Feature: An open and shut case

31 January 2006. By Giles Newton, science editor at the Wellcome Trust.

The only important thing about DNA is its sequence, right? Wrong. Modifications to DNA and histones add additional complexity, by controlling access to DNA sequence.

Modifications to DNA and histones add additional complexity, by controlling access to DNA sequence. Think of DNA and the famous double helix usually springs to mind. With DNA packaging, it's histones and condensed chromosomes of cell division.

But the higher-order structure of DNA is about more than just cramming 2 metres of DNA into a nucleus. By controlling DNA packaging, a cell can exercise exquisite control over which of its genes are active. And interference with this fine control can have a profound impact on human health.

In recent years much has been learnt about the mechanisms involved. At least four modifications are known to influence DNA superstructure: DNA methylation, histone modification, histone variants and RNA-mediated silencing.

DNA methylation

In animals, large swathes of the genome have methylated cytosine (C) residues, usually when they are next to a G (termed CpG). About 80 per cent of human CpGs are methylated, for example. CpGs are rare in the genome, except near genes - clusters of CpGs, known as 'CpG islands', are found upstream of about 60 per cent of human genes, suggesting that they have a role in gene regulation.

In general, methylation is used to silence DNA. For some genes, methyl-cytosines block the binding of other proteins directly, and so interfere with transcription. More commonly, the methylated

DNA attracts methyl-binding proteins, which then recruit histone modification machinery to package the DNA into a closed, silent form. Conversely, undermethylated regions of the genome are usually (but not always) associated with active genes.

For some mammalian genes, termed 'imprinted genes', the pattern of methylation depends on a chromosome's parental origin. Genes active only on the paternal chromosome tend to promote growth of the embryo, while maternal genes protect the mother from the embryo's demands - a kind of 'genomic conflict'.

In any case, methylation patterns appear to be an essential component of the embryo's developmental programme - without the correct patterns, the embryo will not develop properly. This could be one reason why it is so difficult to clone animals. The methylation associated with imprinted genes has to be reset every generation, during the production of eggs and sperm.

Histone modification

Rather than being inert scaffolding, as once thought, histone proteins are now known to have a key role in controlling the access of other proteins to DNA.

The first level of DNA packaging involves the nucleosome core, a bobbin made up of eight histone proteins. The nucleosome cores are then lined up and, assisted by other factors, including the linker histone H1 protein, are themselves coiled into fatter ropes.

While the main part of each histone is within the nucleosome, they also have tails that hang outside. Modifications to amino acids in these tails form recognition signals for other enzymes and proteins to close or open up chromatin.

Can any patterns be discerned in this complex range of modifications? A 'histone code' has been proposed, in which particular modifications always have a specific effect. However, if there is a code it remains undeciphered, and it is unclear why there are so many different forms of histone modification.

Histone variants

The core histones are highly conserved, varying little between many different types of organism. It has recently become clear, however, that variants of these 'standard' histones exist - and have important biological consequences.

Some variants have been associated with particular forms of chromatin, such as the highly compact forms found at centromeres or looser arrangements around active genes. Crucially, it appears that the variants can be swapped in and out of chromatin (histones are usually added when DNA is copied and were thought to be permanent fixtures). This provides a mechanism for dynamically controlling chromatin structure - for example when genes are activated. Although the full implications of this exchange are not clear, it adds yet another potential mechanism for controlling gene activity.

RNA

In mammals, the most extensive silencing of all occurs in females, where one X chromosome is shut down. Xist, a non-coding RNA, is produced from the 'X inactivation centre' and coats the chromosome, triggering DNA methylation and a wave of histone modifications. It was thought originally that the chromosome was completely silenced, but recent studies indicate that this is not so.

RNA interference (RNAi), a system used to control the output of genes, also appears to have a role in DNA silencing. The system is mediated by small interfering RNAs (siRNAs), which promote the degradation of other RNAs (see Small RNA: Big news). It has emerged that they can also silence certain genes by DNA methylation and chromatin modification.

Polycombs

Research in another completely different area has also shed light on epigenetic mechanisms. The Polycomb family of proteins was first discovered in fruit-fly research. Many relatives have been discovered, and their function seems to be to maintain the higher-order structure of DNA, thereby 'fixing' gene activity and maintaining cell fate.

Polycomb family proteins seem to maintain chromatin in a silenced state while members of a second family, the trithorax group, maintain it in an active state. The Polycomb system seems to be linked to DNA methylation, in at least one case directly recruiting the enzymes that methylate DNA.

Implications for health

The surge of interest in DNA and histone modifications has coincided with a recognition of their importance to health. Biological function is dependent on fine control of gene activity. Anything that disrupts this control - including epigenetic changes - has the potential to cause disease.

Abnormal methylation of the control region of the FMR1 gene, for example, leads to fragile X syndrome. There are hints that some autoimmune disorders (such as lupus) and neuropsychiatric conditions (such as autism) may sometimes have epigenetic origins. Perhaps most significantly, though, cancer has been discovered to have a significant epigenetic component. Overall, perhaps half of all cancers are due to epigenetic errors, rather than mutations.

Cancer cells usually show abnormally low levels of DNA methylation. This may activate genes inappropriately and disrupt cell proliferation, or destabilise the entire genome. But excessive DNA methylation can also cause problems, switching off genes that normally keep cell division in check.

If the systems are going wrong, can they be righted? Epigenetic alterations are potentially reversible, so DNA demethylating agents are being tested as cancer treatments, and histone deacetylase inhibitors and methods for targeting gene-specific epigenetic modifications are being developed.

The environment

A recent study of identical twins also hinted at how ageing and the environment might affect the silencing systems. Although twins are epigenetically indistinguishable during the early years of life, older twins had remarkable differences in their genome-wide DNA methylation and histone acetylation.

More generally, compared with changes to genetic sequence, epigenetic systems provide a more dynamic mechanism for linking the environment to changes in gene activity. Some environmental factors - such as low levels of certain nutrients or toxins - affect the enzymes that methylate DNA.

And genetic and environmental influences can also be mutually dependent. Some genetic variants can render people more susceptible to environmental factors that alter DNA methylation. For example, the effects of a variation in the MTHFR gene depend on environmental factors: it can increase the risk of breast cancer three-fold in premenopausal women but may be protective in postmenopausal women taking hormone replacement therapy.

Examples such as these emphasise the interplay between 'nature' and 'nurture'. The two are interdependent, and epigenetic effects are one of the key ways in which they interact.

Modification biochemistry

Methyl groups are transferred to C residues in DNA by enzymes called DNA methyl tranferases (DNMTs), which also ensure that the marks are transferred faithfully when DNA is replicated. Six families of these enzymes have been identified, some of which methylate naked DNA while others carry out maintenance work.

Histone tails can be modified in several difference places, and in several different ways. For example, if lysine residues on histone H3 or H4 are acetylated, chromatin generally becomes more relaxed and accessible to other factors. Conversely, if lysines on H3 are methylated, the chromatin tends to become more compact. These modifications are carried out by a range of histone acetylases and deacetylases, methyl tranferases and methylases. In sum, the combination of modifications recruits specific proteins that influence chromatin structure.

One gene, two syndromes

Prader-Willi and Angelman syndromes arise from errors in imprinting of a gene on chromosome 15. The gene sequence is the same, but a gene inherited from the father gives one set of symptoms (Prader-Willi syndrome) while a gene from the mother leads to another (Angelman syndrome).