The major epigenetic modifications studied to date include DNA methylation, histone modifications and non-coding RNA (both long and short). In vertebrates, DNA methylation is found almost exclusively on CpG dinucleotides, which are generally underrepresented across the genomic landscape except within so-called CpG-rich ‘islands’. As these are frequently associated with proximal gene promoter sequences,Reference Bird¹ the emergent perception has been that DNA methylation is associated with the silencing of gene transcription.Reference Keshet, Yisraeli and Cedar^2, Reference Schubeler, Lorincz and Cimbora³ Consistent with this view, genome-wide surveys of genomic sites, which exhibit varying levels of methylation across normal cell populations, appear to support a generalized inverse correlation between methylation and more ‘densely’ packed heterochromatin regions, and a low level of methylation in regions, which have a more ‘open’ configuration of euchromatin.Reference Bell, Tiwari, Thoma and Schubeler⁴ Thus, a considerable body of data appears to reaffirm the prevailing view that a strategic loss or increase in DNA methylation is linked with a corresponding activation or inactivation of specific genes, respectively, an association that has been distilled predominantly from work on embryogenesis and cancer models.Reference Jones and Baylin⁵

This rather simplistic view of life has, however, long been dogged by inconsistencies, which are evident at the scale of intra-gene domains, and have led even pioneers in the field to conjecture about the ‘DNA methylation paradox’, which suggests ‘methylation in the transcribed region is often correlated with expression, in contrast to the inverse correlation seen at the site of transcriptional initiation’.Reference Jones⁶ And indeed, the higher resolution and deeper coverage afforded by genome-wide-based technologies has begun to throw up exceptions to the broad stroke representative principles outlined above, hinting at methylation relationships that may be more complex than have been initially appreciated.Reference Suzuki and Bird⁷ For example, high levels of gene activity are frequently correlated with low methylation relatively near the transcription start site, but also correlated with higher methylation states within gene body regions.Reference Wagner, Busche and Ge⁸ In addition, many hypomethylated gene promoters remain inactive, and a lack of methylation at CpG-rich promoters often aligns with a capacity for gene expression, rather than expression per se. Furthermore, comparisons of normal and cancerous colon cells and primary fibroblasts have revealed largely similar methylation profiles between all of these cell types, and found only a small subset of CpG-island promoters that are variably methylated in the cancer cell populations when compared with the non-cancer cell populations,Reference Weber, Davies and Wittig⁹ suggesting that aberrant methylation on CpG-island promoters in tumours may not be an obligatory requirement for altered gene expression to arise in pathology as is widely hypothesized. Most significantly, genome-wide mapping reveals the seemingly paradoxical observation that euchromatin regions (which associate with active gene transcription) are often highly methylated in the human genome,Reference Weber, Davies and Wittig⁹ which argues against a role for DNA methylation in the formation of heterochromatin (often associated with transcriptional repression) and rather, suggests a role as an epigenetic mark, which serves to restrict access to DNA on very local genomic scale.

The concept of DOHaD posits that environmental factors act during embryonic, foetal or early postnatal development to influence future health in adulthood, and do so by modifying the ‘epigenetic phenotype’ of the individual, which is most often defined in terms of DNA methylation variance.Reference Lavebratt, Almgren and Ekstrom^10–Reference Inadera¹² As such, the DOHaD field has much to contribute to the debate around the functional significance of DNA methylation. Notably, in a recent genome-wide study of the effects of maternal protein restriction in rats, there was a significant lack of overlap between differentially expressed genes and genes associated with differentially methylated regions in the progeny, when compared with the progeny of rats fed on control dietsReference Altobelli, Bogdarina, Stupka, Clark and Langley-Evans¹³ and is in agreement with several other recent studies.Reference Zhou, Yuan, Mash and Goldman^14–Reference Sinclair, Allegrucci and Singh¹⁶ Moreover, the differentially methylated regions were largely annotated to non-promoter regions of the genome, consistent with previous studies which report that a very low proportion of CpG islands and promoters are differentially methylated.Reference Lister, Pelizzola and Dowen^17, Reference Weber, Hellmann and Stadler¹⁸ Although relatively underappreciated in the context of DOHaD phenomena, the importance of non-CpG-island methylation variation is generally accepted in complex disease research. Interestingly, CpG methylation varies during cellular differentiation, particularly in regulatory regions outside of core promotersReference Meissner, Mikkelsen and Gu¹⁹ raising the possibility that these regions may encode cis-acting regulatory elements (such as enhancers or repressors) that regulate gene expression, specifically in response to environmental stimuli. Collectively, these DOHaD findings are in line with other studies comparing the methylomes of pluripotent and differentiated cells, which demonstrate that CpG-rich and poor regulatory regions may undergo distinct epigenetic regulation.Reference Jones and Baylin^5, Reference Weber, Hellmann and Stadler^18, Reference Meissner, Mikkelsen and Gu¹⁹ More importantly, they imply that the susceptibility of CpG sites outside core promoter regions are particularly responsive to in utero or environmental cues during development, and modulate long-term offspring development or metabolic status.

The DOHaD paradigm is well placed to add substantively more to the DNA methylation debate because of its distinct temporal perspective to organismal biology. In this regard, the careful mapping of progressive epigenetic marks on the promoter of the pancreatic islet development gene Pdx1, during normal and intra-uterine growth restriction (IUGR) compromised rat embryos, is perhaps seminal.Reference Park, Stoffers, Nicholls and Simmons²⁰ These ongoing studies provide striking evidence that during cellular and phenotypic commitment during organogenesis, gene expression levels are clearly set ahead of measurable changes in histone chemistry, and most significantly, that DNA methylation profiles at this genomic locus are the final epigenetic mark to be established. Although recognized previously for the case of chromosome X inactivation, evidence for gene silencing before DNA methylation marking in the context of DOHaD is sparse indeed. Further, the IUGR phenotype is propagated by dynamic histone modifications, which continue throughout neonatal development and presumably serve to recruit different members of DNA methyltransferase (DNMT) family of enzymes along the developmental time course. As a consequence, the Pdx1 gene in IUGR remains hypermethylated and stably suppressed, pancreas growth remains insufficient and diabetes results with adulthood.Reference Park, Stoffers, Nicholls and Simmons²⁰

Critically, these data suggest a hierarchy of epigenetic events may underpin the progressive determination of gene expression patterns, and suggest DNA methylation may represent a final ‘signature’ of the physiological response to an environmental cue. Implicit in this interpretation is the realization that while DNA methylation may reflect the level of gene activity arrived at by the system under any given circumstances, it is less clear that methylation is causally required to determine gene expression levels. Indeed, the recent observation that chemical inhibitors of histone modifying HDAC enzymes can induce re-expression of genes, even through a hypermethylated promoter and without any change in DNA methylation levels, in several types of cancer cells is perhaps most informative in this regard.Reference Raynal, Si and Taby²¹ Thus, it is chromatin state that may determine the immediate potential for gene expression, whereas DNA methylation perhaps confers stability to the degree of silencing over physiologically extended periods.Reference Raynal, Si and Taby²¹ Such an interpretation would be consistent with the notion that histone modifications are chemically more labile, a property that makes them more amenable to change by the cellular regulatory machinery, and thus quickly able to orchestrate changes in transcriptional activity of genes as and when required. DNA methylation, however, would seem suited to providing longer term stable epigenetic marks that are more refractory to transient change and requiring more persistent stimulus for modulation to occur. Such a notion is exemplified in the case of chromosome X inactivation.

The complex interplay between histone and DNMTs during the regulation of chromatin remodelling and gene transcription is now widely appreciated (comprehensively reviewed by Ikegami et al.Reference Ikegami, Ohgane, Tanaka, Yagi and Shiota²²). Yet, the challenges to understanding epigenetic process continue to grow as the complexities of these interactions emerge. For example, DNA methylation has only recently been shown to regulate microRNA expression in the early embryonic development of the mouse,Reference Lee, Chen, Maurya, Su and Tzeng²³ and in a DNMT knockout derivative population of a colon cancer cell line.Reference Han, Witmer, Casey, Valle and Sukumar²⁴ This suggests that the debate over the role of DNA methylation in gene regulation will yet take further unexpected turns and provide yet deeper questions for the DOHaD paradigm to tackle. Wherever possible, aligned measurements of gene expression, key epigenetic indicators and phenotype outcomes will need to be made to achieve the clarity of mechanistic process that we seek. The temporal aspects of DOHaD paradigm provide a unique context for achieving this level of understanding that we aspire to.