Aberrant epigenetic silencing of tumor suppressor genes in human cancer: The roles of DNA hypermethylation and the histone code

Aberrant DNA hypermethylation is a critical event in the silencing of many essential tumor suppressor genes in human cancer and is known to involve histone deacetylation. More recently, methylation of key lysine residues of histone H3 has been shown to associate with or determine active versus inactive transcription states. We show here that the two processes are critically linked. We have mapped key elements of the histone code along the DNA hypermethylated, silenced hMLH1 gene promoter in RKO colorectal cancer cells compared to the DNA unmethylated, active hMLH1 gene promoter in SW480 colorectal cancer cells. We show that the di- form of methyl-H3-K9 and the di- and tri-forms of methyl-H3-K27, as well as the histone methyltransferase (HMT) responsible for catalyzing tri-methyl-H3-K27, EZH2, are strikingly enriched along a critical region of the hypermethylated inactive gene promoter but severely depleted in the same region along the unmethylated active promoter. The mono- and tri-forms of methyl-H3-K9, and the mono-form of methyl-H3-K27 to a lesser extent, show similar trends, although they are limited to a key part of the region examined, which spans the region of highest CpG density along the promoter. Conversely, acetyl-H3 (K9 and K14) and di-methyl-H3-K4 are remarkably enriched along the unmethylated active hMLH1 gene promoter but depleted to virtually undetectable levels along the hypermethylated, transcriptionally inactive gene promoter. Furthermore, we show that upon pharmacologic inhibition of the DNMTs with 5-Aza-dC, only key histone modifications are reversed along the hypermethylated gene promoter, namely acetyl-H3 (K9 and 14) and di-methyl-H3-K4 are increased and di-methyl-H3-K9, as well as the EZH2 HMT, are lost; however, the other marks do not change appreciably. Examination of the dynamics upon 5-Aza-dC treatment reveals that DNA demethylation occurs first and is followed next by reactivation of gene transcription, which precedes histone code reversal. This sequence of events, the fact that only key histone marks are altered, and recent findings from others suggest that DNA demethylation causes reactivation of gene transcription, which leads to histone replacement, and subsequently, to key histone modification changes. In contrast to the results after inhibition of DNA methylation, depletion of the EZH2 HMT in RKO cells did not lead to reactivation of gene transcription, to loss of DNA methylation, or to changes in other histone modifications, although it resulted in a striking loss of the tri-methyl-H3-K27 mark, both globally and at the hypermethylated silenced hMLH1 gene promoter. These results suggest that the DNA hypermethylation is the dominant force in maintaining the "locked in" irreversibly silenced state of tumor suppressor genes in human cancer. We propose, then, that histone modifications, including the EZH2-mediated tri-methyl-H3-K27 mark, may play an essential role earlier in the aberrant epigenetic process, possibly in initiating epigenetic silencing or in converting a reversible silent state to an irreversibly silent state.