Junk DNA: A Journey Through the Dark Matter of the Genome (17 page)

You and I, dear readers, are masterpieces of epigenetics. The 50–70 trillion cells in a human body pretty much all contain exactly the same genetic code.
^a
Whether they are salt-secreting cells in our sweat glands, the skin cells on our eyelids or the cells that produce the shock-absorbing cartilage in our knees, they all contain exactly the same DNA. They just use the information in those genes in different ways, depending on the tissue. For instance, the neurons in the brain express the receptors for neurotransmitters but switch off the genes for haemoglobin, the pigment that carries oxygen in our red blood cells.

These are all examples of situations we have referred to for decades as epigenetic phenomena. Yes, exactly the same word as for the modifications, and it makes sense. These are all situations where something else is happening in addition to, or as well as, the genetic code.

The discovery of DNA methylation finally gave us a mechanism to understand how epigenetic phenomena happen. In a neuron, the genes responsible for producing haemoglobin become heavily methylated and are switched off. They stay switched off through life. In the cells that give rise to red blood cells, however, these genes are not methylated and haemoglobin is created. But the genes that code for neurotransmitter receptors are switched off using this epigenetic mechanism in these cells.

DNA methylation is pretty stable. It’s surprisingly difficult to remove this modification. This is a good thing if your cells need to keep certain genes switched off for long periods. But often our cells need to respond to short-term changes in their environment, if we
drink alcohol or are stressed out by a job interview, for example. Here they turn to a second system. They add modifications to the histone proteins adjacent to genes. Changing the histone modifications can turn genes off, but because these modifications are relatively easy to remove, the cell has the option of turning the genes back on fairly quickly if it needs to. The histone modifications can also be used to modulate the expression of a gene – turn it on a little, quite a bit, quite a lot, a heck of a lot and so on. At a simplistic level we can think of DNA methylation as the on/off switch and histone modifications as the volume control.

The reason histone modifications can act as the fine-tuning mechanism for gene expression is because there are lots of different ones. If DNA is black-to-white with perhaps a few shades of grey depending on the level of methylation, histone modifications are glorious technicolour. There are multiple amino acids that can be modified on histone proteins, and there are at least 60 different chemical groups that can be added to the various amino acids. That creates an extraordinary degree of complexity because at different genes, or the same gene in different cell types, there are thousands of possible combinations of histone modifications. These will be interpreted by the cell in different ways, because they will attract different complexes of proteins that control the gene expression levels and patterns. Some combinations will drive up gene expression, others will drive it down.

Finding a place on the genome

But for years we were faced with a puzzle. The enzymes that add modifications to histone proteins are blind to DNA sequence. They don’t bind DNA and they can’t distinguish one DNA sequence from another. And yet, in the presence of a relevant stimulus, whatever that might be, the enzymes were very precise in how they modified specific histones. They would add (or remove)
modifications at the histones positioned at relevant genes, but ignore nearby histones associated with irrelevant genes.

It’s now starting to look as if one of the roles of long non-coding RNAs is to act as a kind of molecular Blu-Tack, attracting histone-modifying enzymes into the vicinity of selected genes. One of the pieces of evidence that this might be the case came from the work analysing the effects of certain long non-coding RNAs in human ES (embryonic stem) cells that was presented in Chapter 8. The researchers showed that about a third of the long non-coding RNAs they examined bound to complexes of proteins that included histone-modifying enzymes. To examine if this binding of long non-coding RNAs to the proteins had any functional consequences, they knocked down expression of the histone-modifying enzyme in the complex. In almost half the cases, the effects on the cell and on gene expression were the same as if they knocked down the long non-coding RNA itself. This suggested that the long non-coding RNA and the histone-modifying enzymes really were working together in the cell.
²

Many of the investigations of this cross-talk between the long non-coding RNA and epigenetics systems have focused on a specific epigenetic enzyme. This enzyme deposits a specific histone modification that is strongly associated with switching off genes. We can refer to this enzyme as the major repressor.
^b
This has been shown to interact with lots of different long non-coding RNAs.

The long non-coding RNA from a gene targets the major repressor to that gene. The major repressor enzyme then creates repressive modifications on the histones, driving down expression
of the genes. The repressive modifications attract other proteins, which bind and repress the gene even further.

This control by the major repressor epigenetic enzyme is frequently used to control genes that code for other epigenetic enzymes. Often, these will be genes that have the opposite effect to the major repressor, i.e. they tend to turn genes on. The overall effect is that the major repressor has a strong influence on general patterns of gene expression.
³
It represses genes directly, but also indirectly by preventing expression of epigenetic enzymes that normally switch other genes on. An epigenetic double-punch.

Usually this is a completely normal part of the control of gene expression that happens in our cells, and the system is doing exactly what it’s supposed to, making sure that all the complex cellular pathways run in an integrated fashion. But if one part of the complex interaction between long non-coding RNAs and the epigenetic machinery goes out of kilter, problems may develop.

Unfortunately, this seems to be exactly what is happening in some cancers. The major repressor is over-expressed in certain cancers, such as subsets of prostate
⁴
and breast
⁵
cancer, and this over-expression is associated with poor prognosis. In certain types of blood cell cancer, the major repressor has mutated, making it abnormally active.
⁶
The outcome in each case appears to be that the ‘wrong’ genes are repressed. This creates an imbalance where proteins that drive the cell into proliferation outrun those that usually act as a brake, promoting a cancerous state. Drugs that inhibit the activity of the major repressor are in early clinical trials.
⁷

The major repressor works as part of a large complex of proteins
^c
, and various long non-coding RNAs have been shown to be
associated with this complex, suggesting there may be multiple ways of targeting the repressive modifications, depending on the cell type and its behaviour. In Chapter 8 we met a long non-coding RNA whose over-expression drives prostate cancer (
see page 108
). It has been shown to bind to the major repressor and direct it to certain genes, including ones that normally hold back cell proliferation.
⁸
This finding reinforces the concept that there is a delicate balance of long non-coding RNAs and epigenetic modifiers and that disturbing the equilibrium may be dangerous for a cell or an individual. So do similar data around binding of the long non-coding RNA that is involved in skeletal deformities and a range of cancers, which we encountered in the same chapter (
see pages 106, 108
). It binds to the complex containing the major repressor, and simultaneously to another epigenetic enzyme that can deposit an additional repressing modification.
⁹

One of the features implicit in the above explanation is that the long non-coding RNA is transcribed at or near the gene whose histones will be targeted by the major repressor or by other epigenetic enzymes. Although it’s difficult to investigate this, the existing data suggest that this is indeed the case. The major repressor can bind to all sorts of long non-coding RNA molecules. The complex containing the major repressor can recognise different types of histone modifications, depending on the components of the complex. These components can vary from cell to cell. As they ‘scan’ the nearby histones, the complexes can recognise various modification patterns and reinforce these by adding the major repressive modifications. Alternatively, if the region is very rich in modifications that lead to gene expression, the complex may be inhibited and the major repressor will leave the histones alone. This is another of those scenarios where it is a disadvantage to think in purely linear terms, of what came first. Instead, patterns are often maintained or created as a consequence of the histone modification combinations that are already present on the genome.
^10
,
11

This also seems to hold for the opposite effect, where active regions remain active. Long non-coding RNAs have been reported to be expressed from regions where protein-coding genes are switched on. These long non-coding RNAs stay moored to the genome region where they are produced, possibly by forming a third strand to accompany the double helix of DNA. These long non-coding RNAs bind to the enzymes that place methyl modifications on DNA and stop them doing their job. This keeps the genes in that region in an active state.
¹²

If you’re inactive, you stay inactive

Xist, which is critical for switching off expression from one of the X chromosomes in a female cell, was one of the first functional long non-coding RNAs to be identified. Perhaps it’s no surprise that it’s the one whose cross-talk with the epigenetic system has been shown most clearly. As Xist spreads along the X chromosome it attracts other proteins. Many of these are epigenetic enzymes that add chemical modifications to either the DNA or the histone protein. They include the major repressor of histones, and also the enzymes that add methyl groups to DNA.
¹³
The epigenetic modifications they produce strengthen the shutdown of genes and ultimately lead to hyper-compaction of the inactive X chromosome, and the formation of the Barr Body that we encountered in Chapter 7 (
see page 84
).

It may seem puzzling that the epigenetic modifications always get re-established on the correct X chromosome after cell division. It may be easier to imagine this using a physical example. You have two wooden baseball bats, and you coat one of them with magnetic paint, which represents Xist. After the paint has dried you drop both bats into a tub containing little iron discs. One side of each disc is coated with hooked Velcro. The discs represent the epigenetic proteins that bind to the Xist-coated
chromosome. These discs will stick to the bat that has a magnetic covering, but not to the other one. After that, you drop each bat into a tub containing pretty fabric flowers, each backed with a piece of looped Velcro. These represent the modifications. Clearly, the flowers will only stick to the bat that was originally coated with magnetic paint, even though they aren’t magnetic themselves.

You could even take this slightly bizarre thought experiment further. Even if you take the flowers off the bat, if you drop it into another tub containing Velcroed blooms, it will be covered again. You could even take off the little iron discs, and as long as you put the bat back into the first and second tubs, it will get covered in flowers again.

In fact, the only way in which you can prevent the bat being covered in flowers when you drop it into the two tubs is to remove the magnetic paint. This is essentially what happens when women make eggs. The inactivating marks are all removed from the X chromosomes and all the daughter cells, i.e. all the eggs are ‘fresh’ in the sense that they won’t pass on inactivation to their offspring. The magnetic paint has to be applied anew to one of the two X chromosomes during early development.

Keeping the ancient aliens quiet

Long non-coding RNAs clearly interact with and help regulate the function of epigenetic proteins. But it would be a mistake to think this is the only way in which junk talks to the epigenetic system. Far from it. We saw in Chapter 4 that the human genome has been invaded by vast numbers of repetitive DNA elements and how important it is that these are kept switched off. Some researchers have gone so far as to speculate that epigenetic control of gene expression may originally have evolved to keep certain junk regions under control.
¹⁴
It was only later that
the epigenetic system struck out into new territory of regulating normal endogenous genes.

A really striking example of the interplay between junk DNA, epigenetics and the final appearance and behaviour of a mammal can be found in a mouse strain called the Agouti viable yellow mouse. All the mice in this strain are genetically identical, but they can look very different. Some are fat and yellow, some are thin and brown, and others are somewhere in between. The differences in their appearance are due to variable epigenetic regulation of a junk DNA region. In these mice, a repetitive DNA element has become inserted into the genome in front of a particular gene. The DNA element can undergo varying and random degrees of methylation. The heavier the methylation, the more the activity of the repetitive DNA element is repressed, and this affects the expression of the nearby gene.
¹⁵
It’s the expression levels of the nearby gene that ultimately determine how fat and how yellow the mouse will be. This is summarised in Figure 9.1.