The Epigenetics Revolution

Chapter 13. The Downhill Slope

Time moves forward, we age. It’s inevitable. And as we get older, our bodies change. Once we’re past our mid-thirties most of us would agree that it gets harder and harder to sustain the same level of physical performance. It doesn’t matter if it’s how fast we can run, how far we can cycle before needing to stop for a break, or how quickly we recover from a big night out. The older we get, the harder everything seems to become. We develop new aches and pains, and succumb more easily to annoying little infections.

Ageing is something we are good at recognising in the people around us. Even quite small children can tell the difference between the young and the very old, even if they are a bit hazy on everyone in the middle. Adults can easily tell the difference between a 20-year-old and a 40-something individual, or between two people who are 40 and 65.

We can categorise individuals instinctively into approximate age groups not because they give off an intrinsic radio signal about the number of years they have been on earth, but because of the physical signs of ageing. These include the loss of fat beneath the skin, making our features more drawn and less ‘fresh-faced’. There are the wrinkles, the fall in muscle tone, that slight curvature to the spine.

The growth of the cosmetic surgery industry appears to be relentless and shows how desperate we can be to fight the symptoms of ageing. Figures released in 2010 showed that in the top 25 countries covered in a survey by the International Society of Aesthetic Plastic Surgery, there were over eight and a half million surgical procedures carried out in 2009, and about the same number of non-surgical procedures, such as Botox and dermo-abrasion. The United States topped the list, with Brazil and China fighting for second place[234].

As a society, we don’t seem to mind really about the number of years we’ve been alive, but we dislike intensely the physical decline that accompanies them. It’s not just the trivial stuff either. One of the greatest risk factors for developing cancer is simply being old. The same is true for conditions such as Alzheimer’s disease and stroke.

Most breakthroughs in human healthcare up until now have improved both longevity and quality of life. That’s partly because many major advances targeted early childhood deaths. Vaccination against serious diseases such as polio, for example, has hugely improved both childhood mortality figures (fewer children dying) and morbidity in terms of quality of life for survivors (fewer children permanently disabled as a result of polio).

There is a growing debate around the issue sometimes known as human life extension, which deals with extending the far end of life, old age. Human life extension refers to the concept that we can use interventions so that individuals will live to a greater age. But this takes us into difficult territory, both socially and scientifically. To understand why, it’s important to establish what ageing really is, and why it is so much more than just being alive for a long time.

One useful definition of ageing is ‘the progressive functional decline of tissue function that eventually results in mortality’[235]. It’s this functional decline that is the most depressing aspect of ageing for most people, rather than the final destination.

Generally speaking, most of us recognise the importance of this quality of life issue. For example, in a survey of 605 Australian adults in 2010, about half said they would not take an anti-ageing pill if one were developed. The rationale behind their choice was based around quality of life. These respondents didn’t believe such a pill would prolong healthy life. Simply living for longer wasn’t attractive, if this was associated with increasing ill-health and disability. These respondents did not wish to prolong their own lives, unless this was associated with improved health in later years[236].

There are thus two separate aspects to any scientific discussion of ageing. These are lifespan itself, and the control of late-onset disorders associated with ageing. What isn’t clear is the degree to which it is possible or reasonable to separate the two, at least in humans.

Epigenetics definitely has a role to play in ageing. It’s not the only factor that’s important, but it is significant. This field of epigenetics and ageing has also led to one of the most acrimonious disputes in the pharmaceutical sector in recent years, as we’ll see towards the end of this chapter.

We have to ask why our cells malfunction as we get older, leaving us more at risk of illnesses that include cancer, type 2 diabetes, cardiovascular disease and dementia, amongst a host of other conditions. One reason is because the DNA script in the cells of our body begins to change for the worse. It accumulates random alterations in sequence. These are somatic mutations, which affect the tissue cells of the body, but not the germline. Many cancers have changes in the DNA sequence, often caused by quite large rearrangements between chromosomes, where genetic material is swapped from one chromosome to another.

But as we’ve seen, our cells contain multiple mechanisms for keeping the DNA blueprint as intact as possible. Wherever possible, a cell’s default setting is to maintain the genome in its original state, as much as it can. But the epigenome is different. By its very nature, this is more flexible and plastic than the genome. Because of this, it is probably not surprising that epigenetic modifications change as animals age. The epigenome may eventually turn out to be far more prone to changes with age than the genome, because the epigenome is more naturally variable than the genome anyway.

We met some examples of this in Chapter 5, where we discussed how genetically identical twins become less identical epigenetically as they age. The issue of how the epigenome changes as we age has been examined even more directly. Researchers studied two large groups of people from Iceland and from Utah, who have been part of on-going long-term population studies. DNA was prepared from blood samples that had been taken from these people between eleven and sixteen years apart. Blood contains red and white blood cells. The red blood cells carry oxygen around the body, and are essentially just little bags of haemoglobin. The white blood cells are the cells that generate immune responses to infections. These cells retain their nuclei and contain DNA.

The researchers found that the overall DNA methylation levels in the white blood cells of some of these individuals changed over time. The change wasn’t always the same. In some individuals, the DNA methylation levels went up with age, in others they dropped. The direction of change seemed to run in families. This may mean that the age-related change in DNA methylation was genetically influenced, or affected by shared environmental factors in a family. The scientists also looked in detail at methylation at over 1,500 specific CpG sites in the genome. These sites were mainly associated with protein-coding genes. They found the same trends at these specific sites as they had seen when looking at overall DNA methylation levels. In some individuals, site-specific DNA methylation was increased whereas in others it fell. DNA methylation levels were increased or decreased by at least 20 per cent in around one tenth of the people in the study.

The authors stated in their conclusion that ‘these data support the idea of age-related loss of normal epigenetic patterns as a mechanism for late onset of common human diseases’[237]. It’s true that the data are consistent with this model of epigenetic mechanisms leading to late onset disease, but there are limitations, which we should bear in mind.

In particular, these types of studies highlight important correlations between epigenetic change and diseases of old age, but they don’t prove that one event causes the other. Deaths through drowning are most common when sales of suntan lotion are highest. From this one could infer that sun tan lotions have some effect on people that makes them more likely to drown. The reality of course is that sales of suntan lotion rise during hot weather, which is also when people are most likely to go swimming. The more people who swim, the greater the number who will drown, on average. There is a correlation between the two factors we have monitored (sales of sun block and deaths by drowning) but this isn’t because one factor causes the other.

So, although we know that epigenetic modifications change over time, this doesn’t prove that these alterations cause the illnesses and degeneration associated with old age. In theory, the changes could just be random variations with no functional consequences. They could just be changes in the epigenetic background noise in a cell. In many cases, we don’t even yet know whether the altered patterns of epigenetic modifications lead to changes in gene expression. Addressing this question is hugely challenging, and particularly difficult to assess in human populations.

Having said that, there are some epigenetic modifications that are definitely involved in disease initiation or progression. The case for these is strongest in cancer, as we saw in Chapter 11. The evidence includes the epigenetic drugs which can treat certain specific types of cancer. It also includes the substantial amounts of data from experimental systems. These show that altering epigenetic regulation in a cell increases the likelihood of a cell becoming cancerous, or can make an already cancerous cell more aggressive.

One of the areas that we dealt with in Chapter 11 was the increase in DNA methylation that frequently occurs at the promoters of tumour suppressor genes. This increased DNA methylation switches off the expression of the tumour suppressor genes. Oddly enough, this increase in DNA methylation at specific sites is often found against an overall background of decreased DNA methylation in many other areas of the genome in the same cancer cell. This decrease in methylation may be caused by a fall in expression or activity of the maintenance DNA methyltransferase, DNMT1. This decrease in global DNA methylation may also contribute to the development of cancer.

To investigate this, Rudi Jaenisch generated mice which only expressed Dnmt1 protein at about 10 per cent of normal levels in their cells. The levels of DNA methylation in their cells were very low compared with normal mice. In addition to being quite stunted at birth, these Dnmt1 mutant mice developed aggressive tumours of the immune system (T cell lymphomas) when they were between four and eight months of age. This was associated with rearrangements of certain chromosomes, and especially with an extra copy of chromosome 15 in the cancer cells.

Professor Jaenisch speculated that the low levels of DNA methylation made the chromosomes very unstable and prone to breakages. This put the chromosomes at high risk of joining up in inappropriate ways. It’s like snapping a pink stick of rock and a green stick of rock to create four pieces in total. You can join them back together again using melted sugar, to create two full-length items of tooth-rotting confectionery. But if you do this in the dark, you may find that sometimes you have created ‘hybrid’ rock sticks, where one part is pink and the other is green.

The end result of increased chromosome instability in Rudi Jaensich’s mice was abnormal gene expression. This in turn led to too much proliferation of highly invasive and aggressive cells, resulting in cancer[238][239]. These data are one of the reasons why DNMT inhibitors are unlikely to be used as drugs in anything other than cancer. The fear is that the drugs would cause decreased DNA methylation in normal cells, which might pre-dispose some cell types towards cancer.

These data suggest that the DNA methylation level per se is not the critical issue. What matters is where the changes in DNA methylation take place in the genome.

The generalised decrease in DNA methylation levels that comes with age has also been reported in other species than humans and mice, ranging from rats to humpback salmon[240]. It’s not entirely clear why low levels of DNA methylation are associated with instability of the genome. It may be because high levels of DNA methylation can lead to a very compacted DNA structure, which may be more structurally stable. After all, it’s easy to snip through a single extended wire with a pair of cutters, but much harder if that wire has been squashed down into a dense knot of metal.

It’s important to appreciate just how much effort cells put into looking after their chromosomes. If a chromosome breaks, the cell will repair the break if it can. If it can’t, the cell may trigger an auto-destruct mechanism, essentially committing cellular suicide. That’s because damaged chromosomes can be dangerous. It’s better to kill one cell, than for it to survive with damaged genetic material. For instance, imagine one copy of chromosome 9 and one copy of chromosome 22 break in the same cell. They could get repaired properly, but sometimes the repair goes wrong and part of chromosome 9 joins up with part of chromosome 22.

This rearrangement of chromosomes 9 and 22 actually happens relatively frequently in cells of the immune system. In fact it happens so often that this 9:22 hybrid has a specific name. It’s called the Philadelphia chromosome, after the city where it was first described. Ninety-five per cent of people who have a form of cancer called chronic myeloid leukaemia have the Philadelphia chromosome in their cancer cells. This abnormal chromosome causes this cancer in the cells of the immune system because of where the breaking and rejoining happen in the genome. The fusion of the two chromosome regions results in the creation of a hybrid gene called Bcr-Abl. The protein encoded by this hybrid gene drives cell proliferation forwards very aggressively.

Our cells have therefore developed very sophisticated and fast-acting pathways to repair chromosome breaks as rapidly as possible, in order to prevent these sorts of fusions. To do this, our cells must be able to recognise loose ends of DNA. These are created when a chromosome breaks in two.

But there’s a problem. Every chromosome in our cell quite naturally has two loose ends of DNA, one at each end. Something must stop the DNA repair machinery from thinking these ends need to be repaired. That something is a specialised structure called the telomere. There is a telomere at each end of every chromosome, making a total of 92 telomeres per cell in humans. They stop the DNA repair machinery from targeting the ends of chromosomes.

Telomeres play a critical role in control of ageing. The more a cell divides, the smaller its telomeres become. Essentially, as we age, the telomeres get shorter. Eventually, they get so small that they don’t function properly anymore. The cells stop dividing and may even activate their self-destruct mechanisms. The only cells where this is different are the germ cells that ultimately give rise to eggs or sperm. In these cells the telomeres always stay long, so the next generation isn’t short-changed when it comes to longevity. In 2009, the Nobel Prize in Physiology or Medicine was awarded to Elizabeth Blackburn, Carol Greider and Jack Szostak for their work showing how telomeres function.

Since telomeres are so important in ageing, it makes sense to consider how they interact with the epigenetic system. The DNA of vertebrate telomeres consists of hundreds of repeats of the sequence TTAGGG. There are no genes at the telomere. We can also see from the sequence that there are no CpG motifs at the telomeres, so there can’t be any DNA methylation. If there are any epigenetic effects that make a difference at the telomeres they will therefore have to be based on histone modifications.

In between the telomeres and the main parts of the chromosome are stretches of DNA referred to as sub-telomeric regions. These contain lots of runs of repetitive DNA. These repeats are less restricted in sequence than the telomeres. The sub-telomeric regions contain a low frequency of genes. They contain some CpG motifs so these regions can be modified by DNA methylation, in addition to histone modifications.

The types of epigenetic modifications normally found at telomeres and the sub-telomeric regions are the ones that are highly repressive. Because there are so few genes in these regions anyway, these modifications probably aren’t used to switch off individual genes. Instead, these repressive epigenetic modifications are probably involved in ‘squashing down’ the ends of the chromosomes. The epigenetic modifications attract proteins that coat the ends of the chromosomes, and help them to stay as tightly coiled up, and as dense and inaccessible as possible. It’s a little like covering the ends of a pipe in insulation.

It’s potentially a problem for a cell that all its telomeres have the same DNA sequence, because identical sequences in a nucleus tend to find and bind to one another. Such close proximity creates a big risk that the ends of different chromosomes will link up, especially if they get damaged and opened up. This can lead to all sorts of errors as the cell struggles to sort out chains of chromosomes, and may result in ‘mixed-up’ chromosomes similar to the one that causes chronic myeloid leukaemia. By coating the telomeres with repressive modifications that make the ends of the chromosomes really densely packed, there’s less chance that different chromosomes will join up inappropriately.

The cell is, however, stuck with a dilemma, as shown in Figure 13.1.

^{Figure 13.1 Both abnormal shortening and lengthening of telomeres have potentially deleterious consequences for cells.}

If the telomeres get too short, the cell tends to shut down. But if the telomeres get too long, there’s an increased risk of different chromosomes linking up, and creating new cancer-promoting genes. Cell shut-down is probably a defence mechanism that has evolved to minimise the risk of creating new cancer-inducing genes. This is one of the reasons why it’s likely to be very difficult to create drugs that increase longevity without increasing the risk of cancer as well.

What happens when we create new pluripotent cells? This could be through somatic cell nuclear transfer, as we saw in Chapter 1, or through creation of iPS cells, as we saw in Chapter 2. We may use these techniques to create cloned non-human animals, or human stem cells to treat degenerative diseases. In both cases, we want to create cells with normal longevity. After all, there is little point creating a new prize stallion, or cells to implant into the pancreas of a teenager with diabetes, if the horse or the cells die of telomere ‘old age’ within a short time.

That means we need to create cells with telomeres that are about the same length as the ones in normal embryos. In nature, this occurs because the chromosomes in the germline are protected from telomere shortening. But if we are generating pluripotent cells from relatively adult cells, we are dealing with nuclei where the telomeres are already likely to be relatively short, because the ‘starter’ cells were taken from adults, whose chromosomes are getting shorter with age.

Luckily, something unusual happens when we create pluripotent cells experimentally. When iPS cells are created, they switch on expression of a gene called telomerase. Telomerase normally keeps telomeres at a healthy long length. However, as we get older, the telomerase activity in our cells starts to drop. It’s important to switch on telomerase in iPS cells, or the cells would have very short telomeres and wouldn’t create very many generations of daughter cells. The Yamanaka factors induce the expression of high levels of telomerase in iPS cells.

But we can’t use telomerase to try to reverse or slow human ageing. Even if we could introduce this enzyme into cells, perhaps by using gene therapy, the chances of inducing cancers would be too great. The telomere system is finely balanced, and so is the trade-off between ageing and cancer.

Both histone deacetylase inhibitors and DNA methyltransferase inhibitors improve the efficiency of the Yamanaka factors. This might be partly because these compounds help to remove some of the repressive modifications at the telomeres and sub-telomeric regions. This may make it easier for telomerase to build up the telomeres as the cells are reprogrammed.

The interaction of epigenetic modifications with the telomere system takes us a little further away from a simple correlation between epigenetics and ageing. It moves us closer to a model where we can start to develop confidence that epigenetic mechanisms may actually play a causative role in at least some aspects of ageing.

To investigate this more fully, scientists have made extensive use of an organism we all encounter every day of our lives, whenever we eat a slice of bread or drink a bottle of beer. The scientific term for this model organism is Saccharomyces cerevisiae, but we generally know it by its more common name of brewer’s yeast. We’ll stick with yeast, for short.

Although yeast is a simple one-celled organism, it is actually very like us in some really fundamental ways. It has a nucleus in its cells (bacteria don’t have this) and contains many of the same proteins and biochemical pathways as higher organisms such as mammals.

Because yeast are such simple organisms, they’re very easy to work with experimentally. A yeast cell (mother) can generate new cells (daughters) in a relatively straightforward way. The mother cell copies its DNA. A new cell buds off from the side of the mother cell. This daughter cell contains the correct amount of DNA, and drifts off to act as a completely independent new single-celled organism. Yeast divide to form new cells really quickly, meaning experiments can be run in a few weeks rather than taking the months or years that are required for some higher organisms, and especially mammals. Yeast can be grown either in a liquid soup, or plated out onto a Petri dish, making them very easy to handle. It’s also fairly straightforward to create mutations in interesting genes.

Yeast have a specific feature that has made them one of the favourite model systems of epigeneticists. Yeast never methylate their DNA, so all epigenetic effects must be caused by histone modifications. There’s also another helpful feature of yeast. Each time a yeast mother cell gives rise to a daughter cell, the bud leaves a scar on the mother. This makes it really easy to work out how many times a cell has divided. There are two types of ageing in yeast and these each have parallels to human ageing, as shown in Figure 13.2.

^{Figure 13.2 The two models of ageing in yeast, relevant for dividing and non-dividing cells.}

Most of the emphasis in ageing research has been on replicative ageing, and trying to understand why cells lose their ability to divide. Replicative ageing in mammals is clearly related to some obvious symptoms of getting older. For example, skeletal muscle contains specialised stem cells called satellite cells. These can only divide a certain number of times. Once they are exhausted, you can’t create new muscle fibres.

Substantial progress has been made in understanding replicative ageing in yeast. One of the key enzymes in controlling this process is called Sir2 and it’s an epigenetic protein. It affects replicative ageing in yeast through two pathways. One seems to be specific to yeast, but the other is found in numerous species right through the evolutionary tree, all the way up to humans.

Sir2 is a histone deacetylase. Mutant yeast that over-express Sir2 have a replicative lifespan that is at least 30 per cent longer than normal[241]. Conversely, yeast that don’t express Sir2 have a reduced lifespan[242], about 50 per cent shorter than usual. In 2009, Professor Shelley Berger, an incredibly dynamic scientist at the University of Pennsylvania whose group has been very influential in molecular epigenetics, published the results of a really elegant set of genetic and molecular experiments in yeast.

Her research showed that the Sir2 protein influences ageing by taking acetyl groups off histone proteins, and not through any other roles this enzyme might carry out[243]. This was a key experiment, because Sir2, like many histone deacetylases, has rather loose molecular morals. It doesn’t just remove acetyl groups from histone proteins. Sir2 will take acetyl groups away from at least 60 other proteins in the cell. Many of these proteins have nothing to do with chromatin or with gene expression. Shelley Berger’s work was crucial for demonstrating that Sir2 influences ageing precisely because of its effects on histone proteins. The altered epigenetic pattern on the histones in turn influenced gene expression.

These data, showing that epigenetic modifications of histones really do have a major influence on ageing, gave scientists in this field a big confidence boost that they were on the right track. The importance of Sir2 doesn’t seem to be restricted to yeast. If we over-express Sir2 in our favourite worm, C. elegans[244], the worm lives longer. Fruit flies that over-expressed Sir2 had up to a 57 per cent increase in lifespan[245]. So, could this gene also be important in human ageing?

There are seven versions of the Sir2 gene in mammals, called SIRT1 through to SIRT7. Much of the attention in the human field has focused on SIRT6, an unusual histone deacetylase. The breakthroughs in this field have come from the laboratory of Katrin Chua, a young Assistant Professor at the Stanford Center on Longevity (and also the sister of Amy Chua who wrote the highly controversial mothering memoir Battle Hymn of the Tiger Mother).

Katrin Chua created mice which never expressed any Sirt6 protein, even during their development (they are known as Sirt6 knockout mice). These animals seemed normal at birth, although they were rather small. But from two weeks of age onwards they developed a whole range of conditions that mimicked the ageing process. These included loss of fat under the skin, spinal curvature, and metabolism deficits. The mice died by one month of age, whereas a normal mouse can live for up to two years under laboratory conditions.

Most histone deacetylases are very promiscuous. By this we mean they will deacetylate any acetylated histone they can find. Indeed, as mentioned above, many don’t even restrict themselves to histones, and will take acetyl groups off all sorts of proteins. However, SIRT6 isn’t like this. It only takes the acetyl groups off two specific amino acids – lysine 9 and lysine 56, both on histone H3. The enzyme also seems to have a preference for histones that are positioned at telomeres. When Katrin Chua knocked out the SIRT6 gene in human cells, she found that the telomeres of these cells got damaged, and the chromosomes began to join up. The cells lost the ability to divide any further and pretty much shut down most of their activities[246].

This suggested that human cells need SIRT6 so that they can maintain the healthy structures of telomeres. But this wasn’t the only role of the SIRT6 protein. Acetylation of histone 3 at amino acid 9 is associated with gene expression. When SIRT6 removes this modification, this amino acid can be methylated by other enzymes present in the cell. Methylation at this position on the histone is associated with gene repression. Katrin Chua performed further experiments which confirmed that changing the expression levels of SIRT6 changed the expression of specific genes.

SIRT6 is targeted to specific genes by forming a complex with a particular protein. Once it’s present at those genes, SIRT6 takes part in a feedback loop that keeps driving down expression of the gene, in a classic vicious cycle. When the SIRT6 gene is knocked out, the levels of histone acetylation at these genes stays high because the feedback loop can’t be switched on. This drives up expression of these target genes in the SIRT6 knockout mice. The target genes are ones which promote auto-destruction, or the cell’s entry into a state of permanent stasis known as senescence. This effect explains why SIRT6 knockdown is associated with premature ageing[247]. It’s because genes that accelerate processes associated with ageing are switched on too soon, or too vigorously, at a young age.

It’s a little like a crafty manufacturer installing an inbuilt obsolescence mechanism into a product. Normally, the mechanism doesn’t kick in for a certain number of years, because if the obsolescence activates too early, the manufacturer will get a reputation for prematurely shoddy goods and nobody will buy them at all. Knocking out SIRT6 in cells is a little like a software glitch that activates the inbuilt obsolescence pathway after, say, one month instead of two years.

Other SIRT6 target genes are associated with provoking inflammatory and immune responses. This is also relevant to ageing, because some conditions that become much more common as we age are a result of increased activation of these pathways. These include certain aspects of cardiovascular disease and chronic conditions such as rheumatoid arthritis.

There is a rare genetic disease called Werner’s syndrome. Patients with this disorder age faster and at an earlier age than healthy individuals. The condition is caused by mutations in a gene that is involved in the three-dimensional structure of DNA, keeping it in the correct conformation and wound up to the right degree of tightness for a specific cell type[248]. The normal protein binds to telomeres. It binds most effectively when the histones at the telomeres have lost the acetyl group at amino acid 9 on histone H3. This is the precise modification removed by the SIRT6 enzyme. This further strengthens the case for a role of SIRT6 in control of ageing[249].

Given that SIRT6 is a histone deacetylase, it might be interesting to test the effect of a histone deacetylase inhibitor on ageing. We would predict that it would have the same effects as knocking down expression of the SIRT6 enzyme, i.e. it would accelerate ageing. This might give us pause for thought when we plan to treat patients with histone deacetylase inhibitors such as SAHA. After all, an anti-cancer drug that makes you age faster isn’t that attractive an idea.

Fortunately, from the point of view of treating cancer patients, SIRT6 belongs to a special class of histone deacetylase enzymes called sirtuins. Unlike the enzymes we met in Chapter 11, the sirtuins aren’t affected by SAHA or any of the other histone deacetylase inhibitor drugs.

All of this begs the question of whether we are any closer to finding a pill we can offer to people to increase longevity. The data so far don’t seem promising, especially if it’s true that many of the mechanisms that underlie ageing are defences against developing cancer. There’s not a lot of point creating therapies that could allow us to live for another 50 years, if they also lead to tumours that could kill us in five. But there is one way of increasing lifespan that has proven astonishingly effective, from yeast to fruit flies, from worms to mammals. This is calorie restriction.

If you only give rodents about 60 per cent of the calories they would eat if given free access to food, there is a dramatic impact on longevity and development of age-related diseases[250]. The restricted calorie intake must start early in life and be continued throughout life to see this effect. In yeast, decreasing the amount of glucose (fuel) in the culture from 2 per cent to 0.5 per cent extended the lifespan by around 30 per cent[251].

There’s been a lot of debate on whether or not this calorie-restriction effect is mediated via sirtuins, such as Sir2 in yeast, or the versions of Sir2 in other animals. Sir2 is regulated in part by a key chemical, whose levels are affected by the amount of nutrition available to cells. That’s the reason why some authors have suggested that the two might be connected, and it’s an attractive hypothesis. There’s no debate that Sir2 is definitely important for longevity. Calorie restriction is also clearly very important. The question is whether the two work together or separately. There’s no consensus as yet on this, and the experimental findings are very influenced by the model system used. This can come down to details that at first glance might almost seem trivial, such as which strain of brewer’s yeast is used, or exactly how much glucose is in the culture liquid.

The question of how calorie restriction works might seem much less important than the fact that it does. But the mechanism matters enormously if we’re looking for an anti-ageing strategy, because calorie restriction has severe limitations for humans. Food has enormous social and cultural aspects, it’s rarely just fuel for us. In addition to these psychological and sociological issues, calorie restriction has side effects. The most obvious ones are muscle wasting and loss of libido. It’s not much of a surprise that when offered the chances of living longer, but with these side-effects, the majority of people find the prospect unattractive[252].

That’s one of the reasons that a 2006 paper in Nature, led by David Sinclair at Harvard Medical School, created such a furore. The scientists studied the effects of a compound called resveratrol on health and survival in mice. Resveratrol is a complex compound synthesised by plants, including grapes. It is a constituent of red wine. At the time of the paper, resveratrol had already been shown to extend lifespan in yeast, C. elegans and fruit flies[253][254].

Professor Sinclair and his colleagues raised mice on very high calorie diets, and treated the mice with resveratrol for six months. At the end of this six-month period, they examined all sorts of health outcomes in the mice. All the mice which had been on the high calorie diets were fat, regardless of whether or not they had been treated with resveratrol. But the mice treated with resveratrol were healthier than the untreated fat mice. Their livers were less fatty, their motor skills were better, they had fewer diabetes symptoms. By the age of 114 weeks, the resveratrol-treated mice had a 31 per cent lower death rate than the untreated animals fed the same diet[255].

We can see immediately why this paper garnered so much attention. If the same effects could be achieved in humans, resveratrol would be a get-out-of-obesity-free card. Eat as much as you like, get as fat as you want and yet still have a long and healthy life. No leaving behind one-third of every meal and losing your muscles and your libido.

How was resveratrol doing this? A previous paper from the same group showed that resveratrol activated a sirtuin protein, in this case Sirt1[256]. Sirt1 is believed to be important for the control of sugar and fat metabolism.

Professor Sinclair set up a company called Sirtris Pharmaceuticals, which continued to make new compounds based around the structure of resveratrol. In 2008 GlaxoSmithKline paid $720 million for Sirtris Pharmaceuticals to gain access to its expertise and portfolio of compounds for treating diseases of ageing.

This deal was considered expensive by many industry observers, and it hasn’t been without its problems. In 2009, a group from rival pharmaceutical company Amgen published a paper. They claimed that resveratrol did not activate Sirt1, and that the original findings represented an artefact caused by technical problems[257]. Shortly afterwards, scientists from Pfizer, another pharmaceutical giant, published very similar findings to Amgen[258].

It’s actually very unusual for large pharmaceutical companies to publish work that simply contradicts another company’s findings. There’s nothing much to be gained by doing so. Pharmaceutical companies are ultimately judged by the drugs they manage to launch successfully, and criticising a competitor in the early stages of a drug discovery programme gives them no commercial advantage. The fact that both Amgen and Pfizer went public with their findings is a demonstration of how controversial the resveratrol story had become.

Does it matter how resveratrol works? Isn’t the most important feature the fact that it has such dramatic effects? If you are trying to develop new drugs to treat human conditions, it unfortunately matters quite a lot. The authorities who license new drugs are much keener on compounds when they know how they work. This is partly because this makes it much easier to monitor for side-effects, as you can develop better theories about what to look out for. But the other issue is that resveratrol itself probably isn’t the ideal compound to use as a drug.

This is often an issue with natural products such as resveratrol, which was isolated from plants. The natural compounds may need to be altered to a greater or lesser extent, so that they circulate well in the body, and don’t have unwanted side effects. For example, artemisinin is a chemical derived from wormwood which can kill malarial parasites. Artemisinin itself isn’t taken up well by the human body so researchers developed compounds that were variants of the chemical structure of the original natural product. These variants kill malarial parasites, but are also much better than artemisinin at getting taken up by our bodies[259].

But if we don’t know exactly how a particular compound is working, it’s very hard to design and test new ones, because we don’t know how to easily check if the new compounds are still affecting the right protein.

GlaxoSmithKline is standing by its sirtuin programmes, but in a worrying development for the company they have stopped a clinical trial of a special formulation of resveratrol in a disease called multiple myeloma, because of problems with kidney toxicity[260].

The progress of sirtuin histone deacetylase activators is of keen interest to all the big players in the pharmaceutical industry. We don’t know yet if these epigenetic modifiers will set the agenda, or sound the death knell, for development of therapies specifically aimed at increasing longevity or combatting old age. So, for now, we’re still stuck with our old routine: lots of vegetables, plenty of exercise and try to avoid harsh overhead lighting – it does nobody any favours.