Functional Inferences Derived from Defining the Interactome of H3K9me2 Writers and Readers

Abstract

The H3K9me2‐driven histone code pathway has emerged as key for cancer initiation, progression, and a potential target for therapeutic strategies. The H3K9me2 mark is written by the lysine methyltransferases, G9a and GLP, and read by HP1 proteins. The goal of the current work was to develop a comprehensive interactome for these proteins to reveal linked protein partners and complexes, in which these writers and readers function during cell proliferation in human cervical cancer epithelial cells. We immunopurified each of these proteins individually and performed mass spectrometry of the associated proteins at two different phases of the cell cycle, namely G1/S and G2/M. We validated and modeled the H3K9me2‐interacting protein networks using several data science‐based algorithms. Our analyses reveal known and novel interacting proteins for both writers and readers of the H3K9me2 mark. In addition, we show that each of these proteins form distinct protein networks in a cell cycle‐specific manner. The protein networks formed by the writers and readers, which are all known regulators of the same histone mark, overlap in their protein complexes less than previously anticipated. Moreover, we show that these proteins not only formed networks within chromatin but also with other non‐chromatin proteins, such as nuclear receptor coactivator proteins, as well as mediator and polymerase subunits. Furthermore, we discovered that many of the non‐chromatin proteins have the presence of a H3K9me‐like, histone mimicry‐based recognition domain, suggesting another level of interaction and regulation with the H3K9me2‐driven histone code pathway. Mining of large‐scale mass spectrometry data show this H3K9me‐like domain in several G9a‐interacting proteins, including CDC73, ZNF644, and GTSE1. Ontological analyses and network reconstruction suggest that these H3K9me2‐associated interacting protein networks regulate various functions, including gene silencing, gene activation, elongation, DNA replication, DNA repair, and chromosome structure. Thus, these findings extend the current paradigm on pathways that link, directly or indirectly, to the H3K9me2 mark through computational modeling of proteomic‐derived networks. These results extend our understanding of the biochemistry underlying cell cycle‐associated epigenomic processes that, like the H3K9me pathway, are relevant to human cancer.