16-WT tumors. (G) Suppressive activity of CD11b⁺Gr1^int cells from B16-IDO or B16-WT tumors of bilateral flank tumor-bearing mice and representative histograms. (H) MFI for IL-4Rα expression within CD11b⁺Gr1^int and CD11b⁺Gr1^high gated populations. (I) Frequencies of CD11b⁺ and CD3⁺ of CD45⁺ total cells and CD3⁺/CD11b⁺ ratios in bilateral flank B16-IDO and B16-WT tumors, and representative plots. (J) Frequencies of CD8⁺, CD4⁺Foxp3⁻, and CD4⁺Foxp3⁺ cells of CD45⁺ total cells in bilateral flank B16-IDO and B16-WT tumors. (K) Mean tumor growth of αCTLA-4 + αPD-1 and/or IDOi treated B16-IDO and B16-WT tumors in bilateral flank tumor models. (L) Migration of CD11b⁺Gr1^int splenocytes purified from B16-IDO and B16-WT tumor-bearing mice towards homogenized tumors, sorted tumor cells or tumor cell lines as indicated. (M) Suppressive activity of CD11b⁺Gr1^int splenocytes of B16-IDO and B16-WT tumor-bearing mice pre-cultured with homogenized tumors, sorted tumor cells or tumor cell lines. CD11b⁺Gr1^int cells were pre-cultured in contact with tumor cells or placed in 0.4 μm cell culture inserts as specified. (N) Relative levels of soluble factors in supernatants of B16-IDO tumors compared to B16-WT tumors, as determined by Luminex. Data are represented as mean ± SEM.

Recruitment of CD11b⁺Gr1^int cells and activation of immunosuppressive properties are dependent on the tumor microenvironment

The findings above suggested that while B16-IDO tumors induce systemic expansion of the CD11b⁺Gr1^int cells, the tumor microenvironment might be necessary to render them suppressive. To explore this possibility, we established bilateral flank tumor models bearing B16-IDO tumors in one flank and B16-WT tumors in the contralateral flank (Fig. 6D). Remarkably, B16-IDO tumors resulted in enhanced growth of B16-WT tumors on the contralateral flank (Fig. 6E), which was associated with increase in CD11b⁺Gr1^int cells in the B16-WT tumors (Fig. 6F), and acquisition of immunosuppressive properties by these cells (Fig. 6G). Consistent with their suppressive activity, these cells were characterized by increased expression of IL-4Rα (Fig. 6H). In addition, we observed lower percentages of CD3⁺ T cells in these tumors and as a result, a decrease in the intratumoral ratio of CD3⁺ T cells to CD11b⁺ cells (Fig. 6I). The observed decrease in CD3⁺ T cells was due to reduction of both CD4⁺Foxp3⁻ and CD8⁺ T cells, but not CD4⁺Foxp3⁺ Tregs (Fig. 6J). Notably, the antitumor effect of αCTLA-4 + αPD-1 combination therapy in the B16-WT tumor model was also completely lost in the B16-IDO/B16-WT bilateral flank tumor model, and this could be overcome by therapeutic inhibition of IDO with IDOi (Fig. 6K). Altogether, these results demonstrate that the IDO-driven expansion of CD11b⁺Gr1^int cells is indeed systemic, but require the effect of the tumor microenvironment for activation of their suppressive activity.

To understand the mechanisms of expansion of CD11b⁺Gr1^int cells in tumors, we analyzed Bromodeoxyuridine (BrdU) incorporation in CD11b⁺Gr1^int cells in bone marrow (BM), blood, spleen, and tumor tissue of B16-IDO tumor-bearing mice. High percentages of BrdU⁺ CD11b⁺Gr1^int cells were mainly detected in BM and in the spleens but not in tumors (Fig. S6F), suggesting that their accumulation in tumors is likely due to the recruitment from the periphery and less likely due to proliferation at the tumor site. In support of this hypothesis, splenic CD11b⁺Gr1^int cells exhibited an increased ability to migrate through collagen-coated culture inserts toward B16-IDO tumors in vitro, which was not observed with B16-WT tumors (Fig. 6L). This effect required all components of the tumor microenvironment to be present, as migration was significantly diminished toward tumor cells sorted from B16-IDO tumors and completely lost when B16-IDO tumor cell lines cultured in vitro were used as attractants (Fig. 6L).

These findings thus suggest that the CD11b⁺Gr1^int cells likely traffic into the tumor microenvironment from the periphery. To determine whether CD11b⁺Gr1^int cells are functionally activated by the B16-IDO tumors, CD11b⁺Gr1^int splenocytes purified from B16-IDO or B16-WT tumor-bearing mice were cultured with B16-IDO or B16-WT tumors and were tested for the capacity to suppress CD8⁺ T-cell proliferation. We found that CD11b⁺Gr1^int cells cultured in the presence of B16-IDO tumors but not B16-WT tumors were able to suppress CD8⁺ T-cell proliferation (Fig. 6M). The establishment of suppressive capabilities did not require direct contact between CD11b⁺Gr1^int MDSCs and tumors as suppression was also observed when CD11b⁺Gr1^int MDSCs were separated from the tumors by a porous membrane (Fig. 6M). CD11b⁺Gr1^int cells co-cultured with B16-IDO tumor cells sorted from tumors also acquired suppressive capability, albeit lower, while co-culture with the in vitro-cultured B16-IDO cell line had no effect (Fig 6M). These results indicate that CD11b⁺Gr1^int cells gain their suppressive capacity at the tumor site, and suggest that this is mediated by soluble factors, likely produced by different components of the tumor microenvironment.

The B16-IDO tumor microenvironment contains high levels of pro-inflammatory cytokines and chemokines

To identify the potential mediators of the preferential accumulation and activation of CD11b⁺Gr1^int MDSCs in the B16-IDO tumors, the expression of chemokines, inflammatory cytokines, and growth factors in tumors was analyzed, revealing increased expression of multiple factors in the B16-IDO tumors (Fig. 6N). The most differentially expressed factors were TNFα, IL-10, IL-4, IL-6, IL-2, IFNγ, IL-13, MIG, MCP-1, MIP-1α, IP-10 and 2 growth factors: VEGF and GM-CSF (p values < 0.01)(Fig. 6M). Several of these factors have been described to be implicated in recruitment of MDSCs and monocytes as well as expansion, activation and function of MDSCs, and are known to be produced by tumor cells, T cells or by activated MDSC themselves (Gabrilovich and Nagaraj, 2009). Interestingly, despite being immunosuppressive, the B16-IDO tumor microenvironment appeared to exhibit markedly higher levels of pro-inflammatory cytokines, suggesting that the recruitment and/or activation of MDSCs in the B16-IDO tumor could be a part of an adaptive immune response.

Accumulation of MDSCs in tumors is dependent on adaptive immunity

These results prompted us to examine the accumulation and functional capacity of CD11b⁺Gr1⁺ cells in B16-IDO tumors grown in Rag^−/− mice, deficient in T and B cells (Mombaerts et al., 1992). We observed that the absolute number of CD11b⁺Gr1⁺ cells in tumors was significantly reduced when B16-IDO tumors were implanted in Rag^−/− mice (Fig. 7A), while this reduction was not apparent in B16-WT tumors implanted in Rag^−/− mice (Fig. 7A). Furthermore, CD11b⁺Gr1⁺ cells purified from B16-IDO tumors harvested from Rag^−/− mice were no longer able to suppress proliferation of CD8⁺ T cells (Fig. 7B). Finally, the migration of splenic CD11b⁺Gr1⁺ cells towards B16-IDO tumors grown in Rag^−/− mice was completely lost (Fig. 7C). Together, these data show that the adaptive immune response, i.e. T and/or B cells, are likely required for recruitment and functional activation of MDSCs in B16-IDO tumors.

Open in a separate window

Figure 7

Accumulation of MDSCs in tumors is dependent on the adaptive immune response

(A) Absolute number of CD11b⁺Gr1⁺ cells per gram of tumor in B16-WT and B16-IDO tumors from WT and Rag^−/− mice. (B) In vitro suppressive activity of CD11b⁺Gr1^int cells purified from B16-IDO tumors of WT or Rag−/− mice measured in CD11b⁺Gr1^int to CD8⁺ T cell ratios of 1:1. (C) Migration of CD11b⁺Gr1^int splenocytes of B16-IDO tumor-bearing mice towards B16-IDO tumors harvested from WT or Rag−/− mice. (D) Frequency of Foxp3⁺ Tregs of total CD3⁺ T cells in B16-IDO and B16-WT tumors of Foxp3^GFP mice. Representative dot plots showing GFP-Foxp3⁺ expression within CD45⁺ gated populations. (E) Suppressive properties of Foxp3⁺ Tregs sorted from B16-IDO and B16-WT tumors of Foxp3^GFP mice. Representative histograms of CD8⁺ T-cell proliferation are shown in Foxp3⁺ Tregs to CD8⁺ T cell ratio of 1:1. (F) Frequency of CD11b⁺Gr1^int cells of total CD45⁺ cells in tumor, TDLNs and spleens of B16-IDO tumor-bearing Foxp3^DTR mice injected with DT or PBS. (G) Suppression assay with CD11b⁺Gr1^int cells purified from B16-IDO tumors of Foxp3^DTR mice injected with DT and PBS. Representative histograms showing CD8⁺ T-cell proliferation in CD8⁺ to CD11b⁺Gr1^int ratios of 1:1. (H) Migration of CD11b⁺Gr1^int splenocytes towards homogenized B16-IDO tumors from Foxp3^DTR, WT or Rag−/− mice. Tumors were pre-incubated with Treg-blocking antibody or tumor-infiltrating Treg and/or Teff cells in vitro, as indicated. (I) Frequency of CD8⁺CXCR3⁺ T cells of total CD3⁺ T cells B16-IDO tumors of Foxp3^DTR mice injected with DT or PBS. (J) Mean growth rate and weight of B16-IDO tumors in Foxp3^DTR mice injected with DT or PBS. (K) Percentage of CD4⁺Foxp3⁺ Tregs and CD69⁺CD8⁺ T cells in cell suspensions of IDO⁺ and IDO⁻ human metastatic melanoma tumors. (L) Representative IHC staining of IDO, CD14 and Foxp3 in tissue from metastatic human melanoma tumors. Data represents mean +/− SEM (A–E) or one representative experiment (F–J). Data in K–L are representative of 22 patients.

In previous studies, IDO expression has been demonstrated to induce and activate Foxp3⁺ Tregs (Munn and Mellor, 2007). As the only adaptive population that increased in the B16-IDO tumors was Foxp3⁺ Tregs (Fig. 3B), we hypothesized that the IDO-mediated activation of Tregs may be responsible for the observed MDSC expansion. To examine the role of Tregs in IDO-mediated MDSC recruitment, we first tested the frequency and suppressive capacity of the Tregs in the B16-IDO tumors. For this purpose, we used transgenic mice expressing Foxp3-GFP fusion protein (Foxp3^GFP knock-in mice), which allowed us to specifically detect and sort live Foxp3⁺ Tregs (Fontenot et al., 2003). Similar to what we observed in WT mice (Fig. 3B), the frequency and absolute number of Foxp3⁺ Tregs was significantly higher in B16-IDO tumors than B16-WT tumors when tumors were harvested from Foxp3^GFP mice (Fig. 7D). Moreover, Tregs purified from B16-IDO tumors were significantly more potent in suppressing CD8⁺ T-cell proliferation than Tregs from B16-WT tumors (Fig. 7E). These data show that tumor IDO expression changes both the quantity and the suppressive function of tumor-infiltrating Tregs.

Isolation of tumor-infiltrating cells and lymphoid tissue cells

Single cell suspensions from mouse tissue were generated and stained with indicated antibodies for immediate analysis and/or flow sorting as described in SEP.

T cell suppression assay

CD8⁺ cells were purified from single cell suspensions from spleens of naive mice using anti-CD8 (Ly-2) microbeads (Miltenyi Biotech) according to manufacturer’s protocol and labeled with appropriate CellTrace™ Reagent. The labeled CD8⁺ T cells were then plated onto round bottom 96-well plates coated with 1 μg/ml anti-CD3 (clone 1454-2C11) and 5 μg/ml anti-CD28 (clone 37N). Purified suppressor cells (i.e. MDSCs or Foxp3⁺ Tregs) were added in indicated ratios and plates were incubated at 37°C. Percent CD8⁺ T-cell proliferation was measured by assessing dilution of the CellTrace Dye by flow cytometry after 48–72 hours of culture (LSRII flow cytometer, BD Biosciences). Controls were wells without suppressor cells (Stim+) and wells without suppressor cells and anti-CD3/CD28 antibody (Stim−). Details and specific conditions are provided in SEP.

Cell migration

Dissociated tumor cells were placed in the bottom chamber of a 24-well cell culture plate. Purified MDSC splenocytes were labeled with 1 mM CellTrace™ Far Red DDAO-SE (Thermo Fisher Scientific) and place in the upper chamber (3 μm cell culture inserts with PET membranes, BD Falcon). MDSCs were allowed to migrate to the bottom well for 24 hours. Migrated and labeled cells were then analyzed by flow cytometry on a LSRII Flow Cytometer (BD Bioscience) and quantified using Count Bright Absolute Counting Beads (Invitrogen). Percentage migration was calculated as: (number of migrated cells) × 100/(total cells added per well). IL-6, IL-10, MCP-1, IFNγ, TNF, and IL-12p70 recombinant proteins (500 pg/ml, BD Bioscience) (+) or media alone (−) were used for control chambers. Details and specific conditions are provided in SEP.

In vitro differentiation of MDSCs

Purified MDSC splenocytes were labeled with CellTrace™ Far Red DDAO-SE (Thermo Fisher Scientific) and cultured in complete RPMI media with or without dissociated tumor cells at indicated numbers. Two days later, cells were collected and DDAO-SE-labeled MDSCs were isolated by flow (FACSAria™ II Cell Sorter, BD Biosciences). Dead cells were excluded using DAPI (Invitrogen). The flow-sorted MDSCs were subsequently tested in T cell suppression assays as described above.

Cytokine measurements

Cytokine levels in homogenized tumors were measured using Luminex Mouse Cytokine Magnetic 20-Plex Panel (Invitrogen) according to the manufacturer’s instructions and as described in SEP.

In vivo BrdU staining

Tumor-bearing mice were injected with 2 mg BrdU i.p. After 1 hour, mice were sacrificed and BM, blood, spleen and tumor suspensions were prepared. BrdU incorporation in MDSCs was analyzed using the BD FITC BrdU Flow Kit (BD Bioscience) according to the manufacter’s instructions.

Arg1, NO and ROS activity

FACS-sorted MDSCs were washed and analyzed for Arg1, NO and ROS production as described in SEP.

Taqman gene expression analysis

Total RNA extraction and real-time analysis were performed as described in SEP.

Isolation of Pmel lymphocytes and adoptive transfer

Adoptive transfer of in vitro activated pmel cells was performed in mice 5 days after tumor challenge as decribed in SEP.

Treg depletion in mice

Foxp3^DTR mice were injected i.p. on 5 consecutive days with DT (250ng/mouse, Sigma-Aldrich). DT treatment was initiated 8–9 days after tumor inoculation when tumors exhibited a volume of approximately 100mm³.

Patient material and analysis

Patient samples were collected on a tissue-collection protocol approved by the Institutional Review Board. Details on processing and analysis are provided in SEP.

We would like to thank Hong Zhong, Beatrice Yin, and Yuri Igarash for technical assistance. In addition, we thank the Molecular Cytology Core Facility and the Flow Cytometry Facility at Memorial Sloan Kettering Cancer Center.

FUNDING

This work was supported by the Swim across America, Ludwig Institute for Cancer Research, and Breast Cancer Research Foundation. RBH is the recipient of postdoctoral fellowships through The Danish Cancer Society and The Carlsberg Foundation, Denmark. DZ is a Bart A. Kamen Fellow of the Damon Runyon Cancer Research Foundation and has received funding from the Bladder Cancer Awareness Network.