IL10 Release upon PD-1 Blockade Sustains Immunosuppression in Ovarian Cancer

Programmed cell death-1 (PD-1) and program cell death ligand 1 (PD-L1) constitute a major immune regulatory axis that blocks antitumor effector responses in the tumor microenvironment. PD-1 is an inhibitory receptor expressed on several adaptive and innate immune cells, including T cells, B cells, monocytes, macrophages, natural killer (NK) cells, and dendritic cells (DC; refs. 1–4). Known PD-1 ligands include PD-L1 (B7-H1 or CD274) and PD-L2 (B7-DC or CD273; ref. 1). PD-L1 is constitutively expressed on many cells types like DCs and macrophages. PD-L1 is inducible on T cells, B cells, tumor cells, epithelial cells, and endothelial cells. Similarly, PD-L2 is primarily induced, depending on the cytokine milieu, on DCs, B cells, macrophages, and monocytes (1, 5–7). The ligation of PD-1 by its ligands, during T-lymphocyte activation, has been shown to result in the tyrosine phosphorylation of its cytoplasmic tail, leading to recruitment of Src homology domain-containing phosphatase-2 (SHP-2). Activated SHP-2 subsequently dephosphorylates T cell receptor (TCR) proximal signaling complexes such as ZAP70/CD3ζ and PI3K and inhibits Akt phosphorylation culminating in the inhibition of T cell proliferation, cytotoxic functions, and cytokine release, and promotion of apoptosis (8, 9).

PD-1 expression on lymphocytes is dynamic and is upregulated following TCR or B cell receptor (BCR) ligation. Studies also have shown that T cells are programmed epigenetically upon persistent antigenic stimulation for rapid PD-1 expression upon restimulation (10, 11). One crucial observation that is made frequently is that the expression of PD-1 on T cells infiltrating in tumors is very high compared with their peripheral or normal tissue counterparts (12, 13). This high expression of PD-1 on tumor-infiltrating T cells not only identifies them as tumor antigen-specific but also correlates with blunted responsiveness to further stimulation (13, 14). Furthermore, infiltration by PD-1⁺ lymphocytes often correlates with distant metastatic relapse and poor prognosis in many tumor types (12, 14, 15).

In many malignant tissues, it has been observed that up to 40% of the infiltrating immune cells are tumor-infiltrating murine and human myeloid dendritic cells (TIDC) and disease progression correlates with the phenotypic switch of TIDCs from immune stimulatory to immune suppressive (4, 16). We have shown in a murine model of ovarian carcinoma that TIDCs also become increasingly PD-1⁺ as disease progresses rendering them ineffective as immune stimulators by inhibiting the cytokine release, co-stimulatory molecules expression, and antigen presentation capacity (4). It is now clear from many recent studies that the TME is adept at inducing PD-1 expression on the infiltrating immune effector cells, rendering them ineffective. The identification of TME-associated factors and mechanisms responsible for the unique regulation of PD-1 expression on infiltrating immune cells remains underdeveloped.

As we and others have shown, ovarian cancer creates a pronounced immune suppressive TME (17) and of many factors, the cytokine IL10 has been correlated with volume of ascites, poor survival, and relapse (18–20). In the ID8 murine model of ovarian cancer, we have seen that IL10 is present at high levels in both the blood serum and ascites. Additionally, the ascites and tumor-infiltrating immune cells, including the TIDCs, express very high levels of PD-1 compared with their peripheral counterparts (4). Given the immune regulatory role of IL10 and its correlation with disease severity and poor prognosis in ovarian cancer, we sought to explore if IL10 induces PD-1 expression on the TIDCs. Our results show that IL10 treatment of TIDCs results in increased expression of PD-1 in a STAT3-dependent manner. Antibody-mediated blockade of PD-1 on TIDCs subsequently leads to increased release of IL10 into the TME of treated mice. Coblockade of both PD-1 and IL10 results in significant enhancement in survival of tumor-bearing mice, reduced tumor burden, and augmented adaptive antitumor immunity.

Four- to 12-week-old female C57BL/6J (B/6J) mice, obtained from an in-house breeding program or from The Jackson Laboratory (Bar Harbor, ME), were used for experimentation. Animal care and use was in accordance with the Institutional Animal Care and Use Committee at Mayo Clinic.

Leukocytes were cultured in modified RPMI 1640 media as previously described (4). ID8 tumor cells, obtained from Dr. K. Roby, University of Kansas (Lawrence, KS) in 2005, were derived from immortalized ovarian epithelial cells generated by repeated passage in culture and were grown in complete DMEM media (4). They were last authenticated as mouse origin by IDEXX BioResearch in 2014. ID8 cells used in the described experiments were used within 12 passages upon receipt.

ID8 tumor cells (5 × 10⁶ cells) were injected i.p. at a volume of 500 μL in PBS. Tumor and ascites were harvested from the mice when moribund, usually 50 to 100 days following challenge. For comparison of tumor burden, the weight of omentum was measured from tumor-bearing mice as a surrogate. This was done to reduce variability associated with harvesting diffuse nodules scattered throughout the peritoneal cavity. We found in preliminary studies that the weight of the tumor bearing greater omentum is linearly proportional to the total tumor burden (R² = 0.81, P < 0.0001, n = 56). In order to more accurately represent tumor burden, the weight of omentum of age-matched healthy mice was subtracted from the omentum of the tumor-bearing mice.

Mice were inoculated with ID8 cells and on day 25 received either 200 μg hamster IgG (Jackson ImmunoResearch Laboratories) or 200 μg G4 clone PD-1–blocking monoclonal antibody i.p. as described (21, 22). Mice were treated until moribund up to 11 treatments (8–11 times up to 7 weeks). IL10-neutralizing antibody (BioLegend) and IL10R antagonist antibody (BD Pharmingen) were injected i.p. alternating between the two antibodies at a concentration of 100 μg and 200 μg per mouse, respectively. The same amount of isotype matched antibodies (Rat IgG1κ or Rat IgG1χ) from the respective suppliers was injected in the control groups. Tumor and ascites were harvested as necessary when mice were moribund.

Leukocytes from ascites of tumor-bearing mice were isolated by discontinuous Ficoll gradient as described previously (23). Tumor-infiltrating leukocytes (TIL) were harvested by mincing the tumor through a 70-μm nylon cell strainer and separation with a discontinuous Ficoll gradient. From single-cell suspensions, DCs were magnetically isolated based on the cells of interest using CD11c microbeads obtained from Miltenyi Biotec. Naïve mouse leukocytes were obtained from BL/6J mice spleens by grinding the spleen through a 70-μm nylon cell strainer. The splenocytes were centrifuged at 3,000 rpm for 10 minutes and resuspended in 4 mL (ammonium–chloride–potassium) ACK buffer to lyse red blood cells. The remaining cells were then washed and resuspended in culture media. Bone marrow–derived dendritic cells (BMDC) were obtained from BL/6J by flushing femurs and tibias with media and centrifuging cells at 1,500 rpm for 10 minutes, followed by ACK treatment. Cells were resuspended with media containing 10 ng/mL mGM-CSF and 1 ng/mL mIL4 and plated into 6-well plates. Culture media was changed with fresh media containing 10 ng/mL mGM-CSF and 1 ng/mL mIL4 at 48 and 96 hours. The cells were matured with 40 ng/mL TNFα on day 5 of cell culture. After 2 additional days, the cells were washed with PBS, resuspended in fresh media and cytokines (VEGF-A, Genway Biotech; IL4, Peprotech); IL10, Peprotech; IL12, Peprotech) or control vehicle were added to respective wells.

Cell-surface molecule staining and flow cytometry were done essentially as previously described on whole populations and/or enriched populations (21, 22). For flow cytometric analysis, a similar number of events, usually 20,000 to 200,000 were collected for all groups. Antibodies against CD3, CD8, CD11c, F4/80, PD-1, CD45, PD-L1, PD-L2, CD40, CD27, CD69, CD80, and CD86 were purchased from eBioscience. Antibodies against CD4 and CD25 were purchased from BD Pharmingen. Appropriate isotype-matched nonspecific antibodies were used as controls.

For STAT3 inhibition, STAT3 Inhibitor VI (InSolution Calbiochem) was used at 150 μmol/L. IL10 was added to the respective wells 1 hour following the addition of the STAT3 inhibitor. STAT3 siRNAs (siRNA I and siRNA II) and the control siRNA were purchased from Cell Signaling Technologies and were used at a concentration of 150 nmol/L. Mission siRNA transfection reagent (Sigma Aldrich) was used as transfection agent. SiRNA-mediated reduction in STAT3 protein was confirmed using Western blot analysis after 48 or 72 hours. Forty-eight hours after addition of siRNA, cells were either left untreated or treated with IL10 (10 ng/mL) for an additional 48 hours. Cells were then stained for flow cytometric analysis.

BMDCs in respective experiments were lysed in lysis buffer A (150 mmol/L NaCl, 20 mmol/L Tris Cl (pH7.4), 5 mmol/L EDTA, 1 mmol/L CaCl₂, 10 mmol/L NaF, 1% Triton X 100, 5% glycerol, HALT protease/phosphatase inhibitor (Thermo Scientific). Protein (10 μg) was loaded and separated by SDS-PAGE and analyzed by immunoblotting for PD-1 (Novus Biologicals, AF1021), STAT3 (79D7) and p-STAT3 (D3A7; Cell Signaling Technology), and β-actin (LI-COR). Corresponding anti-goat, anti-mouse, and anti-rabbit secondary antibodies conjugated to IRDyes were purchased from LI-COR. Signal was detected using LI-COR Odyssey Imaging System.

RNA was isolated from purified TIDCs using Qiagen RNeasy Plus Mini Kit (Qiagen). First-strand cDNA was synthesized using Superscript First-Strand Synthesis System III (Invitrogen). PD-1–specific quantitative PCR was performed using the PD-1 (NM_008798.2) primer/probe set (Mm01285676_m1) purchased from ThermoFisher. GAPDH was used as internal control using the primer/probe set (Mm99999915_g1) from ThermoFischer. IL10 (NM_010548.2) specific PCR amplification was performed using the following primers: forward primer 5′-ATGCCTGGCTCAGCACTGCT-3′, reverse primer 5′-CATTCATGGCCTTGTAGACACCTTGG-3′. β-Actin (NM_007393.5) was used as a control. The following primers were used for β-actin specific PCR amplification: Murine forward primer 5′-GTCCCTCACCCTCCCAAAAG-3′, murine reverse primer 5′-GCTGCCTCAACACCTCAAC CC-3′. Bands were resolved by running PCR products on a 1% agarose gel.

TIDCs or BMDCs were treated with 10 μg/mL anti–PD-1 antibody (G4 clone), 10 μg/mL isotype control IgG, or left untreated for 24 or 48 hours, after which the supernatants were collected for IL10 and soluble PD-L1 assessments. TIDCs, serum, and ascites samples were also collected for IL10 and PD-L1 assessments. To measure mouse IL10, a Ready-Set-Go ELISA kit from eBioscience was used according to the manufacturer's protocol. Soluble PD-L1 ELISA was performed on the cell culture supernatants according to the manufacturer's protocol (My BioSource).

ELISA plates were coated in bicarbonate coating buffer with recombinant mouse FRα (R&D Systems) at a concentration of 1 μg/mL. Wells that did not receive any serum were used to determine the background. Serial dilutions of purified mouse IgG were used as a standard curve to convert the O.D. reading to a concentration of antigen-specific IgG. After overnight incubation at 4C, the plates were washed with 0.05% Tween in PBS, blocked with 1% BSA in PBS for 1 hour at room temperature, washed again, and then serum samples were added. Plates were then incubated overnight at 4°C followed by wash with PBS/Tween. Anti-mouse HRP secondary Ab was then added to each well and incubated for 1 hour at room temperature before washing and developing with TMB substrate. OD was read using an ELISA reader after reaction was stopped by reducing the pH of the reactions with acid.

Statistical analysis was performed using GraphPad Prism version 4.00 (GraphPad; http://www.graphpad.com). The Student t test, Mann–Whitney U test, or two-way ANOVA test was performed to determine statistically significant difference. A P value ≤0.05 was considered statistically significant.

Our recent studies have shown that PD-1, which has been primarily studied in T cells, is also expressed on the TIDCs (4). In the present study, we wanted to determine what factors (e.g., cytokines), commonly found in tumor microenvironment, lead to the increased expression of PD-1 on the TIDCs. Because TIDCs already have upregulated PD-1 and in general are difficult to access, we developed an in vitro model using BMDCs, which consistently have a much lower level expression of PD-1 relative to TIDCs (4). TNFα-activated BMDCs were treated with a variety of well-described immunologic agonists (IL10, 10 ng/mL; IL12, 20 ng/mL; VEGF-A, 25 ng/mL; poly: IC, 50 μg/mL; CpG, 50 μg/mL; or IL4, 5 ng/mL) for 48 hours after maturation, followed by flow cytometric analysis of PD-1 expression. As shown in Fig. 1A, of the agents tested, only IL10 was able to induce significantly elevated expression of PD-1 on BMDCs. IL10 dose assessments (Fig. 1B) showed dose dependency with maximal effective concentration of 5.1 ± 1.3 ng/mL (n = 18). Fig. 1C shows a flow cytometry histogram of PD-1 expression on BMDCs upon treatment with 10 ng/mL Il10. Time course experiments showed PD-1 was maximally upregulated at 48 hours following exposure to IL10 (10 ng/mL; Fig. 1D). To determine the validity of the BMDC, we also observed the effects of IL10 on expression of PD-1 on the surface of murine ovarian TIDCs derived from ascites. As shown in Fig. 1E and F, while TIDCs express high levels of PD-1, treatment with IL10 for 48 hours led to a further increase in PD-1 expression by approximately double. Increased expression of PD-1 induced by IL10 was due to augmented pdcd1 gene expression, which resulted in increased total cellular levels of PD-1 protein (Fig. 1G).

We previously showed that PD-1 maintains an immune suppressive phenotype of TIDCs and that a primary mediator of T cell suppression is TIDC-expressed PD-L1 (4, 24). Thus, as a readout for immune suppressive potential, we also analyzed PD-L1 expression following IL10 treatment of BMDCs. As shown in Fig. 2H and I, IL10 also led to an increased expression of PD-L1 on the surface. We also observed that IL10 treatment of BMDCs induced the release of soluble PD-L1 (sPD-L1) by these cells (Fig. 1J). In contrast, DCs express only low levels of PD-L2, which was not increased following treatment with IL10 (Fig. 1K and L). The effects of IL6 were also examined in comparison with IL10 because both cytokines have common proximal signaling elements (e.g., JAK-STAT3; ref. 25). IL6 was able to induce PD-1 expression in BMDCs but was not nearly as effective as IL10 (Supplementary Fig. S1A). In fact, the effect of IL6 was biphasic, peaking at 30 ng/mL and then ineffective at higher concentrations. In contrast, IL6 appeared to be more efficient at driving PD-1 expression on CD4 T cells during T-cell activation, whereas IL10 had no effect (Supplementary Fig. S1B). Similar effects were observed with CD8 T cells, although much less pronounced (Supplementary Fig. S1C). Thus, IL10, which is typically at high levels in the ovarian cancer TME, is a key driver of PD-1 expression on TIDCs.

Various transcription factors have been implicated in either promoting or blocking PD-1 expression primarily in T cells (26, 27). We found that pSTAT3 and STAT3 levels were increased in the DCs isolated from ascites of ID8 ovarian tumor–bearing mice and correlated with increased PD-1 expression (Fig. 2A). Because signaling emanating from IL10 receptor involves the activation of STAT3 molecules via tyrosine phosphorylation, subsequent dimerization, and nuclear translocation, resulting in binding to the DNA and gene transcription, we next asked if STAT3 activity has a role in IL10-mediated upregulation of PD-1 expression on DCs. BMDCs in this experiment were treated with STAT3 inhibitor (which prevents the dimerization and hence the nuclear translocation of STAT3) 1 hour prior to addition of IL10. As shown in Fig. 2B and C, STAT3 inhibition completely reversed the IL10-mediated increase in expression of PD-1 on the BMDCs. In a separate approach, we suppressed STAT3 in the BMDCs with siRNA (Fig. 2D) prior to the addition of IL10, which resulted in a decrease in PD-1 expression (Fig. 2E and F). These results are consistent with the recent findings that implicate activated STAT3 as an important transcription factor in PD-1 expression (28).

PD-1 blockade has been shown to be therapeutic in mouse models as well as in humans for certain types of cancers, yet a majority of patients among various tumor types do not benefit (29). Although the response of T and B cells to PD-1 therapy continues to be extensively explored, the effects of PD-1 blockade on DCs are not well established (21). We have shown that TIDCs respond to PD-1 blockade by increasing costimulatory molecule and MHC class I expression as well as increasing antigen presentation, the effects that were primarily dependent on release of inhibition on the NFκB pathway (21). Here, we show that disruption of the PD-1–PD-L1 axis on TIDCs using PD-1 blocking or PD-L1 blocking antibody leads to increased release of the immune regulatory cytokine IL10 (Fig. 3A). We also show that PD-1 blockade on TIDCs leads to an increase of IL10 mRNA, suggesting that the increased production of IL10 upon PD-1 blockade is not merely due to enhanced release of preexisting intracellular stores of IL10 (Fig. 3B). The effect of different concentrations of PD-1 blocking antibody on IL10 release by TIDCs was also measured. As shown in Fig. 3C, progressive increases in the concentration of the blocking antibody beyond 10 μg/mL did not progressively increase the production of IL10. In all the experiments onward, 10 μg/mL antibody was used in the in vitro experiments. From time course experiments, a significant difference in IL10 release was detected as early as 7 hours (Fig. 3D). In the ID8 tumor–bearing mice, a significant increase in the IL10 levels was detected both in the blood serum and ascites of mice that were treated with PD-1 blocking antibody compared with the untreated ones (Fig. 3E).

The effect of PD-1 blockade on BMDCs was also evaluated. PD-1⁺ BMDCs (BMDCs post treatment with IL10 as shown in Fig. 1C) were treated with PD-1 blocking Ab for 24 hours. PD-1 blockade on these cells also led to significantly increased release of IL10 (Fig. 3F). In order to check if IL10, released upon PD-1 blockade, can in turn lead to increased expression of PD-1 on surface of BMDCs, we collected the cell culture supernatants from the BMDCs treated as in Fig. 3F. Cell culture supernatants were then either left untreated or pretreated with IL10-neutralizing Ab before adding to fresh BMDC cultures. As shown in Fig. 3G and H, IL10 preneutralization prevented the increase in PD-1 expression on BMDCs treated with conditioned media from BMDCs treated with anti–PD-1. This shows that PD-1 and IL10 form a feedback loop that amplifies upon PD-1 blockade. These results identify a mechanism in DCs whereby the blockade of PD-1 on these cells leads to a compensatory release of IL10, hence potentially maintaining and fostering the suppressive environment despite PD-1 blockade.

Given the complex interdependent interaction between IL10 and PD-1 that we observed in the above studies and their role in immune suppression in tumor microenvironment, we looked to see if combining these two therapies (PD-1 and IL10/IL10R blockade) has any effect in the generation of antitumor immunity. As shown in Fig. 4A, anti–PD-1 or anti-IL10(R) as monotherapy had minimal effect on the overall survival or tumor burden; but the combination of these significantly delayed tumor growth and enhanced the survival of the tumor-bearing mice. Combination treatment also significantly reduced the tumor burden in the mice (Fig. 4B) compared with the control group.

In order to determine if the combination treatment modulated the quantity and quality of infiltrating immune cells, we looked for T- and B-cell infiltrates in the ascites of tumor-bearing mice. As shown in Fig. 5A, there was a significant increase in the total number of T cells in the ascites of tumor-bearing mice that were treated with the PD-1 blocking antibody alone or the combination regimen of PD-1 blocking and IL10(R)-neutralizing antibodies. A slight increase in the fraction of CD8⁺ T cells and a slight decrease in the frequency of CD4⁺ T cells among the total CD3⁺ T cells compared with other treatment groups were observed in the ascites of mice from the combination treatment group (Fig. 5B and C). Importantly and consistent with an inhibitory role of IL10, we observed a robust increase in both the numbers of activated CD4⁺ and CD8⁺ T cells in the combination treatment group (Figs. 5D and E), whereas those levels were not different from controls in the monotherapy groups.

Similar to the changes in T-cell numbers in the ascites, a significant increase in the frequency of B cells (CD19⁺B220⁺) was observed in the ascites of tumor-bearing mice that were treated with PD-1 blocking antibodies alone or the combination regimen (Fig. 6A and B). Thus, we sought to determine if tumor antigen-specific antibody response was enhanced in the combination treatment group. Folate receptor alpha (FRα) is a well-known tumor antigen that has been shown to be overexpressed in various epithelial derived cancer cells, including ovarian cancer, and deemed capable of driving oncogenic transformation (30). In order to determine if an antigen-specific antibody response was generated, the sera from different treatment groups were assayed for FRα-specific antibodies. As shown in Fig. 6C, there were significantly increased levels of circulating FRα-specific antibodies in the serum of mice from combination treatment group. Also, increased levels of FRα-specific antibodies were observed in ascites of mice from the combination treatment group compared with the other groups (Fig. 6D). These results demonstrate that the combination of PD-1 blockade and IL10(R) neutralization modulates the adaptive immunity by enhancing the infiltration of activated and antigen-specific immune effectors into the TME.

Immune suppressive cells present in the TME are adept at suppressing the effector responses against the tumor and hence promote tumor escape and progression. Myeloid-derived suppressor cells (MDSC) are a subset of such suppressive cells that have been shown to play a major role in immune suppression in the ovarian cancer microenvironment (31). In a murine model of breast cancer, we have shown that a combination regimen of peptide vaccine (MHC class I binding immunogenic peptides from rat neu (neu), legumain, and β-catenin targeting tumor cells, cancer stem cells, and tumor-associated macrophages) and PD-1 blocking antibody leads to a decrease in the frequency of MDSCs in the tumor (22, 32). In this study, we found that although anti–PD-1 treatment alone did not change the frequency of MDSCs in the ascites, anti-IL10, and the combination treatment significantly reduced the frequency of MDSCs in the ascites of tumor-bearing mice (Fig. 7A and B).

In the present work, we have identified the TME-associated cytokine IL10 as a critical regulator of the PD-1–PD-L1 axis in the TME. First, we found that IL10, a cytokine whose expression in increased in the TME of several cancers, is capable of increasing PD-1 surface expression in a STAT-3–dependent manner. Second, we found that blockade of PD-1, with an antagonistic monoclonal antibody, on DCs led to increased release of IL10 by DC, which is further capable of increasing PD-1 expression on DCs, hence creating a vicious immune escape loop. Lastly, we show that PD-1 blockade and IL10 signal antagonism as a combination therapy, using blocking antibodies, enhances the antitumor T- and B-cell immunity in ovarian cancer-bearing mice, leading to significantly improved survival and decrease in tumor burden.

Although the critical role of DCs in initiating and shaping the response of immune system against pathogens and malignancies is widely acknowledged, the contribution of this subset, among tumor-infiltrating immune cells, in cancer malignancies remains obscure (33). Despite comprising up to 40% of infiltrating immune cells in the tumors, studies of the role of DCs in the TME have been conflicting. Recent studies have shown that DCs may switch phenotype from immune stimulatory to immune suppressive along the course of disease progression and PD-1 expression among TIDCs potentially contributes to this process by paralyzing the cells in a persistent immune suppressive state (4, 16). Various cytokines factors have been shown to upregulate PD-1 expression on T cells, including IFNα, IFNγ, and most of the γ chain cytokines (26, 34–37). To the best of our knowledge, the present work is the first demonstration that cytokines, particularly IL10, is able to drive increased expression on DCs derived from either BMDCs or TIDCs. IL10 is a cytokine whose levels have been shown to be elevated in the TME and blood of tumor-bearing mice as well as ovarian cancer patients (18, 38–40). Despite elevated levels of IL10 in sera of both tumor-bearing mice and humans, PD-1 is selectively expressed on the TIDCs and not their peripheral DCs (4, 21). A potential reason for this observation is that IL10, while elevated in sera, may not be at the levels needed to activate pdcd1 gene expression. Indeed, our recent work shows that while elevated, IL10 levels reach only about a mean of 0.65 ng/mL in the blood of patients, which may not be enough to mediate PD-1 expression and/or systemic immune suppression (38). The IL10-mediated increase in PD-1 expression on DCs was dependent on STAT-3 activity, as use of STAT-3 inhibitor or siRNA mediated knockdown of STAT3 both resulted in inability of IL10 to increase PD-1 expression on DCs. These results support the previous findings implicating STAT-3 in mediating PD-1 expression on CD3/CD28 stimulated T cells upon IL6 treatment (28).

A unique finding from our studies of PD-1 biology on TIDC is that the blockade of PD-1 results in augmented cytokine release including high level release of IL10 (4). Thus, the observation that IL10 drives PD-1 expression and that blockade of PD-1 results in highly elevated IL10 production and release reveals a potentially rapid acting compensatory or adaptive immune regulatory loop that may foster escape from anti–PD-1 monotherapy. Indeed, we find that PD-1 blockade-induced IL10 induces further PD-1 expression on the DCs favoring continued functional paralysis. Furthermore, IL10 may also result in maintaining the immune suppressive phenotype of ovarian TIDCs by upregulating surface PD-L1 expression, which as we previously showed is dominant means that TIDCs are able to mediate immune suppression in the ovarian cancer microenvironment (24). Although PD-1 and PD-L1 bind each other, they each have other ligands they bind (1). In this regard, future studies should examine if PD-L1 blockade in vivo, with or without IL10(R) blockade, results in similar therapeutic efficacy in ovarian cancer as seen with the PD-1 blockade. Lastly, another potentially important finding is that IL10 activated release of soluble PD-L1 (sPD-L1), a form of PD-L1 that induces apoptosis of T cells and has been repeatedly shown to be associated with poor outcome in patients with advanced cancers (41–45).

In addition to increased PD-L1 expression, the increased IL10 can also act on many if not all immune cells, and blunt their effector responses (46). Interestingly, Sun and colleagues showed that blockade of PD-1 signaling led to an increase in IL10R expression on tumor antigen-specific CD8⁺ T cells, hence rendering them increasing susceptible to IL10-mediated inhibition and suggesting that compensatory mechanisms can also become activated in T cells following PD-1/PD-L1 axis blockade (47). That IL10 has a major role in evasion of immune-mediated regression of tumor following checkpoint blockade in the present model is demonstrated by the observation that combination treatment, blockade of PD-1 and IL10(R) significantly increased the survival of tumor-bearing mice and reduced the tumor burden. One potentially key finding from the current study is that although PD-1 was largely ineffective alone as a monotherapy it was able to nearly quadruple the numbers of T cells and double the number of B cells in the TME. Despite that, both subsets of effector lymphocytes appeared unable to activate as indicated by CD69 and antibody production for T and B cells, respectively, unless the mice also received anti-IL10(R) therapy. This indicates that, in this model, IL10 is a major compensatory suppression mechanism. These observations suggest that while PD-1 blockade alone may be ineffective as therapy in some tumors (29) owing to the induction of secondary suppressive mechanisms such as the one driven by IL10, a combination of PD-1 blockade and disruption of IL10/IL10R signaling enhances the endogenous antitumor immunity, resulting in improved survival and reduced tumor burden. Although the paucity of intraperitoneal tumor tissue made it difficult to assess for the functional changes in the T and B cells induced by the combination therapy, it is notable that in our prior breast cancer studies that both Th1 and Th2 responses were associated with tumor regression in mice treated with the combination of vaccine and anti–PD-1 (22). These latter results suggest that other phenotypes may contribute to regression if appropriately coordinated with Th1 cell–mediated immune responses.

We have previously shown that PD-1 blockade on TIDCs leads to an increase in IL6 release (21). In the current study, we show that IL6 is able to induce PD-1 expression on DCs, albeit the effect was much less pronounced compared with IL10. Although not studied in ovarian cancer, the combination of IL6 and PD-1/PD-L1 blockade has been evaluated in HCC and PDAC, with results showing synergistic antitumor effects (48, 49). Future studies should determine if IL6 blockade combined with PD-1/PD-L1 blockade results in therapeutic efficacy in ovarian cancer comparable with the effects of IL10 and PD-1 blockade.

The increase in FRα-specific IgG may contribute to antitumor efficacy by directly blocking FRα-dependent tumor cell proliferation and indirectly inducing antibody-dependent cell-mediated cytotoxicity (ADCC) by binding to Fcγ receptors on NK cells as shown in a preclinical study of antifolate receptor alpha antibody (50). Although we showed that IL10 is released by TIDCs at high levels following PD-1 blockade, there are other IL10 sources, such as macrophages, Bregs, and Tregs, that could also contribute to the IL10 mediated immune suppression (46). Future studies needs to evaluate if these cells, similarly to TIDCs, respond to PD-1 blockade by increasing IL10 release. IL10 mediates its immune suppressive effects both directly suppressing lymphocyte responses and indirectly by blocking DC functions (51–54). It is notable, however, that IL10 may under conditions have antitumor activity by enhancing the activity of cytotoxic T cells (55–57).

In conclusion, the approval of anti–PD-1 and anti–PD-L1 for the treatment of several cancers has generated widespread excitement in the field of tumor immunotherapy. But anti–PD-1 treatment is still suboptimal in various cancer subtypes and has not yet been tested in late-stage clinical trials for ovarian cancer. In a recent clinical study of 20 patients with platinum-resistant ovarian cancer, an overall response rate of 15% was achieved (58). Although these results are encouraging, there is still need for improvements in these therapies. Our results show that anti–PD-1 can induce a compensatory production of IL10 by DCs infiltrating the ascites of ovarian cancer. Given the presence of IL10 in high levels in the patients with ovarian carcinoma and its role in immune suppression, our results implicated it in the regulation of PD-1 expression and consequently PD-1 blockade, resulting in enhanced IL10 production is significant. These results open up possibilities to explore if such combination therapy is a viable option in clinical settings for added benefits in patients receiving PD-1–based therapies. Other mechanisms of resistance to PD-1 therapy, including induction of other checkpoint molecules, lack of T-cell infiltration, and low tumor mutational load, have also been identified in various tumors and combinatorial approaches to target these mechanisms are being explored (59).

M.S. Block reports receiving other commercial research support from Merck. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Lamichhane, B. Shreeder, J. Krempski, E.L. Goode, K.L. Knutson

Development of methodology: P. Lamichhane, L. Karyampudi, K.L. Knutson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Lamichhane, B. Shreeder, D. Bahr, J. Daum, K.R. Kalli, K.L. Knutson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Lamichhane, L. Karyampudi, B. Shreeder, J. Daum, E.L. Goode, K.L. Knutson

Writing, review, and/or revision of the manuscript: P. Lamichhane, L. Karyampudi, B. Shreeder, J. Daum, K.R. Kalli, E.L. Goode, M.S. Block, M.J. Cannon, K.L. Knutson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Bahr, M.J. Cannon, K.L. Knutson

Study supervision: K.L. Knutson

The authors are grateful for the technical support of Laura Lewis-Tuffin and the Mayo Clinic Florida Flow Cytometry and Cell Analysis Shared Resource.

This work was supported by the Minnesota Ovarian Cancer Alliance (K.L. Knutson and L. Karyampudi), the Fred C. and Katherine B. Andersen Foundation (K.L. Knutson and K.R. Kalli), the Marsha Rivkin Center for Ovarian Cancer Research (L. Karyampudi, K.L. Knutson, and M.J. Cannon), Mayo Clinic Comprehensive Cancer Center grant P30-CA015083 (R. Diasio), and the Mayo Clinic Ovarian Cancer SPORE (P50-CA136393 to K.L. Knutson, E.L. Goode, M.S. Block, and M.J. Cannon).

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