Hello, everyone. I'd like to welcome you to the Endo 2021 Symposium, Pharmacogenomics of Hormone Receptor Signaling. Thank you to all of you for joining us. This promises to be a fantastic session. We're ready for three outstanding presentations by our esteemed guest speakers. I'm a co-chair of this session. My name is Ursula Kaiser. I'm a professor of medicine and chief of endocrinology at Brigham and Women's Hospital and Harvard Medical School in Boston. Our presentations today have been pre-recorded by our speakers. But before we begin, I'd like to turn it over to my co-chair to introduce himself and then introduce our first speaker. So thank you very much. And good afternoon, everyone. My name is David Calabiro. I'm professor of microendocrinology at the University of Birmingham, UK. And I'm a co-chair in this session. Just a housekeeping announcement before we go into the session. There will be live Q&A sessions after each presentation. So please be sure to submit or upvote your question during the session so that we can ask those questions to the speakers. And without further ado, it's my pleasure to introduce the first speaker, Madame Babu. Madame Babu is Endowed Chair in Biological Data Science and, in fact, the new director of the Center of Excellence for Data Driven Discovery at St. Jude's Children's Research Hospital in Memphis. And Madame is going to give a talk about the pharmacogenomics of the protein-couple receptors, a topic he has been very active in over many years. And without further ado, please, Madame. I'd like to thank you for that introduction and the Endocrine Society for inviting me to participate in ENDO 2021. What I'd like to do in the next 20 minutes or so is talk to you about some of our recent work on how we use data science approaches to understand variation in GPCR signaling in the human population and in different human tissues, and also discuss the implications for drug discovery and development. And here are my disclosures. Technological advancements have allowed us to generate a huge amount of information describing both biological entities and processes at many different scales and resolutions. Now, this presents us with a unique opportunity to use computational methods to gain fundamental insights and discover new scientific principles. The current estimate, according to the NIH and the European Bioinformatics Institute, is that we have access to over 150 petabytes of processed biological data. Now, given this, what drives us scientifically is to find out how we could discover new biological insights and knowledge from large volumes of data. So the mission of my group is to develop data science approaches to make fundamental discoveries and reveal principles of living systems, and the emphasis will be on human biology and pediatric catastrophic diseases. So one of our primary research interests is to understand human biology and diseases through data-driven approaches, and for this, we investigate biological systems at three distinct levels of complexity. The atomic level, the cellular level, and the population level. And we bridge these by investigating the impact of mutations across all these three different levels. And to this end, we develop interdisciplinary computational and experimental approaches that generate, integrate, mine, and analyze disparate data sets. And these include data describing sequences, structures, expression, networks of interactions, and many more. This audience needs no introduction, but I would like to briefly highlight why GPCRs are important. The GPCRs are the largest family of membrane protein receptors in the human genome, and more than 800 different genes code for GPCRs, and they regulate almost every aspect of human physiology and health. Not surprisingly, they have also been implicated in several diseases, including cancer. Importantly, GPCRs are also the targets of over one-third of all FDA-approved drugs. Now, in the last few years, there have been enormous progress in GPCR structure determination. So we now have over 500 structures of about 90 different receptors. Thanks to the amazing developments in Cryo-EM, we also are now generating many more structures of GPCRs in complex with many different adapter proteins and ligands. So what have we done in this particular space? So in our group, we have developed computational methods to investigate protein structures as networks, and have integrated them with diverse genome-scale datasets to provide mechanistic insights into GPCR signaling, all the way from ligand binding recognition sites, to receptor activation mechanisms, to understanding how GPCRs selectively bind to G-proteins, and uncovering the allosteric network resulting in G-protein activation, and ultimately in GDP release. Now, in another major study from our group, we integrated data on receptor structures with variant information from over 60,000 individuals in the human population. And this essentially allowed us to demonstrate that GPCRs that are targeted by FDA-approved drugs are highly variable in the human population, and highlighted the implications for adverse drug response and personalized medicine. Specifically, we showed that GPCRs that are targeted by drugs show extensive variation in the human population at ligand binding sites. Variation occurs not just in ligand binding sites, but also in functional sites, such as post-translation modification sites, effector binding sites, which may result in altered drug response. And we also showed that understanding GPCR genetic variation may help reduce global healthcare expenses through computational calculations of expenses associated with adverse drug reaction. So clearly, there's much more to be explored in this emerging field of pharmacogenomics, and further characterization of some of these GPCR variants that we describe in the study could potentially increase precision prescription and also improve the quality of patients' life. So it's a goal of advancing the field of personalized medicine and fewer pharmacogenomics research, we have created a comprehensive resource within the G-protein coupled receptor database, which is the database for GPCR research that's maintained by our collaborator David Gloriam. And this resource can be accessed at www.gpcrdb.org. Now, in the remaining 15 minutes or so, what I'd like to do is talk to you about a recent work where we try to understand tissue-specific differences in GPCR signaling responses by investigating transcriptome data from over 30 different human tissues. And this was a work that was spearheaded by extremely talented Mary Curie postdoctoral fellow, Dr. Maria Marti, in my group. So given that there are over 50 different organ systems that are made up of more than 200 different cell types, and over 37 trillion cells in the human body, we were interested in the following two questions. Do all cells and tissues respond the same way to a given ligand when a receptor is expressed? What are the mechanisms that diversify GPCR signaling in different tissues, and how prevalent are these? Now for this, we focused on alternative splicing as a mechanism to create protein sequence diversity. And as we all know, alternative splicing is a molecular mechanism where a single gene product can give rise to multiple processed RNA sequences. And if alternative exons are included or excluded, this can result ultimately in generation of protein sequences that are different from each other. Now developments in next-gen sequencing have clearly highlighted that over 90 percent of the human genes undergo alternative splicing, and a large fraction of these isoforms indeed have tissue-specific distribution. How exactly this affects GPCR signaling proteins on a systems level was something that was not known when we embarked on this study, although individual examples of isoforms have been very well documented and established in the literature. So we became interested in the following three questions. How many GPCRs have more than one isoform, and how exactly do they differ in their structure and therefore their likely function? How does their expression contribute to diversification of signaling response in the different tissues? And more importantly, what does this mean for drug response and for drug development? So to address the first question, what we did was to systematically integrate a large-scale dataset from what's called the GTEx consortium, which provides information about transcript abundance level and transcript sequence diversity from 30 different human tissues from about 700 diseased human individuals from the human population. We integrated that with the G-protein coupled receptor database on the human GPCR sequences and developed a computational pipeline that allowed us to show that there are about 154 of the receptors, about 43 percent of all GPCRs, that have two or more different isoforms. Now where exactly do these receptors change in the sequence, and what are the likely functional impact, and do they affect all the different classes of the GPCRs in the same way? So these are some of the questions that we explored, and what we were able to find was that the most dominant isoform, as expected, is the isoform that has all the seven transmembrane helices and all the extracellular and C-terminal sequences, as well as the loops, completely intact. But right after that, the most significant variation happens where the N-terminal sequences have either an altered sequence or have truncations. The next prominent sequence changes happens in receptors where the C-terminus is significantly altered in terms of the sequence, and alternative sequences encoded, or they are truncated. These are very interesting regions simply because these are the regions that do contact the ligand, can get phosphorylated, or can be post-translationally modified, and can affect how long the receptor stays on the membrane, and thus can affect the duration of the signaling response on the cell membrane. All the different classes are equally affected, and not surprisingly, when you look at receptors that have the largest N-terminal variation, these tend to be class B receptors which also have an N-terminal domain. And interestingly, for the C-terminal variation, you also find almost all the receptors having representative members that have an alteration in that particular region of the protein sequence. Now, as we all know, receptors get post-translationally modified in the intracellular loops and C-terminal sequences, and this can affect how they interact with different intracellular adaptive proteins such as arrestin, and thus can influence the downstream signaling response. So what we then did was to map all the known post-translation modification sites from published mass spec experimental results, and what we were able to show was for a number of different receptors, there are altered C-terminal sequences that seem to not have these phosphorylation sites. And what this essentially means is that you could have different sub-populations of the same receptor that are expressed in the cell, but one population may have the right phosphorylation site and hence may have the right downstream signaling response, whereas the other sub-population may have an altered sequence which doesn't have the phosphorylation site, which essentially means that they can respond to the same ligand, but may not be able to engage downstream signaling adapters as effectively or as efficiently, and hence can give rise to all sorts of interesting downstream signaling variation in a cell type or a tissue type specific manner where exactly this particular isoform is expressed. So as I mentioned, a number of different receptors have been systematically and quite extensively characterized in the literature, so we also did an extensive literature mining to identify what are the documented structural and functional changes that have been observed by researchers in the past, and these include where isoforms affect N-terminal sequences that directly affect ligand binding, or instances where you have C-terminal sequences where phosphorylation sites are affected, and these can result in alteration in receptor signaling kinetics and responses. There are also instances where isoforms affect the C-terminal loops that can influence the effective nature by which they can couple with different G-proteins, as well as dimerization interface, because you can have truncations that can act in dominant negative manners. Now in our study, we were able to identify all these types of instances for the known receptors in the literature, but for many of the receptors for which isoforms have not been described, the computational mining exercise has allowed us to discover new receptor isoform variants for those GPCRs where these isoforms have not been characterized in a systematic manner. So to summarize this part of the talk, what I'd like to highlight is that most non-reference isoforms do affect functional hotspots for ligand binding and receptor coupling to different signal transducers, and that characterization of the different structural regions which are affected can suggest the most likely effects of functional impact of these understudied GPCR isoforms. So the key point that I'd like to highlight in this part of the talk is that isoform structural variation can diversify receptor function. So this takes me to the next question. So we have seen that there are all these different isoforms that can affect structure and can influence their function. How exactly are they expressed in the different tissues, and how could they contribute to diversification of signaling responses in the different human tissues? Now to address this, what we did was to also systematically analyze the expression pattern of these receptors and their isoforms in the 30 different human tissues for which we had data available from the GTEx resource. So I'm just going to walk you through a toy example where in this particular instance you have a receptor that has three isoforms. Each box represents one tissue, and here you have all the isoforms are expressed in all these different tissues. So there is no real major complication in the pattern of expression. So here we have a situation where the tissue expression signature is one. Now this is another toy example where there are three receptors, but you can see different combinations of the isoforms are expressed in different tissues. And what this essentially means is that when you expose these tissues to the same ligand or the same concentration, each one of these tissues may show differential downstream signaling response because you have various combinations of isoforms that are expressed. And here is an example where the tissue expression signature is four. So I'm just going to walk you through this using a real example where we focus on the cannabinoid receptor one. So there are three isoforms, and you can see that different combinations of isoforms are expressed in the different human tissues, whereas CD97 also has three different isoforms, but all isoforms are expressed in all these different tissues. So here you have four expression signatures, but you only have one expression signature, and thus what it is likely to be the case is that in the cannabinoid receptor case, the same ligand can give rise to complex expression pattern differences in the different tissues, but in the case of CD97, it's unlikely to vary in the downstream signaling response. Now when we calculated these tissue expression signatures for all the different receptors, we find that many of these receptors display a unique combination of isoform expression in the different human tissues. And what this suggests is that the tissue expression signature can indeed determine how complex the signaling response is likely to be in each one of these different tissues. So we then systematically also try to experimentally validate whether combinations of differential expression of these isoforms can influence downstream signaling. So we focused on the cannabinoid receptor, where different internal isoform sequences have been detected, and through experimental collaboration with Manoj Pradhan Midhu's group, what we were able to show was that co-expression of the different isoform combination can drive distinct cell signaling states. And what this means is that when someone's characterizing a receptor in a particular cell model, if they do not know that an isoform of the receptor is expressed in that particular cell type, then any measurement that gets made is likely to be a cumulative measurement of both the isoforms that are expressed, rather than a readout coming from that particular reference sequence. So when a cell line model changes and the isoform composition changes, same ligand, same concentration can give rise to differences in downstream signaling, which may not be easily explained unless we consider these isoform variations into account. Now we also looked at the problem of signaling bias through collaboration with Graham Ladd's lab in the University of Cambridge. We were able to find that expression of different isoform combination of the GIP receptor can result in signaling bias. So for the same ligand, you can have variation in downstream signaling response in one axis, but not in the other. So to summarize this part, we find that all tissues can express non-reference isoforms, and that reference isoform combination, excuse me, receptor isoform combination, may influence signaling complexity. And the key point is that such isoform combinations can diversify signaling response and may drive physiological signaling bias. So what does this all mean for GPCR drugs? This is a very important consideration to keep in mind. So to address this, we analyzed all the known FDA-approved drug targets, which are GPCRs, and we find that over 40% of drug targets have two or more isoforms. So this may initially appear that it could complicate signaling response, but the flip side of this is that such isoform diversity might also provide an opportunity for targeted responses. Meaning like, you know, we can develop small molecules that can target one isoform that is neatly expressed in the right tissue of interest, but not touch all the other receptors and their isoforms that may be expressed in other tissues, thereby minimizing potential side effects. So we again developed a computational pipeline and identified 42 receptors and about 64 isoforms that show unique expression pattern, as well as also have differences in the N-terminal sequences that are potentially targetable using small molecule ligands. So I'm just going to walk you through a couple of quick examples. So here's an instance of the CCR3 chemokine, where the N-terminal sequence is different in one isoform, but not in the other, and they're expressed differently in different tissues. And this is exactly where they contact the ligand. And the same with the CRF receptor, where you have varying N-terminal sequences, and again, different combinations are expressed that can influence their ability to engage with ligands differently, and thus presents an opportunity for selective targeting using small molecules. Now we also look for genetic evidence for whether isoforms can have unique therapeutic indications. So we mined the largest database that's available, which is a UK biobank that is genotyped over half a million people. And by mining this data, we were able to identify polymorphisms that uniquely associated with phenotypes only in one of the isoforms, but not in the other isoform. Again, suggesting that there is an underappreciated potential for isoform-specific therapeutic indications that could be revealed through data integration and data mining. So in summary to this part of the talk, what I'd like to highlight is that around half of the GPCR drug targets for which there are approved drugs in the market have more than one isoform, and that the genetic evidence links receptor isoform with specific disease phenotypes in the human population. And more importantly, such non-reference isoforms can confer potentially improved tissue selectivity and may represent novel drug targets. And that such rational exploitation of isoform levels drug selectivity may potentially yield new drug candidates with minimized side effects. So all the data that we have generated in this study is also publicly available, again, through the G-protein coupled receptor database that is maintained by our collaborator David Gloriam. And I would really encourage you to visit this website if you're interested in the polymorphism study that I presented or the isoform study that I've just discussed in the remaining part of my talk. So to summarize, what I'd like to highlight is that through a systematic integration of multiple large-scale data sets, we could develop computational approaches that could allow us to assess the prevalence, the functional impact, and the signaling complexity due to the expression of different GPCR isoforms in human tissues. So the key concept that I'd like to put forward is that all of these results are suggesting that we need to move from a canonical view of GPCR signaling, where we think about one receptor coupling to a G-protein and an adapter, such as a rest gene, to a context-specific view of GPCR signaling, where in each one of these tissues, different combinations of GPCR isoforms are expressed, and thus, collectively, the downstream signaling response can be very different depending on which combinations are expressed, even though they may be experiencing the same ligand environment. So you could have varying isoform repertoires that can result in system-specific signaling outputs, or what you may call a physiological signaling bias. So in conclusion, what I'd like to highlight is that GPCR isoforms do affect functional hotspots, which can diversify structure and function, and that expression of unique isoform combination can result in what's called physiological signaling bias, and such isoforms could potentially present novel therapeutic options in terms of targeting them or developing new small molecules that target isoform-specific receptors. All right, so there are a number of discussion points and implications. I won't be getting to the details of this, but I'm very happy to talk about this during the discussion session. But one key point is that these observations could possibly explain how organisms can mount different responses to the same ligand at the tissue level, although leading to a coordinated homeostatic response at the whole organism level. All right, so with this, I'd like to conclude and present some of the key references that captures the science that has been described in an objective manner in these publications. I'd also like to take this opportunity to thank a number of colleagues and co-workers who have been involved in this project. I've had their photographs and names and acknowledged them as we went along. I'd also like to thank the Medical Research Council and St. Jude Children's Research Hospital for their generous support and funding our research. In the last slide, I just wanted to mention that in the newly formed Center for Data-Driven Discovery at St. Jude Children's Hospital, we have a number of new job openings, both experimental and computational, on the GPCR site, particularly on GPCR pharmacology, GPCR cell biology, synthetic biology, functional genomics, and cryo-EM and mammalian protein science, as well as in cancer genetics, human genetics, omics, and data science. So if anyone's interested in working with us, please do let us know. Please do reach out to us. Thank you very much for your attention, and I'll be happy to answer any questions that you may have. So thank you so much, Vadim, for such a fantastic and exciting talk, which is really important, I mean, for many aspects of receptor pharmacology, in fact. I might like to start asking a question while we wait for more questions coming up in the Q&A session. So I was particularly intrigued by your finding that one of the most variable regions is the C-terminus of the receptor, which, as we all know, is very important for receptor trafficking. And you will be also very aware of the recent findings with endosomal, G-protein, and caporeceptor signaling that, of course, complicate to a very high degree the picture. So have you looked into the consequences of some of your isoforms or variants for trafficking and intracellular signaling? David, thank you very much for that question. That's a really important consideration. So we haven't reported that in this publication, but currently in terms of ongoing work, what we are doing is to look for all the different linear motifs in the C-terminal tail to essentially identify whether there are conserved protein-peptide interaction sites that might contribute to differential subcellular localization in the cell. So we do have a list of these receptors with a very high degree of C-terminal variation. So those would be interesting candidates for pursuing questions like what you've just described, David. Thank you very much, Madan. So I'm picking up some questions from the chat. So a first question from Eileen, thank you very much for the beautiful talk. Given many GPCRs are expressed in one cell type and GPCRs can exhibit signal crosstalk with different GPCRs, is it feasible to generate personalized GPCR isoform signatures for multiple receptors at an individual cell level to infer drug responsiveness? Now that's another very interesting question. So one area that we have been again pursuing work that is as yet unpublished is to look at all the single-cell transcriptomic data sets. So not all single-cell transcriptome data sets are amenable for investigation simply because they don't present the entire transcript, so you can't discriminate between isoforms. But then there are a subset of these studies that have identified and have characterized highly abundant transcripts by sequencing the entire transcript. So in this paper we just did a proof-of-concept study where we looked at, you know, human pancreas and there we were able to show that the different sub-cell types like the alpha islets and the different cell types do express unique combinations of very well-known receptors. So it does look like there is very clear segregation of receptor co-expression in the sub-cell types. But to go into the question of personalized medicine, I think we don't have sufficient power to be able to discover what are called the splice QTLs. So these are polymorphisms that affect the expression level of the two different splice variants. Perhaps in recent and in near future when we have more big data sets that are coming out like TopMed and other kinds of big consortia that are generating much more data, a systematic integration effort will allow us to perhaps identify instances where this happens, whether they're physiologically relevant we can't answer up front but these would be good candidates for individual labs to pursue. Fantastic. Another question from Daniel Drucker. How important might human G-receptor variants be for inter-individual differential responses to GIP in different patients? It's again another question unfortunately for which I don't have an answer but what we can say is that the GTEx consortium is a database that has information from 800 different individuals. So one can look at a variation in the expression level between different ethnic groups. We haven't pursued these kinds of questions but the data set and the resource that we have developed will allow individuals who are interested in this to be able to pursue this in much greater detail. So the short answer is I don't have the answer but the answer should be there in the resource that we have. Thank you. Another question. Can you account for effects of GPCR splice variant timers on bias signaling with your data pipeline? Is that something you have been looking at? So I think in the example that we presented in the GIP receptor, there we were able to just co-express two different isoforms and we were able to systematically start seeing signaling bias and that's something that we reported in the paper and we find many such instances where receptor isoforms with altered C-terminals that miss the phosphorylation site in the C-terminal tail might contribute again to like you know the prolonged presence on the membrane which essentially means that in a concentration or a dose dependent manner, you could prolong the signaling depending on how much of the two isoforms are expressed. So theoretical combinations suggest that it is possible and we have identified instances through this systematic computational analysis which receptors are likely to exhibit such a behavior. Whether we have tested them all, we haven't. We showed this for one receptor but for many of the other receptors we do present the data as part of that resource. Yeah, there seems to be still a lot of potential work that can be done. We were excited about this study because it opens up the possibility to share to the GPCR community to be able to pursue or address some of these questions and know which receptor to actually focus on or which receptors are susceptible for these kinds of interesting physiological behavior. Thank you. If I'm allowed another question, so do you have any thoughts or evidence about whether I mean these splice variants or I mean, this complexity is somehow regulated in cells, for instance, during development, or maybe even in the same cell over time, for instance, if you look at different phases of the cell cycle. So have you looked into that, or would it be possible to investigate it with your methods? Thank you, Davide. So we have an ongoing study, and in one of these studies, we've been looking at microRNA-based regulation of the GPCR and splice isoforms. And there, we find evidence that some of these microRNAs uniquely, likely regulate one isoform, but not the other. And many of these microRNAs are also expressed in a developmentally relevant manner. There's a big dataset that's been published by Hendrik Kassemann's group, and we are currently pursuing a study looking at that dataset, because they have a systematic time series, but only from three individuals, not a huge number of individuals, because it's not easy to get developmental time course data for different tissues across the different ages. A mining of that information would hopefully help address some of the questions that you have there. Thank you very much. Greg? Yeah. I'd like to comment that there's a couple of questions from our panelists, and one of them is related to a question that I had, is whether you've had a chance to look at the effectors, intracellular effectors, and whether there are variants there, and as well as the levels of effectors that might affect the responses mediated by these different GPCR isoforms. Great question again, Ursula. So as part of this big project, so we only published one part of the study, the other part, we are writing it up, and that's something that we had already discussed with Davide. For instance, in the case of the GS protein, there is huge number of splice variants, and as you might already know, some of them are also imprinted. So it really is a very, very interesting case where a combination of splicing, microRNA regulation, and the relative expression levels in these different tissues together create a combinatorially very large number of possible ways to explore and explore the space of phenotypic diversity, perhaps all within a healthy range, and then we could start seeing instances where it tips the balance, and then people present themselves with disease. So we're really only hoping that in the follow-up study, which we are currently pursuing, we will be able to more comprehensively characterize the G protein splice variants, the GRK splice variants, the adaptor splice variants on arrestants particularly, and ramps for some of the class B binders. We find very interesting examples there, and hopefully in the next year or so, we'll be able to present some insights into that. Thank you. If I can ask maybe a last question that is also coming from our panelists. So the question is whether you had a chance to somehow compare the isoform patterns that you see with individual sequences, I mean, with variations in the genomic sequences of individuals, if perhaps there is any correlation between the patterns that you see and the genome of those individuals. Now, it's an interesting point, David. So the closest that we came to addressing a question like that was to mine the UK Biobank data. And in the UK Biobank data, they do have information about polymorphisms and associations with the phenotypes. And then when we looked at the more statistically significantly associated polymorphisms with the phenotypes, some of them fall only on one isoform, but not on the other. So that's the first, I would say, instance of indication that gives us some confidence that these are still associations, not causal relationships. But nevertheless, the fact that they're seen only in one part of the, or one of the isoform, but not in the other, and have such a strong association suggests that there are therapeutic indications that can be linked with some of these isoforms. Okay, thank you very much, Madan. There are a couple of more questions coming up in the chat. Maybe we'll have a chance to directly answer those. Otherwise, I think we need to, for the sake of time, we need to move on to the next speaker. Thank you very much again for a fantastic talk. Thank you again, Ursula and David. Thank you. And now I'd like to introduce our second speaker, Dr. Rob Sladek. Dr. Sladek is Associate Professor of Human Genetics and Medicine in Endocrinology at McGill University and the Genome Quebec Innovation Center, where he leads the Diabetes Gene Discovery Group, a project to identify risk loci for type 2 diabetes. The title of his presentation is Using Genetic Associations as a Tool to Dissect GPCR Function in Type 2 Diabetes. Rob. Thanks very much for the introduction, and I'm very grateful for the chance to speak to you today. Here are my disclosures, and I have nothing relevant to disclose. My background is in human genetics, and my interest in G-protein coupled receptors comes from that field. And as a human geneticist, when I think about genetic effects on a receptor or protein function as part of a disease or trade, what I'm interested in are the types of variants, the so-called allelic spectrum, that those genetic changes can reflect. And I consider two things. One is I think about whether the variant has a large effect, whether it's a strong or a weak variant, or whether, in fact, it might actually reduce the disease risk rather than increase it. And I also think about how common the variant is, whether it's, for most of the common diseases I deal with, these variants are gonna affect 10, 20, 30% of individuals in the population. But I'm also quite interested in rare variants. And if we think about this at a population level, these common variants are actually variants that are quite old. For, if we think that a variant can arise in between you and your parents, there's about, oh, maybe a couple hundred new single nucleotide polymorphisms that come up each, in a generation. For those to actually become so common that they affect 10 or 20% of individuals, it takes many, many tens of thousands of years. Now, in my era, the problem with identifying genetic causes of common diseases is that it was a very difficult thing to do until we developed a technique called genome-wide association studies. And these have tended to identify variants that are relatively common in the population. The sweet spot is for variants that affect about 30% of individuals, have an allele frequency of about 30%. And it tends to affect variants that have relatively small effects. And if you think about how those, excuse me, how those variants arose, they will have to have small effects because if they're severely deleterious, selection will push them out of the population. In more recent genetic studies, and in particular the studies I'm gonna talk about today, we're starting to look using sequencing techniques called genome sequencing and exome sequencing techniques at much rarer variants. And because these variants are more recent, in fact, they may only be one or two generations old, it's not uncommon that they can have much larger effects on disease risk in a phenotype. Now, one of the things that we found in our studies of type 2 diabetes was that the genetic loci that seemed to alter disease risk included a number of G-protein coupled receptors. And as you know, these are a fairly large family of proteins and are involved in many metabolic and physiologic responses including some that are quite relevant to diabetes. And we were able to identify quite early coding polymorphisms in the GLP-1 receptor as a risk locus for diabetes. And the fact that we found coding polymorphisms there is quite important because only about 20% of the, well, so far 400 loci we've identified as risk factors for diabetes can actually be tied directly to a gene by looking at that polymorphism. What's also interesting is that the genetic loci that we found included a number of diabetes therapies and the GLP-1R and other G-protein coupled receptors among them. Now, as you know, G-protein coupled receptors signal through a variety of pathways through the cell. And what was attractive to me because we want to look at a lot of proteins is that through collaboration with Stéphane Laporte and Michel Bouvier and colleagues in Montreal, we had a toolkit that could look very specifically at those signaling pathways with a very high signal-to-noise performance. And so using BRET-based biosensors, bioluminescence resonance energy transfer biosensors, we're able to look at the post-receptor pathways of G-protein coupled receptors with exquisite sensitivity and specificity. In the case of G-protein coupled receptors, what this means, in the GLP-1 receptor in particular, what this means is that we can start to look at the activity of competing signaling pathways coming from the receptor. And for GLP-1 receptor, we're very interested in GS-dependent pathways. So I'll show you some results with the EPAC biosensor. And we're also very interested in beta-arrestin, pardon me, beta-arrestin-mediated signaling pathways and the balance between that and G-protein-mediated signaling. So in our studies, basically what we're gonna do is take a large number of genetic variants coming from sequencing studies, do classical dose-response curves looking at GLP-1 signaling through the receptor, measuring beta-arrestin-1 and measuring EPAC activation. From that, we'll get maximum activity levels, which tends to be the things that geneticists are interested in. But we'll also pick up the EC50 for the receptor just in case we have polymorphisms that affect ligand binding. And a lot of the studies I'll talk about today are using the relative activity, which, as you know, is derived from the ratio between the EMAX and the EC50 for each of these signaling curves. And what you can see here is actually the wild-type receptor being studied for beta-arrestin-1 activation and EPAC activation and the negative control is fairly flat. And again, the power of these sensors for us is the high signal-to-noise ratio that they possess. Now, prior functional studies of the GLP-1R had already looked at a number of polymorphisms in the receptor. And so, for example, work by Patrick Satchiston and Denise Wooten had looked at the extracellular loops of the receptor and the ends of several of the transmembrane domains and using a technique, valonine scanning mutagenesis, tried to identify residues that were important for ligand binding in the receptor. We had also, in the past, worked with Philippe Fragel on signaling through the melatonin receptor and Ralph Jokers, who'll speak after me, I expect, was involved in that work. The point is that when we start to, or pardon me, when we start to look at these studies, we get a pretty good view of the signaling pathways that are going on. But because they tend to be hypothesis-directed, they may not find mutations throughout the receptor away from areas that were classically of biochemical interest. Now, within the GLP-1R, genetic studies had basically identified three coding polymorphisms, which affected diabetes risk. Two of these lie in the transmembrane domain of the receptor, and one, the R131Q mutant, lies in the extracellular, the unstructured domain of the receptor. R131Q was found in a Korean population. It's very uncommon in Caucasians, whereas A316T is more common in Caucasians, but all of these are pretty rare SNPs. And as you can see, they have differing effects on diabetes risk. So in a traditional pharmacogenomics approach, what you could do with these is basically using different ligands, look at the effect on a number of post-receptor signaling pathways, and determine whether the particular mutations cause a signaling bias, or whether the specific ligands may correct or worsen that signaling bias. And I guess the one thing that becomes remarkable, or that's most remarkable about this figure, is that when we look at it, it's hard to see the blue line for the wild-type receptor, but when we compare the mutant receptors, the risk-associated T149 polymorphism really stands out from the other two receptors, regardless of what ligand is put on. And so in a classical pharmacogenomics sense, for common polymorphisms, this is the sort of thing you might think about for personalized therapy, or for other genetically precision medicine approaches. Now, for our study, we use sequencing data from the TopMed Consortium, which is basically a study set up by NHLBI to sequence about 120,000 individuals eventually. And the data set that we had access to was sequencing whole genome sequencing data from 55,000 individuals approximately, of which 44,000 had known type 2 diabetes status. Now, these individuals include people with different types of heart disease, ischemic heart disease, atrial fibrillation, also asthma, and other conditions. But what this provides us with is a massive set of whole genome sequences for which we can start looking for rare variants. And from this sample, we identified 119 non-synonymous SNPs in the GLP-1 receptor. Most of these are singletons or doubletons. In other words, one or two people were affected. So we're really talking about polymorphisms that are very rare and are not amenable to traditional pharmacogenetics use, pharmacogenomics use. So the question is, what can we do with these SNPs? Now, interestingly, they're pretty much distributed all over the receptor. And my guess is if we sequenced a large enough cohort, we'd pretty much color this snake diagram in yellow. If we look at surface expression, what we can see is that a number of receptors, receptor mutants, severely affects surface expression. And these are distributed narrowly through the receptor. In total, about 12 of the mutants had no surface expression and another six were expressed, but we saw no evidence of signaling or very severely deteriorated signaling for both beta-arrest and EPAC. If we look at the expressed receptors, here's a plot of all the roughly 100 receptors that are left, wild type indicated in blue, showing the activation of these different receptor mutants, looking at Emax in this case. And what we found is that mutants that affected EPAC response were again, distributed pretty much throughout the receptor with no real concentration in the domains that we expected. And this was quite interesting to us because what that meant was that these mutants were acting allosterically. And in a world where we often do genetic studies, assuming a function of mutants based on computational predictions or species-based homology mapping or structural predictions, it was interesting to us to see that many of the mutants that caused a severe impairment of the receptor function would not have been identified by those computational techniques, which makes our genetic studies a lot more difficult because we can't use those computational models when we start to determine whether a particular variant, one of the 100 that we find, actually has an effect on the receptor function. In addition, we looked at SNPs that affected a beta-arrestant response. And here you can see plotted across the top is EPAC activation for three different polymorphisms we found, and you can see in each case, the EPAC activation is pretty much preserved. EC50 is the same, Emax is the same, or in a couple of cases, maybe a little increased, but the beta-arrestant response is completely absent in these mutant receptors. And again, these are distributed throughout the transmembrane domains of the receptor, as well as in the extracellular unstructured domain, suggesting that their effect on this could be allosteric or rather than a direct effect on beta-arrestant binding. This shows the Emax values for these receptor mutants. I've indicated the ones that are non-significant because basically pretty much of all of the receptor mutants that we, or many of the receptor mutants we studied had severe impairment of Emax for beta-arrestant recruitment. And if we look at the log-r relative activity for these receptors, again, we can see that a large number of the receptors have impaired relative activity. So what this led us to was to find molecular phenotypes for the GLPR1 mutants that we found. So a small number of receptors, in fact, only about seven, six or seven, had unchanged dose-response curves when treated with GLPR1. And we were able to identify groups of receptor with reduced activity and receptors which had selective impairment of beta-arrestant. Enhancement of EPAC, but we weren't able to identify receptors that had deficient EPAC activation. That may be a function of the construct we looked at, but certainly we can identify pathway-specific mutants for these polymorphisms. So using these classes of genetic mutants, and particularly the EPAC biased group and the beta-arrestant reduced group, we went back and used special genetic tests that are designed to work with rare genetic variants. Because the NHLBI TopMed cohort includes people of different ancestries, we corrected for this. And in addition, we corrected for sex, the age at last exam, and also the study from TopMed that they recruited from just in case there was a bias for other diseases that were involved. And what we were able to show was that the beta-arrestant reduced group, and particularly, had the strongest association with diabetes risk, where the EPAC biased group had a slightly less significant association with disease outcome. So what we learned from this study, and I think what was most remarkable for me, was the number of SNPs that significantly alter GLPR, GLP1R function in vitro. And particularly those SNPs that are acting allosterically, and those SNPs which can introduce a signaling bias in the receptor. Because this occurred, we were able to use genetic data that could correlate these in vitro phenotypes with disease risk in human subjects. And this to me is the real power of this type of approach, because what this means is that we can study the effects of genetic variation at timescales and in environments, and by that I mean diet, lifestyle, and other stressors that are relevant to human populations. It lets us start to do not quite knockout studies, but certainly significant and meaningful genetic studies in human subjects. And let us look at the, because complex diseases involve an interaction between genes and environment, let us look at the function of these genes in the context of a meaningful environment stimulus. I think our studies are starting to show that genetic variance, and particularly rare coding variants can help us dissect the pathogenesis of a complex disease. And in addition, these can provide receptor archetypes that severely affect a post-receptor pathway that can be useful in pharmacologic and ligand studies. But where I think the big advantage of a study like this is that it allows us to take these mutations and other mutations that are identified in large biobank samples that have been deeply phenotyped for a large number of diseases, and start dissecting down into the interaction between diabetes and heart disease, for example, or diabetes and kidney disease, and other outcomes that may be determined by GLP-1R or other loci, risk loci, that we've identified, which have sequence polymorphisms, missense polymorphisms that change protein function. I'll end with pointing you to three publications. The first from Jennifer Wessel, looking at the genetics of GLP-1R mutants in diabetes and of rare variants in diabetes. The second paper by Patrick Sexton and Denise Wooten's group that looks at the complexity of signaling at GLP-1 receptor. And a paper from Michelle Bouvier's group that talks about biased agapinism and its role in drug development and discovery. If you're interested in the field, I'd strongly recommend those papers for you, because I think they give good coverage of what's there. I'd like to end by thanking my collaborators in the study. The GLP-1R work that I presented was done mainly by Lama Yamani and Chifang Dang in my group, helped by Aaron Cho and Yung-Neng Kung, working for Stephane Laporte, who's my collaborator at McGill. The statistical analysis that I presented was done by Tim Mahajeranian, who works for Alyssa Manning, and I'm also collaborating with Jennifer Wessel and James Meeks, because one thing we've learned from this study is that we're gonna have to get much more detailed and insightful statistical models to make the most sense of the data. Thanks very much for your time, and I'd be happy to answer questions going forward. Thank you very much, Dr. Sladek, for a really fantastic talk, really interesting data, makes us think about what our future will look like. There's a question from Dan Drucker, says, hi, Rob, nice talk. Have you seen any evidence that genetic variation in the GLP-1 receptor associates with differential responses in terms of glucose levels or food intake or body weight to acute or chronic GLP-1 receptor agonists in humans? The short answer to that is yes, I'd love to do a study like that, but the difficulty is accessing the data. The study actually came out of something called the Accelerating Medicines Partnership, which was a collaboration with Pharma to look at genetics and diabetes. And one of the, I think the disappointing aspect of that consortium, everything else was fantastic, but the disappointing aspect is we couldn't get RCT data with genotypes, because that would be absolutely beautiful to answer that sort of, that question. That's what we'd really like to know. And I think you showed in vitro data that there was no difference between the different GLP-1 receptor agonists in terms of their ability, you know, one, these variants, at least in a subset of the variants, didn't change the responses, or am I not remembering that correctly? Yeah, it's hard to say a lot there. And we didn't see statistically significant differences. The rosettes look different, but nothing that I'd want to stand up and swear to. So I guess I'll ask another question. You know, are we looking ahead to a future when everyone gets their genome sequenced, which is probably not so far into the future, and you could predict that these people might have increased risks of developing type 2 diabetes, you know, what are the implications for that? Would that mean that you might treat them preventatively, or, you know, we might be able to predict one treatment or another? So that, you know, that's obviously the goal of a pharmacogenomics study, and it's a really good question to ask. Other panelists may have some comments on this, but the one thing that really struck me is how rare the genetic variation in some of these receptors is, particularly GLP-1R. So this receptor, in all populations looked at so far, has about six common polymorphisms. Four of them, as far as we can tell in the lab, do almost nothing, or virtually nothing. And two of them have some effects on beta-arrestin signaling. So the short answer is, for this receptor, I don't think we'll be doing it yet, but I would love to see some work coming out of, say, UK Biobank, where they've got large samples and slightly less common polymorphisms to see if there's something there. But at the present time, no, I don't see traditional pharmacogenetic tests being useful for this, or for the other metabolic GPCRs that we found in the GWAS studies for diabetes. There's another question. Jenny Visser asked, she commented, interesting data and approach. Have you also studied more frequent variants? So I will- Frequent variants might be more relevant to the general population? Yeah, I'll study any variant. But the short answer is, the more frequent ones we didn't find as much interesting, and certainly not the extreme bias signaling that we saw with the rare variants. There is maybe a question coming from our panelists, whether you looked also downstream of cyclic MPE signaling in your studies. So have you looked, for instance, at transcriptional effects or any other, do you have any evidence of bias downstream of- So the short answer is not yet, but that's obviously something that's interesting to us. So what we're using, the plan is to take the archetypes that we found, not the whole set of receptors that we found, and get them into some beta cells to try to dissect out these more specialized pathways. Thank you. Am I allowed maybe one question from my side? So do you think you could learn something, or at least could you speculate in terms of what would be, I mean, based on your data, I mean, what we might learn in terms of pharmacological applications of whether, for instance, you can tell whether a certain type of bias might be beneficial for patients, and maybe whether perhaps one day we might get GLP-1 bias agonist that only activate beta-resin versus, or preferentially activate beta-resin versus GS signaling. So do you have any idea based on what you're finding in patients on what might be the ideal GLP-1 agonist? So I think that the work could go several ways that would be useful pharmacologically. And so as Dan Drucker asked, we'd like to know the interaction with drugs for some of these. More importantly, I guess I'd also be curious about other GPCRs that are sitting on the beta cells. And for example, if I know that a pathway is associated with, say, a better response to drugs or a delayed onset of diabetes, transition from diabetes to prediabetes, I might be curious to look for agonists that work along that pathway to see if they're better. But I can also go a step further and, you know, particularly in this world where we're talking about double therapy, triple therapy, this type of thing, I can also start to ask questions about some of the biased agonists that people are using to target both GLP-1 and GIPR. And ideally then I can look for drugs that are gonna push the pathway I want to push in terms of survival data, in terms of outcome data. And perhaps that's a way to influence drug discovery, drug design. And can I ask another question? Was it possible in this cohort to see whether these SNPs that you've studied, they correlated with increased risk of type 2 diabetes? Did they correlate with obesity or BMI as well? That's something that we're looking at, yeah. It turns out because it's a multi-ethnic cohort, the methods for doing rare variant analysis are effectively counting methods. And you can do a little bit of correction for covariates, but it's a bit hard. It's not like the regular association tests that we do for GWAS, where you can just set up a model and everything falls out the way it should. But we're definitely interested in effects on obesity, particularly as we move into receptors like GIPR to try to dissect that out further. So a lot of where we're heading right now is not just to do the assays, but with Alyssa Manning, James Meigs in Boston, Jennifer Wessel in Indiana, to see if we can actually dissect out the statistics better in terms of the populations. And another question from our panelists, you maybe partially answered it, but do the molecular phenotypes that you identified in terms of signaling differences for these variants, do they converge on a small number of cellular responses? Have you looked at other cellular? Yeah, so that's exactly what we're looking at is to see how much overlap there is. Obviously the key things would be beta cell proliferation, beta cell survival, insulin secretion, these types of things. And we'd like to know, does the beta rest in signal, does it actually start to synergize or interfere with the GS signal that we're seeing at a later step in the cells? We don't think it will because the risk output, if it did, why would these be risk variants? But on the other hand, yes, that's exactly what we'd like to know. So we've got time, I think, just for one or two more questions. So have you looked at all at SNPs in non-coding sequences to see if they are associated with risk of type 2 diabetes? Yeah, so most of what we get out of a GWAS is non-coding, about 80% of the SNPs. And there the challenge really is to try to take that back to the gene because linking a promoter or a SNP even to a promoter or enhancer to a gene transcript turns out to take a lot of work. What's interesting is for a lot of these genes, and particularly, if you think about a gene like MODY or a gene that's involved in MODY, we actually see a cluster of different types of mutations there. And some will be the very strong autosomal dominant early onset in kids kind of problem that you'll see. And then the second will be the more nuanced, environmental interacting transcriptional affecting SNPs. And then finally, usually rarer, we'll see coding polymorphisms in the same gene. So there's a whole sort of hierarchy of effects. And each one of those needs a different substrate, families versus populations, and also different lab techniques to sort them out. Great, well, thank you very much, Rob, for a really wonderful talk and very thoughtful questions and answers. And we're gonna pass it back over to Davide to now introduce our final speaker. Thank you, thank you very much. Many thanks, Dr. Sladek, also from my side. It's now my pleasure to introduce the last speaker of this session, who is Ralf Schokez. Ralf is a leader of the functional pharmacology and pathophysiology of member receptors team at the Institut Cauchon in Paris, and is going to talk about type two diabetes associated with melanotonic receptor two. Please, Ralf. Thank you for the nice introduction. And it's my pleasure to participate in this Congress. So what I would like to do today is to talk about type two diabetes, and particularly its association with the melatonin MT2 receptor. So these are my disclosures. And this is my outline. So I will briefly introduce you type two diabetes, then talk about common intronic variants of the MTNR1B locus, and give you some basics about melatonin receptors. In the results section, I will talk about the rare variants of the coding region actually of this gene, the MTNR1 gene, which is coding for the melatonin MT2 receptor. I will talk about the functional characterization of these variants, their association with type two diabetes, and we had the chance to perform a pilot study on 15 patients, and we characterized them in terms of circadian, sleep, and caloric intake. So type two diabetes is a global health problem. There are currently suffering more than 460 million people of this disease, and with an expected increase, overall increase of 51%, with a huge variability depending on the continents and countries. Back in 2009, there were three back-to-back papers in Nature Genetics that found out that by genome-wide association studies that intronic variants in the MTNR1B gene is associated with an increased type two diabetes risk, an increased fasting plasma glucosa levels, and decreased early phase insulin response. So these are frequent variants with minor allergic frequency of about 30%, and an odd ratio of 1.1, 1.15. So over the last couple of years, then a lot of genes have been associated with type two diabetes risk, as you can see here, or associated traits, and MTNR1B is one of them. So to understand the effect of melatonin, particularly in metabolic processes, you have to know that melatonin can have different types of effects. It can have very classical effects, which means immediate effects. To understand this, you have to know that melatonin is actually secreted in a circadian manner with peak levels only during the night, and this happens by the pineal gland. So apart from this immediate effects, melatonin can also have perspective or delayed effects. So this melatonin secreted during the night can have functional consequences during the day. Another interesting aspect of melatonin is that it can have chronobiotic effects, because melatonin has its receptors on the hypothalamic circadian clock, the master clock of the body in the SCN, so where it can actually regulate the circadian rhythm. Another effect is seasonal effects of melatonin, since melatonin is secreted during the night, so it actually can sense the duration of the night, which varies with season. So melatonin can have effects either through the central receptors or also through insulin-sensitive tissues in the periphery, like skeletal muscles, adipose tissue, the pancreas, and liver. So taking all this together, we can expect that melatonin can have two major effects on metabolic processes, either by regulating the circadian rhythm, since circadian disalignment has been associated with metabolic diseases, or through its effect on insulin-sensitive peripheral tissues. Melatonin can act through two receptors, the MT1 receptor and the MT2 receptors. They are typically GI-coupled receptors, also both coupled to beta-restin. There's an additional coupling of the MT1 receptor to the GQ pathway. We are lucky enough that the structure of these receptors has now been solved. Here is the overlay of the MT1 and MT2 receptors, so you see they are quite similar, with the ligand-binding pocket here in the upper part of the transmembrane domain here as a focus, and so actually all non-melatonin receptor ligands accommodate to this general scheme, where we have a hydrophobic pocket here and a hydrophilic part here, and as you can see here, you can classify all the ligands according to this scheme. So back in 2012, we then, we asked the question, can we identify variants in the coding region of this MTNR1B gene? And we did this together with geneticists of the lab of Philip Frogan in Lille by sequencing in more than 7,000 Europeans, including 2,000 type 2 diabetics, and could identify indeed 40 rare variants, which were actually distributed all over the coding region of the receptor. We then performed a preliminary functional study on these receptor mutants, and could then classify them into those which had a loss of function and those which didn't have an obvious phenotype. And this allowed us to show that actually it's only the loss of function variants which are associated with disease risk with an odd ratio of about 5.6. We then went on and wanted to know which pathway more precisely is actually defective and defective and associated with the disease. To do this, we generated a signaling profile of different G proteins, particularly GI and GZ proteins, the beta-restorin recruitment, the cyclic AMP pathway and the ERG pathway, and we're having a look what are the modifications in terms of constitutive activity and in terms of ligand-induced activity. As you can see here, those are in concentration response curves. So we found different flavors, so complete loss of melatonin binding or all pathway defective mutations down to more pathway-specific mutations. And there was also one mutation which had a slight gain of function. So by doing this, we came up with the following scheme. So once again, we were looking on spontaneous responses and melatonin responses. And so interestingly, we could show that G protein activation was associated with disease risk in terms of melatonin-induced response. So all these G protein activation had a reduced response, as you can see here. Even more interestingly, also the beta-restorin pathway was associated with disease risk. And here, actually it was not the melatonin-induced response but rather the spontaneous response, as you can see here. So as we can conclude here, so the detailed function analyzes of this rare variance mutations confirmed the link of the loss of function with disease. And it's not only the cyclic AMP pathway that is affected, but also the beta-restorin pathway. And even more interestingly, it's not only the melatonin-induced response but also the spontaneous receptor response. And that's quite interesting because as I told you earlier, melatonin is secreted in a circadian manner only during the night. So this spontaneous activity is actually present all over the time. And this means that there's a potential importance of this activity also during the day. Furthermore, it opens additional pharmacological opportunities by using inverse agonists. So in the second part, I would like now to come to our pilot study where we had the chance to follow 14 type 2 diabetics that had actually rare carriers of the MT2 mutants. And we could compare them with type 2 diabetics that didn't have this mutation. So these people were wearing actigraphs and had to perform daily food logs. And we were following them for four weeks with the aim to determine the circadian, sleep, and caloric intake phenotypes. So as you can see here, these are the actigrams. So here we have a patient that is very regular. So these bars represent inactivity. So this is the sleep during the 24-hour course. And these are the number of days. So this person is going to bed at a very regular time and waking up always at the same time. And here you have the opposite. This is a patient that has very irregular sleep and a very fragmented sleep pattern. So if you now compare the patterns of the normal type 2 diabetics and the mutation carriers, we can see that mutation carriers have actually a later sleep onset and later mid-sleep, as you can see here. They also have a less regular sleep index, so showing that they have a clear modified sleep pattern. They have also a composite phase deviation, which is actually a measure for circadian disalignment, as you can see here, which is increased in these mutation carriers. According to the caloric intake phenotype, so we can see here over the 24-hour cycle that we have three main meals as expected, breakfast, lunch, and dinner. Here we have a small snack in the afternoon. And when you then superimpose the controls and the mutant carriers, you can see that the mutant carriers actually eat over a longer time during the breakfast and the lunch. So they eat longer and they have more frequent caloric intakes, as you can see here. So these patients have more caloric episodes during the 24-hour cycle. So as we have seen here, if you sum up, so these mutant carriers that I have to remind you, they're actually heterocycles, show circadian sleep and caloric intake phenotypes. They have a more irregular sleep, higher behavioral circadian misalignment, more frequent caloric intake. And this is likely to participate in the increased risk for type 2 diabetes. So as a general take-home message, so I could show you that HH is a loss of function of this receptor, which is linked to the increased type 2 diabetes risk. There are two key signaling events which are associated with the G-portioning signaling. We have the GIGZ pathway. This is likely to operate during the night when melatonin is secreted. And also interestingly, we have a spontaneous component, which is most obvious for the beta-resting signaling pathway. And this is expected to operate during the night and during the day. So this pilot study that I presented you with this variant mutant carriers shows that they have more irregular sleep, circadian misalignment, and more frequent caloric intake, most likely participating in the phenotype. Since we are talking about rare carriers, this work has to be put in the context of precision medicine that will hopefully in the future help to treat these patients. So this work has been carried out by my team in Paris. And there are the key peer persons of this were Angeliki and Natalie. This was also a highly collaborative work. So starting with our geneticists from NIL, Philippe Froger's lab, we did the functional studies together with the lab of Michel Bouquet. Bioinformatics studies were done in Baylor, Olivier Dichtard's lab, and the behavioral studies with the patients were done in collaboration with Sylviette Vetter and Tilde Urneberg's lab. And with this, I just show you the three key references associated with this work. And I thank you for your attention. Thank you very much, Ralph, for such a great talk and clear presentation. While we wait for the chat to warm up and more questions to come, I might start with one or two questions from my side. I would like to start, if I understood correctly, those mutations or those variants were found in heterozygous state, right? Can you comment on what you think might be the mechanism of action of loss of function mutations that are in a heterozygous state? So is that because there is maybe some dominant negative effect or have you looked into that? Thank you for the question. Yes, it's absolutely relevant, especially for the patients. So we had two obvious hypotheses. Indeed, it could be a dosage effect. Or we also had the, since we are GPCR people and we've worked on dimerization, we also had the idea that it might be a dimerization and yes, a dominant effect within dimers. We have followed up on this. So yes, for some, it seems to be dimer driven, but actually it's not that trivial because actually we don't have actually an idea of which of the, whether it was the Y type or the variant is equally expressed at the same time. So we can try to mimic this in vitro, but we might not be close to reality. So yes, we have some hints, but we don't know actually the real in vivo situation. Okay, thank you. And maybe related to this, have you looked at the subcellular localization of the variants? I mean, it's rather common in case of loss of function GPCR mutants suppose, for instance, retention of receptors in the endoplasmic reticulum, which could also be a mechanism of dominant negative effects has been shown in the past. Is there something maybe you looked into? Yes, this was indeed one of the obvious things and the result was really surprising. Actually, none of these variants had a significant effect on the surface expression. That was, I mean, knowing as a variant, as the receptors, we heard the example of GFP1 receptor today, but there are others, vasopressin and others. This was really surprising to see. So at least in 293 cells, let's say there is no obvious trafficking or export phenotype. Okay, fascinating. You also, I mean, of course you cover melatonin receptor ones and twos. I mean, so how redundant are those two receptors or is there a cross talk? Are they cooperating? Are they just two parallel systems? I mean, can you learn something from the patients and the variants you've identified? Yeah, that's another intriguing question. Indeed, we learned from geneticists that yes, you have a phenotype or defect in one gene of a pathway or system. And then you would expect them maybe also to have a similar phenotype or similar variants in another component of the pathway like melatonin synthesis genes or the other receptor. So we actually had a look in the other receptor and the MT1 is not published yet, but it's a surprising result was that there is no, there's also a lot of variants, but actually no impact on disease, on type 2 diabetes. It's most likely related to the expression pattern, which is actually not overlapping. So you're suggesting that they are not expressing the same cells essentially? Yes, for example, the MT2 seems to be much more prominent in the beta pancreatic cells. And also they don't have redundant functions actually in the brain, in the SCN. So it's most likely because of different functions. Okay, so I have, we have a couple of questions from the audience. I'm gonna read a couple of them. So the first question by Peter Liu saying, in the human study, it looked as if there were, there was a lot of napping suggesting sleep was altered. Was sleep architecture looked at formerly by EEG also relevant to metabolism? Was physically activity different in a 24 hour period between the carriers and the normals? No, unfortunately this was a rather simple study. So people stayed at home. So there was no EEG or other more sophisticated techniques. So it was rather activator and self-reporting. Physical activity was also simply monitored by the activator. So we did not know which kind of activity, whether it was really exhausting or just moving. So we, unfortunately we did not, we were unable to dig deeper into these phenotypes. Okay, a couple of other questions. I think one you've already answered, which was about whether, but maybe you can add one word to that, whether you think MT1 can rescue the loss of function of MT2 mutations? I think you've already answered that. Rescue mean rescue. Yeah, rescue, yes. Yeah, no, so we don't see this. And it's, we don't see it on the level of the rare variants and we don't have, there was nothing equivalent coming out from the GWAS study where we had this very frequent SNP. So that it seems to be really MT2 selective. Okay. Another question, which I find intriguing in a way, it's a different way of looking at things, which is, is there any upside perhaps to having any of those loss of function mutations? So is there maybe any benefit besides the association with the DTCs? Maybe we were not looking for the benefits. You have to, actually, the question is tricky. And for example, you may know that many of the mice strains that we are working with, C57, for example, actually doesn't produce melatonin in significant amounts in the mice. Actually doesn't produce melatonin in significant amounts and for sure not in a circadian manner. But they are not so bad. So it's, melatonin is not something that is a matter of life or death. It's a regulator. So it's a function would be rather at a regulatory level. That's what I see. Okay. Then we have a long, short question by Eileen. At least she, that's how she calls it. And who is writing, during menopause, women can experience frequent waking during the night and increased caloric intake. Is there involvement of differential inspection of melatonin receptor variants with aging? Ah, with aging, no idea. We did, I mean, one of the limitations and we already heard this previously from Rob is that we are dealing with rare variants. So they are not only heterozygous, but they are also rare. So the statistics are very painful. So to split the few that we have even up into different age categories is yet another challenge. What one should know is that melatonin levels or the hormone levels are actually declining with age. So we have a peak level when we are young and then they are declining rapidly. So there might be a component that is actually hormone dependent. And that's why many people in the US especially, they try to compensate for this loss. But on the receptor level there, there is no data. Okay, so maybe a last question from the audience again, maybe to conclude maybe with a clinical tone, which says, this is by Ron Neufeld, who says, I have a patient without diabetes that takes 90 milligrams of melatonin and is not sleepy. Should he be genotyped? 90 milligrams is a huge amount. So to, yes, maybe he should be genotyped. So we also had to really, as you have seen, loss of function or loss of melatonin binding is something that we see in a significant number of these rare variants. Yes, it's an option. Okay, many thanks, Ralph, I mean, for the very lively discussion and again, for the fascinating talk. I think for the sake of time, it's now time to close the session. And I would like to thank from my side, all the speakers again, for their participation and all the participants in the discussion. And I will say last few words as we close the session. I'd like to express also my thanks to Ralph for your very interesting presentation and to all of our speakers for a very enlightening session and to all of our participants for participating and for the interesting questions. So this concludes our session. Thank you again and enjoy the rest of your day and the small amount left of the meeting. Thank you.

G-protein coupled receptors

genetic variation

drug response

drug development

GPCR signaling

ligand binding sites

functional sites

personalized medicine

tissue-specific differences

GPCR isoforms

type 2 diabetes

functional studies