The Epigenetics Revolution

Chapter 7. The Generations Game

Sometimes, the best science starts with the simplest of questions. The question may seem so obvious that almost nobody thinks to ask it, let alone answer it. We just don’t challenge things that seem completely self-evident. Yet occasionally, when someone stands up and asks, ‘How does that happen?’, we all realise that a phenomenon that seems too obvious to mention, is actually a complete mystery. This is true of one of the most fundamental aspects of human biology, one we almost never think about.

When mammals (including humans) reproduce, why does this require a male and a female parent?

In sexual reproduction the small, very energetic sperm swim like crazy to get to the large, relatively sedentary egg. When a winning sperm penetrates the egg, the nuclei from the two cells fuse to create the zygote that divides to form every cell in the body. Sperm and eggs are referred to as gametes. When gametes are produced in the mammalian body, each gamete receives only half the normal number of chromosomes. This means they only have 23 chromosomes, one of each pair. This is known as a haploid genome. When the two nuclei fuse after a sperm has penetrated the egg, the chromosome number is restored to that of all ordinary cells (46) and the genome is called diploid. It’s important that the egg and the sperm are both haploid, otherwise each generation would end up with twice as many chromosomes as its parents.

We could hypothesise that the reason why mammals all have a mother and father is because that’s what we need to introduce two haploid genomes to one another, to create a new cell with a full complement of chromosomes. Certainly it’s true that this is what normally happens but this model would also imply that the only reason why biologically we need a parent of each sex is because of a delivery system.

In 2010 Professor Robert Edwards received the Nobel Prize in Physiology or Medicine for his pioneering work in the field of in vitro fertilisation, which led to the so-called test tube babies. In this work, eggs were removed from a woman’s body, fertilised in the laboratory, and re-implanted back into the uterus. In vitro fertilisation was hugely challenging, and Professor Edwards’ success in human reproduction was built on years of painstaking work in mice.

This mouse work laid the foundation for a remarkable series of experiments, which demonstrated there’s a lot more to mammalian reproduction than just a delivery system. The major force in this field is Professor Azim Surani, from Cambridge University, who started his scientific career by obtaining his PhD under the supervision of Robert Edwards. Since Professor Edwards received his early research training in Conrad Waddington’s lab, we can think of Azim Surani as Conrad Waddington’s intellectual grandson.

Azim Surani is another of those UK academics who carries his prestige very lightly, despite his status. He is a Fellow of the Royal Society and a Commander of the British Empire, and has been awarded the prestigious Gabor Medal and Royal Society Royal Medal. Like John Gurdon and Adrian Bird, he continues to break new ground in a research area that he pioneered over a quarter of a century ago.

Starting in the mid 1980s, Azim Surani carried out a programme of experiments which showed unequivocally that mammalian reproduction is much more than a matter of a delivery system. We don’t just need a biological mother and a biological father because that’s how two haploid genomes fuse to form one diploid nucleus. It actually matters enormously that half of our DNA comes from our mother and half from our father.

Figure 7.1 shows what a just-fertilised egg looks like, before the two genomes meet. It’s simplified and exaggerated, but it will serve our purpose. The haploid nuclei from the egg and the sperm are called pro-nuclei.

^{Figure 7.1 The mammalian egg just after it has been penetrated by a sperm, but before the two haploid (half the normal chromosome number) pronuclei have fused. Note the disparity in size between the pronucleus that came from the egg, and the one that originated from the sperm.}

We can see that the female pronucleus is much bigger than the male one. This is very important experimentally, as it means that we can tell the different pronuclei apart. Because we can tell them apart, scientists can transfer a pronucleus from one cell to another, and be certain about which one they transferred. They know if they transferred a pronucleus that came from the father’s sperm (male pronucleus) or from the mother’s egg (female pronucleus).

Many years ago Professor Gurdon used tiny micropipettes to transfer the nuclei from the body cells of toads into toad eggs. Azim Surani used a refinement of this technology to transfer pronuclei between different fertilised eggs from mice. The manipulated fertilised eggs were then implanted into female mice and allowed to develop.

In a slew of papers, mainly published between the years of 1984 and 1987, Professor Surani demonstrated that it’s essential to have a male and a female pronucleus in order to create new living mice. This is shown graphically in Figure 7.2.

^{Figure 7.2 A summary of the outcomes from the early work of Azim Surani. The pronucleus was removed from a mouse egg. This donor egg was then injected with two haploid pronuclei and the resulting diploid egg was implanted into a mouse surrogate mother. Live mice resulted only from the eggs which had been reconstituted with one male and one female pronucleus. Embryos from eggs reconstituted with either two male or two female pronuclei failed to develop properly and the embryos died during development.}

To control for the effects of different DNA genomes, the researchers used inbred mouse strains. This ensured that the three types of fertilised eggs shown in the diagram were genetically identical. Yet despite being genetically identical, a series of experiments from Azim Surani and his colleagues[53][54][55], along with other work from the laboratories of Davor Solter[56] and Bruce Cattanach[57] were conclusive. If the fertilised egg contained only two female pronuclei, or two male ones, no live mice were ever born. You needed a pronucleus of each sex.

This is an absolutely remarkable finding. In all three scenarios shown in the diagram, the zygote ends up with exactly the same amount of genetic material. Each zygote has a diploid genome (two copies of every chromosome). If the only factor that was important in the creation of a new individual was the amount of DNA, then all three types of fertilised eggs should have developed to form new individuals.

This led to a revolutionary concept – the maternal and paternal genomes may deliver the same DNA but they are not functionally equivalent. It’s not enough just to have the correct amount of the correct sequence of DNA. We have to inherit some from our mother and some from our father. Somehow, our genes ‘remember’ who they come from. They will only function properly if they come from the ‘correct’ parent. Just having the right number of copies of each gene, doesn’t fulfil the requirements for normal development and healthy life.

We know that this isn’t some strange effect that only applies to mice, because of a naturally occurring human condition. In about one in 1500 human pregnancies, for example, there is a placenta in the uterus but there is no foetus. The placenta is abnormal, covered in fluid-filled, grape-like lumps. This structure is called a hydatidiform mole, and in some Asian populations the frequency of these molar pregnancies can be as high as 1 in 200. The apparently pregnant women gain weight, often more quickly than in a normal pregnancy and they also suffer morning sickness, often to a quite extreme degree. The rapidly-growing placental structures produce abnormally high levels of a hormone which is thought to be responsible for the symptoms of nausea in pregnancy.

In countries with good healthcare infrastructure, the hydatidiform mole is normally detected at the first ultrasound scan, and then an abortion-type procedure is carried out by a medical team. If not detected, the mole will usually abort spontaneously at around four or five months post-fertilisation. Early detection of these moles is important as they can form potentially dangerous tumours if they aren’t removed.

These moles are formed if an egg which has somehow lost its nucleus is fertilised. In about 80 per cent of hydatidiform molar pregnancies, an empty egg is fertilised by a single sperm, and the haploid sperm genome is copied to create a diploid genome. In about 20 per cent of cases the empty egg is fertilised simultaneously by two sperm. In both cases the fertilised egg has the correct number of chromosomes (46), but all the DNA came from the father. Because of this, no foetus develops. Just like the experimental mice, human development requires chromosomes from both the mother and the father.

This human condition and the experiments in mice are impossible to reconcile with a model based only on the DNA code, where DNA is a naked molecule, which carries information only in its sequence of A, C, G and T base-pairs. DNA alone isn’t carrying all the necessary information for the creation of new life. Something else must be required in addition to the genetic information. Something epigenetic.

Eggs and sperm are highly specialised cells – they are at the bottom of one of Waddington’s troughs. The egg and the sperm will never be anything other than an egg and a sperm. Unless they fuse. Once they fuse, these two highly specialised cells form one cell which is so unspecialised it is totipotent and gives rise to every cell in the human body, and the placenta. This is the zygote, at the very top of Waddington’s epigenetic landscape. As this zygote divides, the cells become more and more specialised, forming all the tissues of our bodies. Some of these tissues ultimately give rise to eggs or sperm (depending on our sex, obviously) and the whole cycle is ready to start again. There’s effectively a never-ending circle in developmental biology.

The chromosomes in the pro-nuclei of sperm and eggs carry large numbers of epigenetic modifications. This is part of what keeps these gametes behaving as gametes, and not turning into other cell types. But these gametes can’t be passing on their epigenetic patterns, because if they did the fertilised zygote would be some sort of half-egg, half-sperm hybrid when it clearly isn’t this at all. It’s a completely different totipotent cell that will give rise to an entirely new individual. Somehow the modifications on eggs and sperm get changed to a different set of modifications, to drive the fertilised egg into a different cell state, at a different position in Waddington’s epigenetic landscape. This is part of normal development.

Almost immediately after the sperm has penetrated the egg, something very dramatic happens to it. Pretty much all the methylation on the male pronucleus DNA (i.e. from the sperm) gets stripped off, incredibly quickly. The same thing happens to the DNA on the female pronucleus, albeit a lot more slowly. This means that a lot of the epigenetic memory gets wiped off the genome. This is vital for putting the zygote at the top of Waddington’s epigenetic landscape. The zygote starts dividing and soon creates the blastocyst – the golf ball inside the tennis ball from Chapter 2. The cells in the golf ball – the inner cell mass, or ICM – are the pluripotent cells, the ones that give rise to embryonic stem cells in the laboratory.

The cells of the ICM soon differentiate and start giving rise to the different cell types of our bodies. This happens through very tightly regulated expression of a few key genes. One specific protein, for example OCT4, switches on another set of genes, which results in a further cascade of gene expression, and so on. We have met OCT4 before – it is the most critical of all the genes that Professor Yamanaka used to reprogram somatic cells. These cascades of gene expression are associated with epigenetic modification of the genome, changing the DNA and histone marks so that certain genes stay switched on or get switched off appropriately. Here’s the sequence of epigenetic events in very early development:

The male and female pronuclei (from the sperm and the egg respectively) are carrying epigenetic modifications;

The epigenetic modifications get taken off (in the immediate post-fertilisation zygote);

New epigenetic modifications get put on (as the cells begin to specialise).

This is a bit of a simplification. It’s certainly true that researchers can detect huge swathes of DNA demethylation during stage 2 from this list. However, it’s actually more complicated than this, particularly in respect of histone modifications. Whilst some histone modifications are being removed, others are becoming established. At the same time as the repressive DNA methylation is removed, certain histone marks which repress gene expression are also erased. Other histone modifications which increase gene expression may take their place. It’s therefore too naïve to refer to the epigenetic changes as just being about putting on or taking off epigenetic modifications. In reality, the epigenome is being reprogrammed.

Reprogramming is what John Gurdon demonstrated in his ground-breaking work when he transferred the nuclei from adult toads into toad eggs. It’s what happened when Keith Campbell and Ian Wilmut cloned Dolly the Sheep by putting the nucleus from a mammary gland cell into an egg. It’s what Yamanaka achieved when he treated somatic cells with four key genes, all of which code for proteins highly expressed naturally during this reprogramming phase.

The egg is a wonderful thing, honed through hundreds of millions of years of evolution to be extraordinarily effective at generating vast quantities of epigenetic change, across billions of base-pairs. None of the artificial means of reprogramming cells comes close to the natural process in terms of speed or efficiency. But the egg probably doesn’t quite do everything unaided. At the very least, the pattern of epigenetic modifications in sperm is one that allows the male pronucleus to be reprogrammed relatively easily. The sperm epigenome is primed to be reprogrammed[58].

Unfortunately, these priming chromatin modifications (and many other features of the sperm nucleus), are missing if an adult nucleus is reprogrammed by transferring it into a fertilised egg. That’s also true when an adult nucleus is reprogrammed by treating it with the four Yamanaka factors to create iPS cells. In both these circumstances, it’s a real challenge to completely reset the epigenome of the adult nucleus. It’s just too big a task.

This is probably why so many cloned animals have abnormalities and shortened lifespans. The defects that are seen in these cloned animals are another demonstration that if early epigenetic modifications go wrong, they may stay wrong for life. The abnormal epigenetic modification patterns result in permanently inappropriate gene expression, and long-term ill-health.

All this reprogramming of the genome in normal early development changes the epigenome of the gametes and creates the new epigenome of the zygote. This ensures that the gene expression patterns of eggs and sperm are replaced by the gene expression patterns of the zygote and the subsequent developmental stages. But this reprogramming also has another effect. Cells can accumulate inappropriate or abnormal epigenetic modifications at various genes. These disrupt normal gene expression and can even contribute to disease, as we shall see later in this book. The reprogramming of the egg and the sperm prevent them from passing on from parent to offspring any inappropriate epigenetic modifications they have accumulated. Not so much wiping the slate clean, more like re-installing the operating system.

But this creates a paradox. Azim Surani’s experiments showed that the male and female pro-nuclei aren’t functionally equivalent; we need one of each to create a new mammal. This is known as a parent-of-origin effect, because it essentially shows that there are ways for a zygote and its daughter cells to distinguish between chromosomes from the mother and father. This isn’t a genetic effect, it is an epigenetic one, and so there must be some epigenetic modifications that do get transmitted from one generation to the next.

In 1987 the Surani lab published one of the first papers to give an insight into this mechanism. They hypothesised that parent-of-origin effects could be caused by DNA methylation. At that time, this was really the only chromatin modification that had been identified, so it was an excellent place to start. The researchers created genetically modified mice. These mice contained an extra piece of DNA that could get inserted randomly anywhere in the genome. The DNA sequence of this extra bit wasn’t particularly important to the experimenters. What was important was that they could easily measure how much DNA methylation was present on this sequence, and whether the amount of methylation was transmitted faithfully from parent to offspring.

Azim Surani and his colleagues examined seven lines of mice with this random insertion. In one of the seven lines, something very odd happened. When a mother passed on the inserted DNA, it was always heavily methylated in her offspring. But when a male mouse passed it on to his offspring, the mouse pups always ended up with low methylation of this foreign DNA. Figure 7.3 demonstrates this.

^{Figure 7.3 Mice generated in which a particular foreign piece of DNA was either methylated or not methylated. Black represents methylated DNA, and white represents unmethylated. When a mother passed on this foreign DNA, the DNA was always heavily methylated (black) in her offspring, regardless of whether she herself had been ‘black’ or ‘white’. The opposite was found for males, whose offspring always had unmethylated ‘white’ DNA. This was the first experimental demonstration that some regions of the genome can be marked to indicate if they were inherited via the maternal or the paternal line.}

Black represents the methylated inserted DNA, whereas white represents unmethylated DNA. Fathers always give their offspring white, unmethylated DNA whereas mothers always give their offspring black, methylated DNA. In other words, the methylation in the offspring is dependent on the sex of the parent who passed the inserted DNA onto them. It’s not dependent on what the methylation was like in the parent. For example, a ‘black’ male will always have offspring with ‘white’ DNA.

What this paper from Azim Surani[59], and another published at the same time[60], demonstrated was that when mammals create eggs and sperm, they somehow manage to barcode the DNA in these cells. It’s as if the chromosomes carry little flags. The chromosomes in sperm carry little flags that say, ‘I’m from Dad’ and the chromosomes in eggs carry little flags that say, ‘I’m from Mum’. DNA methylation is the fabric that these flags are made from.

The description that is used for this is imprinting – the chromosomes have been imprinted with information about which parent they came from originally. Imprinting and parent-of-origin effects are something we will explore in more detail in the next chapter.

What was happening to the foreign DNA in the experiments, which kept changing its methylation status as it was transmitted from parent to offspring? It had, quite by chance, got inserted into a region of the mouse DNA that carried one of these flags. As a consequence, the foreign DNA also started getting DNA methylation flags stuck to it when it was passed down the generations.

The fact that only one of seven mouse lines showed this effect suggested that not all of the genome carries these flags. If the whole genome was marked in this way, we would have expected that all the lines that were tested would show the effect. In fact, the one in seven rate suggests that these flagged regions are the exception, not the rule.

In Chapter 6 we saw that sometimes animals do inherit acquired characteristics from their parents. The work of Emma Whitelaw, amongst others, shows us that some epigenetic modifications do indeed get passed between parent and offspring, via the sperm and the egg. This type of inheritance is pretty rare, but it does strengthen our belief that there must be some epigenetic modifications that are special. They don’t get replaced when the egg and sperm fuse to form the zygote. So, although the vast majority of the mammalian genome does get reset when the egg and the sperm fuse, a small percentage of it is immune from this reprogramming.

Only 2 per cent of our genome codes for proteins. A massive 42 per cent is composed of retrotransposons. These are very odd sequences of DNA, which probably originated from viruses in our evolutionary past. Some retrotransposons are transcribed to produce RNA and this can affect the expression of neighbouring genes. This can have serious consequences for cells. If it drives up expression of genes that cause cells to proliferate too aggressively, for example, this may nudge cells towards becoming cancerous.

There’s a constant arms race in evolution, and mechanisms have evolved in our cells to control the activity of these types of retrotransposons. One of the major mechanisms that cells use is epigenetic. The retrotransposon DNA gets methylated by the cell, turning off retrotransposon RNA expression. This prevents the RNA disrupting expression of neighbouring genes. One particular class, known as IAP retrotransposons, seems to be a particular target of this control mechanism.

During reprogramming in the early zygote, the methylation is removed from most of our DNA. But IAP retrotransposons are an exception to this. The reprogramming machinery has evolved to skip these rogue elements and leave the DNA methylation marks on them. This keeps the retrotransposons in an epigenetically repressed state. This has probably evolved as a mechanism to reduce the risk that potentially dangerous IAP retrotransposons will get accidentally re-activated.

This is relevant because the two best-studied examples of transgenerational inheritance of non-genetic features are the agouti mouse and the AxinFu mouse, which we met in the previous chapter. The phenotypes in both these models are a consequence of the methylation levels of an IAP retrotransposon upstream of a gene. The DNA methylation levels in the parent get passed on to the offspring, and so does the phenotype caused by the expression levels of the retrotransposon[61].

We met other examples of transgenerational inheritance of acquired characteristics in Chapter 6, including the effects of nutrition on subsequent generations, and the transgenerational effects of environmental pollutants such as vinclozolin. Researchers are exploring the hypothesis that these environmental stimuli create epigenetic changes in the chromatin of the gametes. These alterations are probably in regions that are protected from reprogramming during early development after the egg and sperm fuse.

Like John Gurdon, Azim Surani has continued to work highly productively in a field that he pioneered. His work has been focused on how and why eggs and sperm barcode their DNA so that a molecular memory is passed on to the next generation. A large amount of Azim Surani’s initial pioneering work was dependent on manipulating mammalian nuclei by using tiny pipettes to transfer them between cells. Technically, this is a refined version of the methods that John Gurdon used so successfully fifteen years earlier. It’s oddly pleasing to consider that Professor Surani is now based at the research institute in Cambridge that is named after Professor Gurdon, and that they frequently bump into each other in the corridors and the coffee room.