Epigenetics: How Cells Switch Genes On and Off

Same DNA, different cells

Nearly all of the roughly 37 trillion cells in the human body carry an almost identical copy of the same 3-billion-base-pair genome. Yet a pancreatic beta cell manufactures insulin and a neuron does not, despite both having the gene for insulin sitting in their DNA. Genetics studies the sequence itself — the letters A, T, G, C. Epigenetics (Greek "epi", upon) studies the layer of chemical marks sitting on top of that sequence which controls how, when, and in which cell type a gene is actually read.

The defining property of these marks is that they are heritable through cell division: when a liver cell divides, its daughter cells inherit the liver-specific pattern of marks rather than a blank slate. That is what keeps a liver cell's descendants making liver proteins for the rest of an organism's life, without the DNA sequence itself needing to change at all.

DNA methylation

The best-studied mark is DNA methylation: an enzyme called a DNA methyltransferase attaches a methyl group (–CH₃) to a cytosine base, typically at a CpG dinucleotide — a cytosine immediately followed by a guanine. Methylating a gene's promoter region generally silences it, because the methyl groups recruit proteins that compact the surrounding chromatin, physically blocking the transcription machinery from reaching the gene. Unmethylated promoters, conversely, correlate with active genes. The human genome has around 28 million CpG sites, and 70–80% are methylated at any given time — with the conspicuous exception of "CpG islands" at active gene promoters, which stay unmethylated. Abnormal methylation is a hallmark of cancer: tumour-suppressor genes are frequently silenced by hypermethylation, while oncogenes can be switched on by losing methylation they should have kept.

Histones and the nucleosome

DNA in the nucleus is not loose — about 147 base pairs wrap 1.65 times around a spool of eight histone proteins (two copies each of H2A, H2B, H3 and H4) to form a nucleosome, the basic unit of chromatin. Histone tails protruding from that spool carry chemical modifications that either open or close access to the DNA: H3K4me3 marks active promoters, H3K27me3 — placed by the Polycomb complex — silences developmental genes, H3K9me3 marks permanently silenced heterochromatin, and H3K27ac marks active enhancers by neutralising the tail's positive charge and loosening its grip on the DNA. These marks are written by "writer" enzymes, interpreted by "reader" proteins, and removed by "eraser" enzymes — several classes of cancer drug, such as HDAC and EZH2 inhibitors, work by targeting exactly these enzymes.

Chromatin exists in two broad states: open, active euchromatin and compact, silenced heterochromatin. ATP-powered remodelling complexes such as SWI/SNF physically slide or eject nucleosomes to expose or hide the binding sites transcription factors need, and distant regulatory elements called enhancers can loop across millions of base pairs to contact a promoter and switch a gene on.

Development, reprogramming and the environment

A fertilised egg's cells are totipotent — able to become anything — and development is largely a story of cells progressively and stably silencing the genes for every lineage they did not commit to. Remarkably that process can run backwards: in 2006 Shinya Yamanaka showed that adding just four transcription factors (Oct4, Sox2, Klf4, c-Myc) to an ordinary skin cell reprograms it into an induced pluripotent stem cell by erasing most of its epigenetic marks, work that won the 2012 Nobel Prize and launched regenerative medicine.

Because the marks are chemical rather than genetic, they respond to environment in ways DNA sequence cannot. Maternal diet during pregnancy — folate and methionine are methyl-group donors — shapes offspring methylation patterns; early-life stress increases methylation of the glucocorticoid receptor gene, blunting stress response into adulthood; a single bout of exercise rapidly shifts histone acetylation in muscle to upregulate metabolic genes; and methylation drifts predictably enough with age that an "epigenetic clock" built from blood methylation data can estimate biological age to within about four years.

Crossing generations

Most epigenetic marks are erased during germ-cell formation — a reset that, in principle, stops acquired changes from being inherited. But some marks escape that reset. Children of mothers who were pregnant during the Dutch Hunger Winter of 1944–45 showed increased rates of obesity and metabolic disease, an effect that persisted into their own children; in mice, sperm from males subjected to chronic stress carries altered microRNA profiles that transmit anxiety-like behaviour to offspring. This is a narrow, well-documented channel of inheritance, not a general mechanism, and it remains one of the more actively contested areas of epigenetics research — alongside epigenetic therapies such as DNMT and HDAC inhibitors, which are already approved to reactivate silenced tumour-suppressor genes in certain cancers.

Frequently asked questions

Try it live

Explore Epigenetics and see how methylation and histone marks turn genes on and off, and how environmental factors shift the pattern. Everything runs client-side in your browser.