Decitabine Enhances Lymphocyte Migration and Function and Synergizes with CTLA-4 Blockade in a Murine Ovarian Cancer Model

Effective treatment options for ovarian cancer remain limited, as most patients relapse. Immunotherapy can address disease that is refractory to conventional treatment options and represents an exciting new modality to achieve durable patient responses (1). Still, the percentage of such responders could be improved greatly. Improving the immunogenicity of tumors represents one approach to overcome a tumor's evasion of the immune system (2). Similarly, inhibiting immune checkpoint pathways, such as CTLA-4, which attenuates T-cell activation and related antitumor responses, can improve patient outcomes.

CTLA-4 blockade counteracts the immunosuppressive effects of regulatory T cells (Tregs; refs. 3, 4) and stimulates effector T-cell activation, expansion, function, and motility (5, 6). CTLA-4 blockade can also eliminate myeloid-derived suppressor cells (MDSC), which contributes to improving the survival of mice bearing ovarian tumors (7). The therapeutic benefits of targeting checkpoint blockade have been validated in patients with advanced solid tumors (8), including ovarian cancer (9). Although promising, the responses to date benefit only a minority of patients. The administration of an appropriate complementary therapy is expected to improve the magnitude and frequency of responses.

Epigenetic modifiers have gained considerable attention of late in the pursuit of antitumor compounds that do not impart the harsh side effects of cytotoxic chemotherapeutics, as epigenetic patterns differ between normal cells and cancer cells, enabling discrimination (10). The effects of these drugs are thought to be particularly profound in the context of cancer stem cells (11) and drug tolerance heterogeneity (12).

Although DNA hypomethylation is often thought to relieve repression of tumor suppressor genes (13), there is emerging evidence that DNA methyltransferase inhibitors induce expression of immunoregulatory genes in cancer cells (14, 15). Of the cancers investigated to date, this effect is most pronounced in ovarian cancer cells (16). Treatment of ovarian cancer cells with the DNA methyltransferase inhibitor DAC (17) results in expression of cancer–testis antigens and MHC class I, increasing cancer cell recognition by CD8⁺ T cells (18). DAC similarly restores expression of cytokines, cancer–testis antigens, MHC class I and II, and costimulatory molecules in cancer patients (19, 20). These findings support the hypothesis that DAC will synergize with cancer immunotherapies.

In this study, we investigated the impact of DAC on global gene expression in ovarian cancer cells and found the most upregulated class to be immunoregulatory genes. We then confirmed that DAC increases the infiltration and antitumor function of tumor-associated lymphocytes, which are associated with a survival benefit in subcutaneous (s.c.) and orthotopic ovarian tumors. Finally, we combined DAC with immune checkpoint blockade antibodies and found a synergy between DAC and CTLA-4. The mechanism of synergy is related to the kinetics of lymphocyte cytokine production, which is sustained by anti–CTLA-4 following early but transient induction by DAC. The combination of epigenetic modifiers and immune checkpoint blockade is currently being investigated in the treatment of non–small cell lung cancer (NCT01928576), and the data herein suggest that such a combination may benefit ovarian cancer patients as well.

BR5FVB1-Akt cells, generated as previously described (21), were seeded in a 24-well plate overnight and were incubated with 10 mg/mL of DAC for 72 hours, changing the media and drug every 24 hours. Cells were harvested, and total RNA was extracted using the RNeasy Mini Kit (Qiagen). RNA concentration was measured using the Nanodrop (Thermo Fisher Scientific), and whole-transcript gene array (Mouse gene ST 2.0) was performed by the Molecular Biology Core Facility at Dana-Farber Cancer Institute (Boston, MA). The data were analyzed using Affymetrix Transcriptome Analysis Console software. A total of 34,472 genes were included in the analysis. A linear fold change of 2 was used as a minimum threshold value, and Gene Functional Classification Tool (DAVID Bioinformatics Resources 6.7, NIAID/NIH), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, and Gene Ontology map analysis were used to annotate pathways of differentially regulated genes. The data are deposited in the Gene Expression Omnibus (GEO) under the accession number GSE69454.

BR5FVB1-Akt cells were treated with DAC as above. The relative gene expression of Ccl2, Ccl5, Cxcl10, Il6, Irf7, and Tnfrsf18 (GitrL) was measured using the Ambion Cells-to-CT Kit 1-Step Power SYBR Green Kit (Life Technologies) and the Applied Biosystems 7500 Fast Real-Time PCR System and Software. Mouse β-actin was used as the endogenous housekeeping gene for normalization. All quantitative reverse transcription PCR (qRT-PCR) assays were carried out in technical triplicate with biologic duplicate, and the experiments were repeated independently two to three times.

BR5FVB1-Akt cells were cultured in DMEM with 10% FBS and 1% penicillin/streptomycin at 37°C with 5% CO₂. Six-week-old female FVB mice were obtained from Charles River Laboratories. Mice were housed in a pathogen-free facility in accordance with the standards of the Dana-Farber Cancer Institute (Boston, MA). All animal experiments were approved by the Institutional Animal Care and Use Committee. Mice were kept in groups of 3 to 5 per cage in a 12-hour light/12-hour dark cycle and housed at 25°C with 50% relative humidity. For the intraperitoneal (i.p.) model, 4 million BR5FVB1-Akt cells were injected into the peritoneal cavity, and the mice were sacrificed upon development of pronounced ascites. In the absence of treatment, hemorrhagic ascites were formed 20 to 30 days after tumor cell inoculation. For the subcutaneous model, 2 million BR5FVB1-Akt cells were implanted s.c. into FVB mice on the left flank. Tumors formed 30 to 40 days after implantation, and the mice were sacrificed when the tumor size reached 2 cm³.

DAC was dissolved in DMSO (5 μL) and mixed with PBS (95 μL). One mg/kg of DAC (Selleckchem) was injected i.p. on days 4, 6, and 8 after BR5FVB1-Akt cell inoculation for both i.p. and s.c. models. For combinational therapy, 100 μg of anti–CTLA-4 antibody (clone: 9H10) or hamster IgG isotype control in PBS was additionally administered to mice on days 5, 7, and 9 after tumor inoculation. All antibodies were purchased from BioXCell. Mouse ascites and s.c. tumors were monitored daily.

Tumors from solvent control- and DAC-treated groups were collected on day 42 and fixed in 4% paraformaldehyde at 4°C for at least 48 hours, followed by submerging the fixed tissues in 30% sucrose for 24 hours. The tissues were embedded in optimal cutting temperature (OCT) and cryosectioned to slices 10 μm in thickness. Slides were permeabilizied with 0.2% Triton in PBS and blocked in 2% BSA in PBS for 30 minutes prior to incubation with anti-mouse/rat FoxP3 antibody (eBiosciences 11-5773-82), anti-mouse CD8a antibody (Biolegend 100706), and anti-mouse CD45 antibody (Biolegend 103112) for 1 hour at room temperature. To enhance the green fluorescent signal, Alexa Fluor 488 Conjugate (Invitrogen A11096) was added after the primary antibody incubation. Slides were mounted in SlowFade Gold anti-fade reagent with 4′,6-diamidino-2-phenylindole (DAPI) and immediately imaged using a Leica TCS SP5 X confocal laser scanning microscope (Leica Microsystems Inc.).

Tumors were harvested on day 45 after tumor cell implantation, weighed dry, cut into small pieces, and incubated in collagenase solution in a 37°C water bath for 2 hours. GentleMACs was used at end of the incubation to improve homogenization. The dissociated tumors were passed through a 70-μm filter to generate single-cell suspensions, and red blood cells were lysed as necessary. Cells were blocked with 2% FBS and anti-mouse CD16/32 (TruStain fcX; Biolegend) in PBS for 10 minutes at 4°C. Next, they were stained with the following cocktail of fluorescently labeled antibodies: anti-CD3, anti-CD4, anti-CD8a, and anti-Nkp46. The cells were fixed and permeabilized using the FoxP3 buffer kit (eBioscience) for intracellular staining of FoxP3. Cells were then analyzed on a BD Fortessa or BD LSR II using FlowJo software.

The mice were sacrificed 14 or 42 days after tumor cell inoculation. A peritoneal wash was performed by injecting cold saline solution (2% FBS in PBS) into the mouse peritoneum, followed by gentle massage. The cell-containing saline solution was then collected. Again, red blood cells were lysed as necessary. For phenotypic analyses, the following antibodies were used: anti-CD3, anti-CD4, anti-CD8a, anti-CD44, anti-CD62L, anti-Ki67, anti-CD25, anti-Nkp46, anti-F4/80, anti-CD11b, anti-Gr1, anti-Ly6G, anti-CD11c, and anti-FoxP3. For functional assays, the cells were stimulated with phorbol 12-myristate 13-acetate (PMA; 50 ng/mL) and ionomycin (500 ng/mL) with Golgi-Stop (BD Biosciences) in RPMI medium (supplemented with 10% FBS and 1% penicillin/streptomycin). An anti-CD107a antibody was also added during stimulation to evaluate degranulation. After stimulation, surface and intracellular staining were performed using the following antibodies: anti-CD3, anti-CD4, anti-CD8a, anti-Nkp46, anti-FoxP3, anti-IL10, anti-IFNγ, and anti-TNFα. Samples were analyzed on a BD Fortessa or BD LSR II using FlowJo software. All antibodies were purchased from Biolegend, except anti-Nkp46, which was purchased from eBioscience.

Statistical analyses were performed using a two-tailed Student t test (GraphPad Prism). Error bars denote mean ± SD. A P value of less than 0.05 was considered to be statistically significant (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001).

The DNA methyl transferase inhibitor azacitadine has immunomodulatory effects on multiple cancer types, particularly ovarian cancer (16). To explore the impact of the closely related DAC (17) on gene expression in this tumor model, we performed a genome-wide microarray analysis of BR5FVB1-Akt cells treated with DMSO control or DAC for 72 hours. Out of 34,472 unique genes investigated, 83 genes were upregulated and 28 genes were downregulated by at least 2-fold after DAC treatment. Heatmap analyses demonstrate that control- and DAC-treated cells exhibit different global gene expression profiles (Fig. 1A), and the most upregulated Gene Ontology signature is “immune system process” (Fig. 1B). KEGG pathway analysis reveals that the DAC upregulated genes are involved in multiple immune-related signaling pathways, such as cytokine–cytokine receptor interaction, chemokine signaling pathway, and MAPK and TLR signaling pathways (Fig. 1C). Their encoded proteins may contribute significantly to T-cell and NK-cell migration toward the tumor site and subsequent proliferation and antitumor function (22, 23).

The functional groups of the modulated genes were determined using Gene Functional Classification Tool (DAVID Bioinformatics Resources 6.7, NIAID/NIH) and are shown in Supplementary Table S1. To validate the microarray results, we assessed the levels of several noteworthy immune-related genes by quantitative PCR (qPCR). BR5FVB1-Akt cells were treated with DAC, and their RNA was harvested after 72 hours. qPCR analysis confirmed that treatment with DAC upregulates all of the genes inspected (Ccl2, Il6, Cxcl10, GitrL, Ccl5, and Irf7; Fig. 1D). DAC treatment also upregulates surface major histocompatibility complex I (H-2Kq haplotype) on BR5FVB1-Akt cells in a dose-dependent manner (Supplementary Fig. S1). These results revealed an association between DAC treatment and the expression of immune-related genes.

The therapeutic potential of DAC was subsequently examined in vivo using an orthotopic syngeneic tumor model. BR5FVB1-Akt cells were inoculated i.p. into FVB mice, and DAC was administered i.p. at a dose of 1.0 mg/kg on days 4, 6, and 8 after tumor inoculation (Fig. 2A). An equivalent volume (100 μL) of DMSO/PBS (“solvent”) was administered to a second cohort as a negative control. The tumor-bearing mice developed hemorrhagic ascites gradually in the peritoneal cavity and had to be euthanized by day ∼25 in the absence of treatment. Tumors were detectable post-mortem as nodules of various sizes adherent to the omentum, peritoneal wall, or intestinal mesentery. The onset of ascites in DAC-treated mice was delayed by approximately 10 days, and the survival of these mice was significantly improved relative to control mice (P < 0.01, Fig. 2B).

Next, we sought to determine how DAC mediated its antitumor effect. Orthotopic tumors were generated, and tumor-bearing mice were treated with solvent or DAC as above. On day 14 after inoculation, peritoneal fluid was collected, and tumor-associated leukocytes (TAL) were analyzed by flow cytometry. Representative figures of flow-cytometric analysis for various leukocytes, such as CD4⁺ helper T cells, CD8⁺ cytotoxic T cells, NK cells, CD4⁺ Tregs, and MDSCs, are shown (Fig. 3A). The percentage of NK cells increased and the percentage of MDSCs decreased in DAC-treated mice compared with those in control mice (Fig. 3A and B). The percentages of CD4⁺ helper T cells, CD8⁺ cytotoxic T cells, and Tregs did not differ significantly (Fig. 3B), nor did the percentages of macrophages, neutrophils, and dendritic cells (data not shown).

To elucidate the functional capacity of effector lymphocytes in the tumor environment, we evaluated the expression levels of IFNγ and TNFα among CD8⁺ cytotoxic T lymphocytes and NK cells that had been recovered from control- or DAC-treated mice. Representative figures of flow-cytometric analysis of these immune effector molecules are shown for CD8⁺ T cells and NK cells (Fig. 3C). After DAC treatment, the expression of both IFNγ and TNFα was significantly upregulated in both lymphocyte populations (Fig. 3D). These data suggest that DAC treatment recruits NK cells and promotes NK cell and CD8⁺ T-cell antitumor function in the tumor microenvironment.

As with the human presentation of the disease, the orthotopic tumor nodules are spread diffusely throughout the peritoneal cavity; therefore, harvesting sufficient bulk tissue is challenging for flow-cytometric analyses. Thus, to further investigate the phenotypes and functions of immune cells in solid tumor masses, s.c. tumors were generated. On day 45 after inoculation, tumors were collected, photographed, and weighed. Tumor-draining lymph nodes and spleens were also collected to enable a systematic study of the immune compartment. BR5FVB1-Akt cells were inoculated s.c. into FVB mice, and DAC was administered at a dose of 1.0 mg/kg via i.p. injection on days 4, 6, and 8 after tumor inoculation (Fig. 2A). An equivalent volume (100 μL) of solvent was administered to a second cohort as a negative control. More than one third of DAC-treated mice achieved a complete response (Fig. 4A), as evidenced by a lack of palpable tumor over 120 days. To inspect the mechanism underlying this survival benefit, additional cohorts of mice were sacrificed on day 35 after inoculation. Tumors isolated from mice receiving DAC exhibited a reduction in weight relative to tumors from control mice (Fig. 4B). The frequency of tumor-infiltrating CD8⁺ cytotoxic T cells, NK cells, and CD4⁺ helper T cells is increased in DAC-treated mice (Fig. 4C and E). A higher ratio of tumor-infiltrating CD8⁺/Treg, which is associated with favorable prognosis in ovarian cancer (24), was also observed in DAC-treated mice.

We confirmed the increased density of CD8⁺ T cells in tumor tissue by immunofluorescence staining and confocal imaging. Serial tumor biopsies from control and DAC-treated mice were stained for CD8, CD45, and DAPI. We found very few infiltrating CD8⁺ T cells or CD45⁺ leukocytes in tumors from control mice but a reasonably dense infiltrate of these cells in tumors from DAC-treated mice (Fig. 4G and Supplementary Fig. S2), mirroring the flow cytometry findings for CD8⁺ T cells. CD8⁺ T cells were also observed in elevated concentrations in tumor-draining lymph nodes following treatment (Fig. 4D and F). In contrast, their numbers in the spleen were unaltered by DAC treatment (data not shown), indicating that the drug's effects are concentrated at the s.c. tumor site and tumor-draining lymph node despite being administered i.p.

DAC exhibited significant but incomplete therapeutic efficacy as a monotherapy in this ovarian cancer model. This mechanistic study reveals that DAC modulates the function of CD8⁺ T cells and NK cells in the peritoneum and increases their frequency in s.c. tumors. We hypothesized that immune checkpoint blockade would synergize with DAC therapy to combat tumorigenicity and promote antitumor immunity by enhancing CD8⁺ T-cell and NK cell responses. Orthotopic tumors were generated, and tumor-bearing mice were dosed according to the schedule in Fig. 5A. Combining DAC with anti–CTLA-4 substantially delayed the onset of ascites relative to either monotherapy, with a majority of the mice achieving a complete response for a period of 90 days after tumor inoculation (Fig. 5B).

Next, we inspected the mechanism underlying the survival benefit achieved by combining DAC and anti–CTLA-4. Orthotopic tumors were generated, and tumor-bearing mice were administrated with either DAC plus isotype antibody or DAC plus CTLA-4 blockade, following the same dosing schedule shown in Fig. 5A. On day 42 after tumor inoculation, peritoneal fluid was collected and TALs were analyzed. Phenotype analysis of CD8⁺ T cells, CD4⁺ T cells, Tregs, and NK cells showed no significant changes between two groups (data not shown). However, effector memory (CD44^hiCD62L^lo) CD8⁺ and CD4⁺ T-cell populations were highly enriched, with a concomitant decrease in naïve T cells in the combination group relative to DAC monotherapy (Fig. 6A–D). Ki67 expression on TALs was also increased in both CD8⁺ and CD4⁺ T-cell populations (Fig. 6E and F), indicating increased cell proliferation. The frequency of central memory T cells was unaffected by CTLA-4 blockade (Fig. 6A–D).

Production of IFNγ and TNFα by CD8⁺ T cells (Fig. 6G and H) and NK cells (Fig. 6I and J) was elevated in the combination group, suggesting that these cells were functionally engaged in antitumor immunity. Indeed, NK cells exhibited increased degranulation, as determined by CD107a staining (Fig. 6K and L), suggesting that NK cells directly execute cell killing in the context of the increased proinflammatory microenvironment. Notably, CTLA-4 was recently shown to be expressed on NK cells upon their activation and to inhibit their production of IFNγ (25). Consistently, we observed that CTLA-4 blockade promotes IFNγ production by NK cells.

TALs in DAC-treated mice displayed high cytokine production at an early time point (day 14) relative to the control group (Fig. 4A) but relatively low cytokine secretion at a late time point (day 42; Fig. 7A). We therefore investigated TAL function of the DAC plus isotype group and the DAC plus anti–CTLA-4 group at early and late stages of the disease. As described above, peritoneal fluid from mice treated with DAC plus isotype or DAC plus anti–CTLA-4 was collected. Cytokine production from CD8⁺ T and NK cells was analyzed on days 14 and 42 after tumor inoculation. We surprisingly observed a slight decrease of lymphocyte function after the addition of anti–CTLA-4 at the early time point (Fig. 7; P > 0.05 in Fig. 7A–C and P = 0.0129 in Fig. 7D). However, lymphocyte function was greatly decreased at day 42, compared with day 14, in the DAC plus isotype group. In contrast, relatively high levels of cytokine were sustained in the combination treatment group at the later time point (Fig. 7). These data suggest that the immune response initiated in response to DAC treatment can be prolonged by the addition of anti–CTLA-4.

In this study, we evaluated a novel drug combination regimen for epithelial ovarian cancer and explored the underlying mechanisms using an orthotopic mouse model. The approach uses an epigenetic modifier, DAC, in combination with immune checkpoint blockade, anti–CTLA-4. We show a prolonged cytotoxic lymphocyte response in the context of combination therapy and the resultant synergetic effect on survival.

From animal experiments to clinical trials, DAC has been investigated as a sensitizer to immunotherapy (26). Of note, its ability to induce expression of cancer–testis antigens (19, 27), most notably NY-ESO-1, for improved vaccination of ovarian cancer patients (28), is well established. Microarray analysis previously demonstrated that azacitadine, another clinically available hypomethylating agent, yields substantial immunomodulatory effects on ovarian cancer cell lines (16). Our microarray data are in agreement with those from the previous report and show that the most upregulated Gene Ontology signature is “immune system process” after 3 days of DAC treatment. The upregulated genes are involved in several significant pathways that regulate T-cell trafficking and function. For example, the chemokines CCL2, CCL5, and CXCL10 are known to contribute to NK-cell (29, 30) and CD8⁺ T-cell (31, 32) recruitment into the tumor microenvironment. Upregulation of these genes is expected to promote lymphocyte infiltration and antitumor function. Although cellular stress is likely partially responsible for the observed effects, DAC is not particularly cytotoxic at the dosage used (Supplementary Fig. S3), supporting the notion that epigenetic modification is important to the phenotype (14–16).

DAC treatment additionally increases expression of Tnfsf18 (GitrL), which is the ligand for the T-cell costimulatory molecule GITR. GITR ligation enhances the infiltration and activity of effector T cells (33) and NK cells (22, 34) and generates effector T cells that are resistant to the immunosuppressive effects of Tregs (35). These data suggest that DAC not only has cancer cell–intrinsic antitumor effects but also modifies the tumor microenvironment to promote T-cell infiltration and antitumor function. DAC does not affect T-cell proliferation in vitro (data not shown), suggesting that DAC's effect on T cells is indirect, mediated through the drug's action on cancer cells. DAC also directly augments NK-cell responses to activating stimuli, such as IL2 or target cancer cells, and promotes NK cell–mediated cytotoxicity and IFNγ production in vitro by inducing transcription of genes involved in NK reactivity (36). These effects contrast with the impact of the related compound azacytidine, which impairs NK function (36).

We then evaluated the therapeutic benefits of DAC in an orthotopic ovarian cancer model using the BR5FVB1-Akt cell line, which lacks p53 and Brca1 and overexpresses Myc and Akt. These genetic alterations recapitulate the most common genetic alterations observed among ovarian cancer patients (37, 38). Importantly, this animal model resembles the human disease phenotypically as well as genetically. As seen in the human presentation of ovarian cancer, the cancer cells disseminate throughout the peritoneal cavity and metastasize to the peritoneum wall, intestine, liver, and spleen, causing hemorrhagic ascites as well as occasional bowel obstruction and cachexia. Treating mice with DAC significantly delays ascites formation as well as tumor growth. Subcutaneous tumors in mice that received DAC exhibit a significant reduction in growth, with 40% of these mice achieving a complete response.

Immune cell phenotyping shows that DAC-treated mice recruit more NK cells and fewer MDSCs into their peritoneal cavities. MDSCs promote tumor progression via multiple suppressive mechanisms (39). Consistent with our results, it has been observed that the frequency of tumor-infiltrating MDSCs is also decreased in the B16 murine melanoma model in response to DAC treatment (40). NK cells are important in innate immune responses against tumors by direct cytolysis of cancer cells (41) and by antibody-dependent cellular cytotoxicity (ADCC; ref. 42). The low response rate associated with trastuzumab (anti-HER2) monotherapy is thought to be related to impaired NK-cell function in patients with advanced disease, suggesting that treatments that improve NK-cell activity may promote ADCC in addition to direct cytolysis (43). Thus, DAC might be particularly useful in combination with a depleting isotype of anti–PD-L1, promoting infiltration of NK cells while augmenting their ability to kill PD-L1–expressing epithelial cancer cells and MDSCs.

In the orthotopic setting, DAC enhances the function of tumor-associated CD8⁺ T cells and NK cells. Production of the Th1 cytokines IFNγ and TNFα is greatly increased in CD8⁺ T and NK cells isolated from DAC-treated mice. These proinflammatory cytokines play a critical role in surveillance against ovarian tumor development (44). In addition, DAC treatment induces CD80 expression on cancer cells and stimulates tumor-specific cytotoxic T- lymphocyte responses (45). Consistently, CD80 expression on BR5FVB1-Akt cells is increased 3-fold following DAC treatment in vitro (Supplementary Fig. S3). CD80 may be further upregulated in vivo owing to the increased IFNγ and TNFα production from TILs, as surface expression of CD80 can be induced in cancer cell lines and fibroblasts in response to IFNγ and TNFα stimulation (46, 47).

CTLA-4 is a homolog of the costimulatory molecule CD28 and acts to suppress T-cell function by competing with CD28 for their shared ligands CD80 and CD86 on antigen-presenting cells and many cancer cells. Binding of CTLA-4 to these ligands inhibits TCR signaling. Anti–CTLA-4 can reverse such immunosuppression by limiting the activity of tumor-infiltrating Tregs and through FcγR-dependent depletion of intratumoral Tregs (4). CTLA-4 expression is induced on CD4⁺ and CD8⁺ T cells following activation and is constitutively expressed on Tregs (48). A recent study demonstrated that CTLA-4 is also expressed on NK cells upon activation and inhibits their production of IFNγ (25). Together, these findings support the potential synergy between DAC and anti–CTLA-4.

This therapeutic combination yields a much greater antitumor effect than either treatment alone. Mechanistically, the combination promotes the differentiation of naïve T cells into effector CD8⁺ and CD4⁺ T cells and enhances their proliferation in the peritoneal cavity. Tumor-associated CD8⁺ T cells and NK cells also express more IFNγ and TNFα upon addition of the CTLA-4–blocking antibody. A kinetic analysis comparing early- and late-stage disease (14 days vs. 42 days after tumor cell inoculation) reveals that the immune response to DAC is early and transient, but it can be sustained by the addition of anti–CTLA-4. Specifically, the cytotoxic lymphocytes produce high levels of IFNγ and TNFα at day 14, with slightly higher cytokine production in DAC monotherapy than in the combination therapy at this early time point. In contrast, at the later time point, addition of anti–CTLA-4 substantially augments cytokine production, prolonging the benefits of DAC treatment. Recent literature suggests that anti–CTLA-4 enhances CD8⁺ T-cell effector memory formation, function, and maintenance (49), which is consistent with our findings. Of note, the combination also enhances NK cell degranulation, which is a surrogate for direct cancer cell killing. The importance of NK cell and T-cell crosstalk in the tumor microenvironment has been described recently (45). Specifically, the infiltration of both NK cells and CD8⁺ T cells into colorectal cancer is associated with prolonged patient survival.

In summary, the combination therapy achieves its antitumor efficacy by stimulating both innate and adaptive immunity, influencing lymphocyte infiltration, differentiation, cytokine production, and cytolysis. The results provide a rationale for the clinical translation of a regimen combining epigenetic modifiers and immune checkpoint blockade and suggest that this combination may extend the survival of ovarian cancer patients.

No potential conflicts of interest were disclosed.

Conception and design: L. Wang, Z. Amoozgar, J. Huang, M.S. Goldberg

Development of methodology: L. Wang, Z. Amoozgar, J. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Wang, Z. Amoozgar, J. Huang, M.H. Saleh, S. Orsulic

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Wang, Z. Amoozgar, M.H. Saleh

Writing, review, and/or revision of the manuscript: L. Wang, Z. Amoozgar, J. Huang, M.H. Saleh, S. Orsulic, M.S. Goldberg

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Wang, M.H. Saleh, D. Xing

Study supervision: L. Wang, M.S. Goldberg

The authors thank Lina Wu (Peking University Cancer Hospital and Institute) for microarray data analysis and Hye-Jung Kim and Jianmei Leavenworth for helpful discussions.

This work was supported by the Ovarian Cancer Research Fund (Liz Tilberis Scholar), Aid for Cancer Research, Kaleidoscope of Hope, and the Susan F. Smith Center for Women's Cancer.

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.