 Okay, everybody. We're going to get started. Good morning. So this morning we have Dr. D'Angeles going to be talking about gene expression genetics and epigenetics, and it's rolling AMD for us today, so cheers. Okay. Thanks for having me here this morning. And some of this might be a little bit of a review for you. And feel free to ask questions. And I'm not going to take up your whole time this morning, so you can follow up with questions afterwards. Okay, a lot of you probably already know a lot about age-related macular degeneration, that it's the leading cause of blindness in the elderly. And a lot has been done with respect to the genetics of age-related macular degeneration. And a lot of that recently has been done with the AMD Genomics Consortium pulling all these cohorts that are available in trying to get to the genetics. And there's two major loci on chromosome 10, the arms 2, H-traw 1. We don't know if this is two separate genes or possibly one in the same gene, and the complement factor H on chromosome 1 that explain about 65% of the genetic component. We know that smoking's involved, hypertension, cardiovascular disease, as environmental factors, aging of course, and having a first-degree relative which can increase your risk up to 12-fold. So first-degree relative is having a parent or first-degree sibling. And as the patient ages is the patient population ages, we expect close to 3 million to have advanced disease by 2020. So even though we have all these great genetic risk factors, we're not able to reverse vision loss over the long run, and therapies are limited in their applicability. So this is just a picture depicting fundus photos of the intermediate, and I'm showing the age-related eye disease scale here, showing the intermediate or arid's category 3, and the advanced form's geographic atrophy, and the neovascular form which one can lose vision from quickly. And most of you could probably do better describing the phenotype here. One of the approaches we've been using, and some of you have probably already heard about this, is we've used a family-based approach. And I started this work as a postdoctoral fellow when I was working in Boston years ago, over 10 years ago, working with Joan Miller at Mass-Sineer. And this is looking at genes and variants and environmental factors between sibling pairs, because you're controlling for ethnicity and environmental factors, because a lot of studies are, you get a lot of false positives, and you can't replicate these findings from study to study. So what I'm going to talk to you about today is we're getting ready to submit this work. Is looking at gene expression as the driving force at the RNA level, doing pathway analysis, validating it and replicating it at the DNA level, and then looking at it systemically, localize tissue-wise in the eye, and doing a tissue work, because you want to see what's driving gene expression. Is it due to methylation, epigenomics, or is it due to something else? So we looked at it globally at first. So we wanted to look at the sibling pairs and look at the RNA, and we wanted to get RNA, obviously, for my tissue, but you can't do that from living, breathing patients. And the best way to do that is to get it from transformed cell lines. We tried to do RNA from blood, but when we started this, there's a lot of noise involved with looking at RNA directly from blood. And we also controlled for environmental factors that could lead to false positives in your analysis with gene expression studies. And we came up with a set of genes that at least had a two-fold difference between those with neovascular AMD and those without neovascular AMD. And we also wanted to look for other modes of evidence, linkage analysis that weren't found by other studies that didn't have already genes identified for them. And we found some genes of interest, ABCA1, which has evolved in lipid metabolism and a few other genes we wanted to focus on. And this is just showing the full change that is different between those with disease and those without disease. So this means that this gene is decreased in those with disease compared to their unaffected sibling. And this takes, to get to this point, just a lot of programs, a lot of statistical analysis, and a lot of data cleaning. And this is just some of the pathway analysis that you can use to try to visualize the way that these genes might look together or they might function. And the idea is that the way that these genes might function are going to be different in someone with disease than someone without disease. So that you can think about how you would target a pathway in the form of therapies. And red is up-regulated, green is down-regulated. But it also pulls in the genes that don't have colors or genes that also might function in these pathways. These are hypothetical pathways, because we don't know everything about our genome there is to know. And we've already published this work, but we wanted to interrogate this pathway more fully. Ooh, these genes. Okay. So a way to interrogate this a little bit more fully is to look at, we've looked at it at the RNA level. Now we want to look at it at the DNA level. And we only looked at it in a handful of patients. So now we want to expand to it, expand more fully at the DNA level. So we expanded more fully in this discovery family cohort. And we look at SNPs or tagging SNPs so that we can capture full variation within these genes. And those handful of genes that I pointed out, there's five or six of them. And then we want to replicate it in these cohorts. And they're Caucasian cohorts, but they're also ethnically diverse. And then if we want to meta-analyze these cohorts because if it's more strong in meta-analysis, you're more confident that those SNPs are truly associated with disease. We also wanted to use a cases-only model because you have better precision for estimating interactions. And it's more appropriate because we have our major genes already that are associated with susceptibility to disease. So we only expect that these genes are only going to contribute modestly to disease. So the risk alleles that are going to be associated with susceptibility to risk are likely going to be low. So this is just to show you the work involved. And we tested probably 44 SNPs within these disease genes with known SNPs that were associated with disease. We'd already published these. There are well-established risk factors for age-related macular degeneration. And from this, we came up with eight SNPs that we thought were associated with risk. And we tested it in those replication cohorts. And from this, we were able to do meta-analysis. And this is what we came out with. And HTRA in arms two, even though they're significantly associated with susceptibility to age-related macular degeneration, and they're the most strongly associated with all types of age-related macular degeneration, we do not know their function and we do not know their pathway. So this gets us to start thinking about a pathway that might be involved with HTRA in arms two. Green means these are down-regulated and red means these are up-regulated. But what does this mean? So what? These genes interact. We need to maybe figure out why they're interacting. But we also want to show disease relevance with the tissue that's being affected in this disease. So thanks to the Lions Eye Bank, we've been able to end Dr. Hartnett and me Hartnett. We've been able to build our own eye repository and try to focus on the retina and the RPE coroid, the tissues affected by age-related macular degeneration. And we want to look at the macula that's primarily affected, but also at the extra macula, because we want to look at the differences between the two tissue types as well as to see if there's differences between these types. So we look at four geographic regions within the eye. And what we found with the genes that we wanted to focus on these genes that were statistically significantly interacting was that only RGS-13 and Mura were down-regulated in the macular region compared to the extra macular region. And even though H-tri, which is the most in arms two, which are the most strongly significantly associated with susceptibility to risk, weren't changed in their expression between those with disease and those without disease. And here I'm just showing you an example of those with six eyes. We actually did up to 18 eyes and found that this was the case. And then for these genes, ABCA1, which is a lipid transporter gene, and Robo1, those are the extra macula, they were decreased in the RPE. So we found differences between tissue types that were concentrated to the macula. The tissue primarily affected an age-related macular degeneration and we didn't find any other changes for any of the other genes that was statistically significant. We saw trends, but not at a statistically significant level. Now, we see these changes in expression, but what does that mean? We want to look at the mechanism of why these changes are occurring. And one of those reasons could be due to epigenetics. What exactly is epigenetics? What epigenetics means, and some of you already may know this, is changes in gene expression caused by other things other than the base changes, which I've just talked about, which is a single nucleotide polymorphism. And these changes can be due to things like smoking, changes in our diet, or changes toxic things in the environment. And if they occur at the level of the gamete as opposed to a somatic cell, they can be inherited. And I'm going to show you an example of that in mice. These two mice, you see one here that's yellow and heavier looking than this brown mouse. Although these mice look different, they're actually genetically the same. And the reason that they look different can be depicted here. This little brown mouse, and this is due to what's called the agouti gene. This brown mouse, they both express the agouti gene, but as they age, the agouti gene in the brown mouse becomes methylated. In the yellow mouse, the agouti gene is not methylated. So what happens in the yellow mouse, the gene, the yellow mouse develops diabetes and becomes obese and develops cancer at times late in life. If this yellow mouse eats a diet high in folic acid, it will have a healthy brown mouse. So by eating a healthy diet, it can have these healthy brown mice that have methylated agouti genes. So this is just an example of what methylation can do. And we can think about this in age-related macular degeneration because you can have the same gene associated with susceptibility to disease but yet get different phenotypes. You have people that develop just geographic atrophy, neovascular AMD, or only get erodes three but never develop these diseases. So there's... I'm going to just talk about two types. Is DNA methylation and histone modifications. And what we can think about is that these block access that epigenetics or DNA methylation, if you will, blocks access to the DNA. So what happens is you add a methylation at a CPG site or a cytosine site and these CPG sites can occur anywhere, not just in the promoter, but they can occur throughout DNA in the intronic sites but they can occur throughout the gene. We chose to study for the four genes of interest. We looked at all of the genes, the genes that changed expression for ABCA1, ROBO, RGS13, and RPSKA2. And we focused on what we thought were important to where the CPG rich sites. And we did not find... we looked at the cell lines, we looked at the RP and we looked at the retina. And the only thing we found that I'm going to talk about was in the ABCA1 and we found this both systemically and within the tissue. ABCA1 is not novel to age-related macular degeneration risk. It's been found by these groups here to be associated with age-related macular degeneration. It's also involved in cardiovascular disease and lipid transport is shown here. The thing that we found interesting is that we actually found it in one of our significant single nucleotide polymorphisms that we found associated with disease. And we looked at 99 different sites. We found differences in 15 different sites between those with disease and those without disease. And in one of those 15 sites we found three sites with statistical significance. And from one of those sites one of those sites actually contained a single nucleotide polymorphism. So what we found was is that people who were and this was also true with the tissue but we found that people without disease had increased methylation. The thing is we found this true not just for the RPE but for the retina. So the expression change cannot be explained by the methylation alone. So our change in expression is due to... because remember the expression change for ABCA1 which was a decrease in expression was only found in the retina pigment epithelium and we found this methylation this increase in methylation in the normals also in the retinal tissue of the macula. Another way that we can look at what at function of our genotypic changes that we're seeing in our genes is through expression quantitative trait locus or EQTLs. And this is a way to study the relationship between the genome and the transcriptome or the RNA the DNA and the RNA to explain why we're seeing differences in gene expression. So we have cell lines on our SIDPAIR cohort or family based cohort and we have genotype data on these genes that I showed you so we want to try to equate the genotypes in those genes with the expression differences and this is just the type of statistical test that we used and we did find that this promoter change in the H-trod gene affected Robo-1. So this promoter change in the homozygous state of the disease alleles right here affects Robo-1 and because Robo-1 is located on a different chromosome chromosome 3 H-trod resides on chromosome 10 this is called a trans-EQTL because RGS-13 in its homozygous state affects because just because something changes doesn't mean that it can affect expression within the same genes it's a cis-EQTL so it's nice to see that some of the changes that we see in those genes can affect expression of systemic at this systemic level so just to add proof of concept that this pathway might underlie age-related macular generation we see significant gene-gene interaction we've shown differences in expression profiles at the systemic level also at the localized disease level we've been able to show methylation for one of the genes and we've been able to show cis and trans EQTLs we're also doing RNA-seq both on the cell lines and on the eye tissue and nothing would be possible without really good collaborations both within the Moran nationally and internationally and these are all the wonderful collaborators here at the University of Utah here in my laboratory a lot from Margo Morrison the wonderful people Denise, Roseanne, Katie, Katrina and Stefan Hartnett here Emmy Hartnett Ivana and Joan up at Moussineer and Lindsay Farre at Boston University Dr. Park from Seoul and our Greek collaborators and Dr. Sylvester from Queens University so thank you there's a lot of information in a short period of time feel free to ask questions we had systemic genetic only for ABCA1 with the mice was eating a good diet with BOLA and I think we're just starting to learn because bad things can this is just an example we found accidentally that increased methylation was higher yeah right it's very complex because more methylation can be bad but more methylation can be good and I don't think we know enough because you see it both ways depending on the literature and the genes right and we didn't look like we've done one analysis I was looking at is looking at because we did a serum so I'd like to see what the genotype might be protecting them too want to see how that's interacting we've seen a lot of that in this cohort with having a protective genotype but they have a high serum level so I'd like to look at that and maybe it's just the bees normal like it's not complete that's why it'd be good too to have the cell lines from the people who donated their eye system with control by the number of passages so if you try to keep it under a certain number of passages and people have done this successfully for other complex diseases and published successfully for neuropsychiatric diseases for what else has been it for type 2 diabetes and shown correlation say with brain tissue fresh brain tissue and systemically and a good correlation with RNA-seq data use these arguments because that's how I know this that makes sense