Chemoresistance is driven by unique regulatory networks in the genome that are distinct from those necessary for cancer development. Here, we investigate the contribution of enhancer elements to cisplatin resistance in ovarian cancers. Epigenome profiling of multiple cellular models of chemoresistance identified unique sets of distal enhancers, super-enhancers (SE), and their gene targets that coordinate and maintain the transcriptional program of the platinum-resistant state in ovarian cancer. Pharmacologic inhibition of distal enhancers through small-molecule epigenetic inhibitors suppressed the expression of their target genes and restored cisplatin sensitivity in vitro and in vivo. In addition to known drivers of chemoresistance, our findings identified SOX9 as a critical SE-regulated transcription factor that plays a critical role in acquiring and maintaining the chemoresistant state in ovarian cancer. The approach and findings presented here suggest that integrative analysis of epigenome and transcriptional programs could identify targetable key drivers of chemoresistance in cancers.

Significance:

Integrative genome-wide epigenomic and transcriptomic analyses of platinum-sensitive and -resistant ovarian lines identify key distal regulatory regions and associated master regulator transcription factors that can be targeted by small-molecule epigenetic inhibitors.

The American Cancer Society estimates 22,240 new cases of ovarian cancer in 2018 (1). Unfortunately, the five-year survival rate of ovarian cancer remains less than 50%. Thus, nearly 14,000 women in the United States and 160,000 worldwide die of ovarian cancer each year (2). Epithelial ovarian cancers, which account for nearly 90% of all ovarian cancer diagnoses, are associated with worse prognosis (3). They originate mainly from the epithelial cells of fallopian tubes (4, 5) and areas of endometriosis (6), among others. Critically, 75% of the patients with epithelial ovarian cancer are high-grade serous ovarian cancer (HGSOC; ref. 7) that are more challenging to effectively treat. The first-line therapy for ovarian cancer involves the combination of cytoreductive surgery followed by platinum and taxane-based chemotherapy. Platinum-based compounds such as cisplatin induce increased DNA damage through interstrand cross-links and cell death in proliferative cancerous cells (7, 8). Despite the high rate of initial response to therapy, the duration of response declines over time and a vast majority of patients succumb to chemotherapy-resistant ovarian cancer (9–12).

Recent genomic approaches have shed significant light on the genetic risk factors of ovarian cancer. Low-grade ovarian tumors often harbor BRAF, KRAS, BRCA1/2, and PTEN mutations, whereas high-grade tumors are uniformly characterized by TP53 mutations (13, 14). Apart from the antiangiogenic agent bevacizumab, and partially effective PARP inhibitors for patients with BRCA1/2 mutations (15), targeted therapies are lacking for ovarian cancer. Although specific genetic alterations such as reversion of germline BRCA1/2 mutations and inactivating mutations in tumor suppressor RB1, NF1, RAD51B, and PTEN genes were noted in some chemoresistant patients (16), the molecular network that drives and maintains the chemoresistant state in ovarian cancer is largely unknown. In addition to genetic alterations, epigenetic regulation of proximal promoters and distal enhancers are critical determinants of cellular identities. Alterations in the chromatin landscape are increasingly recognized as hallmarks of malignant cellular states (17–19). Because of the technical limitations, previous ovarian cancer epigenetic studies primarily focused on targeted DNA methylation at individual gene promoters. Although these studies implicate differential methylation at multiple genes, such as MLH1 (20), SFRP1 (21), BRCA1 (16), MAL (22), FANCF (23) with chemoresistance, there have been limited attempts to comprehensively map differentially regulated gene promoters and distal enhancers in ovarian cancer.

In this study, we aimed to identify molecular drivers of chemoresistance in ovarian cancer through unbiased epigenomic and transcriptional profiling across multiple cellular models of ovarian cancer. We aimed to map differentially regulated proximal promoters and distal enhancers in multiple cellular models of ovarian cancer. By integrating genome-wide maps of a well-characterized epigenetic mark of active regulatory genomic elements with gene expression profiles, we aimed to identify differentially regulated proximal promoters and distal regulatory elements that are specifically associated with chemoresistance across multiple ovarian cancer cell lines. To this end, we generated multiple isogenic cellular models of cisplatin resistance and performed chromatin immunoprecipitation sequencing (ChIP-seq) analysis of the histone H3, lysine 27 acetylation (H3K27ac) epigenetic mark, which is deposited to active enhancers and promoters. By integrating ChIP-seq maps with RNA-seq gene expression profiles across naïve and chemoresistant cellular counterparts, we found that the chemoresistant state is associated with largely cell-type–specific sets of distal enhancer elements. Critically, we found significant upregulation of distal enhancer clusters known as super-enhancers (SE) in resistant cells. Small-molecule epigenetic drugs that target enhancers and SEs result in significant decrease in the expression of their target genes and an increase in cisplatin sensitivity in chemoresistant HGSOC cells. Our findings identified, in addition to known drivers of chemoresistance, SOX9 as a critical SE-regulated transcription factor (TF) that plays a critical role in chemoresistance across multiple ovarian cancer cell lines.

Cell culture

Human ovarian cancer OVCAR4, CAOV3, OV81, and COV362 cell lines were cultured in complete medium consisting of RPMI1640, 20% heat-inactivated FBS, 1% penicillin/streptomycin. SKOV3 cells were cultured in complete medium consisting of McCoy 5A, 10% heat-inactivated FBS (Sigma Aldrich), 1% penicillin/streptomycin [100 U/mL penicillin, 100 μg/mL streptomycin (PAA Laboratories GmbH)]. Cells were cultured incubator at 37°C in a humidified atmosphere consisting of 5% CO2 and 95% air. The cells were originally obtained from ATCC and monitored periodically for Mycoplasma contamination. The cells were validated using FTA Sample Collection Kits for Human Cell Authentication Service (ATCC).

Creating cisplatin-resistant and resensitized cell lines

Cells were grown in their respective culture media and passaged for at least two generations after thawing to ensure proper viability. When the cells reached 80% confluency, they were split into two 6-cm plates with 40% confluency. Cells were treated with an initial dose of 1 μmol/L cisplatin in 3 mL complete media. After 4 hours, the media for both control and treated cells were aspirated and replaced washed with an equal volume of PBS twice before replacing with drug-free complete media. Cells were allowed to recover for two passages, and treated with the same or increasing dose of cisplatin depending on the viability levels. Once the cells gain resistance, either the dose was increased or cells were periodically treated with cisplatin to maintain the chemoresistant state. Resistant SKOV3 cells reached a maximum concentration of 20 μmol/L cisplatin tolerance, OV81 cells reached a maximum concentration of 40 μmol/L, OVCAR3 cells reached a maximum periodic dose of 18 μmol/L cisplatin, and OVCAR4 cells' largest sustainable periodic dose of cisplatin was 12 μmol/L.

MTT and crystal violet cell proliferation and viability assay

Each cell line was seeded at a density of 4–6 × 102 cells/well in flat-bottomed 96-well culture plate in 100 μL of the culture medium. Stock solutions of cisplatin were subjected to serial dilutions to give final concentrations ranging from 0.1 μmol/L to 250 μmol/L. JQ1 stock suspended in DMSO was first diluted to 10× of the final concentration in DMSO, then further diluted in PBS or complete media to concentrations of 0.25 to 2 μmol/L. The dilutions were added to equal volumes of cell culture in triplicate wells and then the cells were left to incubate. After 24 hours, the media were aspirated and then washed once with 1 volume of PBS, then replaced with 1 volume of the cell's respective complete media and left to incubate. After 48 hours, 10% well volume of stock MTT diluted to 5 mg/mL was added to 100–150 μL of fresh complete media, then added to each well after aspirating the previous media. After 3–4 hours of incubation, 1 equal volume of MTT solubilization media (10% SDS, 0.1% Tris HCl) was added to each well, then covered, and stored at room temperature or in the incubator for 7+ hours. The plate was read on a plate reader that shakes the plate, then reads absorbance at 590 nmol/L. Background absorbance was taken to be the readings of control wells with no cells. These treatments were carried out to determine IC50 values, that is, drug concentrations required for 50% cell kill, as well as synergism between drugs.

For crystal violet assays, fresh media were added every 3–5 days after initial treatment. After 10–14 days, the wells were stained for 30 minutes with crystal violet solution (0.4% crystal violet, 10% formaldehyde, 80% methanol). After staining, the crystal violet solution was removed, and then the stained cells washed once with PBS and 3+ times with water. The plate was inverted overnight and covered to dry the well for imaging with a custom 3D printed insert on an Epson tabletop scanner.

Chromatin immunoprecipitation experiments

SKOV3, OVCAR4, CAOV3, or COV362-naïve, -resistant, or -resensitized cells were grown to 80% confluency in 15-cm plates. A total of 2 × 107 cells were cross-linked in 1% formaldehyde in complete media, or trypsinized, spun down, then resuspended in complete media containing 1% formaldehyde. After 15 minutes, the samples were quenched with glycine to a final concentration of 0.125 mol/L glycine. Cells were then scraped off with a cell scraper and then collected with the mixed quenched media to 50 mL Falcon tubes, where they were spun down and then resuspended in SDS lysis buffer (1% SDS, 10 mmol/L EDTA, 50 mmol/L Tris-HCl, pH 8.1) at a ratio of 1 mL per 2 × 107 cells. Pulse sonications were performed for 9 minutes at 40% amplitude with 30% on/70% off on a Brandon Digital Sonifier (model 250) with a total of 1 mL with maximum 50% SDS Lysis Buffer solution diluted with ChIP Dilution Buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mmol/L EDTA, 16.7 mmol/L Tris-HCl, pH 8.1, 167 mmol/L NaCl). ChIPs were performed with antibodies for K27ac (Abcam, 4729, Lot GR286678-1) and mouse anti-SOX9 (AB5535, Millipore, Lot 2847051). Pulldowns were performed with 50%/50% mixed Dynabeads Protein A (1002D, Lot 00326545) & G (1004D, Lot 00342019). DNA quantities were measured by Qubit 2.0 (Promega QuantiFluor dsDNA System) and Bioanalyzer.

DNA and RNA isolation

RNA was isolated using Qiagen TRIzol (#15596108). DNA was isolated using phenol chloroform extraction. Quantity was measured using the Nanodrop 2000 Spectrophotometer at 260 nm.

PCRs and qPCRs

PCR experiments were performed on an Eppendorf Nexus Gradient equipment. Real-time quantitative PCR was performed on a StepOnePlus Applied Biosystems instrument with SYBR Green or TaqMan polymerase.

ChIP-seq and RNA-seq library preparation

ChIP-seq libraries were prepared using the Illumina TruSeq ChIP Library Preparation Kit. RNA-seq libraries were prepared using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina. Libraries were prepared according to the manufacturer's instruction. Qubit and bioanalyzer measurements were used to determine the library quality.

Apoptosis assay

Cells were seeded into three 6-well plates per sample at 30% confluency and then allowed to settle overnight. Each well was treated with DMSO control, JQ1 (1 μmol/L), cisplatin (3 μmol/L or 6 μmol/L), or both premixed in RPMI complete media. Twenty-four hours later, the wells were washed once with PBS and refreshed with new complete media. Forty-eight hours later, the cells were checked under a microscope, then trypsinized, collected, and washed twice with PBS. All steps further are performed on ice. Unstained and single stained controls are aliquoted and spun down. Controls are stained with either Annexin V binding buffer (eBioscience 00-0055-43) or binding buffer with DAPI or Annexin-FITC, and the samples are stained with both. Data was collected inside the UVA Flow Core facility on a BD Biosciences FACScalibur instrument with 30,000 collected events per sample.

RNA-seq data analysis

Paired-end reads were acquired using HiSeq 2500 (50 bp) or NextSeq 500 (75 bp) system on high-throughput mode from UVA sequencing core. Reads were aligned to the hg19 genome using bowtie2 or hisat2. Read abundance was estimated using either tophat2 or stringtie, depending on if bowtie2 or hisat2 was the aligner, respectively. The counts were then normalized and compared for differential expression as per the DESeq2 R package. Custom R Scripts were used to perform further normalization and quality control. Genes with significant variance between each replicate and with 0 read counts in any of the replicates were removed. Downstream plots used the pheatmap, heatmap.3, and ggplot2 packages. Clustering was performed using kmeans or hclust packages. Downstream pathway enrichment analysis was performed through preranked GSEA and DAVID gene ontology.

ChIP-seq data analysis

Single-end reads were acquired using MiSeq, HiSeq 2500 (50 bp), or NextSeq 500 (75 bp) from UVA sequencing core or Hudson Alpha. Reads were aligned using bowtie or bowtie2, duplicates removed by samtools. Peaks were called using MACS2 (24) or MACS1.4. SEs were defined using K27ac intensity versus rank as published in Whyte and colleagues, 2013. State-specific enhancers for the SKOV3 system were defined as having an intensity of 5-fold change higher over the other chemoresistant state. Normalization and differential peak analysis were performed using a custom R script utilizing edgeR or DESeq2 as referenced in DiffBind R package. The normalized read counts (affinity scores) were used to generate plots through the DiffBind, pheatmap, and ggplot2 packages. Clustering was performed using kmeans or hclust. Gene association was performed through bedtools + DAVID or GREAT analysis defined as proximally associated if within 12.5-kb inclusive window, or the most proximal upstream and downstream genes. SE-associated genes are defined as proximally associated genes whose gene expression goes up or down in the same direction as the SE, and whose expression in the resensitized cells falls in between the naïve and resistant expression values. SOX9-binding site annotation was performed through HOMER.

Western blots

Cells lysates were quantitated with a standard Bradford assay using Bio-Rad Protein Assay Dye Concentrate (catalog no. 500-0006) and BSA as a control. After running the gel, a dry transfer system was used (Life Technologies, IB1001; 7 minutes at 20 volts). Membranes were blocked in 5% nonfat dried milk (Bio-Rad #170-6404). The SOX9 antibody was diluted to 1:500 (AB5535, Millipore, Lot 2847051) α-tubulin (1:2,000; Invitrogen #62204).

CRISPR/Cas9-mediated genetic knockout experiments

Cas9 and sgRNA-expressing plasmid (AddGene 1000000048) was used in all CRISPR experiments. The BsmBI sites were introduced from another plasmid. Twenty nucleotide sgRNAs were designed using the UCSC genome browser and/or GECKO library and checked for off-targets using CROP-IT. Overhangs of 5′-CACC-3′ and 5′-AAAC-3′ were added to the 5′ of the forward and reverse complementary oligos, respectively. Forward and reverse sgRNA oligos were mixed in annealing buffer (10 mmol/L Tris pH 7.5–8, 50 mmol/L NaCl, 1 mmol/L EDTA), then heated to 95°C, annealed in a stepwise thermal decrease and finally ligated with a BsmBI cut p413 plasmid before transformation. Positive sequences confirmed through colony PCR with the U6 promoter primer + reverse sgRNA oligo primer through Sanger sequencing. Vector controls include a nontargeting 20 nucleotide control guide sequences.

HEK293T cells were transfected with PsPAX2 and Pmd2G with FuGene6 (Promega E2691) in a 4:1:5 ratio inside OptiMem media (Gibco #31985070). Media were exchanged to fresh complete DMEM (10% FBS, 1% penicillin/streptomycin) after 4 hours, and then the media collected and replaced after 24 and 48 hours. The collected media were then syringe filtered through 0.22-μm filters, and then stored at 4°C for immediate use or −80°C for long-term storage up to 6 months.

Ovarian cancer cells were seeded at 40%–60% confluency and allowed to attach for 8+ hours. Cells were then infected with the lentiviral mix with 10 μg/mL polybrene or a control media and after 16 hours overnight, and subject to puromycin selection for 72 hours or until all the control cells died. For all cloning purposes, competent DH5α cells were used.

Mouse xenograft experiments and in vivo drug treatment

The mouse studies were conducted in accordance with the guidelines established by the University of Virginia Institutional Animal Care and Use Committee (IACUC). A total of 5 × 106 cells diluted in 200-μL PBS were subcutaneously injected into each hind quarter of 32 mice (Jackson Laboratories). Tumors were allowed to form (25 mm3 or larger), then separated into 4 groups averaging 80 mm3 per tumor per group with 8–10 tumors distributed between 5 and 6 mice. Mice were treated with DMSO control or with JQ1 at 250 μL per mouse (50 mg/kg; Selleckchem S7110 lot 07), cisplatin at 100 μL per mouse (3 mg/kg; TEVA, NDC 0703-5748-11) or with both JQ1 and cisplatin. Except for the first week's Wednesday treatment, JQ1 treatment was applied on Tuesdays. Cisplatin was applied on Fridays of each week. Tumor volumes were measured twice a week using calipers along two axes with the formula: tumor volume = longest diameter × (shortest diameter)2. After the conclusion of the experiment, mice were euthanized according to IACUC guidelines and policies.

Generating cisplatin resistance in ovarian cancer cellular models

To gain molecular insights into cisplatin-induced epigenetic and transcriptional changes, we first generated multiple cellular models of cisplatin resistance. We derived cisplatin-resistant ovarian cancer cells from their sensitive counterparts by periodically exposing OVCAR4, OVCAR3, OV81, and SKOV3 ovarian cancer cell lines to increasing doses of cisplatin treatment. Here, we use the term naïve to indicate relative sensitivity regardless of whether the cell line was originated from a naïve or treated primary patient tumor. The treatments were performed as a 24-hour pulse of cisplatin exposure followed by two cisplatin-free passages (see Materials and Methods; Fig. 1A). Notably, the cell lines used in this study are widely used in ovarian cancer research as model systems for HGSOC. However, recent molecular profiling efforts suggested that the origin of some of these cell lines such as SKOV3 is likely nonserous epithelial ovarian cancer (25, 26). Nevertheless, these in vitro systems together provide a tractable model to study the molecular dynamics of chemoresistance in ovarian cancer. The process was repeated until each cell line developed significant resistance to cisplatin. The acquired chemoresistance was assessed through both short-term MTT cell viability assays (3 days) as well as longer-term (2 weeks) crystal violet assays (Fig. 1B–D; Supplementary Fig. S1). In addition, we also developed “resensitized” cellular counterparts of the SKOV3 system by culturing cisplatin-resistant cells in the absence of cisplatin for an extended period (>6 months). We observed significant restoration of cisplatin sensitivity in these previously resistant cells (Fig. 1B, last panel; Supplementary Fig. S2). Reasoning that these partially resensitized cells could provide additional layers of information about the chemoresistance process, we also profiled epigenomic and transcriptomic signatures of these cells to further filter and identify chemoresistance-associated aberrantly regulated proximal and distal regulatory elements.

Identifying chemoresistance-mediated chromatin state alterations at regulatory genomic regions

To identify differentially regulated genomic elements and their gene targets, we performed chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) against the H3K27ac mark in naïve and chemoresistant counterparts of OVCAR4, OVCAR3, OV81, and SKOV3 cells. H3K27ac marks active promoters as well as enhancer elements for a given cell type (27). Comparative analysis of H3K27ac chromatin state maps across a panel of naïve-matched cisplatin-resistant ovarian cancer cells demonstrates substantial chromatin remodeling in cisplatin resistant cells. To better interpret the epigenetic alterations and associate the changes with proximal gene expression, we also acquired transcriptomic profiles from two different cellular models by acquiring RNA-seq gene expression profiles.

To comparatively analyze the epigenome and transcriptome profiles across samples, we identified genomic regions that are significantly enriched (q < 0.01) for H3K27ac signal using the established MACS2 ChIP-seq peak calling algorithm (24). The significantly enriched overlapping regions in different samples were then merged and their enrichment scores were calculated. The peaks with 30 or fewer reads per region were filtered out. This analysis identified a total of 36,388 distal and proximal regulatory elements across the four cellular models. To stratify the peaks differentially regulated upon platinum resistance, we quantified the relative fold changes of H3K27ac coverage per element separately for each matched naïve/chemoresistant pair. Importantly, we observed that 6,568 (18%) of the regulatory genomic regions undergo significant epigenetic reprogramming during cisplatin resistance. Interestingly, a large fraction (88%) of the differentially regulated genomic elements were cell-type specific [(n = 5,792; Fig. 1D and E]. Integrative analysis of gene expression profiles with the differentially regulated enhancer elements indicates that the cell-type–specific epigenetic changes associate with corresponding transcriptional alterations in the relevant cell types. For example, among the 778 genes that are proximally associated with the active enhancers specific to the resistant OVCAR4 cells, 577 (74%) of them have increased expression in the OVCAR4-resistant cells. In contrast, the same set of genes does not show any expression changes in the SKOV3 cells. In the same cells, 916 (63%) of 1,459 genes proximal to the naïve specific enhancers reduce their expression during chemoresistance (Fig. 1E).

Notably, as expected, the genes that are proximal cell-type–specific clusters are enriched with gene ontology terms related to cell-specific regulatory networks (Supplementary Fig. S3). On the other hand, when genes that are proximal to all enhancers that are up or downregulated in one or more cell types are comprehensively analyzed, we observed gene ontology terms related to cellular phenotypes and signaling pathways that are more relevant to drug resistance (Fig. 1F). For example, genes that are proximal to the resistant-specific enhancers are significantly enriched for epithelial-to-mesenchymal transition (EMT), and TGFβ and WNT signaling pathways, which are known drivers of EMT (28). Furthermore, DNA damage response and drug-resistant genes are also spatially associated with the resistant specific enhancers (Fig. 1F). In contrast, genes near enhancers that are decommissioned in resistant cells are implicated in cellular migration, chemosensitivity, and apoptosis. In line with previous reports (29, 30), these findings suggest that cisplatin resistance is associated with upregulation of mesenchymal and stemness-related genes, and downregulation of migratory and somatic cell phenotypes. We next investigated whether these cisplatin-resistant cells are generally resistant to other chemotherapeutic and drugs. To this end, we tested multiple drugs and chemicals (taxane, JQ1, Et-OH); however, we did not see stronger resistance of these cells to these chemicals (Supplementary Fig. S4). In addition to these phenotypic assessments, comparative transcriptomic analysis revealed that unique sets of genes are differentially up- and downregulated during cisplatin and taxane resistance (Supplementary Fig. S5). These results suggest that unique epigenetic and transcriptomic reprograming takes place during cellular resistance to cisplatin and taxane resistance.

Mapping resistant-specific SEs in HGSOC cells

Recent epigenomic analysis of enhancers shows that certain genomic regions contain clusters of enhancers that have been named as SEs (31–33). Importantly, cell-type–specific TFs, master regulators of disease states, cell fates and oncogenesis tend to be regulated by SEs (31–33). SEs can be computationally defined by the presence of clustered genomic regions marked by intense H3K27ac ChIP-seq signals (31). We therefore aimed to identify SEs and their targets genes in each of our cellular models. In total, we identified 304 and 311 SEs in the OVCAR4- and SKOV3-resistant cells, respectively. Importantly, a large fraction of these SEs only exists in the chemoresistant state. For example, 88% (n = 266) of the SEs identified in the OVCAR4-resistant cells are resistant-specific. These enhancers are either absent or not intense enough to be classified as SEs in the naïve cells (Fig. 2A and B). Interestingly, we observed substantially more (>2.5 fold) resistant-specific SEs than the naïve-specific SEs in the SKOV3 system (Fig. 2B). However, we observed a comparable number of naïve- and resistant-specific SEs in the OVCAR4 system (Fig. 2B). Notably, in the SKOV3 model, we also acquired H3K27ac ChIP-seq epigenomic maps in the cells that lost their cisplatin resistance due to prolonged culture in cisplatin-free media (resensitized cells). Critically, as shown in Fig. 2C for the SIRPA locus, resistant-specific SEs lose nearly half of the H3K27ac signal intensity in the resensitized cells. Globally, around 47% of all resistant-specific SEs return to the naïve state in the resensitized cells, indicating that cisplatin-resistant state is associated with dynamic gain and loss of enhancers and SEs (Fig. 2D). Importantly, the genes that are proximally associated with the SEs (within 12.5 kb or the nearest gene) are expressed at higher levels compared with the genes near typical enhancers (Fig. 2E). In line with this, a set of genes that are associated with state-specific SEs are on average expressed more robustly in cells when these SEs are active (Fig. 2E). Specifically, the genes near the resistant-specific SEs are expressed at higher levels in resistant cells. Similarly, the genes near the naïve-specific SEs are expressed significantly higher in naïve cells. Notably, a highly comparable trend is observed in the SKOV3 system, where we also have the resensitized counterparts. For example, genes near the resistant-specific SEs are expressed at higher levels in the resistant SKOV3 cells compared with the naïve or resensitized cells (Fig. 2E).

Platinum resistance is associated with increased cell signaling but decreased cell metabolism pathways

Cell-type–specific transcriptional programs are tightly regulated by state-specific TFs. To identify differentially expressed genes and potentially their driver factors, we integrated the state-specific epigenomic data with transcriptome analysis across the naïve versus cisplatin-resistant cellular model systems (Fig. 3A). Notably, we observed that approximately 1,000 genes were significantly upregulated and 816 genes significantly downregulated in the resistant cells for both the OVCAR4 and SKOV3 systems (Fig. 3A). Critically, the pathway-level analysis indicates that the genes whose expression is significantly enhanced in the resistant cells are implicated with increased cell signaling. We observed that gene targets of a number of major signaling pathways are enriched and expressed at higher levels in the resistant cells (Fig. 3B). For example, gene targets of TNF-mediated NFκB signaling, IL2/STAT5, and TGFβ, and WNT signaling pathways are all among the genes that are significantly enriched in the genes upregulated in the resistant state across the two cellular models. Furthermore, in line with the genes near the upregulated cis-regulatory elements (Fig. 1F), genes involved in EMT and DNA repair pathways are enriched in upregulated genes. In contrast, the analysis of downregulated genes suggests that metabolic pathways are generally downregulated in the chemoresistant cells (Fig. 3B). Specifically, genes involved in catabolic processes and major metabolic pathways such as oxidative phosphorylation, fatty acid metabolism, and TCA cycle are significantly enriched among the downregulated genes in the resistant cells. Encouraged by these findings, we performed Seahorse-mediated metabolic profiling. In the gene set enrichment analysis, we observed that cisplatin SKOV3 cells have decreased glycolysis and oxidative phosphorylation as measured by overall extracellular acidification rate and oxygen consumption rate (Supplementary Fig. S6). In addition to metabolic genes, genes implicated in apoptosis and p53-related genes are among the significantly downregulated genes in chemoresistant cells (Fig. 3B).

Reasoning that the transcriptional program of the chemoresistant state is governed by a unique set of TFs that are upregulated during chemoresistance, we set out to investigate the major TFs involved in this process. To this end, we integrated epigenomic and transcriptomic analyses to identify those TFs that are potentially regulated by resistant-specific TFs and whose expression is induced during chemoresistance. Given the known role of SEs in driving the expression of key cell-type-specific master regulators, we further prioritized the list of TFs by focusing on resistant-specific SE-regulated TFs. Notably, among the significantly upregulated and SE-associated TFs were multiple previously known players of chemoresistance including ZEB2, E2F7, MYC (31, 33), KLF6, and ELK3 (Fig. 3A; refs. 34, 35). In addition to known TFs, we also identified other SE-driven genes previously implicated in chemoresistance including ALDH1A1 (36), AKAP12 (37), FN1 (38–40), RAD18 (41), VEGFC (42), DNAJB12 (43), and PARK7 (44). Critically, the analysis also identified several TFs such as SOX9, HLX, MYBL1, ZNF430, and ZNF502 that have not previously been implicated in platinum resistance in ovarian cancers (Fig. 3A). Notably, the analysis of gene expression and patient survival in TCGA data shows that higher expression of a number of these TFs such as ELK3, HIC1, ZEB2, and ZNF430, are significantly (P < 0.05) associated with poor patient survival (Fig. 3C; ref. 45). These observations suggest that the identified resistance-associated TFs play significant roles in clinical settings and overall patient survival.

To investigate whether the expression of these TFs is dynamically induced by cisplatin treatment across different HGSOC cell lines, we generated cisplatin-induced chemoresistance in additional cell lines and assessed the expression of the nine TFs that are consistently upregulated in the OVCAR4 and SKOV3 cellular systems. To this end, we periodically treated CAOV3 and COV362 cells with increasing doses of cisplatin, harvested cells, and measured the expression changes for these 9 TFs in cells that are resistant to different doses of cisplatin. Notably, as shown with the representative bar graph for SOX9 mRNA and the corresponding heatmaps for the other TFs, our targeted qRT-PCR results demonstrate that these TFs are consistently upregulated as HGSOC cells become resistant to higher doses of cisplatin (Fig. 3D). Collectively, our results suggest that TFs that are associated with the chemoresistant state are dynamically induced by cisplatin treatment and may function as critical drivers and master regulators of platinum resistance in ovarian cancer.

Targeting cisplatin resistance with the small-molecule epigenetic inhibitor

Targeting TFs with small molecules has been a formidable challenge. However, as SEs are differentially enriched for bromo- and extra-terminal (BET) domain chromatin regulators and Mediator complexes (32, 33), bromodomain inhibitors such as the small-molecule inhibitor JQ1 have been shown to differentially target SE target genes (33). Therefore, small-molecule epigenetic drugs that interfere with the regulatory programs of SEs provide an exciting therapeutic opportunity to selectively target critical TFs that are regulated by SEs (31, 33). We thus tested whether JQ1 will selectively target the expression of the selected SE target genes identified from our RNA- and ChIP-seq analyses. As a control, we also selected gene targets of a typical enhancers that are not associated with SE, but have similar basal- and resistance-mediated gene expression fold change to the SE-target genes. Notably, although we observe that JQ1 reduced the expression of non-SE target genes as well, the SE target genes were impacted significantly more (P = 6.67e-05) after a 3-day JQ1 treatment (Fig. 4A). We next tested whether JQ1 treatment will result in synergistic cytotoxicity in the resistant cells when combined with cisplatin. The results indicate that JQ1 and cisplatin cotreatment resulted in a significant and more profound reduction in cellular viability in the resistant cells compared with the naïve cells (Fig. 4B). In line with the short-term (3 days) MTT assay, longer-term (14 days) crystal violet viability assay also demonstrated significant synergy between cisplatin and JQ1 treatment in the resistant OVCAR4 cells (Fig. 4C). To further corroborate these findings, we tested JQ1 and cisplatin combination on four additional HGSOC cell lines. These results show that JQ1 treatment significantly increases the growth-inhibitory activity of cisplatin across multiple ovarian cancer cell lines (Fig. 4D). To further characterize the synergism between JQ1 and cisplatin, we tested various dose combinations on both naïve and cisplatin-resistant cells and calculated the combination index (CI), a well-established model to study drug synergism (46), where CI values less than one indicate synergism. Although we observe synergistic drug dose combinations in both cell lines, we observe much lower CI values (stronger synergism) in the cisplatin-resistant cells at higher dose combinations (Supplementary Fig. S7). Critically, the combinatorial JQ1 and cisplatin treatment resulted in greater rates of apoptosis than individual treatments in multiple cell lines (Fig. 4E and F; Supplementary Fig. S8), suggesting that the combination results in reduced overall cell proliferation and increased cytotoxicity.

Next, we assessed the clinical relevance of JQ1 and cisplatin combination in vivo on cisplatin-resistant cells in a subcutaneous xenograft model. At the beginning of the treatment, mice were randomly separated into four groups. Notably, single-agent treatment of JQ1 (50 mg/kg/wk) or cisplatin (3 mg/kg/wk) did not have a significant effect on the relative tumor volume (Fig. 4G). In contrast, cisplatin and JQ1 combination treatment resulted in significantly smaller tumors when compared with single-drug treatments or the controls (Fig. 4G). These results are in line with the in vitro data and suggest that small-molecule epigenetic inhibitors that interfere with SE function can synergistically increase the therapeutic efficacy of platinum-based compounds in cisplatin-resistant ovarian cancer cells in vivo.

SOX9 is required for acquiring and maintaining the chemoresistant phenotype

Our results so far suggest that the chemoresistance process is, at least in part, regulated by epigenomic reprogramming and licensing of a specific set of distal enhancers, SEs, and differential expression of their target genes. Notably, targeting these distal enhancers through small-molecule epigenetic inhibitors results in significant reductions in cisplatin resistance and in the expression of SE-associated target genes. Although cellular states are most likely established by the combinatorial actions of a network of TFs (47), we focused our efforts to assess the contribution of SOX9 in platinum resistance. We chose to study SOX9 for two reasons. First, the TF had never been implicated in chemoresistance in ovarian cancer. SOX9 is a known player of chondrogenesis (48) and stemness of neural progenitors (49) and hair follicles (50). Notably, it is overexpressed in several different cancers (51) and believed to be involved in epithelial and mesenchymal transitions. Second, SOX9 is one of the few TFs whose both expression level and putative enhancer elements are consistently upregulated in all the chemoresistance models that we studied (Fig. 5A). Notably, we also observe substantial activation of a SOX9 SE in two naïve-matched cisplatin-resistant patient-derived xenograft tumors (Fig. 5A). In line with the gene expression and distal enhancer activation of the SOX9 locus, we observe a substantial SOX9 protein levels in resistant cells (Fig. 5B). Critically, in the resensitized cells, SOX9 protein levels, enhancer activity and mRNA levels go down, further indicating that high SOX9 expression and protein levels are restricted to the cisplatin-resistant cells (Fig. 5B).

To interrogate the functional role of SOX9 in chemoresistance, we used CRISPR to knock out (KO) SOX9 in both naïve and cisplatin-resistant SKOV3 cells. We performed both population-level KO studies as well as clonal analyses of single cells. Notably, genetic depletion of SOX9 did not result in significant proliferation defects neither in resistant cells nor in naïve cells (Supplementary Figs. S9–S11), suggesting that maintenance of transcriptional programs downstream of SOX9 is not critical for cell proliferation. However, depletion of SOX9 in the cisplatin-resistant SKOV3 or OVCAR4 cells confers significant sensitivity to cisplatin when compared to the SOX9 expressing controls (Fig. 5B; Supplementary Fig. S10). On the other hand, CRISPR-mediated SOX9 depletion does not confer additional sensitivity in naïve cells (Supplementary Figs. S11 and S12). Interestingly, when WT and SOX9 KO-naïve cells are challenged with lower cisplatin doses over a period of 14 days, significantly less number of survivor colonies were observed in SOX9-depleted cells (Supplementary Fig. S12). These findings collectively suggest that SOX9 is not only necessary for the maintenance of cisplatin-resistant state, it also critical for the acquisition of the cisplatin-resistant state in ovarian cancer cells.

To gain further insights into the mechanism of SOX9-mediated acquisition and maintenance of chemoresistance, we used ChIP-Seq to map SOX9-binding sites across the genome in cisplatin-resistant cells (Fig. 5C). As shown for BMP2 and WNT5A loci, SOX9 targets are expressed at higher levels in resistant cells (Fig. 5C). It is notable that WNT5A is a target of a distal SE in the chemoresistant SKOV3 cells. Our enhancer and gene expression analysis highlighted upregulated WNT/β-catenin pathway during chemoresistance in multiple cellular models. It is notable that WNT5A is overexpressed in malignant epithelial ovarian cancers (52), implicated in cancer stem cells (53), and associated with poor prognosis (54–56). In support of our findings, recent publications also highlight significant involvement of WNT pathway in ovarian cancer chemoresistance (57–60). These results collectively highlight that targeting the WNT pathway may have a strong potential to prevent and overcome chemoresistance in ovarian cancer (61).

To understand whether SOX9 is associated with activating or repressive activity at the target genomic regions, we analyzed the epigenomic and transcriptional alterations at SOX9 targets in the naïve and resistant counterparts. Notably, the genomic regions that are bound by SOX9 have increased accumulation of H3K27ac marks, indicating higher activity in resistant cells relative to the naïve counterparts (Fig. 5D). When we analyzed the genomic distribution of SOX9-binding sites, we observed strong enrichment of gene promoters (±2 kb of TSS; Fig. 5E), which is in line with the previous SOX9 ChIP-seq studies (62, 63). The pathway analysis of SOX9 targets indicates that these genes are implicated in invasion, drug sensitivity, cancer stem cell genes, and DNA replication (Fig. 5F). In line with our expectation, SOX9 target genes are expressed at higher levels in the resistant cells compared with the naïve counterparts (Fig. 5G). Critically, when the SOX9-binding intensity is correlated with the gene expression within the resistant cells, we observed significant positive correlation between SOX9-binding intensity and target gene expression (Fig. 5H). These findings indicate that SOX9 is acting as a transcriptional activator while driving and maintaining the chemoresistant state. To further test this hypothesis, we performed targeted qPCR mRNA expression analysis on select genes in WT and SOX9 KO cells. Critically, SOX9 deletion results in significant depletion of multiple other TFs that are significantly upregulated in the resistant cells (Fig. 5I). These results indicate that SOX9 is a critical player of gene expression programs of the cisplatin-resistant cellular states. Furthermore, our data suggest that SOX9 may mediate acquisition and maintenance of cisplatin chemoresistance by coordinating the expression of multiple other TFs and thus controlling their gene targets.

Chemoresistance is a major therapeutic obstacle in the treatment of ovarian cancer. Although specific genetic alterations such as reversion of germline BRCA1/2 mutations and inactivating mutations in tumor suppressor RB1, NF1, RAD51B, and PTEN (16) have been associated with chemoresistance, the molecular network that drives and maintains the chemoresistant state at the chromatin and transcriptional levels is poorly understood. In this study, we integrate epigenome and transcriptome profiling in multiple chemoresistance models to investigate potential drivers of resistance in ovarian cancer. Our findings indicate that chemoresistance is associated with the licensing of a specific set of distal enhancers. Notably, among these genes that are especially upregulated in the cisplatin-resistant cellular state are multiple TFs, whose expression is controlled by unique enhancer clusters known as SEs, which can be potentially targeted by small-molecule epigenetic inhibitors. The data presented here suggest that JQ1 can effectively reduce the expression of such TFs and increase cisplatin sensitivity in the chemoresistant cells. In line with this, there is significant synergy between JQ1 and cisplatin both in vitro and in vivo, suggesting a great translational potential to target chemoresistance in ovarian cancer with chromatin targeting small molecules.

Our findings provide key insights into the epigenetic bases of chemoresistance. Data presented in this article indicate that chemoresistance is associated with large-scale reprogramming and redistribution of H3K27ac histone modifications across the genome. As expected, the reprogrammed enhancer elements are highly enriched near the genes that are associated with DNA damage response, cell signaling, EMT, as well as drug response are aberrantly upregulated in chemoresistant cells. On the other hand, genes implicated in apoptosis signaling, catabolism, and cellular adhesion pathways are downregulated during chemoresistance. Notably, we also observed substantial cell-type–specific epigenetic and transcriptional programs, suggesting a great degree of heterogeneity in the epigenetic state of resistant cells. Critically, despite the epigenetic and transcriptional heterogeneity, our data suggest that small-molecule epigenetic drugs targeting distal enhancers and SEs can be used to alter the aberrant regulation and confer cisplatin sensitivity. BET inhibitors like JQ1 have been shown to release Mediator from cis-regulatory elements at regulatory elements and alter the transcription of target genes. Thus, theoretically, JQ1 can affect both enhancers and SEs. However, because SEs could contain significantly higher levels of H3K27ac and BRD4, they are more sensitive to JQ1 treatment (33). Our experimental analysis supports this notion and explains, in part, the particular effectiveness of JQ1 in our chemoresistant models.

Our findings highlight SOX9 as a critical determinant of the cisplatin resistance. We show that SOX9 is a target of a SE that is strongly activated in resistant cells. Genetic targeting efforts further support the hypothesis that SOX9 is one of the key TFs that maintain the gene expression programs of cisplatin-resistant states. SOX9 depletion results in the downregulation of multiple TFs associated with chemoresistance and increased sensitivity to cisplatin in multiple ovarian cancer lines. It remains to be shown whether the expression SOX9 alone is sufficient to confer cisplatin resistance. It is notable that higher expression of SOX9 is associated with worse prognosis in multiple solid tumors (64). Furthermore, SOX9 is also a critical marker of clinically aggressive disease in metastatic high-grade serous carcinoma (65).

R.M. Campbell has ownership interest (including stock, patents, etc.) in Eli Lilly & Company. P.J. Ebert has ownership interest (including stock, patents, etc.) in Eli Lilly and Company. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Shang, A.A. Jazaeri, R.M. Campbell, T. Abbas, C.N. Landen, M. Adli

Development of methodology: S. Shang, A.A. Jazaeri, A.J. Duval, F. Guessous, T. Abbas, C.N. Landen, M. Adli

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Shang, A.J. Duval, M. Benamar, F. Guessous, I. Lee, P.J. Ebert, T. Abbas, C.N. Landen, A. DiFeo, P.C. Scacheri

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Shang, J. Yang, A.J. Duval, M. Benamar, I. Lee, R.M. Campbell, P.J. Ebert, C.N. Landen, P.C. Scacheri, M. Adli

Writing, review, and/or revision of the manuscript: S. Shang, J. Yang, A.A. Jazaeri, A.J. Duval, T. Tufan, R.M. Campbell, C.N. Landen, P.C. Scacheri, M. Adli

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Shang, A.A. Jazaeri, A.J. Duval, P.J. Ebert, M. Adli

Study supervision: M. Adli

Other (acquisition of preliminary data): N.L. Fischer

The study was supported by the NIH/NCI 1R01 CA211648-01 and UVA Cancer Center pilot award (NCI CCSG P30 CA44579) award (to M. Adli). In addition, the work was supported by the NIH/NCI R01 CA160356 and NIH/NCI R01 CA193677 awards (to P.C. Scacheri). S. Shang was supported in part by the Cancer Training grant T32 CA009109.

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.

1.
American Cancer Society
.
Cancer facts & figures 2018
.
Atlanta, GA
:
American Cancer Society
; 
2018
.
2.
Disease
GBD
,
Injury
I
,
Prevalence
C
. 
Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990–2015: a systematic analysis for the Global Burden of Disease Study 2015
.
Lancet
2016
;
388
:
1545
602
.
3.
Reid
BM
,
Permuth
JB
,
Sellers
TA
. 
Epidemiology of ovarian cancer: a review
.
Cancer Biol Med
2017
;
14
:
9
32
.
4.
Jazaeri
AA
,
Bryant
JL
,
Park
H
,
Li
H
,
Dahiya
N
,
Stoler
MH
, et al
Molecular requirements for transformation of fallopian tube epithelial cells into serous carcinoma
.
Neoplasia
2011
;
13
:
899
911
.
5.
Karst
AM
,
Levanon
K
,
Drapkin
R
. 
Modeling high-grade serous ovarian carcinogenesis from the fallopian tube
.
Proc Natl Acad Sci U S A
2011
;
108
:
7547
52
.
6.
Perets
R
,
Wyant
GA
,
Muto
KW
,
Bijron
JG
,
Poole
BB
,
Chin
KT
, et al
Transformation of the fallopian tube secretory epithelium leads to high-grade serous ovarian cancer in Brca;Tp53;Pten models
.
Cancer Cell
2013
;
24
:
751
65
.
7.
Siegel
R
,
Naishadham
D
,
Jemal
A
. 
Cancer statistics, 2013
.
CA Cancer J Clin
2013
;
63
:
11
30
.
8.
Balch
C
,
Fang
F
,
Matei
DE
,
Huang
TH
,
Nephew
KP
. 
Minireview: epigenetic changes in ovarian cancer
.
Endocrinology
2009
;
150
:
4003
11
.
9.
Giaccone
G
. 
Clinical perspectives on platinum resistance
.
Drugs
2000
;
59
Suppl 4
:
9
17
.
10.
Ozols
RF
. 
Ovarian cancer: new clinical approaches
.
Cancer Treat Rev
1991
;
18
Suppl A
:
77
83
.
11.
Koberle
B
,
Tomicic
MT
,
Usanova
S
,
Kaina
B
. 
Cisplatin resistance: preclinical findings and clinical implications
.
Biochim Biophys Acta
2010
;
1806
:
172
82
.
12.
Galluzzi
L
,
Senovilla
L
,
Vitale
I
,
Michels
J
,
Martins
I
,
Kepp
O
, et al
Molecular mechanisms of cisplatin resistance
.
Oncogene
2012
;
31
:
1869
83
.
13.
Takenaka
M
,
Saito
M
,
Iwakawa
R
,
Yanaihara
N
,
Saito
M
,
Kato
M
, et al
Profiling of actionable gene alterations in ovarian cancer by targeted deep sequencing
.
Int J Oncol
2015
;
46
:
2389
98
.
14.
Cole
AJ
,
Dwight
T
,
Gill
AJ
,
Dickson
KA
,
Zhu
Y
,
Clarkson
A
, et al
Assessing mutant p53 in primary high-grade serous ovarian cancer using immunohistochemistry and massively parallel sequencing
.
Sci Rep
2016
;
6
:
26191
.
15.
Kim
JY
,
Cho
CH
,
Song
HS
. 
Targeted therapy of ovarian cancer including immune check point inhibitor
.
Korean J Intern Med
2017
;
32
:
798
804
.
16.
Patch
AM
,
Christie
EL
,
Etemadmoghadam
D
,
Garsed
DW
,
George
J
,
Fereday
S
, et al
Whole-genome characterization of chemoresistant ovarian cancer
.
Nature
2015
;
521
:
489
94
.
17.
Dawson
MA
. 
The cancer epigenome: concepts, challenges, and therapeutic opportunities
.
Science
2017
;
355
:
1147
52
.
18.
Baylin
SB
,
Jones
PA
. 
A decade of exploring the cancer epigenome - biological and translational implications
.
Nat Rev Cancer
2011
;
11
:
726
34
.
19.
Berdasco
M
,
Esteller
M
. 
Aberrant epigenetic landscape in cancer: how cellular identity goes awry
.
Dev Cell
2010
;
19
:
698
711
.
20.
Zeller
C
,
Dai
W
,
Steele
NL
,
Siddiq
A
,
Walley
AJ
,
Wilhelm-Benartzi
CS
, et al
Candidate DNA methylation drivers of acquired cisplatin resistance in ovarian cancer identified by methylome and expression profiling
.
Oncogene
2012
;
31
:
4567
76
.
21.
Cheng
YY
,
Mok
E
,
Tan
S
,
Leygo
C
,
McLaughlin
C
,
George
AM
, et al
SFRP tumour suppressor genes are potential plasma-based epigenetic biomarkers for malignant pleural mesothelioma
.
Dis Markers
2017
;
2017
:
2536187
.
22.
Lee
PS
,
Teaberry
VS
,
Bland
AE
,
Huang
Z
,
Whitaker
RS
,
Baba
T
, et al
Elevated MAL expression is accompanied by promoter hypomethylation and platinum resistance in epithelial ovarian cancer
.
Int J Cancer
2010
;
126
:
1378
89
.
23.
Taniguchi
T
,
Tischkowitz
M
,
Ameziane
N
,
Hodgson
SV
,
Mathew
CG
,
Joenje
H
, et al
Disruption of the Fanconi anemia-BRCA pathway in cisplatin-sensitive ovarian tumors
.
Nat Med
2003
;
9
:
568
74
.
24.
Feng
J
,
Liu
T
,
Qin
B
,
Zhang
Y
,
Liu
XS
. 
Identifying ChIP-seq enrichment using MACS
.
Nat Protoc
2012
;
7
:
1728
40
.
25.
Beaufort
CM
,
Helmijr
JC
,
Piskorz
AM
,
Hoogstraat
M
,
Ruigrok-Ritstier
K
,
Besselink
N
, et al
Ovarian cancer cell line panel (OCCP): clinical importance of in vitro morphological subtypes
.
PLoS One
2014
;
9
:
e103988
.
26.
Domcke
S
,
Sinha
R
,
Levine
DA
,
Sander
C
,
Schultz
N
. 
Evaluating cell lines as tumour models by comparison of genomic profiles
.
Nat Commun
2013
;
4
:
2126
.
27.
Zhu
J
,
Adli
M
,
Zou
JY
,
Verstappen
G
,
Coyne
M
,
Zhang
X
, et al
Genome-wide chromatin state transitions associated with developmental and environmental cues
.
Cell
2013
;
152
:
642
54
.
28.
Moustakas
A
,
Heldin
CH
. 
Mechanisms of TGFbeta-induced epithelial-mesenchymal transition
.
J Clin Med
2016
;
5
:pii:
E63
.
29.
Shah
MM
,
Landen
CN
. 
Ovarian cancer stem cells: are they real and why are they important?
Gynecol Oncol
2014
;
132
:
483
9
.
30.
Steg
AD
,
Bevis
KS
,
Katre
AA
,
Ziebarth
A
,
Dobbin
ZC
,
Alvarez
RD
, et al
Stem cell pathways contribute to clinical chemoresistance in ovarian cancer
.
Clin Cancer Res
2012
;
18
:
869
81
.
31.
Hnisz
D
,
Abraham
BJ
,
Lee
TI
,
Lau
A
,
Saint-Andre
V
,
Sigova
AA
, et al
Super-enhancers in the control of cell identity and disease
.
Cell
2013
;
155
:
934
47
.
32.
Whyte
WA
,
Orlando
DA
,
Hnisz
D
,
Abraham
BJ
,
Lin
CY
,
Kagey
MH
, et al
Master transcription factors and mediator establish super-enhancers at key cell identity genes
.
Cell
2013
;
153
:
307
19
.
33.
Loven
J
,
Hoke
HA
,
Lin
CY
,
Lau
A
,
Orlando
DA
,
Vakoc
CR
, et al
Selective inhibition of tumor oncogenes by disruption of super-enhancers
.
Cell
2013
;
153
:
320
34
.
34.
Kong
SY
,
Kim
KS
,
Kim
J
,
Kim
MK
,
Lee
KH
,
Lee
JY
, et al
The ELK3-GATA3 axis orchestrates invasion and metastasis of breast cancer cells in vitro and in vivo
.
Oncotarget
2016
;
7
:
65137
46
.
35.
Buchwalter
G
,
Gross
C
,
Wasylyk
B
. 
The ternary complex factor Net regulates cell migration through inhibition of PAI-1 expression
.
Mol Cell Biol
2005
;
25
:
10853
62
.
36.
Yokoyama
Y
,
Zhu
H
,
Lee
JH
,
Kossenkov
AV
,
Wu
SY
,
Wickramasinghe
JM
, et al
BET inhibitors suppress ALDH activity by targeting ALDH1A1 super-enhancer in ovarian cancer
.
Cancer Res
2016
;
76
:
6320
30
.
37.
Chappell
NP
,
Teng
PN
,
Hood
BL
,
Wang
G
,
Darcy
KM
,
Hamilton
CA
, et al
Mitochondrial proteomic analysis of cisplatin resistance in ovarian cancer
.
J Proteome Res
2012
;
11
:
4605
14
.
38.
Wu
W
,
Wang
Q
,
Yin
F
,
Yang
Z
,
Zhang
W
,
Gabra
H
, et al
Identification of proteomic and metabolic signatures associated with chemoresistance of human epithelial ovarian cancer
.
Int J Oncol
2016
;
49
:
1651
65
.
39.
Jinawath
N
,
Vasoontara
C
,
Jinawath
A
,
Fang
X
,
Zhao
K
,
Yap
KL
, et al
Oncoproteomic analysis reveals co-upregulation of RELA and STAT5 in carboplatin resistant ovarian carcinoma
.
PLoS One
2010
;
5
:
e11198
.
40.
Helleman
J
,
Jansen
MP
,
Span
PN
,
van Staveren
IL
,
Massuger
LF
,
Meijer-van Gelder
ME
, et al
Molecular profiling of platinum resistant ovarian cancer
.
Int J Cancer
2006
;
118
:
1963
71
.
41.
Haynes
B
,
Saadat
N
,
Myung
B
,
Shekhar
MP
. 
Crosstalk between translesion synthesis, Fanconi anemia network, and homologous recombination repair pathways in interstrand DNA crosslink repair and development of chemoresistance
.
Mutat Res Rev Mutat Res
2015
;
763
:
258
66
.
42.
Scartozzi
M
,
Giampieri
R
,
Loretelli
C
,
Bittoni
A
,
Mandolesi
A
,
Faloppi
L
, et al
Tumor angiogenesis genotyping and efficacy of first-line chemotherapy in metastatic gastric cancer patients
.
Pharmacogenomics
2013
;
14
:
1991
8
.
43.
Sopha
P
,
Ren
HY
,
Grove
DE
,
Cyr
DM
. 
Endoplasmic reticulum stress-induced degradation of DNAJB12 stimulates BOK accumulation and primes cancer cells for apoptosis
.
J Biol Chem
2017
;
292
:
11792
803
.
44.
Trivedi
R
,
Dihazi
GH
,
Eltoweissy
M
,
Mishra
DP
,
Mueller
GA
,
Dihazi
H
. 
The antioxidant protein PARK7 plays an important role in cell resistance to cisplatin-induced apoptosis in case of clear cell renal cell carcinoma
.
Eur J Pharmacol
2016
;
784
:
99
110
.
45.
Cancer Genome Atlas Research Network
. 
Integrated genomic analyses of ovarian carcinoma
.
Nature
2011
;
474
:
609
15
.
46.
Chou
TC
,
Talaly
P
. 
A simple generalized equation for the analysis of multiple inhibitions of Michaelis-Menten kinetic systems
.
J Biol Chem
1977
;
252
:
6438
42
.
47.
Han
H
,
Shim
H
,
Shin
D
,
Shim
JE
,
Ko
Y
,
Shin
J
, et al
TRRUST: a reference database of human transcriptional regulatory interactions
.
Sci Rep
2015
;
5
:
11432
.
48.
Leung
VY
,
Gao
B
,
Leung
KK
,
Melhado
IG
,
Wynn
SL
,
Au
TY
, et al
SOX9 governs differentiation stage-specific gene expression in growth plate chondrocytes via direct concomitant transactivation and repression
.
PLoS Genet
2011
;
7
:
e1002356
.
49.
Martini
S
,
Bernoth
K
,
Main
H
,
Ortega
GD
,
Lendahl
U
,
Just
U
, et al
A critical role for Sox9 in notch-induced astrogliogenesis and stem cell maintenance
.
Stem Cells
2013
;
31
:
741
51
.
50.
Kadaja
M
,
Keyes
BE
,
Lin
M
,
Pasolli
HA
,
Genander
M
,
Polak
L
, et al
SOX9: a stem cell transcriptional regulator of secreted niche signaling factors
.
Genes Dev
2014
;
28
:
328
41
.
51.
Matheu
A
,
Collado
M
,
Wise
C
,
Manterola
L
,
Cekaite
L
,
Tye
AJ
, et al
Oncogenicity of the developmental transcription factor Sox9
.
Cancer Res
2012
;
72
:
1301
15
.
52.
Badiglian Filho
L
,
Oshima
CT
,
De Oliveira Lima
F
,
De Oliveira Costa
H
,
De Sousa Damiao
R
,
Gomes
TS
, et al
Canonical and noncanonical Wnt pathway: a comparison among normal ovary, benign ovarian tumor and ovarian cancer
.
Oncol Rep
2009
;
21
:
313
20
.
53.
Zhou
Y
,
Kipps
TJ
,
Zhang
S
. 
Wnt5a signaling in normal and cancer stem cells
.
Stem Cells Int
2017
;
2017
:
5295286
.
54.
Qi
H
,
Sun
B
,
Zhao
X
,
Du
J
,
Gu
Q
,
Liu
Y
, et al
Wnt5a promotes vasculogenic mimicry and epithelial-mesenchymal transition via protein kinase Calpha in epithelial ovarian cancer
.
Oncol Rep
2014
;
32
:
771
9
.
55.
Peng
C
,
Zhang
X
,
Yu
H
,
Wu
D
,
Zheng
J
. 
Wnt5a as a predictor in poor clinical outcome of patients and a mediator in chemoresistance of ovarian cancer
.
Int J Gynecol Cancer
2011
;
21
:
280
8
.
56.
Bitler
BG
,
Nicodemus
JP
,
Li
H
,
Cai
Q
,
Wu
H
,
Hua
X
, et al
Wnt5a suppresses epithelial ovarian cancer by promoting cellular senescence
.
Cancer Res
2011
;
71
:
6184
94
.
57.
Niiro
E
,
Morioka
S
,
Iwai
K
,
Yamada
Y
,
Ogawa
K
,
Kawahara
N
, et al
Potential signaling pathways as therapeutic targets for overcoming chemoresistance in mucinous ovarian cancer
.
Biomed Rep
2018
;
8
:
215
23
.
58.
Huang
L
,
Jin
Y
,
Feng
S
,
Zou
Y
,
Xu
S
,
Qiu
S
, et al
Role of Wnt/beta-catenin, Wnt/c-Jun N-terminal kinase and Wnt/Ca(2+) pathways in cisplatin-induced chemoresistance in ovarian cancer
.
Exp Ther Med
2016
;
12
:
3851
8
.
59.
Zhang
C
,
Zhang
Z
,
Zhang
S
,
Wang
W
,
Hu
P
. 
Targeting of Wnt/beta-catenin by anthelmintic drug pyrvinium enhances sensitivity of ovarian cancer cells to chemotherapy
.
Med Sci Monit
2017
;
23
:
266
75
.
60.
Li
J
,
Yang
S
,
Su
N
,
Wang
Y
,
Yu
J
,
Qiu
H
, et al
Overexpression of long non-coding RNA HOTAIR leads to chemoresistance by activating the Wnt/beta-catenin pathway in human ovarian cancer
.
Tumour Biol
2016
;
37
:
2057
65
.
61.
Nagaraj
AB
,
Joseph
P
,
Kovalenko
O
,
Singh
S
,
Armstrong
A
,
Redline
R
, et al
Critical role of Wnt/beta-catenin signaling in driving epithelial ovarian cancer platinum resistance
.
Oncotarget
2015
;
6
:
23720
34
.
62.
Liu
CF
,
Lefebvre
V
. 
The transcription factors SOX9 and SOX5/SOX6 cooperate genome-wide through super-enhancers to drive chondrogenesis
.
Nucleic Acids Res
2015
;
43
:
8183
203
.
63.
Shi
Z
,
Chiang
CI
,
Labhart
P
,
Zhao
Y
,
Yang
J
,
Mistretta
TA
, et al
Context-specific role of SOX9 in NF-Y mediated gene regulation in colorectal cancer cells
.
Nucleic Acids Res
2015
;
43
:
6257
69
.
64.
Ruan
H
,
Hu
S
,
Zhang
H
,
Du
G
,
Li
X
,
Li
X
, et al
Upregulated SOX9 expression indicates worse prognosis in solid tumors: a systematic review and meta-analysis
.
Oncotarget
2017
;
8
:
113163
73
.
65.
Sherman-Samis
M
,
Onallah
H
,
Holth
A
,
Reich
R
,
Davidson
B
. 
SOX2 and SOX9 are markers of clinically aggressive disease in metastatic high-grade serous carcinoma
.
Gynecol Oncol
2019
;
153
:
651
60
.