Histone Deacetylase Inhibitors Synergize with Catalytic Inhibitors of EZH2 to Exhibit Antitumor Activity in Small Cell Carcinoma of the Ovary, Hypercalcemic Type

Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT) is a rare type of ovarian cancer type with poorly differentiated histology, which primarily affects young women with a median age of diagnosis in mid-twenties (1–5). Unlike most common ovarian malignancies, the genome of SCCOHT is diploid with inactivating mutations of the SMARCA4 gene, encoding the ATPase of the SWI/SNF chromatin-remodeling complex, as the only recurrent feature and the likely driver event in the majority of SCCOHT tumors (6–9). In addition, SCCOHT do not express SMARCA2 (10, 11), the alternative ATPase of SWI/SNF chromatin-remodeling complex, a surprising finding given the requirement for SMARCA2 for the survival of most other SMARCA4-deficient cancer cells (12, 13).

Despite the receipt of extensive chemotherapy following surgical debulking, the prognosis of SCCOHT patients is very poor with a 2-year survival rate less than 35% (14, 15), highlighting the urgent demand to develop novel therapeutic options for these patients. Preclinical studies suggest that a subset of patients with SCCOHT may benefit from c-Met inhibitors (16) or oncolytic virus (17). In concordance with the known antagonism between SWI/SNF complex and the polycomb repressive complex 2 (PRC2), we and Chan-Penebre and colleagues have both recently demonstrated that SCCOHT cells are highly sensitive to catalytic inhibition of EZH2, the enzymatic subunit of the PRC2 complex (18, 19). Despite the therapeutic promise of targeting the epigenome of SCCOHT, clinical trial testing of the EZH2 inhibitor EPZ-6438 (tazemetostat) only led to stable disease or partial response in two SCCOHT patients, previously treated with chemotherapy (www.epizyme.com). Therefore, determining whether SCCOHT cells depend on additional epigenetic modulators for survival and whether targeting them can improve the response of SCCOHT cells to EZH2 inhibition remains a priority.

Through regulating the acetylation state of histones, histone deacetylases (HDAC) and acetyltransferases play important roles in the maintenance of chromatin and in regulating many biological processes including transcriptional control, chromatin plasticity, protein-DNA interactions and cell differentiation, growth, and death (20–22). Dozens of HDAC inhibitors, targeting one or several HDACs, have been developed as anticancer agents for reversing aberrant epigenetic states associated with cancer. Most of them induce apoptosis and cell cycle arrest and prevent invasion, metastasis, and angiogenesis. Several pan-HDAC inhibitors, such as SAHA (vorinostat), romidepsin, and panobinotstat, have been approved by the US FDA for treating various hematopoietic malignancies, such as cutaneous T-cell lymphoma. However, treatment with HDAC inhibitors as single agents has often demonstrated limited clinical benefit for patients with solid tumors, prompting the investigation of genetic vulnerability associated with HDAC inhibition and treatment combinations with other cancer therapeutics to improve their clinical utility.

Herein, we demonstrate that that SCCOHT cells were more sensitive to HDAC inhibitors compared with other ovarian cancer lines. Although HDAC inhibitors induced apoptosis and differentiation of SCCOHT cells, the combined treatment with EZH2 inhibitors synergistically suppressed their proliferation, triggered apoptosis, and inhibited their tumor growth in xenograft models.

Cells were cultured in either DMEM/F-12 (BIN67, SCCOHT-1, COV434, and SVOG3e) or RPMI (all other lines) supplemented with 10% FBS and maintained at 37 °C in a humidified 5% CO₂-containing incubator. All cell lines have been certified by STR analysis, tested regularly for Mycoplasma and used for the study within 6 months of thawing. EPZ-6438 (23), quisinostat (24), SAHA, romidepsin, and panobinostat were purchased from Selleckchem for in vitro studies. EPZ-6438 and quisinostat were purchased from Active Biochemku for in vivo studies.

Lenti-GFP (EX-EGFP-Lv102) and Lenti-SMARCA4 (EX-Y4637-Lv102) plasmids were obtained from Genecopeia (EX-Y4637-Lv102). A SMARCA2 gRNA targeting the SMARCA2 genomic region (5′-CTTGTCATGTATACCATCGATGG-3′) was cloned into lentiCRISPR vector (Addgene #49535) to construct lentiCRISPR-SMARCA2. Specific siRNAs used to knockdown HDAC1 were purchased from Dharmacon, including D-003493-04 (#1) and D-003493-09 (#2). Specific siRNAs used to knockdown HDAC2, 5′-GCUACUAAGAUGUGCAAAGAAGACA-3′ (#1) and 5′-CCAGAACACUCCAGAAUAUAUGGAA-3′ (#2), were obtained from Integrated DNA Technology. To produce lentivirus expressing SMARCA4 gene or SMARCA2 gRNA, Lenti-SMARCA4 or Lenti-CRISPR-SMARCA2 was cotransfected with packaging plasmids psPAX2 and pMD2.G into HEK293T cells. Supernatants were collected at 48 and 72 hours for lentivirus preparation and infection.

Cells were seeded in 96-well plates in quadruplicate at a density of 500 to 2,000 cells per well depending on the growth curve of each cell line. Twenty-four hours later, cells were treated with either DMSO or HDAC inhibitors at indicated nine doses (decreasing in two-fold dilution from the highest concentration). For drug combination assays, cells were treated with HDAC inhibitors (i.e., quisinostat, 0–6 nmol/L) and EZH2 inhibitors (i.e., EPZ-6438, 0–1 μmol/L) at various ratios. Six days after treatment, cells were fixed in methanol:acetic acid:water solution (1:1:8) and then stained with 0.5% crystal violet in methanol. The absorbed dye was resolubilized with 10% acetic acid and measured spectrophotometrically at 595 nmol/L. Cell survival was calculated by normalizing the absorbance to that of DMSO-treated controls.

Cells seeded in 96-well plates were exposed to epigenetic drug library (Cayman #11076) at 1 μmol/L. After 6 days, cells were fixed and proceeded to cell viability measurement as described above. Three independent assays were performed to calculate the average cell growth inhibition rate, determined by (100−%cell remaining of drug-treated well)/%survival of cell remaining of DMSO-treated wells.

Cells were seeded in six-well (SCCOHT-1, 2,000 cells/well) or 12-well plates (BIN67: 1,000 cells/well; COV434: 250 cells/well) in triplicate. Twenty-four hours later, cells were treated with either DMSO or 2 nmol/L quisinostat. Drugs were refreshed every 4 to 5 days. Two weeks after treatment, cells were fixed in 10% methanol to 10% acetic acid and then stained with 0.5% crystal violet to determine the numbers of colonies under each condition.

To determine the effect of quisinostat on spheroid formation, SCCOHT cells were mixed in culture media containing 3% matrigel and then seeded into six-well plates precoated with matrigel. DMSO or quisinostat were then added in media with 3% matrigel and refreshed every 3 to 4 days. For analyzing the effect of quisinostat on preformed spheroids, cells (2,000 cells/well) were seeded in 200 μL medium per well containing 2% Matrigel (Corning) in 96-well plates precoated with 1% agarose. After being centrigated at 1,000 × g for 5 minutes, cells were cultured for three days to allow formation of spheroids. Then, culture medium was changed together with adding either DMSO or Quisinostat at indicated dose. After being exposed to drug for 6 days, cell viability was measured by CelltiterGlo (Promega) and normalized to DMSO-treated cells.

Cells treated with DMSO or quisinostat were fixed with 70% ice-cold ethanol, stained for DNA content with propidium iodide, and subjected to flow cytometry analysis to quantitate cell-cycle distribution. To monitor apoptosis, the cell-permeable dye (NucView 488 dye) coupled to an activated caspase-3/7 recognition motif (Essen Bioscience) was added to the culture medium of cells treated as indicated. Upon activation through cleavage by the activated caspase-3/7 during apoptosis, the dye was released and bound to DNA fluorescently, which was detected using the IncuCyte live cell imaging system. The apoptotic index was calculated by dividing the overall fluorescent object counts by cell numbers under each condition.

Whole-cell extracts were obtained for SDS-PAGE electrophoresis as previously described (25). Primary antibodies used were rabbit anti-SMARCA4 (Abcam; ab110641), H3K27Me3 (Millipore; 07-449), Myc (Abcam; ab32072), mouse anti-EZH2 (BD Bioscience; 612667), vinculin (Sigma; v9264), Actin (Sigma; A5441), Histone H3 (Abcam; ab1791), MAP2 (Millipore; MAB3418), TUBB3 (Santa Cruz; sc-51670), BAD (Santa Cruz; sc-8044), p16 (Ventana; 725-4713), and CDKN1A/p21 (BD Pharmacogen; 556430) and horseradish peroxidase-conjugated anti-mouse and anti-rabbit IgG (Amersham). Chemiluminescence was used for detection (Perkin–Elmer Life Sciences). Actin or vinculin were used as protein loading controls.

Cells were lysed in 100 mmol/L HEPES buffer (pH 8.5) containing 1% SDS and 1× protease inhibitor cocktail (Roche). After chromatin degradation by benzonase, reduction and alkylation of disulfide bonds by dithiothreitol and iodoacetamide, samples were cleaned up and prepared for trypsin digestion using the SP3-CTP method (26). In brief, proteins were digested for 14 hours at 37 °C followed by removal of SP3 beads. Tryptic peptides from each sample were individually labeled with TMT 10-plex labels, pooled, and fractionated into 12 fractions by high pH RP-HPLC, desalted, orthogonally separated, and analyzed using an Easy-nLC 1000 coupled to a Thermo Scientific Orbitrap Fusion mass spectrometer operating in MS3 mode. Raw MS data were processed and peptide sequences were elucidated using Sequest HT in Proteome Discoverer software (v2.1.0.62), searching against the UniProt Human Proteome database.

All animal work procedures were performed according to institutional guidelines approved by the Animal Care Committee of the University of British Columbia (A14-0290). In brief, BIN67 or SCCOHT-1 cells (1 × 10⁷ cells/mouse) were injected with a 1:1 mix of Matrigel (Corning) subcutaneously into the back of NRG (NOD.Rag1KO.IL2RγcKO) mice. Mice were randomized to treatment arms once the average tumor volume reached 100 mm³ (BIN67) or 250 mm³ (SCCOHT-1). Quisinostat was formulated in 20% hydroxypropyl-β-cyclodextrin (final pH 8.70) and administrated intraperitoneally at 5, 10, or 20 mg/kg once daily (M/W/F) for 3 weeks. For drug combination, mice were dosed with 5 mg/kg quisinostat (M/W/F) and 200 mg/kg EPZ-6438 once daily (M-Sat, formulated in 0.5% NaCMC (Sigma) containing 0.1% Tween-80) alone or in combination for 3 weeks, followed by drug removal. Tumor size and mouse weight were measured twice or thrice weekly. Tumor volume was calculated as length × (width)² × 0.52. For Quisinostat single drug efficacy studies, mice were terminated when reaching humane endpoint (i.e., tumor reaches 1,000 mm³) or at the end of 3-week treatment, whichever came first. For drug combination study, mice were terminated until reaching humane endpoint.

Student t test was used to evaluate the significant difference between two groups in all experiments except proteomics. The peptide expression change averaging (PECA) analysis (27) was performed for comparing the proteomic profiles of BIN67 cells treated with or without EPZ-6438 to generate signal log ratio values (log fold-change), P values and false discovery rate (fdr) adjusted P values (p.fdr) using peptide level signal values. The PECA analysis (27) was performed for comparing the proteomic profiles of BIN67 cells treated with or without EPZ-6438. Survival curves and IC₅₀ of drug treatment were determined by PRISM software or Microsoft Excel. The drug combination indexes were calculated using the Chou–Talalay method and CalcuSyn software (Biosoft). A P-value or p.fdr-value (for proteomics data) <0.05 was considered significant.

The simplicity of SCCOHT genome suggests that the epigenetic rewiring, in response to defective SWI/SNF chromatin remodeling, may drive tumorigenesis of SCCOHT. To search for potential therapeutic agents that specifically target SCCOHT, we screened an epigenetic drug library that includes 62 compounds targeting various epigenetic modifiers in three SCCOHT cell lines (BIN67, SCCOHT-1, and COV434), two ovarian high-grade serous carcinoma cell lines (OVCAR-4 and ES-2) and an immortalized human granulosa cell line (SVOG3e; Supplementary Table S1). We performed the screen at 1 μmol/L to identify clinically relevant inhibitors for further studies. Although some of the compounds are not effective at this dose, two EZH2 catalytic inhibitors (EPZ-6438 and GSK126) significantly suppressed the growth of SCCOHT cells with little effects on other cell lines as we previously reported (Fig. 1A and B; ref. 19), validating the selectivity of our screen. Notably, multiple HDAC inhibitors displayed preferential suppression of the SCCOHT cells, including SAHA, a clinically used pan-HDAC inhibitor approved for cutaneous T-cell lymphoma (Fig. 1B, P < 0.01).

Next, we validated the cellular response to HDAC inhibitors in three SCCOHT cell lines versus a subset of ovarian epithelial carcinoma cell lines. SAHA, panobinostat, and quisinostat, the three pan-HDAC inhibitors in either clinic use or various trials, effectively suppressed the growth of three SCCOHT cell lines with IC₅₀s significantly lower than other ovarian cancer cell lines examined (Fig. 1C; Supplementary Fig. S1, P < 0.01).

We focused on quisinostat, which has displayed promising results for refractory cutaneous T-cell lymphoma and solid malignancies in recent clinical trials (28, 29) and induced histone H3 acetylation rapidly in all three SCCOHT cell lines (Fig. 1D). Treatment of quisinostat at sublethal doses suppressed SCCOHT cell growth in a time-dependent manner (Supplementary Fig. S2A) and inhibited their colony formation abilities significantly (Fig. 1E). Moreover, treatment of quisinostat completely blocked the formation of BIN67 spheroids at a sublethal dose (Fig. 1F) and dramatically suppressed the growth of preformed spheroids of three SCCOHT cell lines at higher concentration (Fig. 1G). As quisinostat has minimal effect on HDAC6 activity (24), the hypersensitivity of SCCOHT cells to quisinostat can be attributed to inhibition of class I or other class II HDACs. Accordingly, depletion of HDAC1 or HDAC2 completely killed SCCOHT-1 cells and significantly reduced the proliferation of BIN67 cells (Supplementary Fig. S2B–S2D). Simultaneous knockdown of HDAC1 and HDAC2 exhibited significantly stronger inhibitory effect on the growth of BIN67 cells (Supplementary Fig. S2E). These data suggested that inhibition of HDACs, such as class I HDACs, may have therapeutic potential for patients with SCCOHT.

To address whether SMARCA4 and SMARCA2 dual deficiency can predict the sensitivity to HDAC inhibitors in other cell contexts, we assessed the sensitivity of a panel of lung cancer cell lines with various states of SMARCA4 and SMARCA2 expression to HDAC inhibitors. All these cell lines, regardless of SMARCA4 and SMARCA2 status, responded similarly to either SAHA or quisinostat treatment with IC₅₀s much higher than those of SCCOHT cell lines (Supplementary Fig. S3A and S3B). In addition, knockdown of SMARCA4 by shRNAs in MiaPaCa-2 cells, a pancreatic cancer cell line with no detectable SMARCA2 (30), did not increase the cellular sensitivity to HDAC inhibitors (Supplementary Fig. S3C–S3E). Therefore, the greater sensitivity of SCCOHT cells to HDAC inhibition appears to be cell context dependent.

Concordant with growth suppression, BrdUrd incorporation was dramatically reduced by quisinostat treatment in a dose-dependent manner (Fig. 2A), suggesting that quisinostat inhibited the proliferation of SCCOHT cells. Next, we analyzed the effect of Quisinostat on cell-cycle distribution and cell death. Low doses of quisinostat (5 nmol/L) induced cell-cycle arrest at G₁ phase, whereas higher doses of quisinostat (20, 50 nmol/L) significantly triggered cell death (sub-G₁) in all three SCCOHT cell lines 72 hours posttreatment (Fig. 2B). Using a fluorescence-based live cell imaging of activated caspase-3/-7 activity, we further determined that quisinostat elicited apoptosis rapidly in dose- and time-dependent manners in SCCOHT cells (Fig. 2C).

The SCCOHT cells that survived the treatment of 50 nmol/L quisinostat displayed a bipolar and elongated morphology (Fig. 2D). Prolonged treatment of quisinostat at a lower dose (10 nmol/L) also induced morphology change, which largely resembled neurons after 3 weeks of treatment (Supplementary Fig. S4A). Similar morphology changes were observed in SCCOHT cells upon exposure to other HDAC inhibitors, such as SAHA and panobinostat (Supplementary Fig. S4B), suggesting that HDAC inhibition triggers the differentiation of SCCOHT cells.

To elucidate the molecular mechanisms underlying the cellular response of SCCOHT cells to quisinostat, we performed proteomic analysis of BIN67 cells treated with vehicle or 10 nmol/L quisinostat for 24 hours or 72 hours using mass spectrometry. Proteomics analysis revealed that 5.9% (402/6819) and 16.4% (1115/6819) of the BIN67 proteome was significantly altered by at least two folds upon quisinostat treatment for 24 or 72 hours, respectively (Fig. 3A; Supplementary Table S2–S4, p.fdr < 0.05; Fig. 3B). In agreement with the potent growth suppression by quisinostat treatment, ingenuity pathway analysis (IPA) revealed a significant enrichment of genes involved in cell-cycle control and cell death (Supplementary Table S5 and Fig. 3C). Quisinostat treatment significantly increased the levels of CDKN1A, a cyclin-dependent kinase inhibitor required for cell-cycle arrest by HDAC inhibition (Log₂FC = 2.0 and 2.0; ref. 31), and the proapoptotic protein Bak1(Log₂FC = 0.5 and 1.58) in a time-dependent manner. The expression of c-Myc, a key cell-cycle player under the direct control by SWI/SNF chromatin remodeling complex (32), was suppressed in a time-dependent manner (Log₂FC = −1.5 and −1.7). Immunoblotting confirmed these findings and revealed a significant induction of cleaved PARP (Fig. 3E and F), supporting the induction of growth arrest and apoptosis upon quisinostat treatment. Furthermore, in line with the morphology change after quisinostat treatment (Fig. 2D), IPA predicted that the biological processes called “development of neurons” and “morphogenesis of neurons” were activated 72 hours following quisinostat treatment (Fig. 3D). Supporting this, quisinostat treatment induced the expression of MAP2 (Log₂FC = 0.54 and 1.39) and TUBB3 (Log₂FC = 1.15 and 1.85), two markers of neuronal cells, and reduced EZH2 levels in a time-dependent manner (Fig. 3F). Moreover, several biological processes known to play important roles in neuron development, such as organization of cytoskeleton, microtubule dynamics and formation of cellular protrusions, cell migration, and endocytosis, all appeared as activated in the presence of quisinostat in a time-dependent manner (Fig. 3D; Supplementary Table S5). These observations suggest that quisinostat elicited neuronal-like differentiation in SCCOHT cells.

Consistent with our previous observation (10), the levels of SMARCA2 protein and mRNA was induced by quisinostat in a time-dependent manner (Supplementary Table S5, log₂FC = 2.6 and 3.2; Fig. 3F; Supplementary Fig. S5A). To address whether SMARCA2 induction was important for the cellular sensitivity to quisinostat, we used CRISPR/cas9 technology to deplete SMARCA2, which prevented the induction of SMARCA2 upon quisinostat treatment in COV434 cells (Supplementary Fig. S5B). This minimally made COV434 cells more resistant to quisinostat (Supplementary Fig. S5C), suggesting only a small contribution of SMARCA2 induction in the cellular sensitivity of SCCOHT cells to quisinostat.

To gain more insights into the cellular hypersensitivity of SCCOHT cells to HDAC inhibitors, we determined whether HDAC inhibition activated genes that were deregulated by SMARCA4 inactivation. First, we profiled the proteomes of BIN67 cells with ectopic expression of either GFP or SMARCA4. About 5.6% (370/6604) and 1.3% (86/6604) of the identified proteins were significantly upregulated or downregulated, respectively, by at least two folds upon SMARCA4 reexpression in BIN67 cells (Fig. 4A; Supplementary Table S6, p.fdr < 0.05). Pathway analysis of the significantly altered proteins by IPA revealed an enrichment of proteins involved in several molecular and cellular functions, such as cellular movement, cellular assembly and organization, and cell death with a few biological processes being predicted to be activated or inhibited, such as the activation of neuron-like development (Fig. 4B; Supplementary Table S7), which was supported by the altered cell morphology (Fig. 4C). Next, we compared the proteins altered by either quisinostat treatment (72 hours) or SMARCA4 reexpression (96 hours after infection) (|Log₂FC|≥1, p.fdr<0.05). A subset of proteins (194 upregulated, 62 downregulated) was altered by both SMARCA4 reexpression and quisinostat treatment (Fig. 4D; Supplementary Table S8), many of which contributed to induction of cell death, inhibition of cell progression, or the development of a neuronal phenotype and related biological processes as predicted by IPA (Fig. 4E; Supplementary Table S9). These results suggest that loss of SMARCA4 in the unknown precursor cell of SCCOHT may render them dependent on the activity of HDACs for resistance to cell death signals and maintenance of their poorly differentiated state and oncogenic properties.

To further investigate the efficacy of quisinostat in vivo, we tested the potency of quisinostat in xenograft models of SCCOHT. In the BIN67 xenograft model, treatment of quisinostat at either 10 or 20 mg/kg once daily for 3 weeks (M/W/F) was generally well tolerated without body weight loss (Supplementary Fig. S6A). Tumor growth was rapidly and effectively suppressed by either 10 or 20 mg/kg quisinostat (Fig. 5A). The final average tumor weight was also dramatically reduced in quisinostat-treated groups in comparison to the vehicle-treated group after 3 weeks (Fig. 5B, P < 0.001 and 0.01). Western blotting analysis confirmed that the acetylation of histone H3, but not tubulin, was induced by the treatment of either dose of quisinostat and the expression of c-Myc was also significantly reduced in a dose-dependent manner in quisinostat-treated cells (Fig. 5C). However, xenografted tumors expressed substantial amount of CDKN1A that did not differ between vehicle and quisinostat-treated groups, indicating that a significant portion of cells in xenografted tumors may undergo growth arrest after substantial growth and/or other cell populations residing in the tumor may undergo arrest with upregulated CDKN1A expression. As SCCOHT-1 cells were significantly more sensitive to qusinostat than BIN67 cells in vitro (Supplementary Fig. S1), we further tested the effect of 5 mg/kg quisinostat (M/W/F) on the growth of SCCOHT-1 xenograft tumors, which was significantly reduced by the treatment (Fig. 5D). However, all quisinostat-treated SCCOHT-1 tumors progressed during the 3-week treatment period with the median survival time to humane endpoint being increased from 6.5 to 13 days upon treatment (Fig. 5E, P = 0.001), implying that 5 mg/kg of quisinostat, which is comparable to the recommended dose for human (28, 29), can significantly, but not completely, suppress SCCOHT tumor growth.

We and others have previously demonstrated that SCCOHT cells are hypersensitive to the pharmacologic inhibition of PRC2 complex with EZH2 inhibitors (18, 19). It has also been reported that HDACs are recruited by the PRC2 complex to the regulatory regions of PRC2 targets, where they act in concert to silence the expression of their targets (33). Therefore, we determined whether simultaneous inhibition of both HDAC and EZH2 might synergize to inhibit the cell growth. Combined treatment of quisinostat and EPZ-6438 synergistically suppressed cell growth in a wide range of dose combinations in SCCOHT cells (Fig. 6A; Supplementary Fig. S7A) and other SWI/SNF-deficient cells, such as TOV112D (a SCCOHT-like cell line; ref. 18), G401 (a malignant rhabdoid tumor cell line) and SYO-1 cells (a synovial sarcoma cell line; Supplementary Fig. S7B-D) and triggered apoptosis rapidly in BIN67 and COV434 cells than treatment with either drug alone (Fig. 6B). Furthermore, depletion of HDAC2 significantly increased the sensitivity of BIN67 cells to EPZ-6438 treatment (Supplementary Fig. S8), whereas combination of quisinostat and GSK126 or combination of EPZ-6438 and panobinostat or romidepsin (a class I HDAC inhibitor) also synergistically suppressed the growth of SCCOHT cells (Supplementary Fig. S9), suggesting that the synergism was not an off-target effect of quisinostat and EPZ-6438 combination.

Next, we determined the levels of global histone H3K27Me3 and H3K27Ac in SCCOHT cells treated with EPZ-6438 and quisinostat alone or in combination by Western blotting. As shown in Fig. 6C, the level of histone H3K27Me3 was effectively reduced by EPZ-6438, but not quisinostat. Treatment with either quisinostat or EPZ-6438 increased H3K27Ac, whereas combined treatment of quisinostat and EPZ-6438 elicited higher levels of H3K27Ac than either drug alone, suggesting a more profound effect on the activation of their targets. Consequently, a robust induction of CDKN1A (p21) was observed when SCCOHT cells were exposed to both drugs simultaneously (Fig. 6C).

Finally, we determined the efficacy of combined treatment of EPZ-6438 and quisinostat in the BIN67-xenograft model. In agreement with our previous finding (19), daily single-dose treatment of EPZ-6438 at 200 mg/kg failed to suppress BIN67 xenografted tumor growth, whereas 5 mg/kg quisinostat (once daily, M/W/F) only mildly suppressed tumor growth (P < 0.05). Combination of EPZ-6438 and quisinostat completely blocked the growth of tumor (Fig. 6D, P < 0.001) with only a mildly increased mouse body weight loss compared to single treatment of either drugs (Supplementary Fig. S10A). Upon removal of the drug following 3 weeks of treatment, tumors that received combinational treatment started to grow after a short delay with the tumor volume doubling time not significantly different with the tumors in the vehicle-treated group, suggesting that a longer treatment period may be required to fully defeat the tumor growth. The median survival time of the tumor-bearing mice was greatly improved from 44.5 days in vehicle group to 70 days in the group received combinational treatment (Supplementary Fig. S10B, P < 0.001).

SCCOHT is a rare but extremely lethal malignancy of the ovary. There is an urgent demand for effective therapeutic agents to treat patients with SCCOHT. Through a rationale epigenetic drug screen, we identified that pan-HDAC inhibitors, a class of anticancer drugs that regulate gene expression through interrupting the deacetylation of histones and nonhistone proteins, have preferential growth suppressive effects in SCCOHT cells. As our screen is performed at a single but clinically relevant dose (1 μmol/L), we cannot rule out epigenetic modifiers other than HDAC and EZH2 may be putative targets in SCCOHT cells because some of the tool compounds in the library is not effective at 1 μmol/L (Supplementary Table S1). Although a few HDAC inhibitors have been approved for clinical management of several types of hematopoietic tumors, their efficacy in solid malignancies remains unresolved. Our present study demonstrates that pan-HDAC inhibitors displayed a significantly higher preclinical potency in SCCOHT cells than in other ovarian cancer cell lines. Importantly, the IC₅₀ of quisinostat (∼2 nmol/L) in SCCOHT cell is on the lower end of the reported C_max (0.87–3.44 ng/mL, equivalent to 2.2–8.7 nmol/L) in patients who received 12 mg quisinostat, the recommended dosage for phase II clinical trials (29). Furthermore, combined treatment of either pan-HDAC inhibitors or a class I HDAC inhibitor with EZH2 catalytic inhibitors, which were previously reported by us and others to kill SCCOHT cells selectively (18, 19), displayed a robust synergistic effect in growth suppression of SCCOHT cells and xenograft tumors. Therefore, we have provided preclinical evidence that SCCOHT patients may benefit from the treatment of pan-HDAC inhibitors alone or in combination with EZH2 inhibitors.

HDAC inhibitors have been reported to regulate various cellular functions, such as cell proliferation, cell death, DNA repair, and differentiation (34). Accordingly, quisinostat, a hydroxamic acid derivative and second-generation pan-HDAC inhibitor (24), caused cell-cycle arrest, activated differentiation, and induced apoptosis in SCCOHT cells. Proteomic analysis further validated that quisinostat regulated the expression of proteins involved in cell-cycle arrest, death, and neuronal differentiation. Importantly, a subset of these proteins were also regulated by reexpression of SMARCA4 in SCCOHT cells, suggesting that HDAC inhibitors can revert the expression of genes dysregulated by SWI/SNF remodeling complex deficiency in SCCOHT. Interestingly, HDAC inhibitors have been reported to be effective in malignant rhabdoid tumor models and activate the genes that are deregulated due to SMARCB1 loss (35). Therefore, the SWI/SNF complex may be functionally antagonistic to HDACs, in addition to the PRC2 complex (19, 36). Intriguingly, PRC2 complex can also recruit HDACs, via EED, to repress targets transcriptionally (33). Therefore, cancer cells with a deficient SWI/SNF complex will likely rely on the activity of PRC2 and HDAC complex that act in concert to suppress the genes required for differentiation and tumor suppression. This notion is further strengthened by our observation that simultaneous inhibition of HDAC and EZH2 displayed a stronger effect on the acetylation of histone H3K27 globally and the expression of target genes than treatment with either agent alone and elicited a robust synergistic effect in the suppression of SCCOHT cells. Future analysis of the histone acetylation status in the promoter or enhancer regions of those genes upon HDAC inhibitor treatment or restoration of the functional SWI/SNF complex will improve our understanding of the antagonism between SWI/SNF complex and HDAC.

Although SCCOHT cells were highly sensitive to HDAC inhibitors, we did not detect significant difference in the cellular responses to HDAC inhibitors among lung cancer cell lines with loss of SMARCA4 and/or SMARCA2 expression. Depletion of SMARCA4 in a SMARCA2-deficient pancreatic cancer cell line also had little effect on cellular sensitivity to HDAC inhibitors. Thus, the dual deficiency of SMARCA4 and SMARCA2 cannot predict the cellular response to HDAC inhibitors, similar as what was observed for EZH2 inhibitors (19). These results suggest that the high burden of genetic mutations in common malignancies may abrogate their selective sensitivity to epigenetic targeting even in the absence of a functional SWI/SNF complex. Future studies to identify targets that can sensitize cancer cells to EZH2 or HDAC inhibitors may help to improve the efficacy of epigenetic therapy in common malignancies.

The synergistic interaction between EZH2 depletion and HDAC inhibitors has been reported in several cancer types, such as human acute myeloid leukemia, gallbladder carcinoma, and non–small cell lung cancer cells (37–39). However, depletion of EZH2 will affect the noncatalytic function of EZH2 warranting further evaluation of its clinical safety. Our present study for the first time demonstrates that combined suppression of EZH2 catalytic activity and HDACs in a wide range of dose combinations has a synergistic effect in SCCOHT preclinical models without significant adverse events. This synergism is likely due to upregulated transcription as a result of hyperacetylation of H3K27 at the promoter or enhancer of PRC2 targets by simultaneous inhibition of EZH2 and HDAC, and/or hyperacetylation of histone lysine residues by HDAC in addition to increased acetylation of H3K27. Moreover, our finding of the synergism of combined treatment of HDAC inhibitors and EZH2 catalytic inhibitors also holds the promise beyond SCCOHT. TOV112D, a SCCOHT-like cell line with SMARCA4/SMARCA2 dual deficiency and great sensitivity to EZH2 inhibition (18), displayed a strong response to HDAC inhibitors alone or in combination with EZH2 inhibitors. A robust synergistic effect between HDAC and EZH2 inhibitor was also detected in G401 cells, a malignant rhabdoid tumor cell line with high sensitivity to HDAC inhibition (35) or EZH2 inhibition (40), and SYO-1 cells, a synovial carcinoma cell line with SWI/SNF complex deficiency caused by SS18–SSX fusion and great response to quisinostat (41). Therefore, the combinational therapy of EZH2 and HDAC inhibitors may provide a putative therapeutic approach for treating patients with poorly differentiated tumors driven primarily by the loss of activity of the SWI/SNF chromatin remodeling complex, such as SCCOHT, dedifferentiated endometrial carcinoma, rhabdoid tumors, synovial sarcoma, epithelioid sarcoma, and undifferentiated thoracic sarcomas (42–45).

No potential conflicts of interest were disclosed.

Conception and design: Y. Wang, J.D. Lang, W.P.D. Hendricks, T.M. Underhill, G.B. Morin, B.E. Weissman, D.G. Huntsman

Development of methodology: Y. Wang, M.B. Bally, G.B. Morin, D.G. Huntsman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Wang, S.Y. Chen, G. Lambert, C.Y. Shin, N.D. Santos, G.B. Morin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Wang, S.Y. Chen, S. Colborne, K.A. Orlando, M.B. Bally, A.N. Karnezis, R. Hass, G.B. Morin, B.E. Weissman

Writing, review, and/or revision of the manuscript: Y. Wang, S. Colborne, K.A. Orlando, J.D. Lang, W.P.D. Hendricks, A.N. Karnezis, T.M. Underhill, G.B. Morin, J.M. Trent, D.G. Huntsman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.Y. Chen, R. Hass, J.M. Trent

Study supervision: Y. Wang, M.B. Bally, J.M. Trent

Other (conception/design of the in vivo animal experiments): N.D. Santos

We acknowledge the technical support of Sarah Maines-Bandiera, Winnie Yang, Ashley Wang, Nicole Wretham, Dana Masin, Hong Yan, Jenna Rawji, and Chris Ke-dong Wang. We thank Drs. Barbara Vanderhyden, Mikko Anttonen, and Torsten Nielsen for providing BIN67, COV434, and SYO-1 cells, respectively. This work was supported by research funds from the Canadian Cancer Society Research Institute (#703458, to D.G. Huntsman, Y. Wang, M. Bally, and A.N. Karnezis), the National Cancer Institute of NIH (United States; 1R01CA195670-01, to D.G. Huntsman, J.M. Trent, and B.E. Weissman), the Terry Fox Research Institute Initiative New Frontiers Program in Cancer (TFF1021, to D.G. Huntsman), the British Columbia Cancer Foundation (to D.G. Huntsman), and the VGH & UBC Hospital Foundation (to D.G. Huntsman).

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.