[MUSIC] In this module, we're going to continue discussing this particular patient, but from a different point of view. And that is how we use information about lipid regulation to identify new drug targets. And then we'll talk a little bit about other Ways in which genetics can inform the identification of potential new drug targets. A very exciting and interesting area I think. So, I've presented this case to you before, and here is his family tree with lots and lots of Hypercholesterolemia odosominal dominant way, so gets interested sequences is LDLR gene, the gene that encodes the LDL receptor, expecting to find a mutation in the family, and low, and behold nothing is found. So, we see families that had clearly have autosomal dominant familial hypercholesterolemia but do not have a mutation in the LDLR receptors. So where else might a receptor might a mutant mutation be? So one clue might be provided by genome wide association. This is a genome wide association that points out loci that regulate cholesterol in a group of 19,000 patients. And you can see that LDLR receptor is on that plot, and what that means is that there are common variance in the LDLR receptor that contribute in a very, very modest way to variability and LDL cholesterol. This interesting example that will recur throughout our discussions of how common variance in specific genes might have very small effects on a particular physiologic process, and rare variance might have enormous effects on that same process. But this process is also identifies a set of candidate genes that might be interesting in terms of looking at for rare variance in patients with familial hypercholesterolemia that don't have mutations in the LDOR receptor in LDOR. So that was 19,000 patients. And that was in 2009. So through collaborations across the world, investigators have accumulated much much larger data sets, and the results of that are shown here in almost 200,000 individuals. And the relationship between genetics variation and individual genes that control LDL Triglycerides, HDL, and ultimately coronary risk which is not shown here. You can see that there are sets of genes that contribute to variability in LDL but don't do anything else. There are other genes that contribute to variability in LDL and HDL but not anything else. And in this Venn diagram there's overlapping sets that actually encompass almost every combination that you could think of. So this creates a very, very large set of candidate chains for familial hypercholesterolemia. The other way of approaching candidate chains is to understand a fundamental underlying physiology in the process that we're studying and therefore understand what candidates are. So Goldstein and Brown had already identified this idea of LDL receptors as the key mediators of handling LDL from the outside of a cell into the cell. And then signalling within the cell, and then what it turns out that the LDL receptor itself, the protein, it shuttled from the cell surface inside the cell. And then dissociated from the LDL particle, and the receptor itself makes its way back to the cell surface to be used again. And there are a number of proteins inside the cell whose function is to modulate that particular process. One of which is a gene called PCSK9. PCSK9 actually showed up on the G WAS that I just showed you and on that large overlapping set of Van diagrams as a regulator. Of LDL cholesterol. It's function and regulation was actually not well known when this particular story evolved. But its clear that mutations in the PCSK9 chain cause autosomal dominant hypercholesterolemia. The initial idea was tha if you had a mutation, somehow or the other, the function of that gene would get destroyed and perhaps that would mean a change in LDL receptors coming to the cell surface. It turns out that the mutations that cause familial hypercholerstolemia probably result in actually fewer, fewer LDL receptors of the cell surface. It looks like what PCSK9 does is it binds to the cell surface binds to the receptor inside the cell. And once bound, the receptor and the PCSK9 complex are actually degraded. So an overactive PCSK9 results in fewer cell surface receptors. We didn't know that when this paper came out in Nature Genetics. There's a lot of speculation about the mechanism and it turns out that that's probably the mechanism. And here's why we know a little bit more. This is a really interesting experiment. This experiment took place in something called the Dallas Heart Study. A study that recruited thousands of patients from the Dallas Heart area, Dallas area, and studied many, many things. But one of the things they studied was cholesterol handling and also the presence or absence of coronary disease, these are patients who are all middle aged. Now there's a very large African American cohort and there's a large Caucasian cohort. So what the investigators did was they said well, if we think PCSK9 is important for cholesterol handling, let's sequence the PCSK9 gene. And let's find variation in the gene that we know disrupts the function of the encoded protein. So the way you do that is you sequence the gene and you find Polymorphisms that result in early truncation of the protein, so called stop codon. These are nonsense variance, that's the other term for them. And they found a group of 85 patients with nonsense variance compared to about 3200 controls that did not have nonsense variance. And what you can see on one panel is the distribution of LDL cholesterols on the top among the 3,200 without the variants. And on the bottom, among the 82 with the variants. And you can see that the patients who had the variants had much lower LDL cholesterols. They had had low LDL cholesterols all their lives. And the other panel shows that they had much, much less coronary disease. They had a little bit but much, much less in patients who didn't have these non sense variants. So the assumption is that by knocking out PCSK9 function, and these are patients who only have one allele knocked out, the other allele is normal. There's a lowering of LDL cholesterol and a decrease in the incidence of coronary disease. So if you're a drug developer, you look at these data and say well, if I had a drug that could interfere with PCSK9 function, I might actually be able to lower LDL cholesterol And control coronary disease. And in fact this paper appeared in The New England Journal of Medicine in 2006 and in 2012 the results of the first clinical trials of an antibody against PCSK nine were reported in the same journal. And what you can see is that as a function of dose, these are patients who were given a single dose of an antibody against PCSK9, the larger the dose, the longer the effect, but some of the patients developed persistent 70% reductions in LDL cholesterol that last weeks and weeks and weeks after a single dose. So, this proves that inhibiting PCSK9 function either through genetics or through a drug results in striking lowering of LDL and the hope of course is that the drug would then result in a decrease in the prevalence and incidence of atherosclerosis and it's complications like heart attacks. Those trials are underway and of course everybody would like to have a pill rather than an injection of an antibody, but this trial is really important because it demonstrates the proof of principle that that's what that gene Interfering with that genes function does. So we had this story, I've shown you this slide before in one corner we have data obtained from genome wide association studies Is gwass where the variants we find are common across a population and the effect sizes are very modest. An odds ratio of 1.2, 1.5. Maybe 1.7 if you're really lucky too and then at the other end of the spectrum are rare variants that occur in families. We call those mutations that can confer really amazing human phenotypes like LDL cholesterol values of 352. Now, we start to fill in the middle. We're starting to see the idea that there are rare variants, not so rare as the familial hypercholesterolemia that confer large effects. The Dallas Heart study showed that there's a small group of patients among African-Americans, 2%, so that's not really rare, but it's not really common to have pretty large effects, large enough to catch our intention in terms of cholesterol regulation and certainly large enough to catch the attention of a drug developer to develop a new drug based on the genetics. So the story that I've just told you is a story of how rare variation can be a real clue to understanding what a new drug target might look like. These are all promissory notes. So here's another example. This example starts with a GWAS on pro-insulin levels in several thousand patients. You can see there's lots and lots of signals. And one of them is in a gene called SLC30A8. SLC30A8 encodes a zinc transporter. And there is some physiological rationale for thinking zinc transport my modulate risk of developing diabetes. So not only does devariant common variance in SLC30A8 modulate pro insulin levels as you can see in the inset, but they also modulate risk of type two diabetes with an odds ratio that is very, very modest. 1.14 with a P value that's pretty respectable. So, it's a real odds ratio, but it's a very small effect. Now, a group of investigators then asked the question, well what happens if we looked for coding region variants that disrupt the predicted function of SLC 30A8. So those are coding region variance that result in early stop in the protein just like the PCSK9 story, and here's the result of that study. So the first thing to notice is that there is an enormous number of patients that were studied, there were 30,000 cases of type two diabetes and 120,000 Controls. The Allele frequencies for these various stop codon variance are shown on that Y axis and they are as low as 0.05%, very rare Here in the population and as high as 0.4% so then all of these are very rare and that's why it lead very, very large numbers to get statistical significance. There were really three kinds of variance that were studied that are shown on the X axis. Two of them are shown and the other is a group of other stop variance, each one of which occurred once or twice in a population. Most of these patients came from Iceland but a number of them came from other parts of northern Europe, across Europe, and occasionally from other parts of the world. But mostly from caucasian populations. And what you can see is that overall, carrying one of these variants reduces the risk for type two diabetes by about 65, 70%. So again, if you're a drug company and you look at those data, you say type two diabetes is an enormous Medical problem. We have a drug that prevented the development of type 2 diabetes. That would be a blockbuster, important for human health and maybe this is the target. So stay tuned, maybe this will turn out to be a target, maybe this wont. One final example, is was featured in, not in the medical literature but in New York Times. This is a story from 2012 of a young woman in Georgia who cannot feel pain. And you would say, we'll that's a great phenotype to have because you break your arm and you won't be in tears but in fact, you could break your arm and not even know it. So it's not a very nice phenotype and that's showing on the bottom picture, when she was a young child, she had injured her hands and didn't know it. So these children can put their hands into boiling water and not know it. Of course they can still hurt themselves. The actual mutation that causes this particular illness in this child is not listed in the New York Times article, but there are other patients around the world who have been described to have this phenotype, notably in an article in Nature in 2006 that described three families. In which consanguineous marriages resulted in children with the inability to feel pain. The consanguineous marriages part is quite important because what it suggests is that the father and the mother both have a rare variant, that is perhaps prevalent in their family but not prevalent otherwise, and by marrying each other they strikingly increase the risk of a child having two copies of the rare variant and therefore the phenotype. And that's exactly what happened here. The children affected have two copies of a mutation in a gene called SCN9A, which encodes a sodium channel that is important for sensing pain. And that's the variance, all result in total loss of sodium channel. Functioned for that particular version of the sodium channel. And what's interesting is of course if you could develop a drug that could inhibit function of that channel a little bit, maybe you'd have a great analgesic to use after operations, or for anybody with. Chronic pain. Obviously these mutations are tolerated through life. You can see the picture of this young girl on the cover of the New York Times. She's developed normally and otherwise seems like a normal person so you could take that drug. The genetic data suggests that you can take the drugs that inhibit function of that gene and not develop some other amazing side effects. So these are clues, ways in which genetic data often obtained from very, very rare patients with very unusual phenotypes can inform the drug development process. So the take-home messages here are the studies of rare diseases, once again, are highly, highly informative for personalizing medicine, not just for the rare patient who has the really rare disease but for all of us. Where do HMG-CoA reductase inhibitors come from? Where are PCSK9 inhibitors going to come from? Those are all coming and they will be important for large numbers of patients. The other story of course relates to the fact that there are common variants that predispose to diseases like. High cholesterol or heart attack or type 2 diabetes. And those common variants, while they don't tell us very much about risk in an individual subject, understanding the complicated pathways in which those variants modulate, things like LDL cholesterol, can make us smarter, not only in terms of developing drugs, but in terms of predicting exactly what pathways are important for developing high cholesterol and therefore preventing the consequences of high cholesterol.