 Good morning. This morning we have Dr. DeAngelis from her lab is going to be presenting a talk entitled Identifying Pathways Using Geneticist Factors and Their Modifiers to Develop Appropriate Targets for Therapies and Prevention of AMD. So without further ado, Dr. DeAngelis, thanks. Thank you for having me today. Thanks for having me today. And I want you to feel free to ask questions. Some of this may be a review for some of you, but I'm also going to present brand new data today that hopefully will have published within the next couple of months. As most of you may know, age-related macular degeneration is the leading cause of blindness in this country as well as other developed countries. And it's expected to surpass diabetic eye disease and glaucoma combined within the next 20 years. And the important thing to appreciate is the disease heterogeneity with this. And another thing to appreciate is with genetic studies or any disease of complex etiology that is more than one gene associated with the disease like diabetes, cardiovascular disease or cancer or neurological diseases like schizophrenia and autism is to have really good phenotyping. So to work with really good clinicians. And we happen to have that here at the Moran Eye Center. And most people use the age-related eye disease scale. At least we do anyways for the AMD gene consortium sponsored by the National Institute of Health. And this is just showing the grades that we use anyways. You may use something different in the clinic here, but the early intermediate categories two and three and the advanced geographic atrophy which is less frequent than the neovascular form, the more severe form causing blindness. Most therapies right now are directed against this form, but as you know they're limited their applicability and don't reverse blindness over the long term. And they're broad and not very appropriate or direct. And that's what we'd like to use with our genetics and our therapies is to make them more streamlined over the long term. There's been many epidemiological factors associated with age-related macular degeneration as you can see here. The most consistent from study to study is cigarette smoking. Although many of these factors have been associated with age-related macular degeneration over the last 20 years or so. Evidence for genetic risk component comes from studies of families and as with any complex disease we know this because diseases such as this like cancer, diabetes, cardiovascular disease, cluster and families where when you have a first degree relative in the example of age-related macular degeneration now they're putting your risk at six to twelve fold if you have a parent or a sibling with this disease. And twin studies show that concordance rate is greater for an identical twin than a fraternal twin. Now many studies have shown either through candidate gene, genome-wide association studies or linkage studies that there's been several genes they haven't been consistently associated from study to study. The most consistent are the arms 2H tra, locus on chromosome 10 for which the function is not known and the complement factor H on chromosome one for which the function has been well established. Two genes I'm going to focus on today that our lab discovered is the robo one in the rora gene and we believe that they function along with this arms 2H tra 1. We have a paper coming out from the AMD gene consortium the embargo day is March 3rd showing seven new loci on 60,000 individuals and that that's none of these genes here and those remain to be vetted in the function determined. So there's still more work to be done in terms of function and what not in this field. Now have the strongest loci been identified? Likely so. The arms 2 and H tra 1 gene but this always doesn't tell us about the missing heritability specifically the gene-gene interactions and I'm going to show you some examples of that today that we've published on and we're going to be publishing on in the next couple of months and some new methodological approaches to uncover that and disease pathophysiology and uncovering those pathways and there's examples of this with Alzheimer's disease and even cholesterol metabolism. A great example of this is the HMG co-reductase gene where variation in that gene is only responsible for changing cholesterol levels just a very small amount just a fraction but that's what most of the cholesterol lowering drugs are geared against so just because you find variation and it may not be responsible for a large effect a modifier gene may be a great therapeutic target for disease. Now in a complex disease we can't just consider one gene we need to consider groups of genes and we need to consider more than one level not just the DNA but RNA and protein and we need to consider the environment to get to mechanism and hence pathways of genes an example of a genetic approach this is a review for some of you that we use for our discovery cohort and has proved successful for us is using a path is using a family-based approach because this is an aging disease we usually don't have children available because they're too young to manifest the disease and so we've used a seed pair based approach and I started this back when I was a postdoc in Boston years ago back in 2000 1999 and we collected seed pairs extremely discordant seed pairs where one had an individual the pro band had the worst form of the disease who had a sibling who had no form of the disease and was older than the individual who had the disease and we looked for levels of gene expression as well as DNA that were different between the two individuals and that were the same and they were only already matched for race and ethnicity so you didn't have confounding factors that could lead to false positives and analysis this is an example of one of our extremely discordant seed pairs and I have to say the success of this is really due in large part to Joan Miller and Ivana Kim who are still at Massachusetts Eye and Ear of Firmary and this is actually a fraternal pair and unfortunately this pair was one of the first pair recruited in their long deceased but this is an end stage disc of form scarring and this is his brother and of no the brother even though they're fraternal is slightly at an age slightly older but as we can visualize with these fundus photos there's no observable drusen in this fundus photo and the maculi are normal here one of our first approaches with this set of extremely discordant seed pairs and now we have well over 500 of these seed pairs and this is our discovery cohort and you'll find with any genetic study that replication is important on different ethnicities and many different case control sets but we use a family set as our discovery and one thing we wanted to make sure is that each pair when we used our approach we used lymphoblastoid cell lines because we wanted to use our living breathing patients and this was a way to access tissue if you will and we tried blood samples but we had a lot of noise associated with the analysis because we didn't want our patients fasting but we wanted to make sure they were matched for certain epidemiological factors that could affect the end result that could affect gene expression levels and what we found is after cleaning up the data doing principal component analysis and doing several different statistical approaches and this was done in conjunction with your god from Rockefeller University we've got a set of genes that we thought could function together in a pathway and gene expression analysis the results of the genes could infer some functionality and what you see here in green are genes that are decreased at least two fold between an affected sibling or the pro band compared to their unaffected sibling moreover we had linkage data from our lab and other people who work in the field of age-related macular degeneration as well as some genome-wide association studies and you can sort it out by its gene name as well its location to infer some functionality and some of these genes you see here i'm going to focus on robone or first because we have that published and some of these genes you see here is new data that i'm going to present today that will hopefully get published very soon that we also have functional data to support you can also use some specialized software that we have available here at the moran to help to visualize the way its ingenuity pathway analysis to way these to way to visualize the way these genes might function together or the way they interconnect because we don't know a lot about the pathways or the hypothetical pathways in our genome we only know about our classical pathways and you can see that Begev is in this pathway the low density lipo protein is in this pathway and this could be one pathway or several pathways or networks that overlap now remember this is all at the RNA level we need to corroborate this or validate this at another level either at the protein level or the DNA level and we chose to validate this and move further at the DNA level and the ones that are in color are the ones we picked up from our microarray experiments and this is just from the literature and these are additional genes that may function together with our microarray genes and this pathway may function differently or be less robust in a person with AMD than a person who doesn't get AMD another thing the ingenuity pathway analysis does is it ranks possible pathways that your sets of genes can function together and be involved in and the significance of these pathways is the inverse of the p-value or the log of the p-value along here on the x-axis and the y-axis is these functional pathways and as you can see lipid metabolism, cardiovascular disease, not surprisingly functions that have already been implicated and well published in associated with age-related macular degeneration including immunological disease the CFHG but the take-home message here is is that many of these genes have more than one function and many of these genes share the same function so this underscores the redundancy of the human genome and that looking the importance of looking at sets of genes and not one gene in isolation when you think about designing preventative and therapies for any complex disease not just age related macular degeneration so as I said before we wanted to look at the DNA level so one approach is when you look at these sets of genes a way to do it is you don't want to look at every known single nucleotide polymorphism or the difference in DNA between individuals and these exist in our genome quite frequently you want to look at what's called tagging SNPs and these tag for numerous variation in our genome but they encompass they can encompass entire variation within a gene and they're an economical and resourceful approach to look at an entire gene within your cohort and that's what we did and then you can drill down further to try to pinpoint a disease causal single nucleotide polymorphism in order to do that we want we again wanted to look at our extensive Sib pair Sib set because remember we only looked at nine Sib pairs to get our microarray data so we wanted to look at our entire pair of 500 siblings and then try to replicate this in our Greek cohort as well as a cohort from Dr. Deborah Schomburg in Boston which is prospectively based and what we found is that we had a haplotype or a couple of SNPs that are inherited together that we could replicate in this cohort and we published on this data I apologize if I'm glossing over anything but I can follow up with questions later and this is just showing we replicated this in three different cohorts furthermore we were able within that pathway to replicate that these genes were actually interacting mathematically statistically that robo and roar interacted and moreover we were able not only to show that in a get in our Sib pairs but we were able to replicate it in Deborah Schomburg's population as well as the Greek cohort and moreover we were able to show this that these two genes did in fact find via chromatin immunoprecipitation in a mouse model and hence that was recently published now since that work was done we pursued this further and we said is this just relegated is are we just seeing this only in Caucasians because if something's truly going to be a therapeutic target and have global applicability we should be able to see this possibly in other ethnicities well lo and behold with our collaborators Dr. Park in Korea we have a set of siblings not siblings sorry unrelated case controls that are well phenotype from South Korea of almost 1500 individuals and while rora is just relegated or just specific to Caucasians robo can be seen to be significant in the Koreans in an Asian population but more importantly and this is a test of true association the SNP the same SNP is more significant in meta-analysis so that's combining all your individuals and that's a true test of association so once we saw that we said let's test the rest of these genes and the tagging SNPs within them so we said not only are we going to test them for individual association but because of the robo-rora interaction let's also check for epistasis now obviously because rora was not significant in the Koreans we could not replicate the robo-rora interaction that we saw in the Caucasians epistasis the definition is testing for interaction this is also probably responsible for a lot of the missing heritability there's a series right now if you're interested in looking more into what is missing heritability it's not just due to the rare variation you see a lot of focus on exome sequencing it's also due to this gene-gene interaction and eric lander has a series right now in the proceedings of the national academy of science a series of review articles on missing heritability i encourage you to look at it it started last year it's going into this year on this exact topic a way to approach it and looking at gene-gene interaction is not just to look at cases and controls because a lot of gene-gene interaction is going to be a is going to look at modifiers i already told you for disease like age related macular degeneration we probably found the biggest genes that have the largest effect cfh arms 2 and h trauma 1 so the things we're finding are likely the modifier genes the genes in the pathway so we need things they're going to search for things genes or variants that aren't going to have large effect and the controls are likely going to dilute that frequency or wheels there are actually going to be low so this is approach that's been used but not a lot so we took this approach with our cohorts now this slide is a little bit beastly but the take home message here this isn't our discovery cohort of Sims and across the top here you can see our rora the h-tra arms 2 in the robust snips these are the snips that have been published these are the snips the tagging snips that are in those genes you saw in the pathway the things to focus on here are just the yellow ones because these are the ones that would hold up under multiple testing it is very important that you control for multiple testing because you're going to find significance if you run enough tests the blanks mean there was no interaction and you see things that look slightly significant but they're not going to hold up under multiple testing this is the discovery cohort when we pull the interaction and notice we were able to include the koreans the asian cohort here and this is after controlling for multiple testing we have very significant p values and this helps to validate that pathway that you've seen and we see significant interaction with robo rora h-tra this rps ka2 abca1 which functions in lipid metabolism and just a few of these genes not all of these genes are functioning together and the thing also take home message here too it's for all sub types of a and d there's nothing more significant when we look at just neo vascular a and d and this is just a summary and this is showing that there's a handful of genes robo rora and abca1 that are down regulated in rgs 13 and rps k that are up regulated in the affected patients versus the unaffected sibling and we can try to think about disease mechanism and that's what we're testing right now through chromatin immunoprecipitation assays as well as epigenetic assays that we're doing and in fact these are the results this is only one sip here i'm showing here or some of our chromatin immunoprecipitation assay and what i'm just showing here is that robo and h-tra are binding to the rora antibody in so this is validating our statistical interactions at the chromatin immunoprecipitation assay and this is just one example it's not showing any difference between a normal individual and a patient with neo vascular a and d but we're doing a larger cohort i don't have all the data here to see if there's significance between those with a and d and those without a and d see if h is the negative control because we don't expect complement factor h to function in this pathway and then what do we see at the protein level what do we see with robo and rora in the different sub types of a and d when we measure because another thing is is this disease age-related immaculate degeneration is it localized is it systemic so what do we see serum wise well well and behold we don't see anything and this is we happen to have serums only for our sip pairs and when we look at the sub types all together a and d is a whole or separately compared to normals we don't see anything significant however when we parcel out looking at a disease snip this is a beautiful example of interaction when we see when we look at people who have this snip in particular in rora they have their serum levels get lower than those who are normal so we see an interaction for people who who have the g snip they have lower serum levels so this is an example of interaction and we're actually doing something similar with bala and body's lab with flip one we're trying to validate that right now another thing we can do with the ingenuity pathway analysis is go in a little further and see how rora and robo look work together and what we found is that robo and rora supposedly work to function to for try to out to control triglyceride levels as well as levels of lipids because we're trying to figure out how they also work just besides with genes because if they supposedly function in lipid metabolism which was your number one function on that other chart where i showed you biologically how do they do it well supposedly through lipids and try it like triglycerides and ldl so we wanted to look at other serum biomarkers besides the genes themselves when we looked at cytokine sile adhesion molecules and lipo proteins after controlling for the multiple testing only the triglycerides which we found protective for a slight elevation of triglycerides was found to be protective compared to the amd patients again the literature goes both ways in age-related the macular degeneration so with nick bizon from lshu who's one of my former mentors we're trying to suss out what the different lipoproteins are in the triglycerides here because this is a bit bizarre and again the literature is controversial with what's in there from an epidemiological this is just a review of some of the literature with the evidence for in the epidemiological studies with the serum and the reports but there have been genes associated with lipid metabolism cardiovascular disease besides the genes we've reported and others have reported and is this a potential therapeutic pathway possibly but we need to get further in the mechanism and right now we're trying to finish up the chromatin immunoprecipitation studies and working with Dr. Amy Hartnett here at the Moran we're also doing some immunocytochemistry on some fresh eye tissue to try to localize the expression of Robo and Roura between age-related macular degeneration eyes versus normal AMDIs in the RPE and in the retina to try to see where that expression is and we try to put together because we don't know if this is a systemic or localized disease but you have to appreciate you've got all these genes you've got heredity you've got environmental factors like smoking you've got the triglycerides but you've also got this like Roura gene interaction and you've got aging going on and how this affects the eye in the system we've got to put the story together and we're trying to put this all together to get the sequelae of the mechanism of disease because this happens over time and how to put the gene in the environment together to bring on disease is is putting the story together not just focusing on one gene and we couldn't do this without our wonderful collaborators my laboratory here Dr. Hartnett people from Boston Deborah Schomburg Mina Hadar Ivana and Joan people in Boston Juliana Syvestri from Queens University Maria and Ava from University of Thessaly the Greek cohort and Dr. Park from Seoul University and this is a picture of my lab Margo Morrison who did all the statistics and Roseanne who does a lot of the functional work Denise, Katrina, Katie, and Caitlin and nothing would be done without this generous funding especially from the unrestricted grant from here at the Moran so thank you and I'm open for questions yes I mean that that's a great question I none of these cohorts I know have pseudo drusen that these cohorts I study but that would be a great thing to do yeah you know yeah I will I will tell you these normals were followed for 10 years so this this is a family I followed for 10 years and he I followed him up after 10 years and he was still normal no neo vascular after 10 years so so how fast do they go yeah yeah and it's I'm not going to show this picture we we actually dealt with that but a good question