Clinical perspectives of BET inhibition in ovarian cancer

Abstract

Background

Bromodomain and extra-terminal (BET) proteins are epigenetic readers that bind to acetylated lysines of histones and regulate gene transcription. BET protein family members mediate the expression of various oncogenic drivers in ovarian cancer, such as the MYC and Neuregulin 1 (NRG1) genes. BRD4, the most thoroughly studied member of the BET family, is amplified in a significant subset of high-grade serous carcinomas (HGSC) of the ovary. It has been reported that BET inhibitors can attenuate the proliferation and dissemination of ovarian cancer cells by inhibiting oncogenic pathways, such as the FOXM1 and JAK/STAT pathways. BET inhibition can re-sensitize resistant ovarian cancer cells to already approved anticancer agents, including cisplatin and PARP inhibitors. This synergism was also confirmed in vivo in animal models. These and other preclinical results provide a promising basis for the application of BET inhibitors in ovarian cancer treatment. Currently, Phase I/II clinical trials explore the safety and efficacy profiles of BET inhibitors in various solid tumors, including ovarian tumors. Here, we review current knowledge on the molecular effects and preclinical activities of BET inhibitors in ovarian tumors.

Conclusions

BET proteins have emerged as new druggable targets for ovarian cancer. BET inhibitors may enhance antitumor activity when co-administered with conventional treatment regimens. Results from ongoing Phase I/II studies are anticipated to confirm this notion.

Introduction

Ovarian cancer is the fifth most common cause of cancer-related mortality in women and the leading cause of death due to gynecological cancers in the United States [1]. It is estimated that in 2020 308,069 new cases and 193,811 deaths will be reported due to ovarian cancer worldwide. Epithelial ovarian cancer (EOC) accounts for 90% of the cases and is further divided into five histological subtypes: high-grade serous carcinoma (HGSC; 70%), endometrioid carcinoma (EC; 10%), clear-cell carcinoma (CCC; 10%), mucinous carcinoma (MC; 3%) and low-grade serous carcinoma (LGSC; < 5%) [2]. Due to lack of effective screening strategies, approximately two thirds of patients are diagnosed at stage III/IV of the disease. Despite initial high response rates to chemotherapy, most patients eventually relapse and 75% of stage III/IV patients die within 5 years [3]. The high mortality rates of ovarian cancers indicate an emerging need for the identification of new treatment options.

Epigenetic modifications, such as methylation and acetylation of histones, have been found to be substantially altered in cancer. Histone acetylation and deacetylation is mediated by two distinct groups of proteins: histone acetyltransferases (HATs) that transfer acetyl groups to histone lysine residues facilitating the access of transcription factors to DNA, and histone deacetylases (HDACs) that remove them [4]. A third group of proteins that contribute in DNA acetylation are bromodomain and extra-terminal (BET) domain proteins, which serve as the ‘readers’ of DNA acetylation [5]. BET proteins share two N-terminal bromodomains (BD1 and BD2) that interact with acetylated lysine residues on histone tails, and a more divergent C-terminal recruitment domain. The BET family of proteins consists of four conserved members, the ubiquitously expressed BRD2, BRD3a and BRD4, and BRDT, which is mainly expressed in germ cells [5]. BET proteins recognize acetylated lysine residues of histone 4 via their BRDs and, subsequently, recruit other proteins and transcriptions factors to promoters and enhancers of active genes. They also interact with other proteins in a bromodomain-independent manner, including transcription factors such as FLI1, MYB and p53 [6]. Apart from promoters and enhancers, BET proteins have also been found to occupy super-enhancer regions of oncogenes, which are clusters of enhancers associated with high levels of expression. BRD4 binds to promoters and super-enhancer regions of key oncogenes, such as MYC. The C-terminal domain of BRD4 interacts with the P-TEFb complex, resulting in its activation. The P-TEFb complex subsequently phosphorylates RNAPII on serine 2, which is required for RNAPII activation and initiation of transcription [7]. Enhanced BRD4 accumulation results in increased gene expression. As such, BRD4 can regulate the transcription of many M/G1 genes, thereby modulating cell cycle progression from G1 to S and G2 to M phases [8]. BRD4 is the most thoroughly studied BET protein. BRD3 has been shown to regulate transcription via the E2F-Rb pathway, while BRD2 associates with the E2F1 transcription factor and determines E2F1-directed cell cycle progression [9, 10]. The wide implication of BET proteins in cancer-related gene expression has led to the development of small-molecule inhibitors that can efficiently interfere with bromodomain binding to histone acetylated lysine residues and, thus, displace BET proteins from promoter and/or enhancer regions [11]. Currently, several BET inhibitors have been developed and tested in preclinical studies, including JQ1, I-BET151, ABBV-075, I-BET762 and OTX015. Ongoing Phase I and II clinical trials are evaluating the safety and pharmacokinetics of these inhibitors in hematological malignancies and solid tumors. Here, we review all known molecular pathways and cellular functions affected by BET inhibition in ovarian tumors, establishing a rationale for BET inhibitor administration in ovarian cancer treatment.

Preclinical BET inhibition studies

Rationale for BET inhibition

Several lines of evidence indicate that BET inhibition may have therapeutic implications in ovarian cancer. BRD4 gene amplification is more profound in ovarian cancer (~18–19%) than in other cancer subtypes and has been found to be associated with worse overall and progression-free survival rates [12–15]. The Cancer Genomic Atlas (TCGA) data analysis revealed increased BRD4 mRNA levels in 9% of ovarian cancers, consistent with BRD4 gene amplification [14]. TCGA data analysis also revealed that activation of the NSD3-CHD8-BRD4 pathway occurs in 18% of HGSC ovarian tumors and is closely related to shorter progression-free and overall survival rates [16]. NSD3 is a methyltransferase encoded by the WHSC1L1 gene, which serves as a scaffold to connect BRD4 with the CHD8 chromatin remodeler to sustain transcription of key oncogenes [17]. NSD3-CHD8-BRD4 thus constitutes an epigenetic regulatory pathway that is potentially druggable. BRD2 amplifications have been observed in 1.3% of ovarian tumors, whereas BRD3 amplifications have been encountered in 0.8% of ovarian tumors [18]. On the other hand, BRDT expression was found to be absent [15]. BRD4 silencing led to growth suppression of ovarian tumor cells both in vitro and in vivo, highlighting its significance in oncogenesis [15, 19, 20]. BET inhibitors have elicited antiproliferative effects in several ovarian cancer cell lines, regardless histological subtype [12, 13, 19–23]. It should be noted here that ovarian cancer cells are characterized by a greater sensitivity to BET inhibition than other cancer cells [24]. The efficacy of ΒΕΤ inhibitors was confirmed by in vivo studies using murine models and patient-derived xenografts, including chemotherapy resistant cell lines [12, 13, 19, 20, 22, 23, 25].

BET inhibition and cell survival

Preclinical data indicate that BET inhibition attenuates ovarian cancer cell growth via inducing G1 cell cycle arrest rather than causing cell death or apoptosis [13, 22, 24, 26], in agreement with data previously published in other solid and hematological malignancies [27, 28]. However, cell apoptosis has been reported in a number of studies as indicated by increased PARP cleavage and decreased pro-apoptotic BCL-2 and Survivin protein levels [13, 22, 23, 29, 30]. Apoptosis induced by BET inhibition results from a prolonged G1 cell cycle arrest, which forces cancer cells to undergo either senescence or apoptosis [24, 30–32]. BET inhibitor I-BET151 has been found to induce apoptosis via the mitochondrial pathway, as the level of the pro-apoptotic BH3-only protein BIM was elevated [30]. BH3-only protein family members, including BIM, interact with pro-apoptotic effector proteins (BAX and BAK) to permeabilize mitochondrial outer membranes and cause mitochondrial apoptosis [33]. The expression of multiple cell cycle regulating factors was altered by BET inhibitor treatment, including downregulation of Cyclin D1 and CDK4/6 and upregulation of p21 and p16, supporting the effect of BET inhibition on cell cycle progression [22, 29, 30]. Preclinical studies have revealed a robust effect of BET inhibition on cell functions other than proliferation in ovarian tumors. It has been found, for example, that BET inhibition may affect cellular metabolism via disruption of the aerobic glycolysis pathway [22]. BET inhibitor JQ1 has been found to decrease the expression and phosphorylation of lactate dehydrogenase A (LDHA) and, thus, lactate production both in vitro and in vivo [22]. Indeed, siRNA-mediated LDHA knockdown regenerated the JQ1-induced inhibitory effect on glycolysis [22]. Another cell function found to be affected by BET inhibition is the oxidative stress response. JQ1 can dysregulate mitochondrial function and induce endoplasmic reticulum (ER) stress, as indicated by upregulation of PERK, BiP and Calnexin, endoplasmic reticulum (ER) stress markers in ovarian cancer cell lines [22]. Glycolysis and oxidative stress responses are vital processes inevitably related to cell viability, underlying two essential mechanisms of BET inhibitor-induced toxicity. Moreover, the expression of a wide range of kinases has been found to be significantly altered upon BET inhibitor treatment in ovarian cancer cells [29]. Kinases that are upregulated by BET inhibition are key components of the Hippo, RIP and TGF-β signaling pathways, including RIPK1, MAPK9, TGFBR1 and MAP3K3 kinases, and are associated with apoptosis [29]. In contrast, kinases regulating cell proliferation and cell cycle progression, like Aurora kinases A and B are significantly downregulated by BET inhibitors, as described below.

BET inhibition and downstream signaling pathways

A major BRD4-dependent oncogene downregulated by BET inhibition in cancer studies is MYC. BET inhibition has been found to suppress BRD4-mediated expression of the MYC oncogene in both ovarian cancer cell lines and patient-derived xenograft models dose-dependently [12, 19, 21, 22, 24, 29, 34, 35]. MYC expression levels correlate with sensitivity to BET inhibition treatment, with MYC overexpressing tumors being the most sensitive ones [19, 34]. Chemotherapy resistant ovarian cancer cells and patient-derived xenograft models characterized by MYC amplification displayed high sensitivities to BET inhibitors [35]. MYC gene amplification has been observed in the majority of primary (74%), metastatic (78%) and recurrent chemotherapy-resistant (82%) ovarian tumors [35]. Since MYC represents a key effector in tumorigenesis, its down-regulation is fundamental for the efficacy of BET inhibition [24, 35, 36]. However, MYC suppression is not always observed in ovarian cancer cells treated with BET inhibitors, suggesting that multiple other genes may be involved in BET inhibitor-associated toxicity [13, 37]. Other genes regulated by BRD4 include E2F1 and VEGF, which may also be involved in BET inhibition [12]. Genes often associated with BRD4 amplification include NOTCH3, SMARCA4, ILF3 and AP1M1 [14]. Apart from MYC downregulation, other signaling pathways may also be altered by BET inhibitor treatment. We have previously shown that BET inhibition affects the JAK/STAT oncogenic pathway in ovarian cancer cells through downregulation of JAK2 and STAT5 protein expression [34]. JQ1 treatment decreased the expression of BRD4-regulated genes, including IL7R, FOSL1 and HEXIM1, the latter of which has been established as a BET inhibitor efficacy biomarker in ovarian cancer cells [29]. Forkhead Box M1 (FOXM1) is a transcription factor essential for the activation of DNA damage response genes, such as DNApol, POLE2 and BRCA2 and the regulation of chromatin structure enabling DNA repair [38]. In addition to this, FOXM1 is involved in cell cycle progression by promoting G1/S and G2/M transition through regulating the transcription of cell cycle-related genes such as Cyclin B, Cyclin D1, Aurora kinase A and B (AURKA, AURKB) and polo-like kinase 1 (PLK1) [39, 40]. The Cancer Genome Atlas (TCGA) data revealed alteration of FOXM1 expression in ~87% of HGSC cases [41]. BET inhibitors were found to suppress FOXM1 expression and its downstream pathways via impairing BRD4 accumulation at FOXM1 promoters and enhancers [13, 29]. Maintenance of high FOXM1 expression levels was found to be associated with an increased BET inhibitor resistance and a survival benefit. In vivo, JQ1 administration efficiently downregulated FOXM1 expression in mice inoculated with ovarian cancer cells [13]. BET inhibition not only downregulated FOXM1 expression, but also supressed the expression of its target genes, including Cyclin B, PLK1, AURKA and AURKB in ovarian cancer cells [21].

Another key target of BET inhibition in ovarian cancer is the SWI/SNF chromatin remodelling complex [12]. The SWI/SNF complex is a conserved mammalian chromatin-remodelling complex composed of protein subunits including SMARCA4, SMARCA2, ARID1A, ARID1B, SMARCB1, SMARCC1 and SMARCC2, which is implicated in cancer development [42, 43]. The SWI/SNF complex functions as a tumor-suppressor via interaction with key oncogenes and tumorsuppressor genes, such as RB, BRCA1, MYC [43] and BRD4 [44]. Synchronous inhibition of the BRD4 and SWI/SNF pathways showed a synergistic effect [45]. BET inhibition was found to interfere with the BRD4-SWI/SNF interaction in ovarian cancer cells, resulting in an antiproliferative effect that was phenocopied when the SWI/SNF subunit SMARCA4 was silenced [12]. It is well known that SMARCA4 and BRD4 bind to the same MYC enhancer sites [44]. In addition to this, BET inhibition was found to suppress the expression of multiple SWI/SNF family members, including SMARCC2 and SMARCE1 in ovarian cancer cells, resulting in a robust deregulation of the complex [37].

Another subunit of the SWI/SNF complex is ARID1A, which preferentially binds to adenine-thymine enriched DNA regions acting as a tumor suppressor. ARID1A gene mutations seem to implicate up to half (46%) of CCCs and 30% of ECs [46, 47]. Loss of ARID1A is associated with a poor prognosis in terms of progression-free and overall survival in CCC and, thus, there is a need for targeted therapies against ARID1A-mutated tumors [48]. ARID1A loss is supposed to be an early event in CCC development. BET inhibition was found to suppress the growth of ARID1A mutated CCC tumors, but not of ARID1A wild-type tumors [37]. These results were confirmed by in vivo studies in animals and patient-derived xenograft models, which exhibited an ARID1A mutation-dependent sensitivity to BET inhibition [37]. It has recently been shown that ARID1A and ARID1B are mutually exclusive in SWI/SNF complexes, and that ARID1B is upregulated when ARID1A is inactivated [49]. Loss of ARID1B impairs formation of the SWI/SNF complex, thereby decreasing the viability of ARID1A mutated cells. Therefore, a synthetic lethality relationship exists between the ARID1A and ARID1B genes, making ARID1B a potential therapeutic target [49, 50]. It has been reported that BET inhibitors can induce cell death in ARID1A mutated tumors both in vitro and in vivo by downregulating ARID1B expression in a BRD2-dependent manner [37]. This anti-proliferative activity is BRD2 selective since BRD4 does not associate with the ARID1B promoter [37]. These results were confirmed by in vivo studies in mice bearing CCC tumors and patient-derived xenografts which exhibited an ARID1A mutated phenotype [37].

Another gene found to be oncogenic in ovarian cancer is Neuregulin 1 (NRG1) [51]. NRG1 is a growth factor that contains an epidermal growth factor (EGF)-like domain that stimulates ErbB, mainly ErbB3 (HER3), in ovarian cancer [52]. Interestingly, it has been found that NRG1 is overexpressed in ovarian carcinomas, especially in the serous subtype, and that ErbB3 mediates its function [53]. BET inhibition suppressed NRG1 expression in BRD4-amplified patient-derived xenografts, thereby inhibiting tumor growth [12]. Furthermore, BRD4 and NRG1 expression levels positively correlated with the susceptibility to BET inhibition, indicating that NRG1 serves as a mediator of BET inhibitor activity [12]. Another target of BET inhibition is ALDH1A1, a member of the aldehyde dehydrogenase (ALDH) family [20]. ALDH1A1 overexpression is related to platinum and taxane resistance and a poor outcome in ovarian cancer [54]. ALDH1A1 depletion resensitizes resistant cells to chemotherapy [54]. It has been found that BET inhibitors can efficiently suppress ALDH activity in a dose-dependent manner by disrupting BRD4 occupancy of the ALDH1A1 super-enhancer [20]. ALDH suppression was confirmed in vivo in a JQ1-treated mouse model [20].

BET inhibition and genomic instability

BRCA1/2 germline mutations are identified in 15–17% of HGSC tumors and have been found to be mutually exclusive with CCNE1 mutations, which are found in 30% of HR-proficient tumors. Data extracted from multiple independent cohorts showed that a BRCA mutant-like profile, as defined by Large-scale State Transitions (LST) genomic signature, is mainly characterized by gene alterations on chromosomes 3 and 8, whereas gene amplifications on chromosome 19 are mutually exclusive with BRCA mutations [55, 56]. Indeed, chromosome 3 and 8 amplifications are preferentially found in BRCA mutant-like tumors, while chromosome 19 amplifications, including the CCNE1 (19q12), BRD4 (19p13.1) and PAK4 (19q13.2) genes, are mainly observed in BRCA wild type tumors [56]. Specifically, BRD4 was found to be amplified in 18% of BRCA wild-type tumors and in none of BRCA mutant-like tumors [56]. This observation offers new insight into the treatment of BRCA wild-type HGSC tumors that do not benefit from PARP inhibition via targeting amplified genes, such as CCNE1 and BRD4. Indeed, it has been shown that BET inhibition impairs homologous recombination (HR) capacity and thus the response to DNA damage in BRCA wild-type ovarian cancer cells [15, 18, 57].

BET inhibitors synergize with PARP inhibitors (Olaparib and Rucaparib) in both HR-proficient and HR-deficient ovarian cancer cells [15, 18, 57, 58]. Specifically, BET inhibition disrupts BRD2/3/4 binding to promoter regions of the RAD51 and BRCA1 genes, reducing their transcription in a dose-dependent manner [15, 18, 57]. As a result, the formation of BRCA1 and RAD51 complexes required for HR was impaired, rendering BRCA wild-type cells vulnerable to PARP inhibitors [15, 18]. Synchronous BET and PARP inhibition resulted in enhanced apoptosis as indicated by increased PARP cleavage [18].

In addition to BRCA1 and RAD51, other genes downregulated by BET inhibitors in BRCA wild-type ovarian tumors include WEE1 and topoisomerase IIβ-binding protein 1 (TOPBP1) [57]. Synchronous downregulation of WEE1 and TOPBP1 by BET inhibitors could sensitize BRCA wild-type tumors to PARP inhibition. WEE1 is a tyrosine kinase that acts as a G2-M checkpoint regulator which allows transition to mitosis only if DNA is not substantially damaged [59]. In the presence of DNA damage, WEE1 keeps Cyclin-dependent kinase 1 (CDK1) inactivated via phosphorylation, thereby blocking entry into mitosis [59]. Inhibition of WEE1 promotes the entry into mitosis, which results in death of cancer cells due to the accumulation of epigenetic alterations. Thus, WEE1 inhibition sensitizes tumor cells to DNA damaging agents, like DNA crosslinking agents or topoisomerase inhibitors [59]. This effect is more profound in HGSC cells that are TP53 mutated [59, 60]. TP53 somatic mutations have been identified in approximately all (96%) HGSC tumors according to The Cancer Genome Atlas [60]. TOPBP1 is a protein involved in DNA repair that co-localizes with BRCA1 and mediates RAD51 complex formation in case of DNA damage, thus participating in the HR process [61, 62]. TOPBP1 downregulation was found to sensitize ovarian cancer cells to PARP inhibition through impairing HR [62]. In addition, JQ1 reduced both WEE1 and TOPBP1 expression dose-dependently in BRCA wild-type HGSC tumors in vitro and in vivo by attenuating BRD4 binding to the WEE1 and TOPBP1 promoter sites [57]. As a result, JQ1 synergized with the PARP inhibitor Olaparib and enhanced apoptosis in BRCA wild-type cells and mouse models by inducing G2-M checkpoint impairment and mitotic catastrophe [57]. The same effect was observed in PARP inhibitor resistant cells as well as in BRCA2 mutant ovarian cancer cells resistant to PARP inhibition [57]. This finding represents an alternative mechanism of BRCA wild-type ovarian tumor sensitization to PARP inhibition caused by BET inhibitors, in addition to RAD51 and BRCA1 suppression.

As mentioned above, BET inhibition can attenuate ALDH activity by suppressing ALDH1A1 expression [20]. Sustained PARP inhibition was shown to enhance ALDH activity by upregulating ALDH family genes (ALDH1A2, ALDH1A3, ALDH3A1, ALDH3B2), mainly ALDH1A1 through BRD4 recruitment [58]. However, ALDH1A1 overexpression drives resistance to PARP inhibition by promoting microhomology-mediated end joining (MMEJ) [58]. MMEJ is mediated by the PARP1 protein acting as a scaffold for the recruitment of other MMEJ factors to repair double-strand DNA breaks [63]. ALDH1A1 did not affect homologous recombination, but efficiently enhanced MMEJ, which is normally disrupted by PARP inhibition [58]. BET inhibitors may potentially demolish MMEJ-mediated resistance to PARP inhibition by suppressing ALDH1A1-mediated resistance [18, 20, 58].

BET inhibition and metastasis

BET inhibition decreases ovarian cancer cell dissemination and metastasis [21, 30]. It has been found that downregulation of STAT3 phosphorylation and inactivation of the STAT3 pathway caused by BET inhibition may contribute to this effect [30]. The STAT3 pathway has been shown to be implicated in ovarian tumor progression, while pSTAT3 levels were found to be highly elevated in patient ascites and ascites-derived ovarian cancer cells [64]. Therefore, pSTAT3 suppression induced by BET inhibitors is thought to be associated with a reduced metastatic capacity. It is well-known that ovarian tumorigenesis and metastasis are facilitated by epithelial-to-mesenchymal transition (EMT) through downregulation of the epithelial marker E-cadherin and upregulation of mesenchymal markers such as N-cadherin and the transcription factors Twist and Snail [65]. EMT is mediated by cytokine TGF-β via downstream activation of Twist and Snail transcription [66, 67]. This mesenchymal phenotype is associated with increased chemoresistance, especially platinum-based chemotherapy, which is the gold standard of ovarian cancer treatment [66, 68, 69]. BET inhibition has been found to downregulate the expression of stem cell-related genes in ovarian cancer, such as LIM, HES1 and WNT5A, in vitro and in vivo by disrupting BRD4 association with their respective promoter regions [20]. The expression of the EMT markers fibronectin and ZEB2 was found to be efficiently repressed by BET inhibitors [21]. Moreover, BET inhibitors suppressed the expression of the matrix metalloproteinases MMP-9 and MMP-2 required for cell migration and invasion [21, 30]. It is well-known that MMPs enhance the peritoneal adhesion and metastatic capacities of ovarian tumors through cleavage of the ECM proteins fibronectin and vitronectin [70, 71]. Overall, it has been found that BET inhibition efficiently disrupts EMT in ovarian tumors, thereby attenuating their dissemination and metastasis [20, 21].

BET inhibition and immune responses

It has been reported that BET inhibitors can suppress PD-L1 expression in both tumor and immune cells such as macrophages and dendritic cells [25]. Expression of the D274 gene, which encodes PD-L1, was found to be downregulated time- and dose-dependently by JQ1 at the transcriptional level by impairing BRD4 binding to the CD274 promoter [25]. More importantly, BET inhibition was found to suppress PD-L1 expression in an IFNγ-independent manner, since no changes in IFNγ levels were noted. In an in vivo mouse model, JQ1 downregulated PD-L1 expression in both tumor and immune cells, thereby effectively inhibiting tumor growth [25]. In contrast, an increased CD8⁺ cytotoxic activity was observed in vivo, which mediated the cytotoxic effect of JQ1.

BET combination treatment

BET inhibition and Lapatinib

Anti-HER2 treatment with Lapatinib has been found to exhibit synergistic activity with the BET inhibitor AZD5153 in ovarian cancer cells via inhibition of the NRG1 signaling pathway [12]. This combination treatment was, however, highly toxic in mice. In addition, BET inhibitors have been found to suppress the expression of receptor tyrosine kinases (RTKs) and the activity of their downstream pathways. However, feedback activation of some RTKs often occurs via adaptive reprogramming conferring resistance to BET inhibition-sensitive cells [29]. This resistant cell population is dependent on the reactivated signaling pathway and, thus, vulnerable to its blockade with RTK inhibitors, including Lapatinib.

BET inhibition and cisplatin

BET inhibitors have been found to exhibit a synergistic activity with cisplatin in ovarian cancer cells and murine xenograft models, even in platinum-resistant ones [18, 20, 21, 34]. One possible explanation for this effect is ALDH suppression by BET inhibition, which is normally increased by cisplatin treatment [20, 54]. Another mechanism of enhancing cisplatin efficacy may be downregulation of the anti-apoptotic proteins BCL-2 and Survivin [21]. Specifically, it has been shown that BCL-2 inhibition sensitizes chemo-resistant ovarian cancer cells to cisplatin treatment [72]. Finally, cisplatin is a well-known cytotoxic agent that induces double-strand DNA breaks. BET inhibitors sensitize HR-proficient ovarian tumor cells to DNA damage by reducing BRCA1 and RAD51 expression and thus impairing their HR capacity [15, 18]. This effect was exploited to reduce the cytotoxic activity threshold of cisplatin treatment [18].

BET inhibition and PARP inhibitors

Combination treatment with PARP inhibitors such as Olaparib and multiple BET inhibitors (JQ1, I-BET762, OTX015) has been proven to be more effective than either therapy alone in both HR-deficient and HR-proficient cells [15, 57]. As mentioned above, BET inhibition impairs homologous recombination (HR) in BRCA wild-type cells by reducing the expression of BRCA1 and RAD51. In addition to this, BET inhibitors downregulate WEE1 and TOPBP1 expression and reduce ALDH1A1-mediated reactivation of MMEJ, thereby sensitizing BRCA wild-type cells to PARP inhibition [57, 59]. Combination treatment with JQ1 and Olaparib has been applied in vivo in an ovarian cancer xenografted mouse model, which showed decreased tumor progression and ascites development [15, 57].

BET inhibition and MEK inhibitors

BET inhibitors have shown synergistic effects when combined with MEK inhibitors (Trametinib, PD0325901) both in vitro and in vivo [26]. Trametinib blocked feedback activation of the MAPK signaling pathway induced by sustained BET inhibition, which limits the antitumor activity of BET inhibitors as previously stated [26]. Inversely, BET inhibitors reversed resistance to Trametinib caused by feedback reactivation of RTKs in neurofibromin 1 (NF1) deficient ovarian cancer cells [73]. NF1 is a RAS GTPase activating protein that acts by suppressing RAS pathways [74]. Loss of NF1, which is observed in 12% of epithelial ovarian tumors, results in the relief of upstream RAS negative feedback and MEK/ERK pathway reactivation [41, 74, 75]. NF1 deficiency is exploited by Trametinib treatment, which targets the activated MEK/ERK pathway. However, sustained Trametinib treatment led to FOSL1 degradation and reactivation of multiple RTKs and their downstream pathways, including the PI3K-AKT, JAK/STAT and MEK/ERK signaling pathways [73]. This RTK reprogramming provided Trametinib-treated cells with resistance and reduced growth inhibition achieved by MEK inhibition [73]. It was shown that BRD2 and BRD4 are required for the transcription of a wide range of RTKs, including DDR1, EGFR, PDGFRB, ERBB2, ERBB3 and IGF1R. In this context, BET inhibitors synergized with Trametinib in NF1-deficient ovarian cancer cells by attenuating Trametinib-induced expression of the aforementioned RTKs [73]. It is worth mentioning that NF1-proficient cells displayed no benefit from this combination treatment [73]. Overall, cotreatment with JQ1 and Trametinib increased ovarian cancer cell apoptosis as indicated by elevated cleaved PARP and Caspase 3 levels, while single agent treatment only had a cytostatic effect [26, 73].

BET inhibition and Aurora kinase inhibitors

Drug screening led to the identification of Aurora kinase inhibitors as potential agents for BET inhibitor combination treatment in ovarian cancer [26]. Indeed, Aurora kinases A and B (AURKA and AURKB) were shown to mediate the BET inhibitor response in triple negative breast cancer cells by inducing polyploidy and, thus, apoptosis or senescence [32]. As yet, combination treatment with BET inhibitors remains to be tested in ovarian cancer.

BET inhibition and Ponatinib

Ponatinib is a multi-kinase inhibitor of BCR/ABL variants, but also other tyrosine kinases such as fibroblast growth factor (FGF), vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF) [76]. Combined BET inhibitor and Ponatinib treatment resulted in increased ovarian cancer cell apoptosis compared to either therapy alone by reducing MYC expression [24].

BET inhibitor resistance

Resistance to BET inhibition has been reported in ovarian cancer and could potentially limit their clinical application. One way of acquired resistance to BET inhibitors is the induction of autophagy via AKT/mTOR pathway inhibition [23]. Autophagy is a catabolic process through which cytoplasmic materials are transferred to autophagosomes for destruction [77]. Autophagy is involved in multidrug resistance by blocking drug-induced apoptosis [78]. It has been shown that excessive apoptosis can provide a protective mechanism against drug cytotoxicity via abolishing damaged cell components and recycling amino acids, fatty acids and nucleotides to restore cell homeostasis and viability [78]. BET inhibition has been found to induce autophagic activity in JQ1-resistant ovarian cancer cells, as indicated by increased LC3-II lipidation, Beclin-1 and ATG5 expression levels and decreased P62/SQSTM1 expression levels [23]. This effect was not observed in JQ1-sensitive cells, supporting an involvement in BET inhibitor resistance. This induction of autophagy seems to emerge from AKT/mTOR pathway inhibition by JQ1, since phosphorylation of AKT, mTOR and p70S6K was found to be reduced in resistant cells compared to sensitive cells [23]. Inhibition of autophagy by the autophagic inhibitors 3-MA and chloroquine in JQ1-treated resistant cells and in murine models led to increased apoptosis [23]. Another mechanism of both intrinsic and acquired resistance to BET inhibition is the wide activation of RTKs and their downstream signaling pathways, a phenomenon referred as “adaptive kinome reprogramming” [26, 29]. It has been shown that MEK or AKT inhibition can be circumvented by rebound activation of alternative kinase pathways, which bypass the initial inhibitory effect of treatment [79]. In the same context, ovarian cancer cells inherently resistant to BET inhibition were found to display a dose-dependent increase in expression of RTKs, namely FGFR1–3, EGFR and IGF1R, and their downstream AKT and MEK/ERK signaling pathways [26, 29]. In contrast, these pathways were efficiently inhibited in JQ1-sensitive ovarian cancer cells through reduced expression or phosphorylation of multiple RTKs [29]. In addition to intrinsic BET inhibitor resistance, ovarian cancer cells continuously exposed to BET inhibition may develop acquired resistance via upregulating or phosphorylating RTKs, including FGFR1–4, IGF1R, EGFR, PDGFRA, AKT and TGFBR1, or increasing the expression of receptor ligands, such as PDGFA and TGFA [26, 29]. Concordantly, activation of the MEK/ERK, PI3K/AKT1 and JAK/STAT pathways is commonly seen in BET inhibitor acquired resistance. Combination treatment with RTK inhibitors, including the pan-FGFR inhibitor AZD4547, the PI3K/mTOR inhibitor GDC-0941 and the MEK inhibitor Trametinib, restored the sensitivity to JQ1 in resistant cells, underscoring this notion [26, 29].

Discussion

Advanced ovarian cancer is characterized by a significant recurrence rate. Especially for patients with platinum resistant recurrence of the disease, available therapies are ineffective and the development of novel therapeutic options is an unmet medical need. The identification of BRCA1/2 mutations led to the development of PARP inhibitors, establishing three recently FDA approved drugs (Olaparib, Rucaparib and Niraparib) for ovarian cancer treatment. However, PARP inhibitor resistance is reported in 40–70% of BRCA1/2-mutated ovarian cancers, while PARP inhibitor sensitivity is closely related to platinum response [80]. A further understanding of the significance of post translational modifications has paved the way for new molecular therapies. BET activity has been shown to be highly activated in cancer, regulating cell proliferation, angiogenesis, dissemination and metastasis [81]. BRD4 amplification is commonly observed in ovarian tumors, offering a promising target for molecular therapies [12, 14]. BET inhibitors are small molecules that bind competitively to the acetyl-lysine binding sites of BET-family bromodomains, thereby displacing BET proteins from chromatin [5]. JQ1 is the most extensively studied BET inhibitor, and exhibits a high potency and specificity against BRD4 family member [5]. Antitumor efficacy of BET inhibitors, including JQ1, has been validated in various solid and hematological malignancies in preclinical studies. The safety profile and clinical activity of BET inhibitors are currently being investigated in Phase I/II clinical trials for several cancer types, including ovarian cancer. JQ1 is not being tested in clinical trials due to its short half-life. Clinical studies exploring BET inhibitors in ovarian cancer and other solid tumors, including ovarian subtypes, are listed in Table 1. Data extracted from these trials show a favorable pharmacokinetic profile with a reversible toxicity that should be managed. The main toxicities reported in preclinical studies include Grade 3 and 4 thrombocytopenia, gastrointestinal disorders (diarrhea, nausea, vomiting), and decreased appetite and anemia [82–85]. These toxicities were, however, reversible after treatment discontinuation.

Table 1.

Clinical trials of BET inhibitors in ovarian cancer and other tumors

STUDY	PHASE	DRUG	DESIGN	STATUS	RESULTS
NCT02711137	I/II	INCB057643 ± Gemcitabine Paclitaxel Rucaparib Abiraterone Ruxolitinib Azacitidine	Advanced solid tumors and hematologic malignancies (CRPC, BC, HGSC, CRC, glioblastoma multiforme, Ewing sarcoma, pancreatic adenocarcinoma, AML, MDS)	Terminated (due to safety issues)	Has results
NCT02419417	I/II	BMS-986158 Nivolumab	SCLC, TNBC, Ovarian cancer BRCAwt	Recruiting	No results
NCT02431260	I/II	INCB054329	CRPC, BC, CRC, ovarian cancer, pancreatic cancer, AML, MDS, lymphoma	Terminated (due to PK variability)	Has results
NCT03292172	I	RO6870810 Atezolizumab	Advanced ovarian cancer, TNBC	Terminated (due to portfolio prioritization)	No results
NCT01987362	I	RO6870810	Advanced solid tumors	Completed	No results
NCT02391480	I	Mivebresib (ABBV-075)	Advanced solid tumors and hematologic malignancies	Completed	Has results
NCT02369029	I	BAY1238097	Advanced malignancies	Completed	Has results
NCT02259114	I	Birabresib (MK-8628)	Advanced solid tumors	Completed	Has results
NCT02698176	I	Birabresib (MK-8628)	NUT midline carcinoma (NMC), TNBC, NSCLC, CRPC	Terminated (due to limited efficacy and not due to safety reasons)	Has results
NCT01587703	I	Molibresib (GSK525762/ IBET762)	NUT midline carcinoma (NMC), TNBC, ER+ BC, SCLC, CRPC, GIST	Completed	Has results
NCT03266159	II	Molibresib (GSK525762) Trametinib	SCLC, Ras-mutated CRC, Ras-mutated NSCLC, Ras pathway activated solid tumors	Withdrawn (to fully evaluate impact of changing practice in target population)	No results
NCT02516553	I	BI 894999	SCLC, CRPC, CRC, NUT midline carcinoma	Recruiting	No results
NCT02683395	I	PLX51107	Advanced solid tumors and hematologic malignancies	Terminated (business decision)	No results
NCT02630251	I	GSK2820151	Advanced solid tumors	Terminated (due to development of another BET inhibitor GSK525762)	Has results
NCT03035591	I/II	ODM-207	Advanced solid tumors	Completed	No results

BET inhibitors mainly exhibited a cytostatic activity rather than a cytotoxic one, inducing anti-proliferative but not pro-apoptotic effects in ovarian cancer cells [13, 22, 24, 26]. However, combination treatment with BET inhibitors and a wide range of cytotoxic agents has proven to be efficient in ovarian cancer. This characteristic is expected to lead to their incorporation in combination treatment strategies, rather than their application as mono-therapeutic agents. The synergistic activity of BET and PARP inhibitors was profound in ovarian tumors regardless of BRCA mutation status [15, 57]. Indeed, it has been shown that reactivation of BRCA1/2 genes and subsequent restoration of HR is one of the mechanisms underlying resistance to PARP inhibition [86]. This effect is currently under investigation in clinical trials. An ongoing Phase II study is exploring the combination of the BET inhibitor ZEN003694 and the PARP inhibitor Talazoparib in triple negative breast cancer patients (NCT03901469) [87]. Another Phase I/II study was designed to evaluate combination of the BET inhibitor INCB057643 with standard of care agents, including the PARP inhibitor Rucaparib in ovarian cancer (NCT02711137) [88]. However, this latter study was prematurely terminated due to safety issues. Overall, PARP and BET inhibitor combination treatment may be applied in all HGSC patients regardless BRCA mutation status and may potentially overcome PARP inhibitor resistance. A subgroup of ovarian tumors shares KRAS and BRAF mutations that lead to activation of the MEK/ERK pathway [89]. These mutations occur typically in low grade serous, mucinous and endometrioid ovarian carcinomas and may potentially be exploited for targeted therapies. Interestingly, BET inhibitors were found to synergize with the MEK inhibitor Trametinib in ovarian tumor cells, especially in NF1-deficient cells [26, 73]. Combination treatment of the BET inhibitor I-BET762 and Trametinib is currently being evaluated in a Phase II trial involving RAS-mutated solid tumors, such as SCLC, RAS-mutated colorectal cancer (RMCRC), RAS-mutated non-small cell lung cancer (RM NSCLC), RAS-mutated pancreatic adenocarcinoma (RMPAC) and others (NCT03266159) [90]. Combination treatment with other agents that have shown efficacy in preclinical studies remain promising for in-human studies. Based on preclinical studies, the Aurora kinase A inhibitor Alisertib (MLN8237) as well as the anti-HER2 inhibitor Lapatinib offer potential options for combination treatment with BET inhibitors [12, 26].

BET inhibition has been found to suppress PD-L1 expression in both tumor cells and myeloid macrophages and dendritic cells in an IFNγ-independent manner [25]. PD-L1 is a negative effector of the immune response that enables tumor cells to escape T cell immunity. High PD-L1 levels on tumor cells are associated with a poor prognosis in ovarian cancer [91]. Suppression of PD-L1 expression by BET inhibitors on tumor cells may contribute to T cell tumor recognition and, thus, antitumor immune activity. Importantly, BET inhibitors increased CD8⁺ cytotoxic T cell activity in vivo, thereby enhancing the antitumor immune response [25]. However, suppression of PD-L1 could potentially alter the tumor response to immunotherapy. A Phase I study addressing combination treatment with the BET inhibitor R06870810 and the anti-PD-L1 antibody Atezolizumab in triple negative breast cancer and advanced ovarian cancer has been initiated (NCT03292172) [92]. Another Phase I/II study explores the efficacy of the BET inhibitor BMS-986158 as monotherapy or in combination with Nivolumab in selected advanced solid tumors, including BRCA wild-type serous ovarian cancer (NCT02419417) [93]. The efficacy of synchronous BET inhibition and immunotherapy regimens remains to be evaluated in these ongoing studies. Current Phase I clinical studies are focusing on both BET inhibitor safety and its pharmacokinetic profile. Although the ultimate clinical activity of BET inhibitors in ovarian cancer as single agents or in combination with conventional treatment regimens remains to be determined, the molecular events and epigenetic modifications induced by BET inhibition offer a promising rationale for their clinical efficacy.

Perspectives

Our increasing understanding of epigenetic processes has led to the development of several novel interfering compounds. Bivalent inhibitors of BRD bromodomains have been designed to engage both BRDs (BRD1 and BRD2) simultaneously, achieving a more durable inhibition of BET proteins. AZD5153, MT1, biBET and MS645 are novel bivalent BRD4 inhibitors that exhibit higher potency and antitumor activity levels than their monovalent predecessors [94–97]. Dual inhibition of BRDs offer a higher affinity and increased efficacy in BET inhibition. Furthermore, new technologies have led to the development of molecules that indirectly inhibit BET activity without directly interacting with BET bromodomains. Proteolysis targeting chimeric (PROTAC) molecules are novel agents that bind their targeted proteins and link them to E3 ubiquitin ligase, leading to their rapid and durable degradation via the proteasome [98, 99]. BET PROTACs are based on the chemical structure of BET inhibitors so as to effectively bind to BET proteins. dBET1 and ARV-825 are BET-PROTACs designed based on JQ1 and Thalidomide, while BETd-246 uses BETi-211 as BET inhibitor moiety [100] Two BET-PROTACs, MZ1 and ARV-825, based on the JQ1 and OTX015 BET inhibitor structure, respectively, have been tested in ovarian cancer cells [101]. Both agents induced apoptosis in these cells, but no synergistic activity was observed when combined with docetaxel, cisplatin or Olaparib [101]. Another recent innovation was the design of dual BET and HDAC inhibitor molecules that target the epigenome [102–104]. TW09 is a novel BET/HDAC inhibitor based on JQ1 that induces apoptosis in rhabdomyosarcoma cells [102]. In an attempt to achieve a more universal kinase downregulation, compounds inhibiting BET and multiple tyrosine kinases simultaneously have been developed [105]. A dual BET/kinase inhibitor has e.g. been developed based on the JAK2 inhibitor TG101348 structure which synchronously inhibits BRD4 and JAK2, FLT3, RET and ROS1 kinases [105]. Also, a novel dual BET and Polo-like kinase inhibitor WNY0824 has verey recently been launched that was found to effectively downregulate MYC in preclinical studies [106]. BI2536 is a Polo-like 1 kinase inhibitor with a great affinity towards BRD4 and BRDT [107]. In the same context, the dual PI3K and BET inhibitor SF2523 was found to inhibit tumor growth and metastasis in preclinical models [108]. BET inhibition has opened up new horizons in the area of drug discovery, resulting in the development of several epigenetic factors with a wide application in solid tumors. This fast-growing area of drug development has led, and will continue to lead to the initiation of clinical trials for the in-human exploration of new treatment options for solid tumors, in particular ovarian cancer.

Funding

For this work no specific grants from funding agencies in the public, commercial or not-for-profit sectors were received.

Compliance with ethical standards

Declaration of interests

ML has received honoraria from Roche, Astra Zeneca, Astellas, MSD, Janssen, Bristol-Myers-Squibb and IPSEN. KK has received honoraria from Roche, BMS, MSD and IPSEN. MAD has received honoraria through participation in advisory boards from Amgen, Bristol-Myers-Squibb, Celgene, Janssen and Takeda. FZ has received honoraria for lectures and has served as an advisor for Astra-Zeneca, Daiichi, Eli-Lilly, Merck, Novartis, Pfizer and Roche. The remaining authors declare no conflict of interest.