Identification and annotation of functional silencers
2Department of Human Molecular Genetics and Biochemistry, Sackler School of Medicine, Tel Aviv University, Israel
3Sagol School of Neuroscience, Tel Aviv University, Israel
Regulatory elements (REs) that control transcription, such as enhancers and promoters, have been studied extensively over the past decade. In contrast, silencers received less attention, mainly because it has been harder to validate them experimentally. Furthermore, recent studies predicting putative silencers based on epigenomic patterns, did not focus on bona fide functional silences, i.e., DNA elements that decrease their target gene expression. Thus, computational prediction of functional silencers has been limited.
Recent studies indicate that silencers are bound by the PRC2 complex, which mark these regions with the H3K27me3 repressive histone mark, a reaction that is catalyzed by the EZH2 subunit of the complex.
As a preliminary analysis, we focused on RE-promoter links predicted by the FOCS algorithm (Hait et al., 2018) whose distal REs overlap EZH2 ChIP-seq peaks from ENCODE. As expected, H3K27ac and EZH2 signals in the distal REs show, respectively, positive and negative correlation with the expression of their FOCS-predicted target genes. Methylation data from Roadmap Epigenomics consortium shows, as expected, that functional enhancers and silencers are significantly hypo-methylated compared to non-functional REs in the cell types in which they show their activity. Logistic regression analysis using transcription factor binding sites identified many known repressors, such as REST, ZBTB33, MBD2 and MECP2, which are involved in DNA methylation and deacetylation. Based on these insights, we are building a classification model that uses linked REs and employs their methylation status and DNA sequence as features to pinpoint characteristics of functional silencers.