The Epigenetics Revolution Nessa Carey

Prev Next

The mind is its own place, and in itself can make a Heaven of Hell, a Hell of Heaven.

John Milton

One of the most noticeable publishing trends of the last ten years has been the rise and rise of the ‘misery memoir’. In this genre, the authors recount the tough times of their childhood and how they have risen above them to be successful and fulfilled individuals. The genre can be sub-divided into two categories. The first is the poor-but-happy tale, the ‘we had nothing but we had love’ story. The second, which may or may not also include poverty, tends to be much more disturbing. It focuses on harrowing tales of childhood neglect and childhood abuse, and some of these memoirs have been hugely successful. A Child Called It by Dave Pelzer, possibly the most famous of this category of books, spent over six years on the New York Times bestsellers list.

A substantial amount of their appeal seems to lie in the triumph-over-adversity aspects of these memoirs. Readers seem to take heart from the stories of individuals who, despite a terrible start in life, finally grow up to be happy, well-balanced adults. We applaud those who become winners ‘against the odds’.

This tells us something quite revealing. It shows that, as a society, we believe that early childhood events are extremely important in influencing adult life. It also shows that we believe that it is very difficult to get over the effects of early trauma. As a readership, we possibly value these successful survivors because of what we perceive as their relative rarity.

In many ways, we are correct in our assumptions as it is true that dreadful early childhood experiences really can have a dramatic impact on adult life. There are all sorts of ways in which this has been measured and the precise figures may vary from study to study. Despite this, certain clear trends have emerged. Childhood abuse and neglect result in adults with a three times greater risk of suicide than the general population. Abused children are at least 50 per cent more likely than the general population to suffer from serious depression as adults, and will find it much harder to recover from this illness. Adults who were subjected to childhood abuse and neglect are also at significantly higher risk of a range of other conditions including schizophrenia, eating disorders, personality disorders, bipolar disease and generalised anxiety. They are also more likely to abuse drugs or alcohol¹.

An abusive or neglectful environment when young is clearly a major risk factor for the development of later neuropsychiatric disorders. We are so aware of this as a society that sometimes we almost forget to question why this should be the case. It just seems self-evident. But it’s not. Why should events that lasted for two years, for example, still have adverse consequences for that individual several decades later?

One explanation that is often given is that the children are ‘psychologically damaged’ by their early experiences. Whilst true, this isn’t that helpful a statement. The reason why it’s not helpful is that the phrase ‘psychologically damaged’ isn’t really an explanation at all – it’s a description. It sounds quite convincing but on certain levels it doesn’t really tell us anything.

Any scientist addressing this problem will want to take this description and probe it at another level. What are the molecular events that underlie this psychological damage? What happens in the brains of the abused or neglected children, that leaves them so prone to mental health problems as adults?

There is sometimes resistance to this approach from other disciplines, which work within different conceptual frameworks.

This seems rather puzzling. If we don’t accept there is a molecular basis to a biological effect, what are we left with? A religious person may prefer to invoke the soul, just as a Freudian therapist may invoke the psyche. Both of these refer to a theoretical construct that has no defined physical basis. Moving into such a model system, where it is impossible to develop the testable hypotheses that are the cornerstone of all scientific enquiry, is deeply unattractive to most scientists. We prefer to probe for a mechanism that has a physical foundation, rather than defaulting to a scenario in which there is something which is assumed, somehow, to be a part of us, without having any physical existence.

This can generate a cultural clash, but it’s one that’s based on a misunderstanding. A scientist will expect that observable events have a physical basis. For the topic of this chapter, our proposed hypothesis is that terrible early childhood experiences change certain physical aspects of the brain during a key developmental period. This in turn affects the likelihood of mental health problems in adult life. This is a mechanistic explanation. It’s lacking in details, admittedly, but we’ll fill in some of these in this chapter. Mechanistic explanations often sit uncomfortably in our society, because they sound too deterministic. Mechanistic explanations are misinterpreted and taken to imply that humans are essentially robots, wired and programmed to respond in certain ways to certain stimuli.

But this doesn’t have to be the case. If a system has enough flexibility, then one stimulus doesn’t always have to result in the same outcome. Not every abused or neglected child develops into a vulnerable, unwell adult. A phenomenon can have a mechanistic basis, without being deterministic.

The human brain possesses sufficient flexibility to generate different adult outcomes in response to similar childhood experiences. Our brains contain one hundred billion nerve cells (neurons). Each neuron makes links with ten thousand other neurons to form an incredible three dimensional grid. This grid therefore contains a thousand trillion connections – that’s 1,000,000,000,000,000 (a quadrillion). It’s hard to imagine this, so let’s visualise each connection as a disc that’s 1mm thick. Stack up the quadrillion discs on top of each other and they will reach to the sun (which is ninety-three million miles from the earth) and back, three times over.

That’s a lot of connections, so it’s perfectly possible to imagine that our brains have a lot of flexibility. But the connections are not random. There are networks of cells within the giant grid which are more likely to link to each other than to anywhere else. It’s this combination of huge flexibility, but constrained within certain groupings, that is compatible with a system that is mechanistic but not entirely deterministic.

The child is (epigenetically) father to the man

The reason scientists have hypothesised that the adult sequelae of early childhood abuse may have an epigenetic component is that we’re dealing with scenarios where a triggering event continues to have consequences long after the trigger itself has disappeared. The long-term consequences of childhood trauma are very reminiscent of many of the effects that are mediated by epigenetic systems. We have seen some examples of this already. Differentiated cells remember what cell type they are, even after the signal that told them to become kidney cells or skin cells has long since vanished. Audrey Hepburn suffered from ill-health her whole life because of the malnutrition she suffered as a teenager during the Dutch Hunger Winter. Imprinted genes get switched off at certain stages in development, and stay off throughout the rest of life. Indeed, epigenetic modifications are the only known mechanism for maintaining cells in a particular state for exceptionally long periods of time.

The hypothesis that epigeneticists are testing is that early childhood trauma causes an alteration in gene expression in the brain, which is generated or maintained (or both) by epigenetic mechanisms. These epigenetically mediated abnormalities in gene expression predispose adults to increased risk of mental illnesses.

In recent years, scientists have begun to generate data suggesting that this is more than just an appealing hypothesis. Epigenetic proteins play an important role in programming the effects of early trauma. Not only that, they also are involved in adult depression, drug addiction and ‘normal’ memory.

The focus of a lot of research in this field has been a hormone called cortisol. This is produced from the adrenal glands which sit on top of the kidneys. Cortisol is produced in response to stress. The more stressed we are, the more cortisol we produce. The average level of cortisol production tends to be raised in adults who had traumatic childhoods, even if the individuals are healthy at the time of measurement²^,³. What this shows is that adults who were abused or neglected as children have higher background stress levels than their contemporaries. Their systems are chronically stressed. The development of mental illness is, in many cases, probably a little like the development of cancer. A lot of things need to go wrong at the molecular level before a person becomes clinically ill. The chronic stress levels in the abuse survivors push them closer to that threshold. This increases their vulnerability to disease.

How does this over-expression of cortisol happen? It’s a consequence of events that happen far from the kidneys, in our brains. There is a whole signalling cascade involved here. Chemicals produced in one region of the brain act on other areas. These areas in turn produce other chemicals in response and the process continues. Eventually a chemical leaves the brain and signals to the adrenal glands and cortisol is produced. During an abusive childhood, this signalling cascade is very active. In many abuse survivors, this system keeps signalling as if the person is still trapped in the abusive situation. It’s as if the thermostat on a central heating system has malfunctioned, and the boiler and radiators continue to pump out heat in August, based on the weather from the previous February.

The process starts in a region of the brain called the hippocampus, which gets its name from the ancient Greek term for seahorse, it being shaped a little like this creature. The hippocampus acts as a master switch in controlling how much the cortisol system becomes activated. This is shown in Figure 12.1. In this figure, a plus symbol indicates that one event acts to stimulate the next link in the chain. A minus symbol shows the opposite effect, where one event decreases the level of activity of the next event in the chain.

Because of changes in the activities of the hippocampus in response to stress, the hypothalamus produces and releases two hormones, called corticotrophin-releasing hormone and arginine vasopressin. These two hormones stimulate the pituitary, which responds by releasing a substance called adrenocorticotrophin hormone which gets into the bloodstream. When the cells of the adrenal gland take up this hormone, they release cortisol.

Figure 12.1 Signalling events in response to stress set up a cascade of events in selected regions of the brain that ultimately result in release of the stress hormone cortisol from the adrenal glands. Under normal circumstances, this system is controlled by a set of negative feedback loops that act to dampen down and limit the activation of the stress response pathways.

There’s a clever mechanism built in to this system. Cortisol circulates around the body in the bloodstream, and some of it goes back into the brain. The three brain structures shown in our diagram all carry receptors that recognise cortisol. When cortisol binds to these receptors, it creates a signal that tells these structures to calm down. It’s particularly important for this to happen at the hippocampus, as this structure can send out signals to dampen down all the others involved in this signalling. This is a classic negative feedback loop. Production of cortisol feeds back on various tissues, and the final effect is that the production of cortisol declines. This stops us from being constantly over-stressed.

But we know that adults who suffered traumatic childhoods are actually over-stressed. They produce too much cortisol, all the time. Something must be going wrong in this feedback loop. There are a few studies in humans that show that this is the case. These studies examined the levels of corticotrophin-releasing hormone in the fluid bathing the brain and spinal cord. As predicted, the levels of corticotrophin-releasing hormone were higher in individuals who had suffered childhood abuse than in individuals who hadn’t. This was true even when the individuals were healthy at the time of the experiments⁴^,⁵. Because it’s so hard to investigate this fully in humans, a lot of the breakthroughs in this field have come from using animal models of certain conditions and then correlating them where possible with what we know from human cases.

Relaxed rats and mellow mice

A useful model has been based around the mothering skills of rats. In the first week of their lives, rat babies love being licked and groomed by their mothers. Some mothers are naturally very good at this, others not so much so. If a mother is good at it, she’s good at it in all her pregnancies. Similarly, if she’s a bit lackadaisical at the licking and grooming, this is true for every litter she has.

If we test the offspring of these different mothers when the pups are older and independent, an interesting effect emerges. When we challenge these now adult rats with a mildly stressful situation, the ones that were licked and groomed the most stay fairly calm. The ones that were relatively deprived of ‘mother love’ react very strongly to even mild stress. Essentially, the rats that had been licked and groomed the most as babies were the most chilled out as adults.

The researchers carried out experiments where newborn rats were transferred from ‘good’ mothers to ‘bad’ and vice versa. These experiments showed that the final responses of the adults were completely due to the love and affection they received in the first week of life. Babies born to mothers who were lacklustre lickers and groomers grew up nicely chilled out if they were fostered by mothers who were good at this.

The low stress levels of the adult rats that had been thoroughly nurtured as babies were shown by measuring their behaviour when they were challenged by mild stimuli. They were also monitored hormonally, and the effects were as we would expect. The chilled-out rats had lower levels of corticotrophin-releasing hormone in their hypothalamus and lower levels of adrenocorticotrophin hormone in their blood. Their levels of cortisol were also low, compared with the less nurtured animals.

The key molecular factor in dampening down the stress responses in the well-nurtured rats was the expression of the cortisol receptor in the hippocampus. In these rats, the receptor was highly expressed. As a result, the cells of the hippocampus were very efficient at catching even low amounts of cortisol, and using this as the trigger to subdue the downstream hormonal pathway, through the negative feedback loop.

This showed that levels of the cortisol receptor stayed high in the hippocampus, many months after the all-important licking and grooming of the baby rats. Essentially, events that only happened for seven days immediately after birth had an effect that lasted for pretty much all of a rat’s life.

The reason the effect was so long-lasting is that the initial stimulus – being licked and groomed by the mother – set off a chain of events that led to epigenetic changes to the cortisol receptor gene. These changes occurred very early in development when the brain was at its most ‘plastic’. By plastic, we mean that this is the time when it’s easiest to modify the gene expression patterns and cellular activities. As the animals get older, these patterns stay set in place. That’s why the first week in rats is so critical.

The changes that take place are shown in Figure 12.2. When a baby rat is licked and groomed a lot, it produces serotonin, one of the feel-good chemicals in mammalian brains. This stimulates expression of epigenetic enzymes in the hippocampus, which ultimately results in decreased DNA methylation of the cortisol receptor gene. Low levels of DNA methylation are associated with high levels of gene expression. Consequently, the cortisol receptor is expressed at high levels in the hippocampus, and can keep the rats relatively relaxed⁶.

This is a very interesting model to explain how early life events can influence long-term behaviour. But it seems unlikely that just one epigenetic alteration – even one as significant as DNA methylation levels at a very important gene in a critical brain region – could be the only answer. Five years after the work described above, another paper was published by a different group. This also showed the importance of epigenetic changes but in a different gene.

The later group used a mouse model of early-life stress. In this model, baby mice were taken away from their mothers for three hours a day, for the first ten days of their lives. Just like the baby rats that hadn’t been licked or groomed much, these babies developed into ‘high-stress’ adults. Cortisol levels were increased in these mice, especially in response to mild stress, just like the relatively neglected rats.

Figure 12.2 Strong nurturing of baby rats sets up a cascade of molecular events that result in increased expression of the cortisol receptor in the brain. This increased expression makes the brain very effective at responding to cortisol and down-regulating stress responses via the negative feedback loop described in Figure 12.1.

The researchers working on the mice studied the arginine vasopressin gene. Arginine vasopressin is secreted by the hypothalamus, and stimulates secretion from the pituitary. It is shown in Figure 12.1. The stressed-out mice, those that had suffered separation from their mothers in early life, had decreased DNA methylation of the arginine vasopressin gene. This resulted in increased production of arginine vasopressin, which stimulated the stress response⁷.

The rat and mouse experimental studies show us two important things. The first is that when early life events lead to adult stress, there is probably more than one gene involved. Both the cortisol receptor gene and the arginine vasopressin gene can contribute to this phenotype in rodents.

Secondly, the studies also show us that a particular class of epigenetic modification is not in itself good or bad. It’s where the modification happens that matters. In the rat model, the decreased DNA methylation of the cortisol receptor gene is a ‘good’ thing. It leads to increased production of this receptor, and a general dampening down of the stress response. In the mouse model, the decreased DNA methylation of the arginine vasopressin gene is a ‘bad’ thing. It leads to increased expression of this hormone and a stimulation of the stress response.

The decreased DNA methylation of the arginine vasopressin gene in the mouse model occurred through a different route to the one used in the rat hippocampus to activate the cortisol receptor gene.

In the mouse studies, separation from the mother triggered activity of the neurons in the hypothalamus. This set off a signalling cascade that affected the MeCP2 protein. MeCP2 is the protein we met in Chapter 4, which binds to methylated DNA and helps repress gene expression. It’s also the gene which is mutated in Rett syndrome, the devastating neurological disorder. Adrian Bird has shown that the MeCP2 protein is incredibly highly expressed in neurons⁸.

Normally, MeCP2 protein binds to the methylated DNA at the arginine vasopressin gene. But in the stressed baby mice, the signalling cascade mentioned in the previous paragraph adds a small chemical group called a phosphate to the MeCP2 protein and because of this MeCP2 falls off the arginine vasopressin gene. One of the important roles of MeCP2 is attracting other epigenetic proteins to where it is bound on a gene. These are proteins that all cooperate to add more and more repressive marks to that region of the genome. When the phosphorylated MeCP2 falls off the arginine vasopressin gene, it can no longer recruit these different epigenetic proteins. Because of this, the chromatin loses it repressive marks. Activating modifications get put on instead, such as high levels of histone acetylation. Ultimately, even the DNA methylation is permanently lost.

Amazingly this all happens in the mice in the first ten days after birth. After that, the neurons essentially lose their plasticity. The DNA methylation pattern that’s in place at the end of this stage becomes the stable pattern at this location. If the DNA methylation levels are low, this will normally be associated with abnormally high expression of the arginine vasopressin gene. In this way, the early life events trigger epigenetic changes which get effectively ‘stuck’. Because of this, the animal continues to be highly stressed, with abnormal hormone production, long after the initial stress has vanished. Indeed, the response continues long after the animal would even normally ‘care’ about whether or not it has its mother’s company. After all, mice are not renowned for hanging about to look after their ageing parents.

In the depths

Researchers are gradually gathering data that suggest some of the changes seen in the rodent models of early stress may be relevant in humans. As mentioned earlier, there are logistical, but more importantly ethical, issues which make it impossible to perform the same kinds of studies in people. Even so, some intriguing correlations are emerging.

The original work in the rat model was carried out by Professor Michael Meaney at McGill University in Montreal. His group subsequently performed some interesting studies on human brain samples from individuals who had, sadly, committed suicide. The group analysed the levels of DNA methylation at the cortisol receptor gene in the hippocampus from these cases. Their data showed that the DNA methylation tended to be higher in the samples from people who had had a history of early childhood abuse or neglect. By contrast, the DNA methylation levels at this gene were relatively low in the suicide victims who had not had traumatic childhoods⁹. The high DNA methylation levels in the abuse victims would drive down expression of the cortisol receptor gene. This would make the negative feedback loop less efficient and raise the circulating levels of cortisol. This was consistent with the findings from the rat work, where the stressed-out animals from the less nurturing mothers had high levels of DNA methylation at the cortisol receptor gene in the hippocampus.

Of course, it isn’t just people who have had abusive childhoods who develop mental illnesses. The global figures for depression are startling. The World Health Organisation estimates that over 120 million people worldwide are affected by depression. Depression-related suicides have reached 850,000 per annum and depression is predicted to become the second greatest contributor to the global disease burden by 2020¹⁰.

Effective treatment for depression took a big step forwards in the early 1990s with the licensing by the US Food and Drug Administration of a class of drugs called SSRIs – selective serotonin re-uptake inhibitors. Serotonin is a neurotransmitter molecule – it conveys signals between neurons. Serotonin is released in the brain in response to pleasurable stimuli; it’s the feel-good molecule that we met in our happy rat babies. The levels of serotonin are low in the brains of people suffering from depression. SSRI drugs raise the levels of serotonin in the brain.

It makes sense that drugs that cause an increase in serotonin levels would be useful in treating depression. But there’s something odd about their action. The serotonin levels in the brain rise quite quickly when patients are treated with the SSRI drugs. But it usually takes at least four to six weeks before the terrible symptoms of severe depression begin to lift.

This suggests that there is more to depression than simply a drop in the levels of a single chemical in the brain, which perhaps isn’t that surprising. It’s very unusual for depression to happen overnight – it’s not like coming down with the flu. There’s now a reasonable amount of data showing that there are much longer-term changes in the brain as depression develops. These include alterations in the numbers of contacts that neurons make with each other. This in turn is critically dependent on the levels of chemicals called neurotrophic factors¹¹. These chemicals support healthy survival and function of brain cells.

Researchers in the depression field have moved away from a simple model based on levels of neurotransmitters and into a more complex network system. This involves sophisticated interactions between neuronal activity and a whole range of other factors. These include stress, production of neurotransmitters, effects on gene expression and longer-term consequences for neurons and how they interact with each other. While this system is in balance, the brain functions healthily. If the system moves out of balance, this complicated network begins to unravel. This moves the brain’s biochemistry and function further away from health and closer to dysfunction and disease.

Scientists are beginning to focus their attention in this field on epigenetics, because of its potential to create and sustain long-lasting patterns of gene expression. Rodents are the most common model system for these investigations. Because a mouse or a rat can’t tell you how it’s feeling, researchers have created certain behavioural tests that are used to model different aspects of human depression.

We all recognise that different people seem to respond to stress in different ways. Some people seem fairly robust. Others can react really badly to the same stressful situation, even developing depression. Mice from different inbred strains are like this as well. Researchers exposed two different strains to mildly stressful stimuli. After the stressful situation, the researchers assessed the behaviour of the mice in some of the tests which mimic certain aspects of human depression. One strain was relatively non-anxious, whereas the other was relatively anxious. These strains were called B6 and BALB, but we’ll called them ‘chilled’ and ‘jumpy’, respectively, for convenience.

The researchers focused their studies on a region of the brain called the nucleus accumbens. This region plays a role in various emotionally important brain functions. These include aggression, fear, pleasure and reward. The researchers analysed the expression of various neurotrophic factors in the nucleus accumbens. The one that gave the most interesting results was a gene called Gdnf (glial cell-derived neurotrophic factor).

Stress caused an increase in expression of the Gdnf gene in the chilled mice. In the jumpy strain it caused a decrease in expression of the same gene. Now, different inbred strains of mice can have different DNA codes so the researchers analysed the promoter region, which controls the expression of Gdnf. The DNA sequence of the Gdnf promoter was identical in the chilled and the jumpy strains. But when the scientists examined the epigenetic modifications in this promoter, they found a difference. The histones of the jumpy mice had fewer acetyl groups than the histones of the chilled mice. As we’ve seen, low levels of histone acetylation are associated with low levels of gene expression, so this tied up well with the decreased Gdnf expression in the jumpy mice.

This led the scientists to wonder what had happened in the neurons of the nucleus accumbens. Why had the levels of histone acetylation dropped at the Gdnf gene in the jumpy mice? The scientists examined the levels of the enzymes that add or remove acetyl groups from histones. They found only one difference between the two strains of mice. A specific histone deacetylase (member of the class of proteins which removes acetyl groups) called Hdac2 was much more highly expressed in the neurons of the jumpy mice¹², compared with the chilled out mice.

Other researchers tested mice in a different model of depression, called social defeat. In these experiments, mice are basically humiliated. They’re put in an environment where they can’t get away from a bigger, scarier mouse, although they are removed before they come to any physical harm. Some mice find this really stressful; others seem to brush it off.

In the experiments adult mice underwent ten days of social defeat. At the end of this they were classified as either susceptible or resistant, depending on how well they bounced back from the experience. Two weeks later the mice were examined. The resistant mice had normal levels of corticotrophin-releasing hormone. This is the chemical released by the hypothalamus. It’s the one which ultimately stimulates the production of cortisol, the stress hormone. The susceptible mice had high levels of corticotrophin-releasing hormone and low levels of DNA methylation at the promoter of this gene. This was consistent with the high levels of expression from this gene. They also had low levels of Hdac2, and high levels of histone acetylation, which again fits with over-expression of the corticotrophin-releasing hormone¹³.

It might seem odd that in one model system Hdac2 levels went up in the susceptible mice, whereas in another they went down. But it’s important with all these epigenetic events to remember that context is everything. There isn’t just one way in which Hdac2 levels (or those of any other epigenetic gene, for that matter) are controlled. The control will depend on the region of the brain and the precise signalling pathways that are activated in response to a stimulus.

The drugs do work

There’s more evidence supporting a significant role for epigenetics in responses to stress. The naturally jumpy B6 mice were the ones with the increased expression of Hdac2 in the nucleus accumbens, and decreased expression from the Gdnf gene. We can treat these mice with SAHA, the histone deacetylase inhibitor. SAHA treatment leads to increased acetylation of the Gdnf promoter. This is associated with increased expression of the Gdnf gene. The crucial finding is that the treated mice stop being jumpy and become chilled instead¹⁴ – changing the histone acetylation levels of the gene changed the mouse’s behaviour. This supports the idea that histone acetylation is really important in modulating the responses of these mice to stress.

One of the tests used to investigate how depressed the mice become in response to stress is called the sucrose-preference test. Normal happy mice love sugared water, but when they are depressed they aren’t so interested in it. This decreased response to a pleasant stimulus is called anhedonia. It seems to be one of the best surrogate markers in animals for human depression¹⁵. Most people who have been severely depressed talk about losing interest in all the things that used to make life joyful before they became ill. When the stressed mice were treated with SSRI anti-depressants, their interest in the sugared water gradually increased. But when they were treated with SAHA, the HDAC inhibitor, they regained their interest in their favourite drink much faster¹⁶.

It’s not just in the jumpy or chilled mice that histone deacetylase inhibitors can change animal behaviour. It’s also relevant to the baby rats who don’t get much maternal licking and grooming. These are the ones that normally grow up to be chronically stressed, with over-activation of the cortisol production pathway. If these ‘unloved’ animals are treated with TSA, the first histone deacetylase inhibitor to be identified, they grow up much less stressed. They react much more like the animals who received lots of maternal care. The levels of DNA methylation at the cortisol receptor gene in the hippocampus go down, increasing expression of the receptor and improving the sensitivity of the all-important negative feedback loop. This is presumed to be because of cross-talk between the histone acetylation and DNA methylation pathways¹⁷.

In the social defeat model in mice, the susceptible animals were treated with an SSRI anti-depressant drug. After three weeks of treatment, their behaviour was much more like that of the resilient mice. But treatment with this anti-depressant drug didn’t just result in increased levels of serotonin in the brain. The anti-depressant treatment also led to increased DNA methylation at the promoter of the corticotrophin-releasing hormone.

These studies are all very consistent with a model where there is cross-talk between the immediate signals from the neurotransmitters, and the longer-term effects on cell function mediated by epigenetic enzymes. When depressed patients are treated with SSRI drugs, the serotonin levels in the brain begin to rise, and signal more strongly to the neurons. The animal work described in the last paragraph suggests that it takes a few weeks for these signals to trigger all the pathways that ultimately result in the altered pattern of epigenetic modifications in the cells. This stage is essential for restoring normal brain function.

Epigenetics is also a reasonable hypothesis to explain another interesting but distressing feature of severe depression. If you have suffered from depression once, you are at a significantly higher risk than the general population of suffering from it again at some time in the future. It’s likely that some epigenetic modifications are exceptionally difficult to reverse, and leave the neurons primed to be more vulnerable to another bout.

The jury’s out

So far, so good. Everything looks very consistent with our theory about life experiences having sustained and long-lasting effects on behaviour, through epigenetics. And yet, here’s the thing: this whole area, sometimes called neuro-epigenetics, is probably the most scientifically contentious field in the whole of epigenetic research.

To get a sense of just how controversial, consider this. We’ve met Professor Adrian Bird in this book before. He is acknowledged as the father of the DNA methylation field. Another scientist with a very strong reputation in the science behind DNA methylation is Professor Tim Bestor from Columbia University Medical Center in New York. Adrian and Tim are about the same age, of similar physical type, and both are thoughtful and low key in conversation. And they seem to disagree on almost every issue in DNA methylation. Go to any conference where they are both scheduled in the same session and you are guaranteed to witness inspiring and impassioned debate between the two men. Yet the one thing they both seem to agree on publicly is their scepticism about some of the reports in the neuro-epigenetics field¹⁸.

There are three reasons why they, and many of their colleagues, are so sceptical. The first is that many of the epigenetic changes that have been observed are relatively small. The sceptics are unconvinced that such small molecular changes could lead to such pronounced phenotypes. They argue that just because the changes are present, it doesn’t mean they’re necessarily having a functional effect. They worry that the alterations in epigenetic modifications are simply correlative, not causative.

The scientists who have been investigating the behavioural responses in the different rodent systems counter this by arguing that molecular biologists are too used to quite artificial experimental models, where they can study extensive molecular changes with very on-or-off read-outs. The behaviourists suspect that this has left molecular biologists relatively inexperienced at interpreting real-world experiments, where the read-outs tend to be more ‘fuzzy’ and prone to greater experimental variation.

The second reason for scepticism lies in the very localised nature of the epigenetic changes. Infant stress affects specific regions of the brain, such as the nucleus accumbens, and not other areas. Epigenetic marks are only altered at some genes and not others. This seems less of a reason for scepticism. Although we refer to ‘the brain’, there are lots of highly specialised centres and regions within this organ, the product of hundreds of millions of years of evolution. Somehow, all these separate regions are generated and maintained during development and beyond, and thus are clearly able to respond differently to stimuli. This is also the case for all our genes, in all our tissues. It’s true that we don’t really know how epigenetic modifications can be targeted so precisely, or how the signalling from chemicals like neurotransmitters leads to this targeting. But we know that similarly specific events occur during normal development – so why not during abnormal periods of stress or other environmental disturbances? Just because we don’t know the mechanism for something, it doesn’t mean it doesn’t happen. After all, John Gurdon didn’t know how adult nuclei were reprogrammed by the cytoplasm of eggs, but that didn’t mean his experimental findings were invalid.

The third reason for scepticism is possibly the most important and it relates to DNA methylation itself. DNA methylation at the target genes in the brain is established very early, possibly pre-natally but certainly within one day of birth, in rodents. What this means is that the baby mice or baby rats in the experiments all started life with a certain baseline pattern of DNA methylation at their cortisol receptor gene in the hippocampus. The DNA methylation levels at this promoter alter in the first week of life, depending on the amount of licking and grooming the rats receive. As we saw, the DNA methylation levels are higher in the neglected mice than in the loved ones. But that’s not because the DNA methylation has gone up in the neglected mice. It’s because DNA methylation has gone down in the ones that were licked and groomed the most. The same is also true at the arginine vasopressin gene in the baby mice removed from their mothers. It’s also true for the corticotrophin-releasing hormone gene in the adult mice that were susceptible to social defeat.

So, in every case, what the scientists observed was decreased DNA methylation in response to a stimulus. And that’s where, molecularly, the problem lies, because no-one knows how this happens. In Chapter 4 we saw how copying of methylated DNA results in one strand that contains methyl groups and one that doesn’t. The DNMT1 enzyme moves along the newly synthesised strand and adds methyl groups to restore the methylation pattern, using the original strand as a template. We could speculate that in our experimental animals, there was less DNMT1 enzyme present and so the methylation levels at the gene dropped. This is referred to as passive DNA demethylation.

The problem is that this can’t work in neurons. Neurons are terminally differentiated – they are right at the bottom of Waddington’s landscape, and cannot divide. Because they don’t divide, neurons don’t copy their DNA. There’s no reason for them to do so. As a result, they can’t lose their DNA methylation by the method described in Chapter 4.

One possibility is that maybe neurons simply remove the methyl group from DNA. After all, histone deacetylases remove acetyl groups from histones. But the methyl group on DNA is different. In chemical terms, histone acetylation is a bit like adding a small Lego brick onto a larger Lego brick. It’s pretty easy to take the two bricks apart again. DNA methylation isn’t like that. It’s more like having two Lego bricks and using superglue to stick them together.

The chemical bond between a methyl group and the cytosine in DNA is so strong that for many years it was considered completely irreversible. In 2000, a group from the Max Planck Institute in Berlin demonstrated that this couldn’t be the case. They showed that in mammals the paternal genome undergoes extensive DNA demethylation, during very early development. We came across this in Chapters 7 and 8. What we glossed over at the time was that this demethylation happens before the zygote starts to divide. In other words, the DNA methylation was removed without any DNA replication¹⁹. This is referred to as active DNA demethylation.

This means there is a precedent for removing DNA methylation in non-dividing cells. Perhaps there’s a similar mechanism in neurons. There’s still a lot of debate about how DNA methylation is actively removed, even in the well-established events in early development. There’s even less consensus about how it takes place in neurons. One of the reasons this has been so hard to investigate is that active DNA demethylation may involve a lot of different proteins, carrying out a number of steps one after another. This makes it very difficult to recreate the process in a lab, which is the gold standard for these kinds of investigations.

Silencing the silencer

As we’ve seen repeatedly, scientific research often throws up some very unexpected findings and so it happened here. While many people in epigenetics were looking for an enzyme that removed DNA methylation, one group discovered enzymes that added something extra to methylated DNA. This is shown in Figure 12.3. Very surprisingly, this has turned out to have many of the same consequences as demethylating the nucleic acid.

A small molecule called hydroxyl, consisting of one oxygen atom and one hydrogen atom, is added to the methyl group, to create 5-hydroxymethylcytosine. This reaction is carried out by enzymes called TET1, TET2 or TET3²⁰.

Figure 12.3 Conversion of 5-methylcytosine to 5-hydroxymethylcytosine. 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.

This is highly relevant to the question of DNA demethylation, because it’s the effects of DNA methylation that make this change important. Methylation of cytosine affects gene expression because methylated cytosine binds certain proteins, such as MeCP2. MeCP2 acts with other proteins to repress gene expression and to recruit other repressive modifications like histone deacetylation. When an enzyme such as TET1 adds the hydroxyl group to the methylcytosine to form the 5-hydroxymethylcytosine molecule, it changes the shape of the epigenetic modification. If a methylated cytosine is like a grape on a tennis ball, the 5-hydroxymethylcytosine is like a bean stuck to a grape stuck to a tennis ball. Because of this change in shape, the MeCP2 protein can’t bind to the modified DNA any more. The cell therefore ‘reads’ 5-hydroxymethylcytosine in the same way as it reads unmethylated DNA.

Many of the techniques used until very recently looked for the presence of DNA methylation. They often couldn’t distinguish between unmethylated DNA and 5-hydroxymethylated DNA. This means that many of the papers which refer to decreased DNA methylation may actually have been detecting increased 5-hydroxymethylation without knowing it. It’s currently unproven, but it may be that instead of actually demethylating DNA, as reported in some of the behavioural studies, neurons really convert 5-methylcytosine to 5-hydroxymethylcytosine. The techniques for studying 5-hydroxymethylcytosine are still under development but we do know that neurons contain higher levels of this chemical than any other cell type²¹.

Remember, remember

Despite these controversies, research is continuing into the importance of epigenetic modifications in brain function. One area that is attracting a lot of attention is the field of memory. Memory is an incredibly complex phenomenon. Both the hippocampus and a region of the brain called the cortex are involved in memory, but in different ways. The hippocampus is mainly involved in consolidating memories, as our brains decide what we are going to remember. The hippocampus is fairly plastic in the way that it operates, and this seems to be associated with transient changes in DNA methylation, again through fairly uncharacterised mechanisms. The cortex is used for longer-term storage of memories. When memories are stored in the cortex, there are prolonged changes in DNA methylation.

The cortex is like a hard drive on a computer with gigabytes of storage. The hippocampus is more like the RAM (random access memory) chip, where data are temporarily processed before being deleted, or transferred to the hard drive for permanent storage. Our brain separates out different functions to selected cell populations in different anatomical regions. This is why memory loss is rarely all-encompassing. Depending on the clinical condition, for example, either one of short-term or long-term memory may be relatively lost or remain relatively intact. It makes a lot of sense for these different functions to be separated in our brains. Just try to imagine life if we remembered everything that ever happened – the phone number that we dialled only once, every word a dull stranger said to us on a train, or the canteen menu from a wet Wednesday three years ago.

The complexity of our memory systems is one of the reasons why it is quite a difficult area to study, because it can be difficult to set up experiments where we are absolutely sure which aspects of memory our experimental techniques are actually addressing. But one thing we know for sure is that memory involves long-term changes in gene expression, and in the way neurons make connections with one another. And that again leads to the hypothesis that epigenetic mechanisms may play a role.

In mammals, both DNA methylation and histone modifications play a role in memory and learning. Rodent studies have shown that these changes may be targeted to very specific genes in discrete regions of the brain, as we have come to expect. For example, the DNA methyltransferase proteins DNMT3A and DNMT3B increase in expression in the adult rat hippocampus in a particular learning and memory model. Conversely, treating these rats with a DNA methyltransferase inhibitor such as 5-azacytidine blocks memory formation and affects both the hippocampus and the cortex²².

A particular histone acetyltransferase (protein which adds acetyl groups to histones) gene is mutated in a human disorder called Rubinstein-Taybi syndrome. Mental retardation is a frequent symptom in this disease. Mice with a mutant version of this gene also have low levels of histone acetylation in the hippocampus, as we would predict. They also have major problems in long-term memory processing in the hippocampus²³. When these mice were treated with SAHA, the histone deacetylase inhibitor, acetylation levels in the hippocampus went up, and the memory problems improved²⁴.

SAHA can inhibit many different histone deacetylases, but in the brain some of its targets seem to be more important than others. The two most highly expressed enzymes of this class are HDAC1 and HDAC2. These differ in the ways they are expressed in the brain. HDAC1 is predominantly expressed in neural stem cells, and in a supportive, protective population of non-neurons called glial cells. HDAC2 is predominantly expressed in neuronal cells²⁵, so it’s unsurprising that this is the histone deacetylase that is most important in learning and memory.

Mice whose neurons over-express Hdac2 have poor long-term memory, even though their short-term memory is fine. Mice whose neurons don’t express any Hdac2 have excellent memories. These data show us that Hdac2 has a negative effect on memory storage. The neurons which over-expressed Hdac2 formed far fewer connections than normal, whereas the opposite was true for the neurons lacking Hdac2. This supports our model of epigenetically-driven changes in gene expression ultimately altering complex networks in the brain. SAHA improves memory in the mice that over-express Hdac2, presumably by dampening down its effects on histone acetylation and gene expression. SAHA also improves memory in normal mice²⁶.

In fact, increased acetylation levels in the brain seem to be consistently associated with improved memory. Learning and memory both improved in mice kept in conditions known as environmentally enriched. This is a fancy way of saying they had access to two running wheels and the inside of a toilet roll. The histone acetylation levels in the hippocampus and cortex were increased in the mice in the more entertaining surroundings. Even in these mice, the histone acetylation levels and memory skills improved yet further if they were treated with SAHA²⁷.

We can see a consistent trend emerging. In various different model systems, learning and memory improve when animals are treated with DNA methyltransferase inhibitors, and especially with histone deacetylase inhibitors. As we saw in the last chapter, there are drugs licensed in both these classes, such as 5-azacytidine and SAHA, respectively. It’s very tempting to speculate about taking these anti-cancer drugs and using them in conditions where memory loss is a major clinical problem, such as Alzheimer’s disease. Perhaps we might even use them as general memory enhancers in the wider population.

Unfortunately, there are substantial difficulties in doing this. These drugs have side-effects which can include severe fatigue, nausea and a higher risk of infections. These side-effects are considered acceptable if the alternative is an inevitable and fairly near-term death from cancer. But they might be considered less acceptable for treating the early stages of dementia, when the patient still has a relatively reasonable quality of life. And they would certainly be unacceptable for the general population.

There is an additional problem. Most of these drugs are really bad at getting into the brain. In many of the rodent experiments, the drugs were administered directly into the brain, and often into very defined regions such as the hippocampus. This isn’t a realistic treatment method for humans.

There are a few histone deacetylase inhibitors that do get into the brain. A drug called sodium valproate has been used for decades to treat epilepsy, and clearly must be getting into the brain in order to do this. In recent years, we have realised that this compound is also a histone deacetylase inhibitor. This would be extremely encouraging for trying to use epigenetic drugs in Alzheimer’s disease but unfortunately, sodium valproate only inhibits histone deacetylases very weakly. All the animal data on learning and memory have shown that stronger inhibitors work much better than weak ones at reversing these deficits.

It’s not just in disorders like Alzheimer’s disease that epigenetic therapies could be useful if we manage to develop suitable drugs. Between 5 and 10 per cent of regular users of cocaine become addicted to the drug, suffering from uncontrollable cravings for this stimulant. A similar phenomenon occurs in rodents, if animals are allowed unlimited access to the drug. Addiction to stimulants such as cocaine is a classic example of inappropriate adaptations by memory and reward circuits in the brain. These maladaptations are regulated by long-lasting changes in gene expression. Changes in DNA methylation, and in how methylation is read by MeCP2, underpin this addiction. This happens via a set of poorly understood interactions which include signalling factors, DNA and histone modifying enzymes and readers, and miRNAs. Related pathways also underpin addiction to amphetamines²⁸^,²⁹.

If we return to the starting point of this chapter, it’s clear that there’s a major need to stop children who have suffered early trauma from developing into adults with a substantially higher than normal risk of mental illness. It’s very appealing to think we might be able to use epigenetic drug therapies to improve their life chances. Unfortunately, one of the problems in designing therapies for children who have been abused or neglected is that it’s actually pretty difficult to identify those who will be permanently damaged as adults, and those who will have healthy, happy and fulfilled lives. There are enormous ethical dilemmas around giving drugs to children, when we can’t be sure if an individual child actually needs the treatment. In addition, clinical trials to determine if the drugs actually do any good would need to last for decades, which makes them economically almost a non-starter for any pharmaceutical company.

But we mustn’t end on too negative a note. Here’s a great story about an epigenetic event and behaviour. There is a gene called Grb10 that is involved in various signalling pathways. It’s an imprinted gene, and the brain only expresses the paternally inherited copy. If we switch off this paternal copy, the mouse can’t produce any Grb10 protein, and the animals develop a very odd phenotype. They nibble off the face fur and whiskers of other mice in the same cage. This is a sort of aggressive grooming, a bit like a pecking order in chickens. In addition, if faced with a big mouse that they don’t know, the Grb10 mutant mice don’t back away – they stand their ground³⁰.

Switching off Grb10 in the brain results in what might sound like a rather impressive, kick-ass kind of a mouse. It maybe even seems odd that this gene is normally switched on in the brain. Wouldn’t mice that switched off Grb10 be the butchest, most successful mice? Actually, it’s more likely that they’d be the mice most likely to get themselves beaten up. There are a lot of mice in the world, and they encounter each other pretty frequently. It pays to recognise when you are out-gunned.

When the Grb10 gene is switched off in the brain, it’s like a bad Friday night for the mouse. Let’s put this in human terms so we can see why. You’re down the pub when a person twice your size and all muscle knocks against you and you spill your pint.

When this gene is switched off, it’s as if you have a friend next to you who says, ‘Go on, you can take him/her, don’t wimp out.’ We all know how badly those scenarios tend to play out. So let’s end this chapter by raising a cheer for imprinted Grb10, the gene that likes to say, ‘Leave it mate, it’s not worth it.’

Prev Next