Covalent histone modifications--miswritten, misinterpreted and mis-erased in human cancers

Abstract

Post-translational modification of histones provides an important regulatory platform for processes such as gene transcription and DNA damage repair. It has become increasingly apparent that the misregulation of histone modification, which is caused by the deregulation of factors that mediate the modification installation, removal and/or interpretation, actively contributes to human cancer. In this Review, we summarize recent advances in understanding the interpretation of certain histone methylations by plant homeodomain finger-containing proteins, and how misreading, miswriting and mis-erasing of histone methylation marks can be associated with oncogenesis and progression. These observations provide us with a greater mechanistic understanding of epigenetic alterations in human cancers and might also help direct new therapeutic interventions in the future.

Figure 1. ‘Miswriting’ of histone methylation is associated with the initiation or progression of human cancer

(a) MLL-containing complexes induce H3K4me3 at Hox genes in early hematopoietic progenitor cells. Following terminal differentiation, a transition of chromatin state occurs at Hox, which is characterized by loss of H3K4me3 and gain of EZH2-mediated H3K27me3^, . EZH2 polycomb factors and associated HDACs induce the stable silencing of Hox. MLL-PTD, a MLL rearrangement form that harbors a duplication of MLL exon 4–12, causes an elevated level of H3K4 methylation. (b–d) In leukemia, MLL fusion proteins lose a large carboxyl portion that includes the H3K4me3-‘writing’ SET domain, retain the chromatin targeting factors (Menin and LEDGF), and also acquire aberrant trans-activation mechanisms through its fusion partner. A subset of MLL fusions, MLL-AF10, MLL-ENL and MLL-AF9, directly interact with DOT1L and induce the methylation of H3K79 at Hoxa9 (panel b). Some other MLL fusions, MLL-AF4, MLL-AF5q31 and MLL- ELL1, interact with and recruit p-TEFb transcription elongation complexes to Hoxa9 (panel c). DOT1L-complexes (DOT1L-AF10-AF17-ENL/AF9) associate with p-TEFb complexes via the shared components. Another MLL fusion partner EEN recruits PRMT1 and induce methylation of H4R3 at Hox (panel d). (e) Over-expression of EZH2 in tumor cells silences the tumor suppressor gene such as INK4B-ARF-INK4A. Please note that EZH2 can regulate oncogenes (panel a) or tumor suppressors (panel e) in different cellular contexts.

Figure 2. ‘Reading’ or ‘mis-reading’ the H3K4me3 marks by the PHD finger-containing factors in normal cellular processes and during cancer development

(a) Interaction with histone modification (H3K4me3 recognized by the TAF3 PHD finger^, and histone acetylation by the double bromo domain of TAF1) and the DNA binding (TBP to the TATA box sequences) serve to anchor and/or stabilize the TFIID complex to core promoters, a critical step of the assembly of general transcription initiation machineries for active gene transcription^, . (b) Both recognition of H3K4me3 by the RAG2 PHD finger and binding to the recombination signal sequences (RSS) by RAG1-RAG2 complexes are critical for recruiting and/or stabilizing the RAG1/2 complexes at V(D)J gene segments to be recombined during B and T cell development^, . (c) Chromosomal translocation NUP98-JARID1A or NUP98-PHF23 fuses the N-terminal part of a nucleoporin protein, NUP98, to an H3K4me3-binding PHD finger of JARID1A or PHF23 (left panel). Such a NUP98-PHD finger fusion oncoprotein prevents the removal of H3K4me3 presumably mediated by the JARID1 histone demethylase and associated repressive factors (right panel) and enforces the expression of leukemia oncogenes such as HOX and MEIS1 . Arrows at the bottom indicate the effect of each complex on transcription. (d) Upon insult of DNA damage, H3K4me3 serves as a mechanism to recruit and/or stabilize the ING protein complexes to genes involved in the regulation of cell proliferation or apoptosis, which is then followed by their repression (in case of ING1/2-HDAC complexes) or activation (in case of ING4/5-HAT complexes)^{, ,} . A subset of cancer-associated somatic mutations of ING1 specifically interfere with the binding to H3K4me3/2 marks^{, ,} . (e) Recognition of H3K4me3 by the PHD finger of Pygopus (Pygo), an interacting cofactor of BCL9 and β-catenin, has been suggested to be critical for efficient activation of Wnt signaling pathway.

Figure 3. Physical interaction between histone methylation ‘writers’ and ‘erasers’ ensures a robust response during the transition of chromatin states, and such cooperation is also observed during cancerous transformation

(a) Cooperation between histone methyltransferases and demethylases, exemplified by MLL-JMJD3 interaction and EZH2-JARID1 interaction, underlies a dynamic change in H3K4me3 and H3K27me3 at leukemia-associated oncogenes such as HOX, a process that is perturbed by leukemia oncoproteins such as NUP98-JARID1A or MLL-fusion in leukemia^, . (b) Upon RAS signaling-induced oncogenic stress or the replicative stress, switch of histone methyltransferases and demethylases underlies activation of the tumor suppressor locus INK4B-ARF-INK4A and induction of senescence, a mechanism to prevent cancerous transformation^{–, , ,} . In cancer cells, over-expression of EZH2 and JHDM1, or down-regulation of JMJD3, interferes with such a switch of chromatin state and thus senescence response^{, ,} .