Research Into Congenital Heart Disease

Report of Presentation given by Dr Nick Dennis, Senior Lecturer in Genetics at Southampton General Hospital

Dr Nick Dennis:

Thank you very much. It is very nice to be here and see one or two familiar faces. If I was in the audience I could almost guarantee that I would nod off in the next half–hour or so, especially after such a good lunch. So if anyone does, I shan't be at all cross.

So, this is an illustration of a human embryo at about seven weeks of development. At this stage the architecture of the heart is pretty much laid down. The size of this embryo is about 17 mm. If we were to go back about three or four weeks, the embryo would be about 3 mm in size –– about three weeks old. The heart is developing here. The heart, at this stage, is pretty much a straight tube, it's coming in at the back and coming out at the front. So, in those intervening three or four weeks the heart has developed into quite a complex organ, and if we look at a stage somewhere in between, you can see from the outside that some of the main chambers of the heart are beginning to be recognisable. So it's gone from a straight tube to a kinked tube. The blood is coming in here, it's going around this sort of "S" shape, and going out at the other end. This looping process with the two ends held still and the bit between getting longer, so it forms a loop, and then a loop with a kink, is the basis of the development of the heart in that period. What happens in between is that the common ventricle is sub–divided into a right and a left ventricle, and this in–flow part, the common atrium, is sub–divided into a left atrium and right atrium. The out–flow tract is sub–divided into the aorta and pulmonary artery. By the time you reach the seven–week stage there has been a sub–division of the tube into two channels. The blood is coming into the heart in the right atrium, into the right ventricle, going out through the lungs and then back in to the left atrium, the left ventricle, and then around the body.

Now, why should you want to have a geneticist to talk to you about how the heart develops? Somebody said to me over lunch that congenital heart disease doesn't usually run in families, and I would guarantee that there are not many people here today, or members of the Heart Circle, who have got a family history of congenital heart disease or more than one affected child. Indeed, it doesn't usually run in families, but occasionally there are rare genetic problems that have a major impact on the heart. It is very much a case of you don't know what you've got until it goes wrong. So by identifying these genetic problems and focussing on them we are able to identify formal processes that we wouldn't otherwise be able to see. You can't study these things very easily, but it makes it a lot easier to study them if you've got an individual where one link in the chain has gone wrong and you can focus on that link, and it tells you about something that is happening in all of us but is very difficult to find out about. Of course, all of these processes must be controlled by genes. So what I'm going to be talking about is how we can identify genes which have an effect on these developmental processes, and when these genes go wrong, they may be relevant in the occurrence of congenital heart disease.

I thought I'd spend just five or ten minutes talking about genes. Probably the best place to start is with the chromosomes, as chromosomes are the things that you can see. We've got a set of these in every cell, and this is what they look like. We've got 46 chromosomes in every cell, and they carry the genes. They carry them within the egg and the sperm cell from the parents to the offspring. You can take a picture like this, cut it up and line up the chromosomes into pairs so that here are the two number one chromosomes, one from the person's mother and one from the person's father. When we are looking at a picture like this we are looking at an individual's genetic material. All the genes are present on those chromosomes. But there is a huge difference in scale between a gene and a chromosome. Whereas we've got 46 chromosomes, we're looking there at about 80,000 genes. That's approximately the number thought to be present in humans. It is an estimate, nobody's been able to count them all yet as they haven't all been identified. So, 46 chromosomes, about 80,000 genes, and how do genes fit in to chromosomes? Well, genes are made of DNA. Here is the structure of DNA, this is the famous double helix. I'll come back to that in a minute. So you can imagine one thread of DNA running all the way from one end of the chromosome to the other. That thread of DNA is very long and is packed in a rather intricate way into the chromosome. I thought quite a good model of this was a sock. If you had a sock, a life–size sock, and say it was a chromosome, with one strand of DNA going all the way from one end to the other (represented by the woollen thread of the sock), if you unravelled it, the thread (or DNA strand) would be four miles long, if that was a chromosome – that's the difference in scale between the chromosome and the length of DNA that is contained within that chromosome.

Along the length of that DNA there are regions that are genes. The function of those genes is to specify the structure of proteins. So, here's a gene, this is the DNA strand. The way that it does it is that it makes a copy of one of these strands, and this is the message, also known as messenger RNA. So this is DNA, this is messenger RNA. That message travels out from the nucleus of the cell into the cytoplasm around the nucleus and that's where the different proteins are made. So for every protein you want to make, you have to have a gene in the nucleus that's got the message to make that particular protein. The way the message works is that proteins are made of building blocks called amino acids, and the protein is simply a string of amino acids. If this is the first amino acid, there will be a code in the gene up at this end that says put amino acid number one here, and it'll specify which one it is. And then next to that there'll be another code which will specify amino acid number two. So the sequence of amino acids in the protein is specified by the sequence of code words in the gene. Once you've got the appropriate sequence of amino acids in the protein, the protein will fold into its characteristic shape and it will do whatever job it is supposed to do in a cell. It is proteins that run our bodies and here are the sort of things proteins do: some of them are enzymes that convert one substance into another; some of them are structural proteins, many of these are fibrous structures like you have in your skin, nails, bones, heart valves, muscles etc. So they are important for their physical properties. Many are carrier proteins that go around in the blood picking something up, carrying it around in the blood and then releasing it again, maybe when it gets to some tissue. The best known example of that would be haemoglobin carrying oxygen around in the blood. Another important group of proteins is signalling proteins. These are a bit like a lock and a key, so you'll have a signal, perhaps going around in the blood or maybe diffused from one cell to the cells around it, and that will be having an affect on those cells, but only if the target cells are producing the appropriate receptor proteins. So the target cells have to produce the lock for that key to fit in to, and you need genes to make both.

So most of the things that happen in the body are run by proteins. I've just got a couple of examples of signalling systems here. This is a scheme of a signalling system: there are many different signalling systems that are of this type. Here we've got a cell. From outside the cell one of these signalling molecules (which might be a protein) is coming along and reacting with a receptor in the cell membrane. If that signal comes along it makes a switch in the receptor and that causes conversion of another protein to an activated form, and that in turn activates another protein, and that in turn activates a whole series of proteins. So there is a sort of a cascade mechanism. There are many switches in cells that work in this sort of way whereby a signal from outside the cell can make things happen inside the cell. You need genes to make all these different proteins.

Another type of switch that exists in cells is between one gene and other genes. So some genes produce proteins that regulate other genes. Here is a gene called a G1, and it produces a protein, here called P1, and that can go and sit on the DNA of a number of other genes and switch them on or switch them off. If it switches them on, these genes will then start producing their own protein. So this is a master gene that may control a whole batch of other genes. Some of the things I'm going to be talking about later on come in to this category.

There are some important things to say about the structure of DNA that are relevant to identifying genes. I have shown you that DNA is like a double helix. It is a double–stranded structure with the two strands being wound around each other. But the best analogy for it I think is a zip fastener. Here are the two strands – you can zip them together, and you can zip them apart again. So here is DNA as a zip fastener. It is not quite the same as a zip fastener as there are four kinds of "teeth" on DNA. They are called A, T, G, and C. If you have an A on one side, you've got to have a T on the other side, and if you have a C on one side, you've got to have a G on the other side. So there is a specific pairing mechanism within the two strands of DNA. That means if you unzip them, you can always tell what the partner strand was meant to be by reading this strand, and you can make a new copy of the partner strand. This is how cells reproduce their DNA every time they divide. This is a very important property of DNA. If you think about it, any genetic material has to be copied over and over again without too many mistakes coming in. Otherwise it is not much use as a genetic material. We can exploit this property of DNA for analysis and research. So, here we have a double–stranded bit of DNA where it has been unzipped. If we had made a DNA probe that matches this strand, so that A and T and C and G all pair up correctly, our probe will go and seek out its partner sequence within the original DNA. This is an incredibly powerful technique, and I'll show you some examples of it being used. You've got 3,000 billion of these little As, Ts, Cs and Gs in your DNA, and you can have a little strand of this that is about 20 long, and it will go and seek out its matching sequence from those 3,000 billion of these letters. With four letters you've got four possibilities for a one–base sequence, for a two–base sequence you've got 16 possibilities, and you can see it very quickly mounts up, with over a million possibilities for a ten–base sequence. It is a very simple language, but you can rapidly generate a lot of diversity.

So, that is a bit about how genetics works. A very important property that genes have is that they come in families, and this is an example. These are the genes that make haemoglobin. So you don't just have haemoglobin and the genes for haemoglobin. You have lots of genes for haemoglobin and some of them are on chromosome 11, some are on chromosome 16, and a related protein called maeoglobin is on chromosome 22. These genes have all been derived from an ancestral gene which gave rise to the haemoglobin D gene and the B gene, about 40 million years ago. If you go back about 200 million years, all of the genes appear to be derived from the same ancestral gene. And if you go back far enough, there was an organism that had just a single haemoglobin gene that gave rise to all of these. So, if you were to look at the structure of our genes, you'd find there is a lot of similarity between them, and that they fall in to families. I think the reason we see this is that we are very complex organisms and if you can make 10 different genes out of a single gene, then they all get slightly different functions and all begin to differ a little bit. These are not all identical, but very similar. The way this is used in human development is that foetal haemoglobin is made from the A and the G, and when a person is born, they switch from the G to the B. So adult haemoglobin is made from the A and the B gene. So it is possible to adapt slightly different functional qualities for foetal life and adult life because of this switch in the haemoglobin. That is one way in which gene families help things to work.

This is a picture of a haemoglobin molecule, which may just clarify things. It has got two proteins made by the A gene, there are the Alpha chains, these come from the A genes on chromosome 16. And then it has got two Beta chains from the B genes on chromosome 11. So it has got four sub–units, and this is the backbone of the protein if you like, the chain of amino acids. Within the haemoglobin B gene there is a code for each of these amino acids, and it is in the same sequence as the amino acids on in the protein.

Not only are genes within a species related to each other, but if you find a gene in a human, you can be certain that you will find the same gene in a mouse, and you can be pretty certain that you'll find that gene in a frog. It may be a whole family of genes. We are very similar in our genetic makeup to all other mammals, pretty similar to reptiles, and for some genes that have very important functions, you can find the same genes in yeast. So there is incredible consistency throughout the living world. This provides a way of looking at things that is very powerful, because it means that things that you learn in the mouse about what mouse genes are doing are likely to be very closely reproduced in humans. That is a theme that I will come back to, as it has been a very powerful way of identifying genes.

OK, so on to some things to do with the heart now. I think one must start with Down's Syndrome, as this is the most common genetic disorder that has a strong cardiac element. About 40% of children with Down's Syndrome have congenital heart defects, and about half of those defects are of a particular type, called atrio–ventricular defects, or AV canal defects. This is a relatively rare defect among chromosomaly normal children. In fact if you look carefully at the anatomy of the heart in Down's Syndrome, a lot of the children who don't have AV canal defects do have a heart that is getting that way but didn't quite get there. They have a normally functioning heart, but the way it has developed is a bit like an AV canal defect. So, what is going on here, and can it tell us anything about the way the heart develops?

What we have in Down's Syndrome is three copies of chromosome 21. That means that instead of two copies of these genes on chromosome 21, the child with Down's has three. This is responsible for the problems and the differences in development that they have. The question arises: what is the gene, or what are the genes, on chromosome 21 that produce these various effects, and how are they doing it? So what do we know about the genes on chromosome 21? As you can see, quite a lot of genes have been mapped to chromosome 21. This is just one part of chromosome 21 – the long arm – the short arm up here does not really do very much. Just occasionally, we see children that do not have a complete extra chromosome 21; they just have an extra copy of the long arm. Those children look almost identical to children with Down's Syndrome, so it looks as if most of the genes that are responsible for the features of Down's Syndrome when they are present in three copies are the ones down here at the bottom. That may just mean that there are more genes down here, and in fact I couldn't reproduce this entire picture as there are so many genes in this part of the arm. Each of these abbreviations is a different gene. I can't tell you what they all are – I could given time, but...

So, we don't yet know what the gene or genes is/are that produce Down's. We know it is in this region, but we don't yet know what it is. One of the ways in which people are looking at this is to look at the corresponding mouse chromosome, and it turns out that mouse chromosome 16 contains the same messages as human chromosome 21, and it is possible to create a mouse with three copies of chromosome 16, effectively a Down's Syndrome mouse. By looking at the function of different genes in that region, this may be one of the ways that we may get a lead on what the gene is in human chromosome 21 that causes these heart defects. When that has been found, it will then be possible to ask a new question: do children with AV canal defects that don't have Down's Syndrome have problems in this gene. So that is what I meant by being able to focus on a bit of development that you couldn't focus on until someone came along with a major problem there. The major problem helps you focus in on that in non–Down's Syndrome children.

I'm going to go on from Down's Syndrome to another problem with chromosomes, which some of you may have heard of. This is a historical picture from a fairly old book. Shprintzen was a cleft palate doctor in New York, and he described this syndrome. He was seeing people with cleft palates who had a particular facial appearance, and quite a lot of them had heart problems: you see it say ventricular septal defect 70–75%, Fallot's tetralogy, right aortic arch and so on. I think a bit before that, another doctor who was an immunologist, called Di George, was noticing immune problems, absence of the thymus (one of the main sources of T lymphocytes), and also that many of these children had aortic arch anomalies. So they had some overlap with the Sprintzen cases. But at the time when this book was written, they were thought to be two different syndromes. And then about ten or twelve years ago, it was noticed that some of these children in both categories had small abnormalities on chromosome 22. I don't have a picture of one of my own patients because I don't have a patient who has consented to let me show their picture (just because I haven't asked them, probably), but this is from the textbook. This is a patient illustrating Shprintzen Syndrome, showing the characteristic facial appearance. You may say that this is a normal face, and indeed it is, but to a clinical geneticist there are some unusual features, including the slightly up–slanting eyes, rather tubular–shaped nose, and rather small mouth, plus the rather long spindly fingers. So there are subtle features. It is not as obvious a syndrome as Down's Syndrome, but if you see a few people with this you start to recognise it. So, as I said, they started to find these subtle abnormalities on chromosome 22. It turned out that a lot of these abnormalities were not visible on the normal chromosome microscopy, and it was only because we were getting new techniques at the time that we were able to start seeing these.

So, these are some chromosomes, but they've had a new technique applied – a labelled DNA probe. Here we've got chromosome 22, and here it's had two DNA probes applied – one of them labelled with a green fluorescent dye, and one with a red fluorescent dye. These DNA probes are little bits of DNA that will pair specifically with their partner strands somewhere on the chromosome. You can see on this chromosome 22 that both the red and the green are stuck on. But on this chromosome 22, you only see the green signal – the red signal is missing. And this tells us that the part of the chromosome that the probe ought to be sticking to is absent. However if you looked at that patient's chromosomes down the microscope, you would not see any difference. This approach, the FISH test (which stands for Fluorescent Inset Hybridisation) came along about eight or ten years ago, and this 22q11–deletion syndrome is something we see quite a lot. We did a study at Southampton where we looked at about 50 babies who came through the cardiac unit, and we concentrated on babies with what are called cono–truncal abnormalities, these are outflow tract problems, like Tetralogy of Fallot, interrupted aortic arch, persistent cruncus, some types of ventricular septal defect, and the clinical geneticists went in and examined them all, and said, before we knew the result of the FISH test, whether we thought this baby was going to have a 22q11 deletion or not. Out of that 50, I think we picked out about 15 or so that we thought might have a 22q11 deletion. Of those 15, something like six did. So, we slightly over–diagnosed it, but we didn't miss any cases. So you can pick up these subtle features. Sometimes you pick them up perhaps when they're not genuine, or it may be that some of these patients have smaller deletions that this probe still isn't picking up, and they may turn out to have something in this region.

So now we want to ask the question: within this little deletion, is there a gene, which when you only have one copy instead of two, gives you a particular type of problem with cardiac development. This time it is with those outflow tracts. If you remember, there was a common tube going out at the top end of the heart that has to be subdivided. It comes up from the two ventricles, it spirals round and going into the pulmonary artery and the aorta. That is the area where there are problems in these children. So we can go to the genetic databases and ask what are the genes on chromosome 22. This is the area of the chromosome that we are interested in – this is where the deletion is. We don't yet know which of the genes on this string of genes is the one that makes you have a cono–truncal heart defect if you have one copy deleted. It doesn't do it 100% of the time, incidentally – so you can have a 22q11 deletion and not have a heart defect. So, it is quite a variable syndrome. It is very important to the families that have a child with this, but in a wider context, scientifically, it is even more important because it is going to give us a way in to finding a particular gene which controls an aspect of the heart's development.

This is a paper that was recently published in a journal called the PNAS (the Proceedings of the National Academy of Sciences in the USA). This is a very high–status scientific journal: people who get their work published in it think very highly of themselves, quite rightly, because it is very difficult to get your work published in certain nature/science journals, and they have very high standards for the work they publish. So, this is the human 22q11 region, and they've drawn it out with all the genes that they know, and as this area here is the one we're interested in, they have enlarged it, and they've put all the DNA probes here and these little triangles are genes that they know are in that region. They have compared it with the mouse region, and you can see that the mouse region is not quite the same, but most of the DNA sequences in the human are also in the mouse. However, if you compare the human chromosome and the mouse chromosome, you can see that the positions are somewhat different. So, there is a gene called HCF2 in the human that is in a different position in the mouse. It looks as if there have been a couple of inversions somewhere along the line of evolution between the ancestral mouse and the ancestral human. All the same genes are there, but they are slightly rearranged.

The reason the scientists are doing this is as a way of focussing in on these genes and doing things in the mouse that would not be possible in humans. For example, we can work out exactly when these genes are switched on in development, and what happens if you interfere with these genes. It is now possible to create a mouse with a defect in a specific gene – the so–called "knock–out" mouse –– because you can knock out a gene in the mouse and see what effect it has on it. So that gives you a much more powerful way of finding things. Again, the story isn't finished, because the gene or genes in this region that have an effect on the heart have not been identified.

Williams Syndrome is a condition that some people here may be familiar with. This is also caused by a very small deletion on chromosome 7. The sequence of events was somewhat different in this case. The first thing that happened was that it came through people who had supra–valvular aortic stenosis. This means that there is an abnormal narrowing of the aorta above the aortic valve, at the root of the aorta, outside the heart. This condition can run in the family as a straight inherited disorder – not with Williams Syndrome. It can go from parent to child with a 50–50 risk: what we call dominant inheritance. The gene responsible was discovered through a stroke of luck. One or two families with symptoms were found to have a little chromosome rearrangement running through chromosome 7. So the scientists looked at what was on chromosome 7 at that point and discovered that the elastin gene was there. Elastin is a structural protein, and there's loads and loads of it in your aorta: it is the stuff that makes your aorta resilient. When they looked at the elastin gene, they found abnormalities in the people with supra–valvular aortic stenosis. So that is the gene that, when it goes wrong, gives you that heart defect.

However, people with William's Syndrome have supra–valvular aortic stenosis, and they also have short stature, a characteristic appearance and mild to moderate learning difficulties. They also have a very characteristic personality: they are extremely friendly. So, researchers asked whether it could be the elastin gene and maybe the genes next–door to it too. And that indeed is what happens in William's Syndrome. They have a small deletion on chromosome 7 that knocks out the elastin gene (which results in the heart condition), but also several genes on each side of it are also knocked out, and this produces the other features. We don't yet know which genes give you the personality type, which genes affect the learning abilities, or which genes are responsible for the characteristic physical appearance.

What about when we don't have a clue from any chromosome abnormality? Here is an example of that. This is a family that we have studied with a condition called Ebstein's Anomaly. This is a very rare situation because Ebstein's Anomaly is usually a one–off event in a family. However, in this family, two sisters both had children with this anomaly. Their chromosomes looked entirely normal, and they had some minor skeletal abnormalities (slight difficulty in straightening their fingers – we don't know if it is relevant or not). What we can do in this situation is just to say that here we have several people in a family with the same condition, and presumably this is caused by a gene. Can we identify another gene somewhere on a chromosome, or even just a normal variant on a chromosome, that comes down in a family with this condition. It might just be a blood group or anything that varies in the normal population. So, let's say that we had a gene that was either A or B and that we typed all the people in the family and we studied how it passed down through the family. So we see what is passed along with the Ebstien's anomaly – we discover something that is on the same chromosome that we can also see is being passed on through the family with the condition. This is called a linkage study. So, if we can find something that is passed along with the condition, that will tell us which chromosome this is, and whereabouts on the chromosome it is being carried. So linkage analysis ideally needs big families, but is a useful pointer.

So, we published our study of the family with Ebstein's Anomaly, and shortly afterwards I had an e–mail from a scientist in Texas who is working on linkage in cardiac problems. He is one of the people who identified the gene for a condition called Long QT Syndrome. He was very interested in this family and asked us to send him the DNA. We sent him the DNA so that he could do linkage studies and try to find out where the gene is. He is still working on it, so that is another unfinished story.

In another case this approach was successful. This next condition I want to talk about is called Holt Orum Syndrome. This was published in 1994. By studying families with this syndrome they discovered that the gene is situated on the long arm of chromosome 12 – what we call 12q. Holt Orum Syndrome is dominantly inherited, and gets passed on from parent to child with a 50–50 chance each time. In this family, it was passed from the father to the daughter with this 231 chromosome. Most of the people in the family who have inherited the syndrome have inherited the chromosome. This is a way of localising genes – applying the approach of getting a family, typing genetic markers and trying to find something that is going through the family. Of course, you might type 100 genetic markers and find that none of them are going through the family because they are all scattered across the chromosome, and if you don't know which chromosome to start with, you might spend a long time looking and still never find it.

So, what is Holt Orum Syndrome? It is not a very common condition. It involves a congenital heart defect, usually septal defects, plus missing thumbs, possible short forearms and possible unusual configuration around the shoulders. Now, if you think back to the embryo, I didn't mention it at the time, but at about the time that the heart is completed, the upper limb bud has been growing, the fingers are all joined together. The lower limb bud develops a little later. So the heart and the limb are developing at about the same time. So it may not be surprising that in some syndromes you have problems with the heart, and problems with the upper limb.

The linkage study enabled them to say which chromosome it was on and in which region of the chromosome it was. It took about three years. They found the actual gene on chromosome 12 and called it TBX 5. It is a member of the Brachey–Uri T gene family. The Brachey–Uri T gene was one found in the mouse, and it causes a condition called short tail. So, when they found this gene in the human, it was related to a gene they had known about for a long time in the mouse. So, what does this gene do? The easiest way to find out is to go back to the mouse. This is from a paper I just came across – it is very recent. It is focused on these TBX genes. There are two of them in the mouse – TBX 4 and TBX 5. This is showing you a mouse embryo at a similar stage to the human embryo that I showed you earlier. It looks very similar, because all vertebrate embryos look very similar early in development. So here, they asked the question, when is this gene switched on, and whereabouts in the embryo is it switched on? You remember that the gene makes a message, and that that message is a copy of the DNA sequence in the gene? Well, you can have a DNA probe that will recognise the message, because it has got the pairing sequence, and it is labelled, either with fluorescent dye or radioactivity. So, you get an embryo, you put the probe on the embryo and it will light up where that gene is creating its message. This has shown us that the TBX 4 gene is being produced at this stage of the embryo in the region of the hind limb bud, whereas the TBX 5 gene, the one similar to the Holt Orum gene, is being produced where the front limb bud is developing. So this all fits with it causng effects with the limb and the heart at this early stage of development.

What is the function of the protein that this gene is making? Well, it turns out that this protein is one of those that switch other genes on. It is what's called a transcription regulator. Many of the genes causing developmental abnormalities and congenital abnormalities that have been identified up to now have turned out to be transcription regulators – they are genes that control other genes. So this story is a bit more finished than the others: the gene for Holt Orum has been identified and we know quite a bit about what it is doing.

The last example I want to show is to do with left–right symmetry. You are probably aware that there is an important group of congenital heart defects that are associated with problems in differentiating the left and right sides of the body. If you think back to that heart loop, it is vitally important that it loops the right way. Some heart defects are associated with the heart looping wrongly, and then you get abnormal connections and things like uni–ventricular hearts. Sometimes, you get a clue to the fact that this asymmetry that should be present is not working properly. The anatomy of the two atria may look similar, the bronchial anatomy may be similar, there may be absence of the spleen or you may get spleens on both sides of the body. These belong to the group of conditions know as heterotaxy, where there's been a problem in determining the right and left side of the body. This has been recognised in humans for quite a long time. Within the last few years, half a dozen genes have been recognised in the mouse that seem to be important in determining the symmetry. This is expressed in very early embryos – earlier even than the 3mm one I showed you. By that stage, the embryo knows which are its left and right sides. If something goes wrong with that process, you are liable to get heart and other defects. Of those six genes known in the mouse, one or two of them so far have been shown to link with human problems with determination of symmetry. Although it looks as if quite a lot of the babies who have left–right asymmetry problems may just have them for not very strongly genetic regions. So that is another area where studying other organisms is going to throw light on things.

I've said on the abstract in your programmes that maybe these approaches will not give us the answer to all congenital heart disease. Obviously, if we find the gene for Holt Orum Syndrome, then we've solved that one, in a sense, and that's important for the people who have got that in their families. But we're next going to ask the question: is this gene going wrong in more subtle ways that might be contributing to ventricular septal defects that do not go along with limb problems or run strongly in the family. The same goes for all the other genes that we might identify. Most congenital heart disease occurs as an isolated event, in a child who doesn't have a syndrome. Most of these don't run strongly in the family, although if you do family studies, you can usually show that they show a slight clustering in families: brothers and sisters might have a 2–3% chance of having a similar defect, whereas in the general public that might be a bit under 1%. So probably genes are implicated in these. The genes that we pick up through these strongly genetic conditions may have more subtle effects on some of the not–so genetic situations.

The last overhead that I was going to show you (that seems to have disappeared!) tried to make the point that the conditions that I have been talking about are where you have a major spanner in the works. A bit of chromosome may be missing, or a very important gene has gone wrong. The next stage is where you have a complex process where some of the genes are not working quite as well as they should be. You can get away perhaps with three or four genes not working so well, but once you have four or five not working well, then you might be in trouble. This is how we've thought about a lot of congenital abnormalities for a long time. If you read the books, people talk about multi–factorial or polygenic mechanisms for things – they just mean that there might be a lot of genes involved, each with a fairly small effect. This means that probably for the majority of congenital heart disease cases, it is going to be quite difficult to identify those effects. Obviously, once we've found genes that are having major effects in these strongly genetic situations, then we can look at those in the more common situations. But this is another problem that will probably require large numbers to study. It is similar to finding genes for things like high blood pressure and diabetes: common adult diseases. We're probably looking at a situation where in many cases it is a whole mixture of factors. Some cases may just be that the heart development is so complicated that it can just get knocked off course by a minor environmental factor very early on. So we may never find a gene explanation for some cases.

The main message I want to pass on to you, I think, is that studying the rare abnormalities is a very fruitful way of finding out how the normal system works. Also, I would like to emphasise that the genetic messages in the living world are very much the same all the way from yeast up to humans and that this provides us with a very fruitful way of getting a handle on some of these. It is just beginning to bare fruit, and the pace of advances is accelerating all the time. I'm sure we'll know much more about some of the genes that I've been talking about within the next few years. The aim is to prevent congenital abnormalities by finding out what goes wrong, when, and whether there is any measure you can take to correct things. The example of folic acid in neural–tube defects is a small example of how one might find a way of putting things back on the right course or prevent things from going off course. So I'll stop there I think. Thank you!

[Applause]

Questions:

Q. You've explained how the research is done but where does the research end and genetic engineering start? When do you start doing something about these genes?

A. The line at the moment is that gene therapy or gene replacement is only legal for sematic cells, so in other words you are not supposed to do anything to a gene that might get in to the egg or the sperm and go in to the next generation. So at the moment society feels, and the scientific community feels, that it is too early to start doing things to genes that might get passed on to the next generation. So all the examples of gene therapy that are being looked at at the moment are in bits of the body where you can get at them – like the lungs in CF and the bone marrow in immune defects etc. With developmental abnormalities, where something is happening in foetal life, it is difficult to see how you would get at it early enough or see that something is going wrong early enough at the moment. So, it is looking just a bit too far ahead to know how intervention would work. If you were able to know that something is going to go wrong in the next two weeks in a three–week embryo that will result in a congenital heart defect, and you knew which gene it was, and you knew that by putting something in you might correct that imbalance – that might be a way in which this knowledge might have an impact on prevention. But there are too many ifs there to imagine it at the moment – it is a generation away, which in scientific terms is probably 10+ years.

Q....but there's actually one part of a gene that can affect character or the mental make–up of a child – so where do you draw the line?

A. That's right. Even after you've identified a gene and you know that if it goes wrong in a certain way and gives you a certain clinical condition, you don't know all the effects that it is having in the body and that is one reason why people are not keen on replacing genes that might get passed on to the next generation.

Q. Have there been increased levels of heart defects in places where there have been mutations or likely mutations, for example Hiroshima, or in workers who have been exposed to radiation?

A. I don't think there have, and in the case of the atomic bomb explosions, effects were looked for quite hard, and they didn't actually find evidence of an increase in mutations, which is a bit surprising. One might have to wait another generation, but so far no survey has shown an increase in mutations from nasty environmental things that people might have been in contact with. But one has to bear in mind that most epidemiological data on heart defects is not of good enough quality to pick it up. People often assume that we have a good register of all the babies that are born with a heart defect in Wessex or wherever, but there isn't. It would be quite a major undertaking to make such a register. I think it would be worthwhile, and there are some registers being set up. We have one at Southampton General, where we're beginning to get all the heart defects that come through the cardiac unit plus all the ones that are picked up on pre–natal ultrasound scanning. But within the framework of normal clinical practice, those sorts of information systems just don't exist. So, the information isn't very good, but as far as it goes it doesn't show any relationship to environmental situations.

Q. You talked about environmental factors having a possible physical effect, affecting the growth of the heart while it is developing, then you talked about genes, where you thought they were causing a particular thing not to happen, or something to go wrong – for example you were talking about the elastin. One of the biggest dilemmas for the family is "Oh, so–and–so had a heart problem," or "My mom had a heart problem," or whatever: are you saying that it is unlikely to be genetic if they are different conditions?

A. Well, I think with a question like that, and it's the sort of question we deal with every day in our counselling clinics, you just have to look very carefully at what the precise abnormality is in the family. The first question you then ask is, "does this fit a pattern that might be genetic?" So it might well be, if we found that the parent had an absent thumb and a heart defect, then we'd be worried about Holt Orum Syndrome, in which case it might be a gene, but equally you can have a combination with a VSD, abnormalities of the thumb, but also with vertebral abnormalities and maybe renal tract abnormalities, and then you might be thinking of something called Vater Association, which is a collection of things that tend to go wrong at the same time, and usually are not very genetic. So it depends on making a precise diagnosis.

Q....But in a way I suppose it is worth not knowing too much about it because you can't do anything about it anyway?

A. Well, people come along wanting to have this information, and usually we will be reassuring people in our clinic: people come along worried about a high risk, but quite often we can identify that it is not a high risk. We might have to tell them that it is a slightly increased risk, but if they come along thinking it might be 50% and we tell them it is only 3% then that is helpful. That 3% will have come along not from any theory but from empirical studies – were someone's gone out and found all the families they can with, say, aortic stenosis or whatever, and interviewed the family and found out all about the relatives. Those studies are very helpful to us. They are very labour–intensive, obviously, and it is difficult to get bodies like the MRC to pay for them now, but a lot of such studies were done 15–20 years ago and that information is still useful.

Q. We have two daughters with heart defects. We had the chromosome 22 test and it turned out negative. Just how reliable is that test, and does the result mean that we don't expect there to be a genetic problem in our family, or are we still at a stage where we're not sure?

A. Well, it is a case for individual assessment. The 22q test will just exclude one type of heart abnormality, and in fact most of the 22q11 deletions do not run in the family they just occur in the individual child. There are types of familial heart abnormality that have nothing to do with 22q, but of course we haven't got tests for all of them yet. Someone would have to look at the precise heart defect, how similar the children are, and try to arrive at some sort of risk estimate. It may also be that there are some children who have problems within this 22q region but it may be too small an abnormality for us to pick up on at the moment. So, if your children looked, clinically, as if they had a 22q abnormality, then we would be rather worried about that possibility, and we'd be looking for things like facial appearance, mild learning difficulties, palate abnormalities, nasal speech and so on. If we found those things then we might say that this looks like a 22q abnormality that we haven't picked up on in the test. That would make it very interesting as people are looking for exactly this sort of patient so they can focus down on the fine detail of what's happening in that region. But I suspect that that is not the case for you, and that your daughters' conditions have nothing to do with the 22q deletion. We'd have to look at their heart conditions and just come up with some sort of risk estimate. To be honest, a lot of the risk estimates we give out at the clinic are guestimates: it is an educated guess really, based on some information, family studies, and our thoughts on looking at the individual child.