Successful immunotherapy with IL-2/anti-CD40 induces the -mediated mitigation of an immunosuppressive

Jonathan M. Weissa, Timothy C. Backa, Anthony J. Scarzelloa, Jeff J. Subleskia, Veronica L. Halla, Jimmy K. Stauffera, Xin Chena, Dejan Micica, Kory Aldersonb, William J. Murphyb, and Robert H. Wiltrouta,1

aLaboratory of Experimental Immunology, Cancer and Inflammation Program, NCI Frederick, Frederick, MD 21702; and bDepartment of Dermatology, University of California, Davis, CA 95616

Edited by Thomas A. Waldmann, National Institutes of Health, Bethesda, MD, and approved September 28, 2009 (received for review August 25, 2009) Treatment of mice bearing orthotopic, metastatic tumors with associated with the recruitment of mononuclear cells capable of anti-CD40 antibody resulted in only partial, transient anti-tumor producing tumor promoting factors (8, 9), as well as MDSC that effects whereas combined treatment with IL-2/anti-CD40, induced contribute to tumor progression through the inhibition of effec- tumor regression. The mechanisms for these divergent anti-tumor tor cell functions (8, 10). responses were examined by profiling tumor-infiltrating leukocyte We reported previously that IL-2 and agonistic antibody to subsets and chemokine expression within the tumor microenvi- CD40 (␣CD40) synergize for the regression of metastatic tumors ronment after immunotherapy. IL-2/anti-CD40, but not anti-CD40 in mice (11). Although we identified CD8ϩ T cells and host IFN␥ alone, induced significant infiltration of established tumors by NK expression as critical components of this therapeutic approach -and CD8؉ T cells. To further define the role of in (11), the specific mechanisms underlying the IL-2/␣CD40 syn leukocyte recruitment into tumors, we evaluated anti-tumor re- ergistic anti-tumor responses within the microenvironment re- sponses in mice lacking the , CCR2. The anti- main unclear. We demonstrate in a murine model of metastatic tumor effects and leukocyte recruitment mediated by anti-CD40 renal cancer that ␣CD40 may be limited by its dependency upon ؊ ؊ alone, were completely abolished in CCR2 / mice. In contrast, MCP-1 and an inability to remove Tregs and MDSC specifically IL-2/anti-CD40-mediated leukocyte recruitment and reductions in from within the tumor microenvironment, allowing for eventual ؊ ؊ primary tumors and metastases were maintained in CCR2 / mice. tumor progression. In contrast, synergistic anti-tumor responses Treatment of mice with IL-2/anti-CD40, but not anti-CD40 alone, and protection achieved by IL-2/␣CD40 are associated with the also caused an IFN-␥-dependent increase in the expression of expression of Th1 chemokines that are associated with favorable multiple Th1 chemokines within the tumor microenvironment. prognosis in RCC (5, 6), an augmentation of effector leukocytes Interestingly, although IL-2/anti-CD40 treatment increased Tregs in and concomitant removal of suppressive cells specifically within ␥ the spleen, it also caused a coincident IFN- -dependent reduction the tumor microenvironment. in CD4؉/FoxP3؉ Tregs, myeloid-derived suppressor cells and Th2 chemokine expression specifically within the tumor microenviron- Results ment that was not observed after treatment with anti-CD40 alone. CCR2 Expression Is Required for ␣CD40, but Not IL-2/␣CD40 Mediated Similar effects were observed using IL-15 in combination with Anti-Tumor Responses. Our previous study showed that IL-2/ anti-CD40. Taken together, our data demonstrate that IL-2/anti- ␣CD40 exhibited strong synergy for treatment of established CD40, but not anti-CD40 alone, can preferentially reduce the metastatic tumors in mice, as compared to IL-2 or ␣CD40 as overall immunosuppressive milieu within the tumor microenviron- single agents (11). Furthermore, we found that ␣CD40 treatment ment. These results suggest that the use of anti-CD40 in combina- of Renca-bearing mice induced significant reduction in tumors IMMUNOLOGY tion with IL-2 or IL-15 may hold substantially more promise for in association with high levels of systemic MCP-1 levels, sug- clinical cancer treatment than anti-CD40 alone. gesting a possible role for MCP-1 in leukocyte recruitment into ͉ ͉ tumors and CD40-dependent anti-tumor effects (12). To deter- chemokines tumor immunotherapy CD40 mine the relative contribution of MCP-1 to the ␣CD40- and IL-2/␣CD40-mediated anti-tumor responses, we compared tu- any strategies for cancer treatment use combinations of mor outcomes in treated WT and mice deficient in CCR2, the Mimmunotherapeutic agents for enhanced anti-tumor re- receptor for MCP-1 (13). Treatment of WT mice with ␣CD40 sponses. However, these approaches are often complicated by a significantly reduced primary tumor areas, but a substantially need to overcome tumor-induced immune suppression in the greater reduction was observed after IL-2/␣CD40 treatment tumor microenvironment. In this regard, T regulatory (Treg) (Fig. 1A). Interestingly, the effects of ␣CD40 were abrogated in cells and myeloid-derived suppressor cells (MDSC) have been CCR2Ϫ/Ϫ mice, whereas the efficacy of IL-2/␣CD40 treatment identified as functional suppressor cells within tumors (1, 2). The was maintained in CCR2Ϫ/Ϫ mice. most effective immunotherapeutic regimens are likely to consist Since the orthotopic tumor model that we used forms spon- of agents that restructure, within the tumor microenvironment, taneous tumor metastases in the lungs, we next examined the the composition of tumor-infiltrating leukocytes away from these inhibitory elements in favor of effector cells, such as NK ϩ cells and CD8 T cells. Author contributions: J.M.W. designed research; J.M.W., T.C.B., X.C., and D.M. performed Chemokine expression can regulate the polarization of im- research; A.J.S., J.J.S., V.L.H., and J.K.S. contributed new reagents/analytic tools; J.M.W., mune responses (3). For example, CXCR3 and CCR5 are A.J.S., V.L.H., K.A., W.J.M., and R.H.W. analyzed data; and J.M.W. wrote the paper. preferentially expressed on Th1 T cells and M1 and The authors declare no conflict of interest. their respective ligands are associated with enhanced cell- This article is a PNAS Direct Submission. mediated immune responses (3–5) and favorable prognosis in Freely available online through the PNAS open access option. human RCC (5, 6). Another chemokine, chemoat- 1To whom correspondence should be addressed. E-mail: [email protected]. tractant (MCP)-1 activates macrophages for enhanced This article contains supporting information online at www.pnas.org/cgi/content/full/ anti-tumor activities (7), however MCP-1 expression is also 0909474106/DCSupplemental.

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0909474106 PNAS ͉ November 17, 2009 ͉ vol. 106 ͉ no. 46 ͉ 19455–19460 Downloaded by guest on September 30, 2021 Fig. 2. ␣CD40 induces CCR2-dependent leukocyte recruitment into tumors, whereas IL-2/␣CD40 maximally induces CCR2-independent increases in tumor- infiltrating leukocytes. (A) Wild-type and (B) CCR2Ϫ/Ϫ tumor-bearing mice Fig. 1. CCR2 is required for ␣CD40, but not IL-2/␣CD40-mediated anti-tumor were treated, as indicated. On day 22, mice were euthanized and the primary responses. (A) WT and CCR2Ϫ/Ϫ tumor-bearing mice were treated as indicated. tumor was dissected. Total leukocytes were counted and further identified by On day 22, mice were euthanized, the primary tumor was dissected and cell surface staining with the indicated antibodies and flow cytometry. The measured. (B) Lungs from these treated mice were collected on the same day total number of each cell subset was calculated by multiplying the total Ͻ and fixed in Bouin’s solution. The number of lung metastases was counted number of leukocytes by the percent expressing each surface marker (*, P 0.05; NS ϭ not significant). Data are derived from two tumors per treatment under a dissecting microscope (*, P Ͻ 0.05 and **, P Ͻ 0.005 as compared to control-treated mice). (C) For studies of tumor progression, mice underwent group in each of three separate experiments. surgical removal of the tumor-bearing kidney followed by treatment with IL-2 ␣ and/or CD40. Survival analysis was plotted according to the Kaplan-Meier ␣ method and statistical differences determined using the log- test (*, P Ͻ (Fig. S2). Thus, as compared to IL-2 or CD40 alone, IL-2/ 0.0001). For all tumor studies, the results from one experiment consisting of 10 ␣CD40 maximally enhances leukocyte recruitment into the mice/treatment group are shown and results are representative of three tumor microenvironment. similar experiments. To determine the relative contribution of MCP-1 to the ␣CD40- and IL-2/␣CD40-mediated leukocyte recruitment into tumors, we analyzed the profile of leukocytes in treated ␣ ␣ effect of CD40 and IL-2/ CD40 treatment on lung metastases. CCR2Ϫ/Ϫ mice (Fig. 2B). The IL-2/␣CD40 mediated increases ␣ ϩ Similar to primary tumors, CD40 alone significantly reduced in CD8 T cells, B cells and macrophages were maintained in Ϫ/Ϫ the number of metastases in WT, but not CCR2 mice (Fig. CCR2Ϫ/Ϫ mice. In contrast, the ␣CD40-mediated increase in 1B). In contrast, the reduction in metastases achieved by IL-2/ macrophages was lost in CCR2Ϫ/Ϫ mice (Fig. 2B). We also found ␣ Ϫ/Ϫ CD40 was maintained in CCR2 mice. that NK cell recruitment in response to IL-2/␣CD40 was abol- To evaluate the efficacy of the different treatments on survival ished in CCR2Ϫ/Ϫ mice. of mice bearing metastatic disease, we performed a unilateral The results above were further confirmed by immunohisto- nephrectomy of the tumor-bearing kidney, followed by treatment of Ϫ/Ϫ Ϫ Ϫ chemical analyses on tumors from treated WT or CCR2 mice residual metastatic disease. IL-2/␣CD40-treated WT and CCR2 / (Fig. S3). Whereas IL-2/␣CD40 indued and CD8ϩ mice had significantly prolonged survival, with Ͼ75% of the mice T cell recruitment in WT and CCR2Ϫ/Ϫ mice, the ␣CD40- remaining disease-free for Ͼ80 days (Fig. 1C). Surviving mice were induced recruitment of macrophages was lost in CCR2Ϫ/Ϫ mice rechallenged s.c. with Renca on day 92 and these mice demon- (Fig. S3A). By computer-assisted quantitation of the immuno- strated complete resistance to tumors (Fig. S1). Although ␣CD40 reactive areas, the IL-2/␣CD40 mediated increase in F4/80ϩ mediated significant reduction in primary tumors (Fig. 1A), neither macrophages and CD8ϩ T cell recruitment was significant (Fig. ␣CD40 nor IL-2 treatments alone prolonged survival (Fig. 1C). S3 A and B). These results indicate that CCR2 is required for the Thus, ␣CD40 mediates CCR2-dependent anti-tumor responses, ␣CD40, but not IL-2/␣CD40-mediated recruitment of leuko- whereas the more substantial, durable responses induced by IL-2/ cytes into tumors. ␣CD40 are CCR2 independent. IL-2/␣CD40 Increases the Expression of RANTES, MIP-1␥, MIG, and CCR2 Is Required for ␣CD40, but Not IL-2/␣CD40-Mediated Recruit- IP-10. Our data indicated that leukocyte recruitment into the ment of Leukocytes into Tumors. Next, we characterized the profile tumor induced by the ␣CD40 treatment was dependent upon of tumor-infiltrating leukocytes in response to IL-2, ␣CD40, and CCR2-mediated responses, whereas recruitment caused by IL- IL-2/␣CD40 immunotherapy. In contrast to IL-2 or ␣CD40 2/␣CD40 was CCR2-independent. To determine whether this alone, the tumors from IL-2/␣CD40-treated mice contained was due to the increased expression of additional chemoattrac- significant increases in the numbers of infiltrating CD8ϩ T cells, tants induced by IL-2/␣CD40, we analyzed tumors for chemo- NK, and B cells (Fig. 2A). In contrast, CD4ϩ T cells did not kine expression by qPCR. IL-2/␣CD40 significantly in- increase after any treatment. Treatment of tumor-bearing mice creased CXCL9/MIG, CXCL10/IP-10, CCL5/RANTES, and with either ␣CD40 or IL-2/␣CD40 also induced significant CCL9/MIP-1␥ expression within tumors (Fig. 3). Furthermore, macrophage recruitment into tumors. These cellular changes the IL-2/␣CD40 mediated increase in chemokine expression was were still evident when they were normalized for tumor area completely lost in GKO mice (Fig. 3). Interestingly, CXCR3, the

19456 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0909474106 Weiss et al. Downloaded by guest on September 30, 2021 Fig. 3. IL-2/␣CD40 induces the IFN␥-dependent increase in chemokine ex- pression within tumors. WT and GKO tumor-bearing mice were treated, as indicated. On day 22, mice were euthanized and the primary tumor was dissected and placed in RNALater. Total RNA was extracted from tumors and analyzed by qPCR for (A) CXCL9, (B) CXCL10, (C) CCL5, and (D) CCL9 chemokine . For each gene, the results from saline control-treated sam- ples were normalized to 1. Each closed circle represents results from one experiment using wild-type mice (n ϭ 5) and the p values on each graph were derived from the comparison of ␣CD40 to IL-2/␣CD40 treated, wild-type mice. No statistical differences were observed using GKO mice (open triangles and dotted mean lines; n ϭ 3). Fig. 4. IL-2/␣CD40 treatment selectively reduces the number of CD4ϩ/FoxP3ϩ cells and Treg-associated chemokines within tumors. WT and GKO tumor- receptor for MIG and IP-10, and CCR5, a RANTES receptor, bearing mice were treated as indicated. On day 22, mice were euthanized and were also significantly upregulated (Fig. S4). Taken together, the spleens and tumors harvested. The total number of Tregs in each micro- these data suggest that IL-2/␣CD40 induces the IFN␥-dependent environment was determined by multiplying the total number of (A) spleno- upregulation of several chemokines within tumors that may cytes or (B) tumor-associated leukocytes by the percentage of CD4ϩ/FoxP3ϩ facilitate leukocyte recruitment. cells in each sample (*, P Ͻ 0.03 for both spleen and tumor). Data are derived from two tumors per treatment group in each of three separate experiments. IFN Gamma Is Required for the IL-2/␣CD40-Mediated Regulation of (C) CCL17 and (D) CCL22 gene expression in tumors was analyzed by qPCR. The results from saline control-treated samples were normalized to 1. Each closed ␥ .Tumor-Infiltrating CD4؉ and CD8؉ T Cells Since IFN expression circle represents results from one experiment using wild-type mice (n ϭ 3) and was required for chemokine induction within the tumor micro- p-values on each graph were derived from the comparison of ␣CD40 to environment, we next examined its role in IL-2/␣CD40-induced IL-2/␣CD40 treated, wild-type mice. No statistical differences were observed leukocyte recruitment. IL-2/␣CD40 treatment of WT mice using GKO mice (open triangles and dotted mean lines; n ϭ 3). (E) CCL17 and ϩ significantly increased the number of tumor-associated CD8 T CCL22 protein expression in tumors was analyzed by ELISA of five individual cells (Fig. S5A). This response was markedly reduced in GKO tumor lysates (*, P Ͻ 0.05 and **, P Ͻ 0.02 as compared PBS treated groups; mice, although it was still significant compared to control mice. NS ϭ not significant). (F) Tumor-bearing mice were treated beginning day 12 IL-2 and ␣CD40 as single agents also increased the number of with PC61 alone or in combination with IL-2 or ␣CD40. On day 22, mice were IMMUNOLOGY Ͻ Ͻ CD8ϩ T cells in WT, but not GKO mice. IL-2/␣CD40 also killed, the primary tumor was dissected and measured (*, P 0.03; **, P 0.002 as compared to PBS alone). Each treatment group had nine mice. reduced the number of CD4ϩ T cells within tumors in WT, but not GKO mice (Fig. S5A). These cellular changes were still evident when they were normalized for tumor area (Fig. S5B). tumor responses (15), also reduced tumor-associated Tregs (Fig. ␣ Taken together, IL-2/ CD40 significantly increased the 4B). However, unlike IL-2, IL-15 by itself did not increase Treg CD8:CD4 ratio of tumor-infiltrating T cells and this was com- numbers in tumors. pletely lost in GKO mice (Fig. S5C). We hypothesized that the reduced numbers of tumor- ␣ associated Tregs could be due in part to a reduction in che- IL-2/ CD40 Selectively Reduces the Number of Tumor-Infiltrating moattractant signals for these cells. Therefore, we analyzed Regulatory T Cells and Treg-Associated Chemokines Within Tumors. tumor lysates for the expression of CCL17 and CCL22, two Since the studies above demonstrated an overall reduction in ϩ ␣ chemokines implicated in the association of Tregs with human CD4 T cells by IL-2/ CD40, we hypothesized that part of the ␣ anti-tumor effect could be due to reduction in CD4ϩ Tregs. We (1, 16, 17) and murine (18) tumors. IL-2/ CD40 treatment ␣ ␣ significantly reduced the gene expression of both chemokines therefore compared the ability of IL-2, CD40, and IL-2/ CD40 ␣ treatments to regulate the numbers of CD4ϩ/CD25ϩ/FoxP3ϩ (Fig. 4 C and D); whereas either IL-2 or CD40 alone had no ␣ Tregs in tumors and spleens. Consistent with our previous effect. The IL-2/ CD40-mediated reduction of both chemokines finding (14), IL-2/␣CD40 increased the number of Tregs within was abolished in GKO mice. Next, we analyzed tumor lysates for the spleen (Fig. 4A). However, we also found that the same CCL17 and CCL22 protein expression by ELISA. Consistent treatment selectively reduced the numbers of tumor-associated with our gene expression studies, IL-2/␣CD40 significantly de- Tregs (Fig. 4B). IL-2 and ␣CD40, as single agents, had no creased the protein expression of both chemokines (Fig. 4E). significant effect upon the number of Tregs in either the spleen Interestingly, in GKO mice, CCL17 protein expression was or tumor. Furthermore, the IL-2/␣CD40-mediated reduction in significantly increased in response to ␣CD40 or IL-2/␣CD40 Treg numbers was lost in GKO mice. IL-15/␣CD40, another treatments (Fig. 4E). To determine whether the increases in combination immunotherapy recently shown to induce anti- splenic Tregs could be attributed to alterations in CCL17 or

Weiss et al. PNAS ͉ November 17, 2009 ͉ vol. 106 ͉ no. 46 ͉ 19457 Downloaded by guest on September 30, 2021 efficiently than those isolated from the periphery (Fig. S7), which indicates a tumor-promoting role for Tregs, in our model.

IL-2/␣CD40 Treatment Selectively Reduces the Number of Myeloid- Derived Suppressor Cells, Arginase Expression, and Associated Che- mokines Within Tumors. Since IL-2/␣CD40 increased the number of tumor-associated macrophages, we hypothesized that it might also alter macrophage subsets present in the tumor microenvi- ronment. Thus, we compared the ability of IL-2, ␣CD40, or IL-2/␣CD40 treatment to regulate the numbers of myeloid- derived suppressor cells (MDSC) in tumors and spleens. MDSC were identified as CD45ϩ, CD11bϩ, Gr1Lo, CD124ϩ cells (2). IL-2/␣CD40 significantly increased the number of MDSC within spleens (Fig. 5A) but significantly reduced them within tumors (Fig. 5B). In contrast, IL-2 and ␣CD40, as single agents, had no effect on MDSC in either site. The ability of IL-2/␣CD40 to reduce the number of tumor-infiltrating MDSC was lost in GKO mice. IL-15/␣CD40 treatment similarly mediated a significant reduction in tumor-derived MDSC (Fig. 5B). To functionally characterize the effects of IL-2/␣CD40 treat- ment on MDSC, we analyzed arginase expression within tumor lysates. IL-2/␣CD40 reduced tumor-associated arginase expres- sion in WT, but not GKO, mice (Fig. 5C). We also quantitated changes in arginase expression by sorting F4/80ϩ macrophages from the tumors of treated mice and found that those cells isolated from the tumors of IL-2/␣CD40 treated mice had reduced arginase expression on a per cell basis (Fig. S8). These results indicate that the reduction in MDSC achieved by IL-2/ ␣CD40 therapy also results in functional differences in monocyte populations within the tumor microenvironment. ␣ Fig. 5. IL-2/ CD40 treatment selectively reduces the number of myeloid- Since IL-2/␣CD40 mediated a reduction in tumor-associated derived suppressor cells, arginase expression, and associated chemokines within tumors. Wild-type and GKO tumor-bearing mice were treated as MDSC, we next analyzed tumor lysates for the expression of indicated. On day 22, mice were euthanized and the spleens and tumors CXCL5/ENA-78, a chemokine involved in the recruitment of these harvested. The total number of MDSC in each microenvironment was deter- cells (19). IL-2/␣CD40 treatment significantly reduced CXCL5 mined by multiplying the total number of (A) splenocytes or (B) tumor- expression in tumors (Fig. 5D), whereas IL-2 or ␣CD40 had no associated leukocytes by the percentage of CD11bϩ/Gr1Lo/CD124ϩ cells in each effect. The IL-2/␣CD40-mediated reduction of CXCL5 expression sample (*, P Ͻ 0.05 for spleen and P Ͻ 0.01 for tumor). Data are derived from was lost in GKO mice (Fig. 5D). ELISA analysis of tumor lysates two tumors per treatment group in each of three separate experiments. (C) revealed that IL-2/␣CD40 treatment reduced CXCL5 protein ex- Arginase expression in three separate tumors/treatment group was analyzed pression in tumors isolated from wild-type, but not GKO mice (Fig. Ͻ as described in Materials and Methods (*, P 0.02 as compared to both PBS 5E). Although the number of MDSC increased in the spleens of and ␣CD40-treated wild-type groups). (D) CXCL5 gene expression in tumors ␣ was analyzed by qPCR. The results from saline control-treated samples were IL-2/ CD40-treated mice, we found no significant differences in normalized to 1. Each closed circle represents results from one experiment CXCL5 expression by qPCR analysis of spleen cDNA (Fig. S6C). using wild-type mice (n ϭ 3) and p-values were derived from the comparison Taken together, these results demonstrate that IL-2/␣CD40 therapy of ␣CD40 to IL-2/␣CD40 treated, wild-type mice. No statistical differences were results in the IFN␥-dependent downregulation of CXCL5 expres- observed using GKO mice (open triangles and dotted mean lines; n ϭ 3). (E) sion that may mediate the recruitment of MDSC into the tumor CXCL5 protein expression in tumors was analyzed by ELISA of five individual microenvironment. tumor lysates (**, P Ͻ 0.03 as compared to PBS treated groups; NS ϭ not significant). Discussion Our results demonstrate clear differences in immune and anti- tumor responses achieved by IL-2/␣CD40, as compared to those CCL22 gene expression, we analyzed the expression of each achieved by either IL-2 or ␣CD40 alone. Elevated Tregs and chemokines gene in spleen cDNA isolated from each treatment MDSC have been described recently in human RCC and clinical group. However, no significant change in the expression of either approaches aimed at removing these immunosuppressive cells chemokine gene was identified for any treatment group (Fig. S6 ␣ will likely be an important component of improved immuno- A and B). Overall, these results demonstrate that IL-2/ CD40 therapies (20). Although ␣CD40 alone can mediate significant specifically reduces the number of tumor-associated Tregs and ␥ leukocyte recruitment into primary tumors, it fails to remove that this correlated with the IFN -dependent downregulation of Tregs and MDSC within the tumor microenvironment, which chemokines that may control their recruitment to the tumor site. may allow for eventual tumor outgrowth. Consequently, ␣CD40 Finally, we sought to demonstrate a tumor-supporting role of only marginally improves primary tumor burden and has no ␣ Tregs in Renca-bearing mice. We treated mice with CD25 significant impact upon lung metastases or long-term survival of depleting antibody (PC61) alone or in combination with either treated mice. The complete dependency of ␣CD40 upon CCR2, IL-2 or ␣CD40. On the one hand, treatment of tumor-bearing a receptor associated with the recruitment of potentially tumor- mice with PC61 alone increased the primary tumor area (Fig. promoting leukocytes (8, 10), and the inability of ␣CD40 to 4F). However, when PC61 was used in combination with either remove tumor-infiltrating suppressor cells, was in dramatic IL-2 or ␣CD40 where substantial infiltration of Tregs into contrast to the local immune responses achieved by IL-2/␣CD40 tumors was detected, a significant reduction in primary tumor in which the net result appears to be a conversion of the overall area was observed. Furthermore, CD4ϩ/CD25ϩ Tregs isolated tumor microenvironment from a potentially suppressive milieu from Renca tumors suppress T cell proliferation even more to a predominantly anti-tumor, Th1 environment.

19458 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0909474106 Weiss et al. Downloaded by guest on September 30, 2021 Another important distinction derived from our study was the from IL-2/␣CD40-treated mice can produce more IFN␥ upon divergent cellular responses to IL-2/␣CD40 therapy in the stimulation (11), illustrating the potential for a feedback loop spleen, as compared to that of the tumor microenvironment. within the tumor microenvironment, whereby the ongoing re- Consistent with our previous observations, IL-2/␣CD40 medi- cruitment of beneficial effector cells and the coordinated re- ated a significant increase in splenic Tregs (14). Our ongoing duction of immunosuppressive Tregs and MDSC may be rapidly studies suggest that this increase may be due to expansion, rather amplified. than recruitment of Tregs to the spleen. Since peripheral in- We also investigated the relative contribution of MCP-1 to the creases in Tregs and MDSC did not correlate with reduced recruitment of leukocytes into tumors. MCP-1 expression is anti-tumor responses, we suggest that specific cellular changes elevated in many cancers (29, 30) and has been associated with within the tumor microenvironment are more predictive of the recruitment of M2 polarized, or alternatively activated successful immunotherapeutic outcomes. Consistently, Treg ac- macrophages that may promote tumor progression (8, 30). Our cumulation within other tumors correlated with poor survival study contrasts divergent roles for CCR2 interactions in different (17). We further demonstrate a tumor-supporting role of Tregs immunotherapeutic approaches for treating solid tumors. Al- in Renca-bearing mice. First, CD25ϩ Tregs isolated from tumors though CD40-activated macrophages mediate potent anti-tumor suppressed T cell proliferation even better than Tregs isolated effects in vitro (31), our data suggest that the efficacy of ␣CD40 from the periphery. The reduction in primary tumor area using in vivo might be limited by a dependency upon MCP-1, the IL-2 in combination with PC61 antibody was particularly inter- potential for recruiting M2 macrophages and an inability to esting, since IL-2 alone consistently failed to have any appre- overcome suppressive pathways within the tumor microenviron- ciable effect on primary tumor size. However, anti-CD25 deple- ment. In contrast, the IL-2/␣CD40-mediated shift away from a tion by itself caused an increase in tumor burden, presumably dependency on MCP-1/CCR2 interactions is desirable, in that it due to the removal of CD25ϩ effector cells. Nevertheless, also leads to the induction of multiple Th1 chemokines more anti-tumor responses achieved by IL-2/␣CD40 remained supe- closely associated with Th1 effector cell recruitment and favor- rior to either IL-2 or ␣CD40 in combination with anti-CD25, able anti-tumor responses in RCC (5, 6). presumably due to the ability of IL-2/␣CD40 to mediate other Overall, we suggest that IL-2/␣CD40 and other approaches effects, such as the removal of MDSC, effector cell recruitment, that simultaneously increase effector responses and decrease and synergistic IFN␥ induction that are critical components of counterregulatory responses within the tumor microenviron- IL-2/␣CD40 antitumor therapy (11). We also show that IL-15/ ment may generate more effective anti-tumor responses. IL-2 ␣CD40, another combination immunotherapy recently shown to has been used in patients with metastatic melanoma and RCC induce anti-tumor responses (15), similarly decreased tumor- although durable responses have only been observed in a mi- associated Tregs and MDSC. In this regard, IL-15 may have nority of patients (32). Similarly, objective responses were certain advantages over IL-2, since it did not increase tumor- observed in only a minority of patients with advanced solid associated Tregs by itself as IL-2 did. The expression profile of tumors after treatment with ␣CD40 (33). It is tempting to chemokines within tumors clearly plays an important role in the speculate that the limited efficacy of ␣CD40 is due to an inability redistribution of suppressor cells and represents an attractive to overcome local regulatory mechanisms within the tumor target for immunotherapeutic regimens that seek to alter the microenvironment that ultimately allow the tumor to escape balance of the tumor microenvironment toward a beneficial host immune control. Nevertheless, CD40 is an attractive therapeutic immune response. target because, in addition to its ability to stimulate immune We also demonstrate a central role for IFN␥ in regulating the responses, its ligation by CD40ϩ tumors induces pro- composition of tumor-infiltrating leukocytes after therapy. Pre- duction (12) and enhances Fas-mediated tumor apoptosis (34). viously, we showed that the anti-tumor effect of IL-2/␣CD40 was While ongoing trials involving agonistic CD40 antibodies con- ϩ dependent on CD8 T cells and IFN␥ (11). We, and others, have tinue to provide important insights into the cellular responses to shown that the IFN␥-dependent recruitment of effector NK and developing tumors in vivo, we believe that our results support the CD8ϩ T cells is mediated by Th1 chemokines such as RANTES, ␣ ␣

development of IL-2/ CD40 or IL-15/ CD40 combination ther- IMMUNOLOGY Mig and IP-10 that have been associated with favorable anti- apy for the improved treatment of solid tumors, including RCC. tumor responses in patients (5, 6, 21, 22). Second, and perhaps more interesting, was the requirement for IFN␥ in the reduction Materials and Methods in Tregs and MDSC within the tumor microenvironment. This Mice. BALB/c wild-type mice were obtained from the Animal Production Area selective reduction was closely associated with the downregula- of the National Cancer Institute (NCI) Frederick Cancer Research and Devel- tion of chemokines implicated in recruiting these regulatory cells opment Center. BALB/c CCR2Ϫ/Ϫ mice were provided by Dr. Cara Mack (Uni- Ϫ Ϫ to tumors (1, 16, 18, 19). It is interesting to note that ␣CD40 versity of Colorado Health Sciences Center) (13). BALB/c IFN␥ / (GKO) mice treatment increased CCL17 expression in GKO, but not WT were obtained from Jackson Laboratories. Mutant alleles were confirmed by mice. These findings are consistent with reports that CCL17 PCR genotyping. Mice (8–10 weeks of age) were used in accordance with an approved NCI Frederick Institutional Animal Care and Use protocol. expression can be upregulated by such as TNF␣ (itself ␥ induced potently by CD40 ligation) and downregulated by IFN Cells and Reagents. The renal adenocarcinoma of BALB/c origin (Renca) was (23, 24), which has emerged as a master regulatory cytokine in passaged i.p. as described (35). Recombinant human IL-2 was obtained from our model. Although the regulation of these chemokines by the NCI. Recombinant human IL-15 was from Peprotech, Inc. Agonist rat IFN␥ has been established (23–26), we extend this observation anti-mouse CD40 (clone FGK115B3) was purified from ascites, as described to demonstrate the feasibility of preferentially redistributing (12). Endotoxin was Ͻ1 EU/mg antibody, as determined by chromogenic effector and regulatory cells specifically within the tumor mi- Limulus Amebocyte Lysate (Cambrex). Purified rat IgG was purchased from croenvironment by combination immunotherapy that depends Jackson ImmunoResearch Laboratories. Antibody against IL-2R␣ (anti-CD25; on IFN␥. We further demonstrate the IFN␥-dependent reduc- clone PC61) used for depletion was purified from ascites. tion in arginase expression within tumor-derived leukocytes. ϫ 5 Functional arginase production by MDSC has been demon- In Vivo Tumor Model. Renca cells (1 10 ) were injected under the kidney capsule of mice on day 0. Mice treated with IL-2 received 300,000 IU i.p. twice strated in RCC patients and may result in impaired T cell a day on days 11, 15, 18, and 21 post tumor injection. Mice treated with signaling and tumor-induced tolerance (27). Arginase produc- anti-CD40 received 100 ␮g i.p. on days 11–15 and 18–21 post tumor injection. tion by macrophages is also impaired by IFN␥ and has been Mice treated with IL-15 received 1 ␮g i.p. on days 11–15 and 18–21 post tumor suggested to define a population alternatively activated by Th2 injection. In some experiments, mice received 200 ␮L of PC61 (400 ␮g/mL) cytokines (28). Finally, tumor- infiltrating leukocytes isolated antibody or saline control i.p. on days 11, 15, and 18. On day 22, mice were

Weiss et al. PNAS ͉ November 17, 2009 ͉ vol. 106 ͉ no. 46 ͉ 19459 Downloaded by guest on September 30, 2021 euthanized and primary tumors were collected. Tumor length and width was Arginase Assay. Leukocytes were isolated from tumors, counted and lysed in measured using calipers. Lungs were fixed in Bouin’s solution and lung me- 10 mM Tris-HCl (pH 7.4) containing 0.4% Triton X-100 and Halt protease tastases were counted under a dissecting microscope. In some studies, mice inhibitor (Thermo Scientific). Lysates were spun briefly and arginase activ- received a unilateral nephrectomy of the tumor-bearing kidney on day 11, ity contained within the supernatants quantified using a commercially ␣ followed by treatment with IL-2 and/or CD40 and were monitored for tumor available kit (BioAssay Systems). In some experiments, CD45ϩ, F4/80ϩ tu- progression. For tumor rechallenge experiments, long-term survivors or con- mor-associated leukocytes were sorted and then lysed for arginase activity, trol naïve mice were injected with 7.5 ϫ 104 RENCA s.c. and tumors were measured. as above.

Isolation of Leukocytes from Spleen. Spleens were harvested on day 22, placed ELISA. Tumors were homogenized in RIPA lysis buffer (Thermo Scientific). in HBSS and filtered through a two-chamber sterile Filtra-Bag (Fisher Scien- Chemokine protein expression was determined using a commercially avail- tific). Spleens were gently pressed and the resulting single cell suspension was able ELISA kit following the manufacturer’s instructions (R&D Systems). The collected from the other side of the bag. Splenocytes were counted using a amount of each chemokines was normalized against total protein, as deter- Sysmex KX-21 (Roche Diagnostics). mined using a BCA protein assay kit (Thermo Scientific).

Isolation of Tumor-Infiltrating Leukocytes. Tumors were dissected on day 22, Quantitative PCR (qPCR). Tumor and spleen samples were analyzed by qPCR as filtered through a two-chamber sterile Filtra-Bag (Fisher Scientific) and di- detailed in SI Methods. gested in RPMI containing 5% FCS, 250 U/mL type IV collagenase (Invitrogen), 100 ␮g/mL DNase I (Roche Molecular Biochemicals) and 1 mM EDTA (pH 8.0), Immunohistochemistry. Tumors were analyzed by immunohistochemistry as at 37 °C for 45 min. Then, the homogenate was processed in a tissue stom- acher-80 (Seward) for 30 s, washed with HBSS (BioWhittaker), and resus- detailed in SI Methods. pended in 40% Percoll (Amersham Pharmacia) in DMEM (BioWhittaker). The suspension was underlaid with 80% Percoll and centrifuged for 25 min at Statistical Analysis. Statistical differences were analyzed using a Mann- 1,000 ϫ g. Leukocytes were collected from the interphase, washed, and Whitney U test (GraphPad Prism, GraphPad Software, Inc.). Significance was counted. indicated by P Ͻ 0.05 values.

Flow Cytometric Analysis. Cells were analyzed by flow cytometry as detailed in ACKNOWLEDGMENTS. We thank Drs. Giorgio Trinchieri and John Ortaldo for SI Methods. critically reviewing the manuscript. This work was supported by the Intramural Research Program of the National Institutes of Health National Cancer Insti- Treg Suppression Assay. Treg suppression of T-cell proliferation was analyzed tute and with federal funds from the National Cancer Institute under Con- as detailed in SI Methods. tracts N01-CO-12400 and R01-CA-95572 (to W.J.M.).

1. Jensen HK, et al. (2009) Increased intratumoral FOXP3-positive regulatory immune cells 19. Yang L, et al. (2008) Abrogation of TGF beta signaling in mammary carcinomas recruits during -2 treatment in metastatic renal cell carcinoma. Clin Cancer Res Gr-1ϩCD11bϩ myeloid cells that promote metastasis. Cancer cell 13:23–35. 15:1052–1058. 20. Ko JS, et al. (2009) Sunitinib mediates reversal of myeloid-derived suppressor cell 2. Sica A, Bronte V (2007) Altered macrophage differentiation and immune dysfunction accumulation in renal cell carcinoma patients. Clin Cancer Res 15:2148–2157. in tumor development. J Clin Invest 117:1155–1166. 21. Cozar JM, et al. (2005) Analysis of NK cells and chemokine receptors in tumor infiltrat- 3. Mantovani A, et al. (2004) The chemokine system in diverse forms of macrophage ing CD4 T lymphocytes in human renal carcinomas. Cancer Immunol Immunother activation and polarization. Trends Immunol 25:677–686. 54:858–866. 4. Kim JJ, et al. (1998) CD8 positive T cells influence antigen-specific immune responses 22. Musha H, et al. (2005) Selective infiltration of CCR5(ϩ)CXCR3(ϩ) T lymphocytes in through the expression of chemokines. J Clin Invest 102:1112–1124. human colorectal carcinoma. Int J Cancer 116:949–956. 5. Kondo T, et al. (2006) Favorable prognosis of renal cell carcinoma with increased 23. Iellem A, et al. (2000) Inhibition by IL-12 and IFN-alpha of I-309 and macrophage- expression of chemokines associated with a Th1-type immune response. Cancer Sci derived chemokine production upon TCR triggering of human Th1 cells. Europ J Im- 97:780–786. muno 30:1030–1039. 6. Reckamp KL, et al. (2007) Expression of CXCR3 on mononuclear cells and CXCR3 ligands 24. Xiao T, et al. (2003) Thymus and activation-regulated chemokine (TARC/CCL17) pro- in patients with metastatic renal cell carcinoma in response to systemic IL-2 therapy. duced by mouse epidermal Langerhans cells is upregulated by TNF-alpha and IL-4 and J Immunother 30:417–424. downregulated by IFN-gamma. Cytokine 23:126–132. 7. Zachariae CO, et al. (1990) Properties of monocyte chemotactic and activating factor 25. Schnyder-Candrian S, Strieter RM, Kunkel SL, Walz A (1995) -alpha and (MCAF) purified from a human fibrosarcoma cell line. J Exp Med 171:2177–2182. interferon-gamma down-regulate the production of interleukin-8 and ENA-78 in 8. Mantovani A, et al. (2004) Infiltration of tumours by macrophages and dendritic cells: human . J Leukoc Biol 57:929–935. Tumour-associated macrophages as a paradigm for polarized M2 mononuclear phago- 26. Bonecchi R, et al. (1998) Divergent effects of interleukin-4 and interferon-gamma on cytes. Novartis Found Symp 256:137–145. macrophage-derived chemokine production: An amplification circuit of polarized T 9. Torisu H, et al. (2000) Macrophage infiltration correlates with tumor stage and angio- helper 2 responses. Blood 92:2668–2671. genesis in human malignant melanoma: Possible involvement of TNFalpha and IL- 27. Ochoa AC, Zea AH, Hernandez C, Rodriguez PC (2007) Arginase, prostaglandins, and 1alpha. Int J Cancer 85:182–188. myeloid-derived suppressor cells in renal cell carcinoma. Clin Cancer Res 13:721s–726s. 10. Huang B, et al. (2007) CCL2/CCR2 pathway mediates recruitment of myeloid suppressor 28. Modolell M, et al. (1995) Reciprocal regulation of the nitric oxide synthase/arginase cells to cancers. Cancer Lett 252:86–92. balance in mouse bone marrow-derived macrophages by TH1 and TH2 cytokines. Europ 11. Murphy WJ, et al. (2003) Synergistic anti-tumor responses after administration of J Immuno 25:1101–1104. agonistic antibodies to CD40 and IL-2: Coordination of dendritic and CD8ϩ cell 29. Monti P, et al. (2003) The CC chemokine MCP-1/CCL2 in pancreatic cancer progression: responses. J Immunol 170:2727–2733. Regulation of expression and potential mechanisms of antimalignant activity. Cancer 12. Shorts L, et al. (2006) Stimulation through CD40 on mouse and human renal cell Res 63:7451–7461. carcinomas triggers cytokine production, leukocyte recruitment, and antitumor re- sponses that can be independent of host CD40 expression. J Immunol 176:6543–6552. 30. Varney ML, Johansson SL, Singh RK (2005) Tumour-associated macrophage infiltration, 13. Kuziel WA, et al. (1997) Severe reduction in leukocyte adhesion and monocyte extrav- neovascularization and aggressiveness in malignant melanoma: Role of monocyte asation in mice deficient in CC chemokine receptor 2. Proc Natl Acad Sci USA 94:12053– chemotactic protein-1 and vascular endothelial -A. Melanoma Res 12058. 15:417–425. 14. Alderson KL, et al. (2008) Regulatory and conventional CD4ϩ T cells show differential 31. Buhtoiarov IN, et al. (2005) CD40 ligation activates murine macrophages via an IFN- effects correlating with PD-1 and B7–H1 expression after immunotherapy. J Immunol gamma-dependent mechanism resulting in tumor cell destruction in vitro. J Immunol 180:2981–2988. 174:6013–6022. 15. Zhang M, et al. (2009) Interleukin-15 combined with an anti-CD40 antibody provides 32. Yang JC, et al. (2003) Randomized study of high-dose and low-dose interleukin-2 in enhanced therapeutic efficacy for murine models of colon cancer. Proc Natl Acad Sci patients with metastatic renal cancer. J Clin Oncol 21:3127–3132. USA. 106:7513–7518. 33. Vonderheide RH, et al. (2007) Clinical activity and immune modulation in cancer 16. Gobert M, et al. (2009) Regulatory T cells recruited through CCL22/CCR4 are selectively patients treated with CP-870,893, a novel CD40 agonist monoclonal antibody. J Clin activated in lymphoid infiltrates surrounding primary breast tumors and lead to an Oncol 25:876–883. adverse clinical outcome. Cancer Res 69:2000–2009. 34. Lee JK, et al. (2005) Constitutive expression of functional CD40 on mouse renal cancer 17. Curiel TJ, et al. (2004) Specific recruitment of regulatory T cells in ovarian carcinoma cells: Induction of Fas and Fas-mediated killing by CD40L. Cell Immunol 235:145–152. fosters immune privilege and predicts reduced survival. Nat Med 10:942–949. 35. Williams PD, Pontes EJ, Murphy GP (1981) Studies of the growth of a murine renal cell 18. Mailloux AW, Young MR (2009) NK-dependent increases in CCL22 secretion selectively carcinoma and its metastatic patterns. Res Commun Chem Pathol Pharmacol 34:345– recruits regulatory T cells to the tumor microenvironment. J Immunol 182:2753–2765. 349.

19460 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0909474106 Weiss et al. Downloaded by guest on September 30, 2021