A Guest Blog by Dr Adele Murrell, University of Bath, UK
At the beginning of this year, Jeff Mann (Murdoch’s Children’s institute Australia published a visions and reflections article questioning the definition of “Epigenetics” (Mann, 2014). The concept has gone from the ‘Waddington marble running down a valley to simulate progenitor cells being channelled along alternative routes to differential fates’, to the study of “mitotically stable” changes in gene expression that cannot be explained by changes in DNA sequence. As more and more mechanistic elements of gene regulation are identified, the question of their mitotic stability or maintenance during cell division and hence their suitability for inclusion as components of the “epigenome” arises. Maintaining X inactivation is clearly a mitotically transmitted phenomenon and therefore conforms to the above definition of epigenetics, but the dynamic transient changes of histone acetylation during the cell cycle will not fit this the definition. Jeff Mann proposes we need to introduce another term such as “memigenetics”. Memigenetics will specifically distinguish “the study of mitotically and/or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence” from other epigenetic studies. The prefix memi- is a contraction of ‘memo-epi-’, and is intended to mean a ‘remembering’ of what is ‘over’.
Do we really need another term, or can we continuously change the definition? What do we want the epigenome to do? And what will we do when we find that it has broader functions than the box we are trying to squeeze it into? I rather like the concept that the epigenome is an adaptive multicomponent organelle and would be happy for the term “epigenetic” to include all phenomenona/mechanisms that impact upon DNA function. Adrian Bird described epigenetics as the structural adaptation of chromosomal regions so as to register, signal or perpetuate altered activity states’ (Bird 2007). The various components of the epigenome interact and like signalling pathways there seems to be considerable crosstalk. A mitotically heritable component may impact upon a transient component. For example an imprinted locus with allelic differences in clonally inherited DNA methylation can also have differential transient histone modifications during the cell which result in asynchronous replication of the imprinted alleles.
More important than semantics, is the question of whether epigenetic factors can really be the drivers of disease independently of DNA sequence. Are the methylation changes that we see in cancer either due to genetic mutations at nearby sequences that supposedly guide and recruit the methylation machinery, or are they due to mutations within the methylation machinery? Several cancer studies have identified genetic mutations that drive tumorigenesis. There is very little evidence that aberrant DNA methylation is a driver event in cancer. A recent report on subtypes of ependymomas (tumors of the central nervous system, which often manifest as tumors in the hindbrain in children) reveal that type A posterior fossa ependymomas do not have recurrent DNA mutations that could be potential driver mutations, but instead have several genes that are hypermethylated and silenced (Mack et al., 2014). Many of the CIMP (CpG island methylator phenotype) genes are Polycomb PRC2 target genes, known to be silenced in undifferentiated embryonic stem cells to prevent differentiation. These results point to an epigenetic origin of an embryonal cancer potentially through PRC2 mediated repression of differentiation. These results also conform to an instructive hypothesis that postulates that PRC2 target genes in embryonic stem cells are earmarked for gaining DNA methylation in cancer (Schlesinger et al., 2007; Ohm et al., 2007; Widschwendter et al., 2007). Epigenome profiling shows that DNA methylation and the polycomb mark H3K27me3 are mostly mutually exclusive – suggesting an antagonistic relationship between these two silencing marks (Lindroth et al., 2008; Reddington et al., 2013). Removal of DNA methylation results in an accumulation of PRC2 and H3K27me3 at previously DNA methylated loci (Reddington et al., 2013). However, depletion of PRC2 and reduction of H3K27me3 does not have a reciprocal effect on DNA methylation (Hagerman et al., 2013). A recent insight and perspectives article by Reddington, Sproul and Meehan (Bioessays, 2013) proposes several mechanisms whereby crosstalk between polycomb and DNA methylation can reprogram gene expression in cancer – including epigenetic switching at promoters and the accumulation of H3K27me3 blocks following DNA hypomethylation.
Genomic imprinting continues to provide several paradigms for epigenetic gene regulation. Congenital syndromes associated with aberrant imprinting include Beckwith Wiedemann syndrome (BWS; MIM #130659, KCNQ1OT1 and IGF2-H19 chromosome 11) and Neonatal diabetes (TND MIM #601410, PLAGL1 locus on chromosome 6). Mechanisms whereby imprinted gene expression is disrupted include changes in parent-of-origin-specific mono-allelic DNA methylation due to uniparental disomy (UPD, i.e., both chromosomes come from one parent), copy number variation, mutation of the expressed copy, or epimutation (i.e., changes in DNA methylation that do not seem to be associated with any genetic mutations). A subset of BWS and TDN patients have hypomethylation at multiple imprinted loci across the genome (hypomethylation of imprinted loci (HIL) (Eggermann et al., 2011). In some of these disorders, a shared pattern of methylation derangement can be detected, and underlying genetic mutations in ZFP57 and NLRP2 have been identified; in other cases, the cause(s) remain unknown. A recent study by the MacKay group which profiled HIL patients with TND and BWS that do not have ZFP57 and NLRP2 mutations indicates an overlap between imprinted genes with hypomethylation in both these syndromes and further identifies three more candidate imprinted gene regions. These results highlight the existence of further factors that influence imprint establishment in trans (Docherty et al., 2014); dare I whisper – perhaps there is no such thing as a “driver epimutation”?