Cancer-specific bivalent promoters featuring low-level H3K27me3 signals favor active transcription and govern the cancer cell state transition

Abstract

Bivalent chromatin maintains genes in low-expression, poised states in embryonic stem cells (ESCs). However, bivalent promoters correlate with the transcriptional activation of oncogenic programs in malignancies, a seemingly contradiction that remains to be resolved. Here, we identify a class of cancer-specific bivalent promoters (CSBPs) through the integration of a system-level longitudinal framework. Compared with ESCs, CSBPs are characterized by lower and narrower H3K27me3 deposition alongside abundant H3K4me3, thus permitting the persistent expression of genes critical for cancer stem cell (CSC) formation and maintenance, as exemplified by SOX9. The generation of CSBPs is essentially induced by the acquisition of H3K27me3 during cell state transition, which is mediated by specific binding of PRC2.1 and the de novo recruitment of PRC2.2. Notably, disrupting the bivalency of CSBPs significantly increases H3K4me3 levels, leading to hyperactivation of CSBPs and eventually inhibiting clonal expansion of CSCs and impairing tumorigenesis. Our study not only helps explain the puzzle of transcriptionally active bivalent genes in cancer but also provides insights into the development of therapies targeting phenotypic plasticity.

1. Introduction

Bivalency represents one of the intriguing chromatin landscapes initially discovered in embryonic stem cells (ESCs) and is characterized by the concurrent presence of activating histone H3 lysine 4 trimethylation (H3K4me3) and repressive histone H3 lysine 27 trimethylation (H3K27me3) modifications at the bivalent promoters of cell lineage regulators [1,2]. H3K4me3 is acquired by COMPASS complexes to facilitate transcriptional activation, whereas H3K27me3 is deposited by Polycomb complexes to mediate gene repression [3,4]. This bivalent chromatin state typically maintains genes in a low-expression, poised state, allowing for the precise activation or repression of transcription during subsequent developmental stages [5,6]. Embryonic development and cancer progression share several common features, including rapidly dividing cells, cellular plasticity and a highly vascular microenvironment [7]. Emerging evidence suggests that bivalent genes are critical regulators of malignant potential. Various cancer phenotypes are associated with specific bivalent genes; for example, ZEB1 maintains a bivalent state to increase responsiveness to TGF-β signals during the transition to a stem cell-like phenotype [8]. However, a systematic analysis of the bivalent chromatin landscape in cancer cells is still lacking.

Unlike embryos, tumor cells possess unlimited replicative potential due to their oncogenic mutations [9], which require active transcription for proliferation and survival. In addition, cancer cells display diverse phenotypic states due to their inherent heterogeneity and coexist within a spectrum of differentiated states, ranging from stem-like or progenitor-like cells to fully differentiated cells [10]. Whether bivalent genes in cancer cells remain transcriptionally silent, as they act in embryonic cells, or exhibit an alternative state is still unclear. Recent studies have provided evidence that more bivalent chromatin is present in cancer stem cells (CSCs) than in non-CSCs and that bivalent chromatin is associated with increased transcriptional activity [11]. Coincidentally, bivalent promoters are significantly enriched among upregulated genes in bladder cancer [12], implying that a previously unknown pattern of bivalency may exist in cancer cells, but its exact role has yet to be explored.