A MULTI-OMICS APPROACH TO DISSECT ABERRANT EPIGENETIC MECHANISMS IN TRIPLE NEGATIVE BREAST CANCER

In recent years, experimental evidence has shown that epigenetic changes including DNA methylation and histone post-translational modifications (PTMs) play a role in cancer development and progression. Histone PTMs have been largely studied in the context of cancer and have been described as hallmarks of cancer and markers of diagnosis and prognosis. In addition, the histone modifying enzymes involved in the recognition, deposition, and removal of histone PTMs are often altered and overexpressed in cancer. Therefore, the identification of these enzymes as potential targets for epigenetic therapies is promising for the development of specific treatments. In this project, we focused on the molecular characterization of epigenetic mechanisms in Triple Negative breast cancer (TNBC), which is the most aggressive type of breast cancer, characterized by a high proliferation rate, and associated with poorer patient outcomes compared to other breast cancer subtypes, particularly Luminal A subtype. Because of the absence of specific molecular markers, there are no targeted therapies available for TNBC and only few treatment options are in use, including standard chemotherapy, surgery, and radiation therapy. Therefore, there is an urgent need to identify specific markers for this type of tumor as well as novel potential targets. Preliminary data obtained by our group, by mass spectrometry (MS), highlighted an epigenetic signature distinguishing the TNBC subtype from the other breast cancer subtypes. We confirmed the same pattern in an additional set of samples and we selected three of the identified changes (H3K4me2, H3K9me3, and H4K20me3) for a follow-up mechanistic investigation, using two parallel approaches. We first performed a multi-OMICs analysis to investigate the genomic distribution of H3K4me2 in TNBCs and its effects at the expression level. By integration of ChIP-seq and RNA-seq data from a set of fresh frozen clinical samples we found that the localization of H3K4me2 at promoters and super-enhancer regions results in higher levels of expression and seems to drive expression of genes associated to the TNBC phenotype. By the integration of data from clinical samples and the model breast cancer cell line MDA-MB-231 we identified a set of genes that are candidate TN-specific genes whose expression is regulated by the unique presence of H3K4me2 in their cis-regulatory regions. We then took advantage of a CRISPRi epigenetic tool to genetically manipulate the expression of these genes through the LSD1-mediated removal of H3K4me2 at these regulatory regions. The result obtained revealed a direct correlation between the presence of H3K4me2 at promoters and gene expression. Then, we investigated the possible upstream causes of histone PTM deposition focusing on H3K9me3 modification, which we found to be overexpressed in TN versus Luminal A tumors. By correlating the levels of histone PTMs in TNBC samples with the expression levels of the histone modifying enzymes by RNA-seq analysis, we found a strong correlation between the increase in H3K9me3 levels and the over-expression of the H3K9-selective methyltransferases SETDB1, SUV39H1, and SUV39H2 in TN subtype. Upon pharmacological and genetic manipulation of these methyltransferases, we identified a strong effect at phenotypic level. In addition, the knockdown of SUV39H1 and SUV39H2, but not SETDB1, revealed a decrease in H3K9me3 levels paralleled by an increase in H4K20me3 levels, suggesting a potential mechanism of crosstalk. Hence, we hypothesized that upon depletion of SUV39H1 and SUV39H2 a re-localization of both H3K9me3 and H4K20me3 may occur. To assess this hypothesis, we used immunofluorescence to analyze the localization of H3K9me3 and H4K20me3 on mitotic chromosomes and correlated these data with ChIP-seq analysis of these two epigenetic marks in a set of Luminal A and TN cell lines. At basal condition, we observed a more pronounced localization of these histone marks at centromeric regions in Luminal A cells compared to TN. Interestingly the same analysis carried out in TN cell lines depleted for SUV39H1 and SUV39H2 revealed a re-localization at centromeres of the two PTMs, which acquired a distribution much more similar to that of Luminal A cell lines. Additional experiments are needed to better characterize the molecular mechanism at the base of this potential crosstalk. Overall the results obtained indicate that our integrative approach in the analysis of aberrant epigenetic features, based on the quantitative MS analysis, and the integration of ChIP-seq and RNA-seq data can suggest novel molecular targets in the TNBC subtype, which we studied through pharmacological and genetic modulation.