Chromatin Deregulation in Cancer

Abstract

During cancer development, committed cells are reprogrammed to malignant cell states. Genetic and epigenetic changes to chromatin complexes contribute to the reprogramming by disrupting normal cell identity and activating oncogenic phenotypes. How the complexes regulate these phenotypes, however, is often unclear: perturbing chromatin modifying complexes affects restricted—and often context dependent—sets of genes despite causing global changes to the chromatin state. To understand how general chromatin modifiers regulate specific cancer phenotypes, we altered chromatin remodeling complexes linked to cancer development and studied the relationship between their effects on chromatin and their control of gene expression. We first mapped chromatin accessibility in two dissimilar cancer types, small cell lung cancer(SCLC) and acute myeloid leukemia(AML), to measure the spectrum of regulatory changes in cancer and identify factors underlying the oncogenic chromatin state. We observed frequent expansion in chromatin accessibility at cell type specific DHSs that contain low affinity motifs of diverse lineage specifying transcription factors (TFs). We next inhibited a histone demethylase, LSD1, in a panel of SCLC and AML lines cancer lines to study how different chromatin states direct the effects of a chromatin modifier. Despite a shared mechanism of LSD1 recruitment in both cancers, LSD1 inhibition affected cancer specific expression programs due to the preferential response of labile regulatory elements whose activity was most tightly linked to TF dosage. Finally, we tested the effect of a recurrent somatic mutation in a chromatin remodeler by reactivating the BAF complex member SMARCA4 in a SMARCA4-null lung adenocarcinoma cell line. We observed low affinity TF binding sites were preferentially affected by SMARCA4 reactivation and the domain level organization of such sites directs the effects of global chromatin remodeling to key developmental regulators. Across experiments, we find a common class of regulatory elements is affected by multiple unrelated regulatory complexes. The affected elements contain homotypic, low affinity binding motifs, which make them inherently sensitive to variation in the concentration of bound regulatory factors. During normal development, low affinity sites allow for the precise activation of regulatory elements in response to environmental cues. We therefore propose the intrinsic lability of developmental regulatory elements makes them preferentially susceptible to the changes in the regulatory environment during cancer development and targets the global effects of chromatin remodelers to highly specific developmental phenotypes.