Abstract 4186: An integrative genomic pipeline to target the NCOR1 cistrome with precision

Abstract

Abstract NCOR1 is a co-regulator of multiple transcription factors (TFs) and is mutated in major cancers. We mined publically available data-sets to reveal the diversity, function and druggability of the NCOR1 cistrome. From ENCODE, in K562 chronic myelogenous leukemia cells, NCOR1 ChIP-Seq data were integrated with H3K9me3 (a functional consequence), CpG methylation levels and open and closed chromatin states in a matrix of 45 interactions that were combined with RNA-Seq data. Boot-strapping approaches then revealed significantly affected gene networks. Surprisingly, only the combination of NCOR1, H3K9me3 and an intermediate level of CpG methylation, in open chromatin, significantly associated with a network of 212 repressed genes (lo_gene). By contrast, 15 different NCOR1 combinatorial binding patterns associated with significantly elevated gene expression. For example, NCOR1 and low CpG methylation associated with a network of 908 significantly elevated genes (hi_gene). A simple linear regression model was built on the correlation of NCOR1 and 11 different histone modifications with RNA-seq values, across 81 bins spanning gene loci, to reveal the optimal bin predictor of gene expression. NCOR1 binding at the TSS significantly and positively correlated with gene expression, and mirrored the patterns of the activating histone mark H3K4me3. Transcription factor and GO enrichment in the lo_gene and hi_gene networks revealed these significant associations with different TFs and GO terms. In the lo_genes network the most significantly associated TF was regulatory factor X (RFX1), (FDR=1e-4), whereas in the hi_gene network it was Nuclear factor (erythroid-derived 2)-like 2 (NFE2L2), (FDR=2.7e-23). GO analyses revealed these cohorts were most associated with skeletal development (FDR, 2e-3) and transcription regulation (FDR 1.8e-2) terms respectively, underscoring their diverse functions. Mining the NCI60 database, combined with bootstrap approaches, identified that the NCOR1 hi_gene (upregulated) network was most significantly enriched for genes associated with sensitivity towards HDAC inhibitors, including vorinostat, compared to other drug classes including RTKi. Finally, mining a GEO microarray data set of 59 CML patients revealed that significantly up-regulated genes in Imatinib resistant patients were significantly enriched for targets associated with vorinostat sensitivity in the NCI60. Together this integration pipeline in R, from ENCODE, GEO and NCI60, forms a highly flexible pipeline that can be applied to any DNA binding protein or epigenetic state as a prelude to drug redeployment and/or precision medicine approaches. Specifically, these data suggest that cancer related functions of NCOR1 are surprisingly most associated with sustained gene expression, and these targets, both in CML cell lines and patients, associate with sensitivity to clinically approved HDAC inhibitors. Citation Format: Mark D. Long, Patrick van den Berg, Moray J. Campbell, Prashant K. Singh, Sebastiano Battaglia. An integrative genomic pipeline to target the NCOR1 cistrome with precision. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4186. doi:10.1158/1538-7445.AM2014-4186