The Epigenetics Revolution

Chapter 4. Life As We Know It Now

So far this book has focused mainly on outcomes, the things that we can observe that tell us that epigenetic events happen. But every biological phenomenon has a physical basis and that’s what this chapter is about. The epigenetic outcomes we’ve described are all a result of variations in expression of genes. The cells of the retina express a different set of genes from the cells in the bladder, for example. But how do the different cell types switch different sets of genes on or off?

The specialised cell types in the retina and in the bladder are each at the bottom of one of the troughs in Waddington’s epigenetic landscape. The work of both John Gurdon and Shinya Yamanaka showed us that whatever mechanism cells use for staying in these troughs, it’s not anything to do with changing the DNA blueprint of the cell. That remains intact and unchanged. Therefore keeping specific sets of genes turned on or off must happen through some other mechanism, one that can be maintained for a really long time. We know this must be the case because some cells, like the neurons in our brains, are remarkably long-lived. The neurons in the brain of an 85-year-old person, for example, are about 85 years of age. They formed when the individual was very young, and then stayed the same for the rest of their life.

But other cells are different. The top layer of skin cells, the epidermis, is replaced about every five weeks, from constantly dividing stem cells in the deeper layers of that tissue. These stem cells always produce new skin cells, and not, for example, muscle cells. Therefore the system that keeps certain sets of genes switched on or off must also be a mechanism that can be passed on from parent cell to daughter cell every time there is a cell division.

This creates a paradox. Researchers have known since the work of Oswald Avery and colleagues in the mid-1940s that DNA is the material in cells that carries our genetic information. If the DNA stays the same in different cell types in one individual, how can the incredibly precise patterns of gene expression be transmitted down through the generations of cell division?

Our analogy of actors reading a script is again useful. Baz Luhrmann hands Leonardo DiCaprio Shakespeare’s script for Romeo and Juliet, on which the director has written or typed various notes – directions, camera placements and lots of additional technical information. Whenever Leo’s copy of the script is photocopied, Baz Luhrmann’s additional information is copied along with it. Claire Danes also has the script for Romeo and Juliet. The notes on her copy are different from those on her co-star’s, but will also survive photocopying. That’s how epigenetic regulation of gene expression occurs – different cells have the same DNA blueprint (the original author’s script) but carrying varied molecular modifications (the shooting script) which can be transmitted from mother cell to daughter cell during cell division.

These modifications to DNA don’t change the essential nature of the A, C, G and T alphabet of our genetic script, our blueprint. When a gene is switched on and copied to make mRNA, that mRNA has exactly the same sequence, controlled by the base-pairing rules, irrespective of whether or not the gene is carrying an epigenetic addition. Similarly, when the DNA is copied to form new chromosomes for cell division, the same A, C, G and T sequences are copied.

Since epigenetic modifications don’t change what a gene codes for, what do they do? Basically, they can dramatically change how well a gene is expressed, or if it is expressed at all. Epigenetic modifications can also be passed on when a cell divides, so this provides a mechanism for how control of gene expression stays consistent from mother cell to daughter cell. That’s why skin stem cells only give rise to more skin cells, not to any other cell type.

The first epigenetic modification to be identified was DNA methylation. Methylation means the addition of a methyl group to another chemical, in this case DNA. A methyl group is very small. It’s just one carbon atom linked to three hydrogen atoms. Chemists describe atoms and molecules by their ‘molecular weight’, where the atom of each element has a different weight. The average molecular weight of a base-pair is around 600 Da (the Da stands for Daltons, the unit that is used for molecular weight). A methyl group only weighs 15 Da. By adding a methyl group the weight of the base-pair is only increased by 2.5 per cent. A bit like sticking a grape on a tennis ball.

Figure 4.1 shows what DNA methylation looks like chemically.

^{Figure 4.1 The chemical structures of the DNA base cytosine and its epigenetically modified form, 5-methylcytosine. C: carbon; H: hydrogen; N: nitrogen; O: oxygen. For simplicity, some carbon atoms have not been explicitly shown, but are present where there is a junction of two lines.}

The base shown is C – cytosine. It’s the only one of the four DNA bases that gets methylated, to form 5-methylcytosine. The ‘5’ refers to the position on the ring where the methyl is added, not to the number of methyl groups; there’s always only one of these. This methylation reaction is carried out in our cells, and those of most other organisms, by one of three enzymes called DNMT1, DNMT3A or DNMT3B. DNMT stands for DNA methyltransferase. The DNMTs are examples of epigenetic ‘writers’ – enzymes that create the epigenetic code. Most of the time these enzymes will only add a methyl group to a C that is followed by a G. C followed by G is known as CpG.

This CpG methylation is an epigenetic modification, which is also known as an epigenetic mark. The chemical group is ‘stuck onto’ DNA but doesn’t actually alter the underlying genetic sequence. The C has been decorated rather than changed. Given that the modification is so small, it’s perhaps surprising that it will come up over and over again in this book, and in any discussion of epigenetics. This is because methylation of DNA has profound effects on how genes are expressed, and ultimately on cellular, tissue and whole-body functions.

In the early 1980s it was shown that if you injected DNA into mammalian cells, the amount of methylation on the injected DNA affected how well it was transcribed into RNA. The more methylated the injected DNA was, the less transcription that occurred[19]. In other words, high levels of DNA methylation were associated with genes that were switched off. However, it wasn’t clear how significant this was for the genes normally found in the nuclei of cells, rather than ones that were injected into cells.

The key work in establishing the importance of methylation in mammalian cells came out of the laboratory of Adrian Bird, who has spent most of his scientific career in Edinburgh, Conrad Waddington’s old stomping ground. Professor Bird is a Fellow of the Royal Society and a former Governor of the Wellcome Trust, the enormously influential independent funding agency in UK science. He is one of those traditional British scientific types – understated, soft-spoken, non-flashy and drily funny. His lack of self-promotion is in contrast to his stellar international reputation, where he is widely acknowledged as the godfather of DNA methylation and its role in controlling gene expression.

In 1985 Adrian Bird published a key paper in Cell showing that most CpG motifs were not randomly distributed throughout the genome. Instead the majority of CpG pairs were concentrated just upstream of certain genes, in the promoter region[20]. Promoters are the stretches of the genome where the DNA transcription complexes bind and start copying DNA to form RNA. Regions where there is a high concentration of CpG motifs are called CpG islands.

In about 60 per cent of the genes that code for proteins, the promoters lie within CpG islands. When these genes are active, the levels of methylation in the CpG island are low. The CpG islands tend to be highly methylated only when the genes are switched off. Different cell types express different genes, so unsurprisingly the patterns of CpG island methylation are also different across different cell types.

For quite some time there was considerable debate about what this association meant. It was the old cause or effect debate. One interpretation was that DNA methylation was essentially a historical modification – genes were repressed by some unknown mechanism and then the DNA became methylated. In this model, DNA methylation was just a downstream consequence of gene repression. The other interpretation was that the CpG island became methylated, and it was this methylation that switched the gene off. In this model the epigenetic modification actually causes the change in gene expression. Although there is still the occasional argument about this between competing labs, the vast majority of scientists in this field now believe that the data generated in the quarter of a century since Adrian Bird’s paper are consistent with the second, causal model. Under most circumstances, methylation of the CpG island at the start of a gene turns that gene off.

Adrian Bird went on to investigate how DNA methylation switches genes off. He showed that when DNA is methylated, it binds a protein called MeCP2 (Methyl CpG binding protein 2)[21]. However, this protein won’t bind to unmethylated CpG motifs, which is pretty amazing when we look back at Figure 4.1 and think how similar the methylated and unmethylated forms of cytosine really are. The enzymes that add the methyl group to DNA have been described as writers of the epigenetic code. MeCP2 doesn’t add any modifications to DNA. Its role is to enable the cell to interpret the modifications on a DNA region. MeCP2 is an example of a ‘reader’ of the epigenetic code.

Once MeCP2 binds to 5-methylcytosine in a gene promoter it seems to do a number of things. It attracts other proteins that also help to switch the gene off[22]. It may also stop the DNA transcription machinery from binding to the gene promoter, and this prevents mRNA messenger molecule from being produced[23]. Where genes and their promoters are very heavily methylated, binding of MeCP2 seems to be part of a process where that region of a chromosome gets shut down almost permanently. The DNA becomes incredibly tightly coiled up and the gene transcription machinery can’t get access to the base-pairs to make mRNA copies.

This is one of the reasons why DNA methylation is so important. Remember those 85 year old neurons in the brains of senior citizens? For over eight decades DNA methylation has kept certain regions of the genome incredibly tightly compacted and so the neuron has kept certain genes completely repressed. This is why our brain cells never produce haemoglobin, for example, or digestive enzymes.

But what about the other situation, the example of skin stem cells dividing very frequently but always just creating new skin cells, rather than some other cell type such as bone? In this situation, the pattern of DNA methylation is passed from mother cell to daughter cells. When the two strands of the DNA double helix separate, each gets copied using the base-pairing principle, as we saw in Chapter 3. Figure 4.2 illustrates what happens when this replication occurs in a region where the CpG is methylated on the C.

^{Figure 4.2 This schematic shows how DNA methylation patterns can be preserved when DNA is replicated. The methyl group is represented by the black circle. Following separation of the parent DNA double helix in step 1, and replication of the DNA strands in step 2, the new strands are ‘checked’ by the DNA methyltransferase 1 (DNMT1) enzyme. DNMT1 can recognise that a methyl group at a cytosine motif on one strand of a DNA molecule is not matched on the newly synthesised strand. DNMT1 transfers a methyl group to the cytosine on the new strand (step 3). This only occurs where a C and a G are next to each other in a CpG motif. This process ensures that the DNA methylation patterns are maintained following DNA replication and cell division.}

DNMT1 can recognise if a CpG motif is only methylated on one strand. When DNMT1 detects this imbalance, it replaces the ‘missing’ methylation on the newly copied strand. The daughter cells will therefore end up with the same DNA methylation patterns as the parent cell. As a consequence, they will repress the same genes as the parent cell and the skin cells will stay as skin cells.

Epigenetics has a tendency to crop up in places where scientists really aren’t expecting it. One of the most interesting examples of this in recent years has related to MeCP2, the protein that reads the DNA methylation mark. Several years ago, the now discredited theory of the MMR vaccine causing autism was at its height, and getting lots of coverage in the general media. One very respected UK broadsheet newspaper covered in depth the terribly sad story of a little girl. As a baby she initially met all the usual developmental milestones. Shortly after receiving an MMR jab not long before her first birthday she began to deteriorate rapidly, losing most of the skills she had gained. By the time the journalist wrote the article, the little girl was about four years old and was described as having the most severely autistic symptoms the author had ever seen. She had not developed language, appeared to have very severe learning difficulties and her actions were very limited and repetitive, with very few purposeful hand actions (she no longer reached out for food, for example). Development of this incredibly severe disability was undoubtedly a tragedy for her and for her family.

But if a reader with any sort of background in neurogenetics read this article, two things probably struck them immediately. The first was that it’s very unusual – not unheard of but pretty uncommon – for girls to present with such severe autism. This is much more common in boys. The second thing that would have struck them was that this case sounded exactly the same as a rare genetic disorder called Rett syndrome, right down to the normal early development and the timing and types of symptoms. It’s just coincidence that the symptoms of Rett syndrome, and indeed of most types of autism, first start becoming obvious at around the same age as when infants are typically given the MMR vaccination.

But what does this have to do with epigenetics? In 1999, a group led by the eminent neurogeneticist Huda Zoghbi at the Howard Hughes Medical Institute in Maryland showed that the majority of cases of Rett syndrome are caused by mutations in MeCP2, the gene which encodes the reader of methylated DNA. The children with this disorder have a mutation in the MeCP2 gene which means that they don’t produce a functional MeCP2 protein. Although their cells are perfectly capable of methylating DNA correctly, the cells can’t read this part of the epigenetic code properly.

The severe clinical symptoms of children with the MeCP2 mutation tell us that reading the epigenetic code properly is very important. But they also tell us other things. Not all the tissues of girls with Rett syndrome are equally affected, so perhaps this particular epigenetic pathway is more important in some tissues than others. Because the girls develop severe mental retardation, we can deduce that having the right amount of normal MeCP2 protein is really important in the brain. Given that these children seem to be fairly unaffected in other tissues such as liver or kidney, perhaps MeCP2 activity isn’t as important in these tissues. It could be that DNA methylation itself isn’t so critical in these organs, or maybe these tissues contain other proteins in addition to MeCP2 that can read this part of the epigenetic code.

Long-term, scientists, physicians and families of children with Rett syndrome would dearly love to be able to use our increased understanding of the disease to help us find better treatments. This is a huge challenge, as we would be trying to intervene in a condition that affects the brain as a result of a gene mutation that is present throughout development, and beyond.

One of the most debilitating aspects of Rett syndrome is the profound mental retardation that is an almost universal symptom. Nobody knew if it would be possible to reverse a neurodevelopmental problem such as mental retardation once it had become established, but the general feeling about this wasn’t optimistic. Adrian Bird remains a major figure in our story. In 2007 he published an astonishing paper in Science, in which he and his colleagues showed that Rett syndrome could be reversed, in a mouse model of the disease.

Adrian Bird and his colleagues created a cloned strain of mice in which the Mecp2 gene was inactivated. They used the types of technologies pioneered by Rudolf Jaenisch. These mice developed severe neurological symptoms, and as adults they exhibited hardly any normal mouse activities. If you put a normal mouse in the middle of a big white box, it will almost immediately begin to explore its surroundings. It will move around a lot, it will tend to follow the edges of the box just like a normal house mouse scurrying along by the skirting boards, and it will frequently rear up on its back legs to get a better view. A mouse with the Mecp2 mutation does very few of these things – put it in the middle of a big white box and it will tend to stay there.

When Adrian Bird created his mouse strain with the Mecp2 mutation, he also engineered it so that the mice would also be carrying a normal copy of Mecp2. However, this normal copy was silent – it wasn’t switched on in the mouse cells. The really clever bit of this experiment was that if the mice were given a specific harmless chemical, the normal Mecp2 gene became activated. This allowed the experimenters to let the mice develop and grow up with no Mecp2 in their cells, and then at a time of the scientists’ choosing, the Mecp2 gene could be switched on.

The results of switching on the Mecp2 gene were extraordinary. Mice which previously just sat in the middle of the white box suddenly turned into the curious explorers that mice should be[24]. You can find clips of this on YouTube, along with interviews with Adrian Bird where he basically concedes that he really never expected to see anything so dramatic[25].

The reason this experiment is so important is that it offers hope that we may be able to find new treatments for really complex neurological conditions. Prior to the publication of this Science paper, there had been an assumption that once a complex neurological condition has developed, it is impossible to reverse it. This was especially presumed to be the case for any condition that arises developmentally, i.e. in the womb or in early infancy. This is a critical period when the mammalian brain is making so many of the connections and structures that are used throughout the rest of life. The results from the Mecp2 mutant mice suggest that in Rett syndrome, maybe all the bits of cellular machinery that are required for normal neurological function are still there in the brain – they just need to be activated properly. If this holds true for humans (and at a brain level we aren’t really that different from mice) this offers hope that maybe we can start to develop therapies to reverse conditions as complex as mental retardation. We can’t do this the way it was done in the mouse, as that was a genetic approach that can only be used in experimental animals and not in humans, but it suggests that it is worth trying to develop suitable drugs that have a similar effect.

DNA methylation is clearly really important. Defects in reading DNA methylation can lead to a complex and devastating neurological disorder that leaves children with Rett syndrome severely disabled throughout their lives. DNA methylation is also essential for maintaining the correct patterns of gene expression in different cell types, either for several decades in the case of our long-lived neurons, or in all daughters of a stem cell in a constantly-replaced tissue such as skin.

But we still have a conceptual problem. Neurons are very different from skin cells. If both cells types use DNA methylation to switch off certain genes, and to keep them switched off, they must be using the methylation at different sets of genes. Otherwise they would all be expressing the same genes, to the same extent, and they would inevitably then be the same types of cells instead of being neurons and skin cells.

The solution to how two cell types can use the same mechanism to create such different outcomes lies in how DNA methylation gets targeted to different regions of the genome in different cell types. This takes us into the second great area of molecular epigenetics. Proteins.

DNA is often described as if it’s a naked molecule, i.e. DNA and nothing else. If we visualise it at all in our minds, a DNA double helix probably looks like a very long twisty railway track. This is pretty much how we described it in the previous chapter. But in reality it’s actually nothing like that, and many of the great breakthroughs in epigenetics came about when scientists began to appreciate this fully.

DNA is intimately associated with proteins, and in particular with proteins called histones. At the moment most attention in epigenetics and gene regulation is focused on four particular histone proteins called H2A, H2B, H3 and H4. These histones have a structure known as ‘globular’, as they are folded into compact ball-like shapes. However, each also has a loose floppy chain of amino acids that sticks out of the ball, which is called the histone tail. Two copies of each of these four histone proteins come together to form a tight structure called the histone octamer (so called because it’s formed of eight individual histones).

It might be easiest to think of this octamer as eight ping-pong balls stacked on top of each other in two layers. DNA coils tightly around this protein stack like a long liquorice whip around marshmallows, to form a structure called the nucleosome. One hundred and forty seven base-pairs of DNA coil around each nucleosome. Figure 4.3 is a very simplified representation of the structure of a nucleosome, where the white strand is DNA and the grey wiggles are the histone tails.

^{Figure 4.3 The histone octamer (2 molecules each of histones H2A, H2B, H3 and H4) stacked tightly together, and with DNA wrapped around it, forms the basic unit of chromatin called the nucleosome.}

If we had read anything about histones even just fifteen years ago, they would probably have been described as ‘packaging proteins’, and left at that. It’s certainly true that DNA has to be packaged. The nucleus of a cell is usually only about 10 microns in diameter – that’s 1/100th of a millimetre – and if the DNA in a cell was just left all floppy and loose it could stretch for 2 metres. The DNA is curled tightly around the histone octamers and these are all stacked closely on top of each other.

Certain regions of our chromosomes have an extreme form of that sort of structure almost all the time. These tend to be regions that don’t really code for any genes. Instead, they are structural regions such as the very ends of chromosomes, or areas that are important for separating chromosomes after DNA has been duplicated for cell division.

The regions of DNA that are really heavily methylated also have this hyper-condensed structure and the methylation is very important in establishing this configuration. It’s one of the mechanisms used to keep certain genes switched off for decades in long-lived cell types such as neurons.

But what about those regions that aren’t screwed down tight, where there are genes that are switched on or have the potential to be switched on? This is where the histones really come into play. There is so much more to histones than just acting as a molecular reel for wrapping DNA around. If DNA methylation represents the semi-permanent additional notes on our script of Romeo and Juliet, histone modifications are the more tentative additions. They may be like pencil marks, that survive a few rounds of photocopying but eventually fade out. They may be even more transient, like Post-It notes, used very temporarily.

A substantial number of the breakthroughs in this field have come from the lab of Professor David Allis at Rockefeller University in New York. He’s a trim, neat, clean-shaven American who looks much younger than his 60 years and is exceptionally popular amongst his peers. Like many epigeneticists, he began his career in the field of developmental biology. Just like Adrian Bird, and John Gurdon before him, David Allis wears his stellar reputation in epigenetics very lightly. In a remarkable flurry of papers in 1996, he and his colleagues showed that histone proteins were chemically modified in cells, and that this modification increased expression of genes near a specific modified nucleosome[26].

The histone modification that David Allis identified was called acetylation. This is the addition of a chemical group called an acetyl, in this case to a specific amino acid named lysine on the floppy tail of one of the histones. Figure 4.4 shows the structures of lysine and acetyl-lysine, and we can again see that the modification is relatively small. Like DNA methylation, lysine acetylation is an epigenetic mechanism for altering gene expression which doesn’t change the underlying gene sequence.

^{Figure 4.4 The chemical structures of the amino acid lysine and its epigenetically modified form, acetyl-lysine. C: carbon; H: hydrogen; N: nitrogen; O: oxygen. For simplicity, some carbon atoms have not been explicitly shown, but are present where there is a junction of two lines.}

So back in 1996 there was a nice simple story. DNA methylation turned genes off and histone acetylation turned genes on. But gene expression is much more subtle than genes being either on or off. Gene expression is rarely an on-off toggle switch; it’s much more like the volume dial on a traditional radio. So perhaps it was unsurprising that there turned out to be more than one histone modification. In fact, more than 50 different epigenetic modifications to histone proteins have been identified since David Allis’s initial work, both by him and by a large number of other laboratories[27]. These modifications all alter gene expression but not always in the same way. Some histone modifications push gene expression up, others drive it down. The pattern of modifications is referred to as a histone code[28]. The problem that epigeneticists face is that this is a code that is extraordinarily difficult to read.

Imagine a chromosome as the trunk of a very big Christmas tree. The branches sticking out all over the tree are the histone tails and these can be decorated with epigenetic modifications. We pick up the purple baubles and we put one, two or three purple baubles on some of the branches. We also have green icicle decorations and we can put either one or two of these on some branches, some of which already have purple baubles on them. Then we pick up the red stars but are told we can’t put these on a branch if the adjacent branch has any purple baubles. The gold snowflakes and green icicles can’t be present on the same branch. And so it goes on, with increasingly complex rules and patterns. Eventually, we’ve used all our decorations and we wind the lights around the tree. The bulbs represent individual genes. By a magical piece of software programming, the brightness of each bulb is determined by the precise conformation of the decorations surrounding it. The likelihood is that we would really struggle to predict the brightness of most of the bulbs because the pattern of Christmas decorations is so complicated.

That’s where scientists currently are in terms of predicting how all the various histone modification combinations work together to influence gene expression. It’s reasonably clear in many cases what individual modifications can do, but it’s not yet possible to make accurate predictions from complex combinations.

There are major efforts being made to learn how to understand this code, with multiple labs throughout the world collaborating or competing in the use of the fastest and most complex technologies to address this problem. The reason for this is that although we may not be able to read the code properly yet, we know enough about it to understand that it’s extremely important.

Some of the key evidence comes from developmental biology, the field from which so many great epigenetic investigators have emerged. As we have already described, the single-celled zygote divides, and very quickly daughter cells start to take on discrete functions. The first noticeable event is that the cells of the early embryo split into the inner cell mass (ICM) and the trophoectoderm. The ICM cells in particular start to differentiate to form an increasing number of different cell types. This rolling of the cells down the epigenetic landscape is, to quite a large degree, a self-perpetuating system.

The key concept to grasp at this stage is the way that waves of gene expression and epigenetic modifications follow on from each other. A useful analogy for this is the game of Mousetrap, first produced in the early 1960s and still on sale today. Players have to build an insanely complex mouse trap during the course of the game. The trap is activated at one end by the simple act of releasing a ball. This ball passes down and through all sorts of contraptions including a slide, a kicking boot, a flight of steps and a man jumping off a diving board. As long as the pieces have been put together properly, the whole ridiculous cascade operates perfectly, and the toy mice get caught under a net. If one of the pieces is just slightly mis-aligned, the crazy sequence judders to a halt and the trap doesn’t work.

The developing embryo is like Mousetrap. The zygote is pre-loaded with certain proteins, mainly from the egg cytoplasm. These egg-derived proteins move into the nucleus and bind to target genes, which we’ll call Boots (in honour of Mousetrap), and regulate their expression. They also attract a select few epigenetic enzymes to the Boots genes. These epigenetic enzymes may also have been ‘donated’ from the egg cytoplasm and they set up longer-lasting modifications to the DNA and histone proteins of chromatin, also influencing how these Boots genes are switched on or off. The Boots proteins bind to the Divers genes, and switch these on. Some of these Divers genes may themselves encode epigenetic enzymes, which will form complexes on members of the Slides family of genes, and so on. The genetic and epigenetic proteins work together in a seamless orderly procession, just like the events in Mousetrap once the ball has been released. Sometimes a cell will express a little more or a little less of a key factor, one whose expression is on a finely balanced threshold. This has the potential to alter the developmental path that the cell takes, as if twenty Mousetrap games had been connected up. Slight deviations in how the pieces were fitted together, or how the ball rolled at critical moments, would trigger one trap and not another.

The names in our analogy are made up, but we can apply this to a real example. One of the key proteins in the very earliest stages of embryonic development is Oct4. Oct4 protein binds to certain key genes, and also attracts a specific epigenetic enzyme. This enzyme modifies the chromatin and alters the regulation of that gene. Both Oct4 and the epigenetic enzyme with which it works are essential for development of the early embryo. If either is absent, the zygote can’t even develop as far as creating an ICM.

The patterns of gene expression in the early embryo eventually feed back on themselves. When certain proteins are expressed, they can bind to the Oct4 promoter and switch off expression of this gene. Under normal circumstances, somatic cells just don’t express Oct4. It would be too dangerous for them to do so because Oct4 could disrupt the normal patterns of gene expression in differentiated cells, and make them more like stem cells.

This is exactly what Shinya Yamanaka did when he used Oct4 as a reprogramming factor. By artificially creating very high levels of Oct4 in differentiated cells, he was able to ‘fool’ the cells into acting like early developmental cells. Even the epigenetic modifications were reset – that’s how powerful this gene is.

Normal development has yielded important evidence of the significance of epigenetic modifications in controlling cell fate. Cases where development goes awry have also shown us how important epigenetics can be.

For example, a 2010 publication in Nature Genetics identified the mutations that cause a rare disease called Kabuki syndrome. Kabuki syndrome is a complex developmental disorder with a range of symptoms that include mental retardation, short stature, facial abnormalities and cleft palate. The paper showed that Kabuki syndrome is caused by mutations in a gene called MLL2[29]. The MLL2 protein is an epigenetic writer that adds methyl groups to a specific lysine amino acid at position 4 on histone H3. Patients with this mutation are unable to write their epigenetic code properly, and this leads to their symptoms.

Human diseases can also be caused by mutations in enzymes that remove epigenetic modifications, i.e. ‘erasers’ of the epigenetic code. Mutations in a gene called PHF8, which removes methyl groups from a lysine at position 20 on histone H3, cause a syndrome of mental retardation and cleft palate[30]. In these cases, the patient’s cells put epigenetic modifications on without problems, but don’t remove them properly.

It’s interesting that although the MLL2 and PHF8 proteins have different roles, the clinical symptoms caused by mutations in these genes have overlaps in their presentation. Both lead to cleft palate and mental retardation. Both of these symptoms are classically considered as reflecting problems during development. Epigenetic pathways are important throughout life, but seem to be particularly significant during development.

In addition to these histone writers and erasers there are over 100 proteins that act as ‘readers’ of this histone code by binding to epigenetic marks. These readers attract other proteins and build up complexes that switch on or turn off gene expression. This is similar to the way that MeCP2 helps turn off expression of genes that are carrying DNA methylation.

Histone modifications are different to DNA methylation in a very important way. DNA methylation is a very stable epigenetic change. Once a DNA region has become methylated it will tend to stay methylated under most conditions. That’s why this epigenetic modification is so important for keeping neurons as neurons, and why there are no teeth in our eyeballs. Although DNA methylation can be removed in cells, this is usually only under very specific circumstances and it’s quite unusual for this to happen.

Most histone modifications are much more plastic than this. A specific modification can be put on a histone at a particular gene, removed and then later put back on again. This happens in response to all sorts of stimuli from outside the cell nucleus. The stimuli can vary enormously. In some cell types the histone code may change in response to hormones. These include insulin signalling to our muscle cells, or oestrogen affecting the cells of the breast during the menstrual cycle. In the brain the histone code can change in response to addictive drugs such as cocaine, whereas in the cells lining the gut, the pattern of epigenetic modifications will alter depending on the amounts of fatty acids produced by the bacteria in our intestines. These changes in the histone code are one of the key ways in which nurture (the environment) interacts with nature (our genes) to create the complexity of every higher organism on earth.

Histone modifications also allow cells to ‘try out’ particular patterns of gene expression, especially during development. Genes become temporarily inactivated when repressive histone modifications (those which drive gene expression down) are established on the histones near those genes. If there is an advantage to the cell in those genes being switched off, the histone modifications may last long enough to lead to DNA methylation. The histone modifications attract reader proteins that build up complexes of other proteins on the nucleosome. In some cases the complexes may include DNMT3A or DNMT3B, two of the enzymes that deposit methyl groups on CpG DNA motifs. Under these circumstances, the DNMT3A or 3B can ‘reach across’ from the complex on the histone and methylate the adjacent DNA. If enough DNA methylation takes place, expression of the gene will shut down. In extreme circumstances the whole chromosome region may become hyper-compacted and inactivated for multiple cell divisions, or for decades in a non-dividing cell like a neuron.

Why have organisms evolved such complex patterns of histone modifications to regulate gene expression? The systems seem particularly complex when you contrast them with the fairly all-or-nothing effects of DNA methylation. One of the reasons is probably because the complexity allows sophisticated fine-tuning of gene expression. Because of this, cells and organisms can adapt their gene expression appropriately in response to changes in their environment, such as availability of nutrients or exposure to viruses. But as we shall see in the next chapter, this fine-tuning can result in some very strange consequences indeed.