What Is Epigenetics? How the Same Genome Builds Different Cells

The term was coined by Conrad Waddington in the 1940s for the study of how a genotype produces a phenotype during development, and it has since narrowed to mean heritable changes in gene activity that do not involve changes to the DNA sequence itself. The heritable part is doing a lot of work and is where most disputes begin, since it can mean passed from a cell to its daughters, which is uncontroversial and well documented, or passed from a parent to offspring, which is a far stronger claim. The uncontroversial core is that a genome is not a program that runs itself but a set of instructions accessed selectively, and that which parts are accessible is regulated, maintained and modified. Every multicellular organism depends on this, because a fertilised egg and every cell descended from it carry the same sequence and must nonetheless become hundreds of distinct cell types.

Several molecular systems do the regulating, and they interact:

• •DNA methylation, the attachment of a methyl group to cytosine bases, typically where a cytosine is followed by a guanine, which generally silences the gene in that region and is copied to daughter cells after division
• •Histone modification, since DNA is wound around histone proteins whose tails can carry acetyl, methyl, phosphate and other groups, and the pattern of these determines how tightly the DNA is packed and therefore how accessible it is
• •Chromatin remodelling, in which protein complexes physically reposition the histone spools to expose or hide stretches of sequence
• •Non-coding RNA, including microRNAs that degrade or block specific messenger RNAs and longer RNAs that recruit silencing machinery to particular regions
• •Three-dimensional organisation, since the genome is folded so that regulatory sequences can contact genes far away along the strand, and the boundaries of these folded domains matter

Several phenomena are established and depend entirely on these mechanisms. Cell differentiation is the primary one: a cell becomes and stays a muscle cell because a particular pattern of methylation and histone marks is established and maintained. X-chromosome inactivation silences one of the two X chromosomes in each cell of a female mammal, chosen at random early in development, which is why tortoiseshell cats have patches, since each patch descends from a cell that silenced a different X. Genomic imprinting silences either the maternal or the paternal copy of certain genes, so that a small number of genes are expressed from only one parent's copy, and disruption of the imprinted region on chromosome 15 causes two very different syndromes depending on which parent's copy is affected. Cancer involves epigenetic changes as consistently as genetic ones, with tumour suppressor genes frequently silenced by methylation rather than mutated, and several approved drugs work by reversing such marks. Cloning by nuclear transfer works at all only because the epigenetic state of an adult nucleus can be partly reset by an egg.

The popular interest is in whether experiences change marks that are passed to children and grandchildren, and this is where care is needed. In mammals, the epigenome is erased twice, once shortly after fertilisation and once in the developing germ cells, which is a strong barrier to transmission, and a small number of regions escape it. Studies frequently cited include work on the Dutch famine of 1944, where people conceived during the famine showed differences in methylation at certain genes six decades later along with raised rates of metabolic disease, and mouse experiments in which an odour paired with a mild shock produced sensitivity to that odour in offspring. These are real findings and their interpretation is contested: prenatal exposure affects the developing foetus and its germ cells directly, which is not the same as transmitting an acquired mark across generations, and demonstrating genuine transgenerational inheritance in mammals requires effects persisting into the great-grandchild generation, which few studies reach. In plants and in the worm Caenorhabditis elegans, transgenerational epigenetic inheritance is well documented and can persist for many generations, which is part of why the mammalian claims seem plausible and why the barrier in mammals is interesting.

The field attracts claims it does not support. It does not mean genes are unimportant or that inheritance has been overturned, since the sequence still specifies what can be made and the regulation determines when. It does not provide a mechanism for wishing away a genetic condition, and commercial services offering to optimise your epigenome by diet or meditation are selling something the evidence does not support at the individual level. It is not a revival of Lamarck in any strong sense, since the mechanisms are themselves encoded by genes and were shaped by selection. And a measured difference in methylation is extremely easy to produce and extremely hard to interpret, since age, smoking, cell type composition in a blood sample and technical artefacts all shift these measurements, which is why the field has had reproducibility problems. What it is, unglamorously, is the regulatory layer without which a genome could not build an organism, and one of the more promising targets in cancer treatment.

Epigenetic mechanisms, chiefly DNA methylation, histone modification, chromatin remodelling and non-coding RNA, control which genes are accessible in a given cell and keep that state through cell division, which is how one genome produces hundreds of cell types. They explain X-chromosome inactivation, genomic imprinting and a substantial part of cancer biology. Transmission of acquired marks across generations is well established in plants and worms and remains contested in mammals, where the epigenome is erased twice between generations.