BRD4 functions as an epigenetic reader and plays a crucial role in regulating transcription and genome stability. Dysregulation of BRD4 is frequently observed in various human cancers. However, the molecular details of BRD4 regulation remain largely unknown. Here, we report that PRMT2- and PRMT4-mediated arginine methylation is pivotal for BRD4 functions on transcription, DNA repair, and tumor growth. Specifically, PRMT2/4 interacts with and methylates BRD4 at R179, R181, and R183. This arginine methylation selectively controls a transcriptional program by promoting BRD4 recruitment to acetylated histones/chromatin. Moreover, BRD4 arginine methylation is induced by DNA damage and thereby promotes its binding to chromatin for DNA repair. Deficiency in BRD4 arginine methylation significantly suppresses tumor growth and sensitizes cells to BET inhibitors and DNA damaging agents. Therefore, our findings reveal an arginine methylation–dependent regulatory mechanism of BRD4 and highlight targeting PRMT2/4 for better antitumor effect of BET inhibitors and DNA damaging agents.

Cancer development is influenced by hereditary mutations, somatic mutations due to random errors in DNA replication, or external factors. It remains unclear how distinct cell-intrinsic and -extrinsic factors affect oncogenesis within the same tissue type. We investigated murine soft-tissue sarcomas generated by oncogenic alterations (Kras ^G12D activation and p53 deletion), carcinogens (3-methylcholanthrene [MCA] or ionizing radiation), and both factors in a potentially novel model (MCA plus p53 deletion). Whole-exome sequencing demonstrated distinct mutational signatures in individual sarcoma cohorts. MCA-induced sarcomas exhibited high mutational burden and predominantly G-to-T transversions, while radiation-induced sarcomas exhibited low mutational burden and a distinct genetic signature characterized by C-to-T transitions. The insertion-deletion/substitution ratio and number of gene copy number variations were high for radiation-induced sarcomas. MCA-induced tumors generated on a p53-deficient background showed the highest genomic instability. MCA-induced sarcomas harbored mutations in putative cancer driver genes that regulate MAPK signaling ( Kras and Nf1 ) and the Hippo pathway ( Fat1 and Fat4 ). In contrast, radiation-induced sarcomas and Kras ^G12D p53 ^–/– sarcomas did not harbor recurrent oncogenic mutations; rather, they exhibited amplifications of specific oncogenes: Kras and Myc in Kras ^G12D p53 ^–/– sarcomas and Met and Yap1 for radiation-induced sarcomas. These results reveal that different initiating events drive oncogenesis through distinct mechanisms.

Autophagy, particularly with BECN1, has paradoxically been highlighted as tumor promoting in Ras-driven cancers, but potentially tumor suppressing in breast and ovarian cancers. However, studying the specific role of BECN1 at the genetic level is complicated due to its genomic proximity to BRCA1 on both human (chromosome 17) and murine (chromosome 11) genomes. In human breast and ovarian cancers, the monoallelic deletion of these genes is often co-occurring. To investigate the potential tumor suppressor roles of two of the most commonly deleted autophagy genes in ovarian cancer, BECN1 and MAP1LC3B were knocked-down in atypical (BECN1+/+ and MAP1LC3B+/+) ovarian cancer cells. Ultra-performance liquid chromatography mass-spectrometry metabolomics revealed reduced levels of acetyl-CoA which corresponded with elevated levels of glycerophospholipids and sphingolipids. Migration rates of ovarian cancer cells were increased upon autophagy gene knockdown. Genomic instability was increased, resulting in copy-number alteration patterns which mimicked high grade serous ovarian cancer. We further investigated the causal role of Becn1 haploinsufficiency for oncogenesis in a MISIIR SV40 large T antigen driven spontaneous ovarian cancer mouse model. Tumors were evident earlier among the Becn1+/- mice, and this correlated with an increase in copy-number alterations per chromosome in the Becn1+/- tumors. The results support monoallelic loss of BECN1 as permissive for tumor initiation and potentiating for genomic instability in ovarian cancer. Life requires an excellent trash and recycling system. One system mammalian cells rely upon is called autophagy. Autophagy clears cells of larger forms of debris and is critical for normal cell function. We previously found that ovarian cancer cells are unable to execute normal cellular recycling during periods of stress, due to losses in core autophagy genes. Since autophagy has previously been implicated as a cancer cell survival factor, we directly investigated the impact of autophagy gene loss in ovarian cancer models. We discovered a reduction in autophagy proteins disabled normal genome quality control during cell division, leading to genomic instability. We document here for the first time whether reducing an autophagy gene in mice influences ovarian tumor formation. We found earlier ovarian tumors in autophagy deficient mice compared with control mice. We conclude that serous ovarian carcinomas, which rank among the most genetically altered cancers, gain their hallmark genomic instability in part by losing autophagy genes.

Background: Genomic instability and chemoresistance can arise in cancer due to a unique form of plasticity: that of polyploid giant cancer cells (PGCCs). These cells form under the stress of chemotherapy and have higher than diploid chromosome content. PGCCs are able to then repopulate tumors through an asymmetric daughter cell budding process. PGCCs have been observed in ovarian cancer histology, including the deadly and common form high-grade serous ovarian carcinoma (HGSC). We previously discovered that drugs which disrupt the cellular recycling process of autophagy are uniquely efficacious in pre-clinical HGSC models. While autophagy induction has been associated with PGCCs, it has never been previously investigated if autophagy modulation interacts with the PGCC life cycle and this form of tumor cell plasticity. Methods: CAOV3 and OVCAR3 ovarian cancer cell lines were treated with carboplatin or docetaxel to induce PGCC formation. Microscopy was used to characterize and quantify PGCCs formed by chemotherapy. Two clinically available drugs that inhibit autophagy, hydroxychloroquine and nelfinavir, and a clinically available activator of autophagy, rapamycin, were employed to test the effect of these autophagy modulators on PGCC induction and subsequent colony formation from PGCCs. Crystal violet-stained colony formation assays were used to quantify the tumor-repopulating stage of the PGCC life cycle. Results: Autophagy inhibitors did not prevent PGCC formation in OVCAR3 or CAOV3 cells. Rapamycin did not induce PGCC formation on its own nor did it exacerbate PGCC formation by chemotherapy. However, hydroxychloroquine prevented efficient colony formation in CAOV3 PGCCs induced by carboplatin (27% inhibition) or docetaxel (41% inhibition), as well as in OVCAR3 cells (95% and 77%, respectively). Nelfinavir similarly prevented colony formation in CAOV3 PGCCs induced by carboplatin (64% inhibition) or docetaxel (94% inhibition) as well as in OVCAR3 cells (89% and 80%, respectively). Rapamycin surprisingly also prevented PGCC colony outgrowth (52–84% inhibition). Conclusions: While the autophagy previously observed to correlate with PGCC formation is unlikely necessary for PGCCs to form, autophagy modulating drugs severely impair the ability of HGSC PGCCs to form colonies. Clinical trials which utilize hydroxychloroquine, nelfinavir, and/or rapamycin after chemotherapy may be of future interest.

Unusually high aneuploidy is a hallmark of epithelial serous ovarian cancer (SOC). Previous analyses have focused on aneuploidy on average across all tumor cells. With the expansion of single-cell sequencing technologies, however, an analysis of copy number heterogeneity cell-to-cell is now technically feasible. Here, we describe an analysis of single-cell RNA sequencing (scRNA-seq) data to infer arm-level aneuploidy in individual serous ovarian cancer cells. By first clustering high-quality sequenced epithelial versus non-epithelial cells, high-confidence tumor cell populations were identified. InferCNV was used to predict segmented copy-number alterations (CNAs), which were then used to determine arm-level aneuploidy at the single-cell level. Control comparisons of normal cells to normal cells showed zero arm-level aneuploidy, whereas a median of four aneuploid events were detectable in cancer cells. A heterogeneity analysis of high-grade tumor cells compared to low-grade tumor cells showed similar levels of cell-to-cell variation between cancer grades. Metastatic tumors potentially showed selection pressure with reduced cell-to-cell variation compared to cells from primary tumors. Minor cell populations with CNAs similar to metastatic cells were identified within the matched primary tumors. Taken together, these results provide a minimum estimate for single-cell aneuploidy in serous ovarian cancer and demonstrate the utility of single-cell sequencing for CNA analysis.

Polyploid giant cancer cells (PGCC) are frequently detected in tumors and are increasingly recognized for their roles in chromosomal instability and associated genome evolution that leads to cancer recurrence. We previously reported that therapy stress promotes polyploidy, and that acid ceramidase plays a role in depolyploidization. In this study, we used an RNA-seq approach to gain a better understanding of the underlying transcriptomic changes that occur as cancer cells progress through polyploidization and depolyploidization. Our results revealed gene signatures that are associated with disease-free and/or overall survival in several cancers and identified the cell cycle inhibitor CDKN1A/p21 as the major hub in PGCC and early progeny. Increased expression of p21 in PGCC was limited to the cytoplasm. We previously demonstrated that the sphingolipid enzyme acid ceramidase is dispensable for polyploidization upon therapy stress but plays a crucial role in depolyploidization. The current study demonstrates that treatment of cells with ceramide is not sufficient for p53-independent induction of p21 and that knockdown of acid ceramidase, which hydrolyzes ceramide, does not interfere with upregulation of p21. In contrast, blocking the expression of p21 with UC2288 prevented the induction of acid ceramidase and inhibited both the formation of PGCC from parental cells as well as the generation of progeny from PGCC. Taken together, our data suggest that p21 functions upstream of acid ceramidase and plays an important role in polyploidization and depolyploidization.