false
zh-CN,zh-TW,en,fr,de,hi,ja,ko,pt,es
Catalog
The Many Faces and Phases of Co-Regulators
The Many Faces and Phases of Co-Regulators
The Many Faces and Phases of Co-Regulators
Back to course
[Please upgrade your browser to play this video content]
Video Transcription
Good morning, everyone. We're about to start the session. So my name is Daniel Joelle. I'm an associate professor in the Department of Microbiology, Immunology, and Cancer Biology at the University of Virginia. And I'm also a co-leader of our NCI Comprehensive Cancer Center for the Cancer Biology Program. Good morning. My name's Denise Okafor. I am an assistant professor at Penn State in the Department of Biochemistry and Molecular Biology. And welcome to today's session this morning, the many phases and phases of co-regulators, which as the name implies, it's all about co-regulators. We have four exciting talks today. Today's talks span a variety of themes, including thinking about how co-regulators interact with nuclear receptors, transcription factors, modifying enzymes. We'll hear about the composition of these transcriptional complexes in various states. We'll learn about the regulatory roles that co-regulators play in metabolism and inflammation and the roles they play in the proliferation of cancer cells in prostate cancer. Each talk will go for about 17 minutes. Then we'll have time for questions, hopefully. And this is a hybrid session today, so we will be fielding questions here, but also online. And so when anyone has a question, please, it's very important that you use the microphone. There is a QR code that will pop up again. And you feel free to scan that to access questions and polls and to leave any feedback. And so without further ado, I'll introduce our first speaker, Dr. Murray Campbell. He's a professor of pharmaceutics and pharmacology at The Ohio State University College of Pharmacy. His lab combines computational genomics and epigenomics approaches to understand transcriptional regulation in prostate and breast cancer, including how co-regulator functions are corrupted in these states. And today, he will tell us about their roles in prostate cancer progression. Thank you. Okay. Nice, 17 minutes has started. Okay, I'm very, very grateful for the invitation. I'm glad to be here. And it's gonna give a disclaimer, isn't it? And actually, I've got two disclaimers. I don't think I know what a co-regulator is. And you'll see why I say that, although I think I've been studying them for a while. I have no financial disclosures. I think this is how we got here. So this is, oh, is that not working? Okay, top left is David, I can't read it from here, Martin Rosenberg, kind of characterizing one of the first co-regulators in a lambda phage. And then 1990, we get through to co-activators in yeast. And then 1995 was a big year. Folks started applying yeast to hybrid approaches. And so David Moore identified it, the trip factors. Bert O'Malley's group found the SRCs. And Ron Evans identified it, which one's that? ENCOR1, I think, or ENCOR2. So we had an explosion of kind of understanding then. And I think that zoomed forward a long time to sort of kind of where we think we are contemporary. All right, so this laser's definitely not working. So on the left, I guess that's what we think we have. When transcription comes together, there's an active crosstalk between a distal enhancer and the basal transcriptional machinery. There's chromatin looping. There are multiple processes occurring there. And then in the silenced or repressed state, many of those things are undone. I want to talk a little bit about numbers. I like numbers. So if we think about just a canonical gene, it's probably got, on average, a half a dozen enhancers. And on average, they're about 160 kb away from the gene. And obviously, many of these regulatory regions are overlapping or shared. And when we measure them, we get a number of different types of numbers. We get the extent of their overlap, the sort of significance of the enrichment by ChIP-seq, and then obviously the distance from the gene. And we can think about integrating those with a range of sort of systromic approaches, ChIP-seq, ATAC-seq, Methyl-seq, and ask how chromatin states shape the systrome and then how that impacts the transcriptome. Then obviously, the transcriptome's highly nuanced. And we can think about these, well, so we should also point out, we can think about how genetic variation impacts that, so structural variance within regulatory regions are gonna impact that. And then that orange triangle is supposed to be a TAD, so we can think about these things in some sort of spatial organization. And then ask questions, further questions about the 3D and the 4D genome and use chromatin capture confirmation to understand that. And then we can ask, well, how do these events change across disease states or in different populations? And obviously, in a big red circle, I put the word data integration. And then the numbers are pretty aggressive. 2,000 transcription factors and co-regulators, perhaps 100 different histone modifications, multiple DNA methylation states and multiple nucleosome changes, multiple different types of transcripts, perhaps 50 million human SNPs, multiple approaches to understand those chromatin capture. And then you can imagine all the different nuances that come in experimental design. So I think it's very challenging to think, how are we really deciding how a protein-protein interaction in cis is impacting a transcriptome? Hence why I say I don't know what a co-regulator is. So I guess the questions we have are, probably their actions are heterogeneous and not exclusive, so it's challenging to map this co-regulator to transcription factor to gene output in a high-fidelity manner. And I think that's gonna need wet and dry lab approaches. Co-regulator definitions, I think, are opaque. We can think about the things that were discovered in 1995 were a direct protein-protein interaction, but what about something that's indirect but still brings about the regulation? And I'm gonna show you three vignettes. So the first one is NCoR2 in prostate cancer. We've been interested in this for quite a while. So NCoR2's altered in prostate cancer, and one of the arguments is that it changes, if it essentially take a break off the androgen receptor and lead to proliferation, and the stand-up to cancer cohort and other folks have shown that very elegantly. I guess we had a different argument that changes in NCoR2 perhaps inappropriately silence these broadly anti-cancer nuclear receptors. And then perhaps NCoR2 just changes other things. I'll show you, this is published. I'll just quickly show you three pieces of data. So we did NCoR2 ChIP-seq in prostate cancer models, so in Lincap and C42, which is an androgen deprivation therapy recurrent variant of Lincap, and then treated them with ethanol DHT. These word clouds are what NCoR2 overlaps with using the Giggle database. And Giggle is a library of about 11,000 ChIP-seq datasets. Yes, AR comes up in there, C42 DHT, that bottom right circle, but you can see there are lots of different transcription factors NCoR2's interacting with. And then we did RNA-seq and, oh, yeah, now it would be really nice to have a laser pointer, but, oh, no, it is working. I just can't point. Okay, so vehicle, and then vitamin, vitamin D, not vitamin D, DHT. So, you know, we're changing genes with DHT, and then we've done a couple of, we've made two stable variants of knockdown of NCoR2. C42 is not responding, really. This is the DHT response. This is the effect of just NCoR2 alone. So it profoundly changes how the system works. But then when we knock NCoR2 down and do it in the CWR22 mouse, loss of NCoR, or reduced NCoR2 makes the disease less responsive to androgen deprivation therapy. So it supports a role for androgen signaling, but these word clouds, oh, this is, suggest that there are multiple things being impacted. And, indeed, the genes that are knocked down by NCoR2 are associated, in this case, with neuroendocrine prostate cancer. So they're associated with very aggressive prostate cancer. So the summary of that, perhaps, is that NCoR2 overlaps with multiple ChIP-seq datasets. And, yes, there's a footprint of the androgen receptor in there, but it's very cell background specific, and it does map to a range of enhancers, including super enhancers. The background of NCoR, the cell background really determines how NCoR2 works. And then this, perhaps, is one of the key take-home points, is that NCoR2 targets are both up and down regulated following knockdown. So genes that we know have an NCoR2 peak, when we knock it down, both go up and down. Not the same gene, separate genes. So it's not a canonical exclusively co-repressor function. NCoR2 levels and, perhaps, mutation impact androgen deprivation and associate with neuroendocrine cancer, but, perhaps, the mechanism is more nuanced. And so the take-home message from this first vignette is that this is a canonical co-repressor, so a canonical co-regulator, and yet there's still so many questions about it. The cell context is really important, and there's diverse roles on transcription. So it's, perhaps, more a mediator than a co-repressor, and it clearly impacts therapy. The second vignette I'm gonna share is about prostate cancer health disparities. So men of African genomic ancestry experience a more aggressive prostate cancer than their European-American counterparts. And now we're gonna talk about vitamin D. And there's a long literature on vitamin D, perhaps, being anti-cancer in prostate cancer. But, actually, all of that preclinical work and clinical work was ultimately equivocal. There are no vitamin D-centered therapies for prostate cancer. But, perhaps, it's actually time to revisit it in the context of prostate cancer health disparities, and I'm not the only person saying this. There are a number of studies supporting that. And so, I guess, our question was, well, how does genomic ancestry impact VDR genomic functions? So this is an ice western of VDR, plus or minus the vitamin D receptor, western plus or minus vitamin D in a range of prostate models. And the ones, RC43N, 43T, 77N, 77T, are African-American prostate samples. The N, or cell lines, the N is a non-malignant model, the T is an isogenic tumor model. And the volcano-type plot on the bottom there is RIME data, so this is proteins that are interacting with the VDR in one of the prostate cancer cell lines. And you can see, well, the first thing you can see is there's about 100 proteins directly interacting with the VDR in one of these cell models. And I've labeled them, I have a canonical list of co-activators, co-regulators, or mixed function co-regulators and transcription factors, and so I've just labeled the proteins that are there. And you can see there are lots of different co-regulators in there. There's NCoA5 in there. There are no SRCs, for example. NCoR2 didn't come down in this complex. So some of these things we perhaps thought would be there aren't there. This is a busy volcano plot, but it's showing the deltas. We're all interested in deltas, and by that I mean how it varies from one to the next. And so comparing RC43N to HPR, so just the non-malignant models, you can see there's lots of things changing. And then on the RC43T versus Lincap, there's also lots of things changing. So this complex is highly dynamic and evolving. So I took those co-regulator lists, and mine TCGA say, well, what co-regulators are exclusively associated, in this case, with TMPRSS2 fusion, African-American prostate cancer. And a couple of ones came out, including BAS1A and SMARCA5, which are part of this WCRF complex, which is perhaps more widely known as SWI-SNF. And so there's an ice western, or BAS1A. Yes, indeed, even in our cell models, there's loss of BAS1A, compared to perhaps the European-American models. And so we've transfected BAS1A back into all the models, and then done RNA-seq. And what these volcano plots show are, yes, there's lots of genes changing in the presence of BAS1A and vitamin D. And the ones that are in red are genes that we know have an ATAC-seq responsive region when we add, wow, add vitamin D. And the circles around them are VDR ChIP-seq genes. And so the question is, are there more ATAC-seq genes in the African-American models, say? And the answer is yes. So BAS1A significantly enhances regulation of genes, more so, or actually only so, in the African-American models. And it changes the GSCA enrichment patterns. So BAS1A is impacting VDR signaling, in this case, and enhancing signaling. It's qualitatively and quantitatively modulating the VDR system transcriptome. Take-home message from this one is that genomic ancestry is really impactful to determine what co-regulators are associated. And proteins that don't directly interact with the VDR still are profoundly regulating how the VDR transcription impacts the genome. Then the final vignette is on RAR-Gamma and a role of potentially bookmarking AR. So again, this is more RIME data for the RAR-Gamma complex. Again, showing you there's lots of co-regulators coming down with them. The blue circle around it, in this case, is genes that are also regulated by the same microRNA. In this case, it's Mer-96. So this complex is being highly targeted by a specific microRNA. We've done ChIP-seq for a whole load of things. Oh, this is where I really need it. What you want to look at are the AR-ChIP-seq peaks. Ah, I can't do it. All right, so AR-MoC with vehicle 768 peaks. So that's just the basal AR in these models. We add DHT in, it goes up to 1368. You add RAR-Gamma expression into the system, we restore it. It goes up to 5,000, 9,000. And then we add in another co-regulator of RAR-Gamma, this guy called TAC-1, and it's going up to 82,000. So it's really, really impacting the amount of AR binding in the system. And this figure just shows it really impacts the significance of the ChIP-seq peaks. So this is another nuclear receptor complex that's driving where the AR can bind and the number of binding sites and the significance of the binding sites. Okay, I've got two complicated slides and then I'm gonna finish. These are the ChIP-seq data in either MoC cells or RAR-Gamma cells, but it's the ChIP-seq for AR, so it's those AR binding sites. And then we've got parallel RNA-seq, either to cells treated with enzalutamide for the AR, so AR-ChIP-seq peaks, but for cells treated with enzalutamide or K27-AR peaks treated with enzalutamide. And the bottom right graph is just showing you all the different chromatin contexts where those peaks are that are associated with genes regulated by enzalutamide. So RAR-Gamma in the system allows the AR to bind at sites that are highly responsive, that are enhancer sites, because it's K27, that are highly responsive to enzalutamide. That's what that says. This says that those genes are also enriched for luminal differentiation, and the pink is luminal target genes, and the GSEA supports that, and then the graphs on the right are showing that there's significant enrichment of luminal target genes when we have these AR-regulated luminal target genes, when we have RAR-Gamma and Taq1 in the system. And then final, final one is, how do we think this loops back to a cancer cohort? I took the standup to cancer cohort, which is about 400-odd aggressive prostate cancer samples, and separated them by upper 20% RAR-Gamma versus lower 20% RAR-Gamma, did differential expression, and then of those genes, correlated those with AR-K27 target genes from our ChIP-seq, and then did partial correlation analysis. So partial correlation analysis is asking the question, if we correlate A to B, what's the impact of C? And the A to B I was taking was AR to these AR-K27 target genes, but the C I took were some of these co-regulators I found in the original RIME data. And the answer is, yeah, look, some of these co-regulators that are in the RAR-Gamma complex are profoundly impacting the responsiveness of target genes, or strengthening the correlation between AR and AR-K27, sorry, AR-K27 target genes in this standup to cancer cohort. Again, underlying the importance, perhaps, of the RAR-Gamma TAC1 complex acting as, as I say, I've called it a bookmarking function. It's not my phrase, I've got that from elsewhere. I'm happy to follow up on that in questions. But that that is regulating, profoundly regulating how the AR responds in this system. Okay, I think I've just said that for the matter of time. And then, overall summary. So I think, to me, definitions of co-regulators are evolving. I think we have to think about direct and indirect mechanisms. The VDR-Bas1A story, that's, you know, if we think this is a nucleosome positioning story, why is that being coordinated with the VDR? And obviously it's not just the VDR, the other nuclear receptors, but how do we coordinate transcriptomic with nucleosome positioning and the specificity of that? I think, perhaps, we need to think about all of these processes, from upstream transcription for factor binding, nucleosome spacing, association with the basal transcriptional machinery. Even specificity of splicing and translation, you could imagine, could occur in a transcript-specific manner. And perhaps that could even be pulled in to this definition of co-regulator. I think in the context of prostate cancer, disease is, disease state is gonna be impactful, or is impactful, but clearly genomic ancestry as well. I think it's a lot of bioinformatics. I'm a strong advocate for bioinformatic approaches. Right, and then the acknowledgements. Folks that did the work, Sajjad and Shahid and Manjunath and Mark Long, lots of other lovely people that I'm fortunate enough to work with, and some people that have been kind enough to fund what we do, and I'll stop there and happily take any questions. Thank you. So we have time for questions. Yes, we do. You can come to the microphone, or online. Sorry. My name's Sajjad McGill. In the context of the RIR gamma type one, is RIR gamma a receptor or co-activators? What's the impact of retinoic acid on this? Yeah, so we used a synthetic retinoid in the system to test that question. The impact is pretty subtle. So it seems to be the basal RIR gamma in a non-ligand-dependent manner that's doing this. And I think that's what we need to think about with these type two nuclear receptors. They're in the nucleus, regardless of ligand, so they're doing stuff. And it binds to RIRE in this region. Yeah, and I didn't show you the ATAC-Seq data, but it's actually, we've got a strong binding at mononucleosomes, so not just nucleosome-free regions, but mononucleosome regions, which perhaps supports this bookmarking idea. So it's able to bind to not actually accessible chromatin, and maybe then facilitate where the AR binds. Maybe one thing that you didn't touch upon is maybe the whole post-translational modification of those co-regulators that probably modified all the ACAs, activators, and ligands, and you didn't touch about that, so. No, I haven't gone there. Post-translational modifications scare me, apart from on histones. Hi. Hi. Morag Young from Melbourne. Thank you, that was a really interesting talk. Thank you. I just wanted to ask a bit about the dynamics of the system, because often when we do these co-IPs and chip, we treat for a certain amount of time, maybe for an hour or half an hour, and then cross-link, and then do these things. So what are your thoughts about the time course and the dynamics of these systems, and are we missing things, because they're kind of... Yeah, that's a great question. I'll tell you two things that will just shake our head in disbelief. So I'm a kind of four, eight hour person, so of course I was loosey-goosey with all the details. You know, the Rhymer at four hours, no, eight hours, the chips at four hours, you know, these sorts of time frames, right? Gordon Hagar, though, you know, God bless him, all that glucocorticoid work he's doing, he's showing 30 seconds, 30 minutes, this thing's really, really dynamic, and then in April there was a paper in Science, this guy had done this amazing, these guys, this group had done this amazing thing where they'd taken a single TAD with one gene in it, I don't know if anybody saw this paper, and did incredible labeling of this TAD so that they could monitor it in real time. It's a murine development gene in an ES cell. The TAD was productive for 20 minutes, twice in a 24 hour period, and only 7% of the cells had a productive TAD at any one time. I don't know how we study anything. If that's really the unit we're studying, you know, it just, 60, 70% of the time it wasn't there, so I'm not sure what time really means, unless you're talking about a productive TAD, so. Yeah, okay, thank you. Okay, there you go. All right, we're good? Yeah, thank you. Thank you so much. Thank you. So our next speaker is Dr. Jennifer Estal. She received her PhD from the University of Toronto and did her postdoctoral training at Harvard in the Spiegelman Lab. She is an associate professor at the Montreal Clinical Research Institute, the ICRM. Her research focuses on the molecular changes involved in the pathogenesis of diabetes and metabolic illnesses. Her work represents the leading edge of molecular mechanisms and experimental models to study NASH and NFLAD. Welcome. Thank you so much. First, I'd like to thank the organizers, of course, for the invitation. I'm gonna try to hopefully get the laser pointer. No, okay. These are my disclosures. Here's the QR code we were asked to put up. Hope you scanned it. So I'm a cell biologist by training, but I really think of myself as a physiologist. So our lab really wants to understand how glucose is regulated in the body. So we think a lot about diabetes, and we think a lot about the organs that regulate glucose, and our lab focuses a lot on how the liver and the pancreas talks to each other. And today, I'm really gonna focus mostly on the liver because I feel it's a very interesting organ. It really responds very quickly. It adapts very fast. The environment it is dealing with is full of hormones, metabolites, lipids. It's seeing signals from the pancreas. It's seeing signals from the periphery, from the adipose tissue, as well as metabolic, as well as inflammatory signals from the gut. So it really has to adapt quickly, and when it doesn't adapt quickly, you get inappropriate glucose production, you get decreased insulin responsiveness, and you get diabetes. And in 70 to 80% of people who have diabetes also have NAFLD and NASH. So I'm gonna talk about my favorite co-activator or co-regulator, which is PGC1-alpha. It was originally, it's called this because it was originally cloned as a co-activator of PPAR-gamma. However, since then, there's really been an abundance of work to show that it co-activates, those lights are crazy, that co-activates a multitude of different transcription factors to modulate different programs. I wanna emphasize that it's a very inducible protein and very short-lived, which is something that people don't really know very much about it. It's induced by metabolic stress, such as exercise, cold, fasting. And this is kind of the transcriptional program that it's really well-known for. It's known to be a very important regulator of mitochondrial biogenesis, oxidative phosphorylation, really the entire program of mitochondrial biology. So you can up-regulate fatty acid oxidation, the TCA cycle, ROS detoxification. It really makes the mitochondria just more efficient when it's needed. So again, it's really an inducible signal. When we think about PGC1 alpha-in-the-liver, there's a lot of groups that have worked on this for many times, but often the first thing that comes to mind with PGC1 alpha-in-the-liver is gluconeogenesis. So it is a regulator of gluconeogenesis, and we've dug a little bit deeper into that recently. And it's through co-activation of FOXO1 and some other transcription factors. But since then, really a lot of work has been shown to show that PGC1 alpha-in-the-liver co-activates a number of different transcription factors you can see here to up-regulate the respiratory capacity of the liver. And this would be important, you can imagine, when the liver has metabolic challenges, when it's fatty, when there's lots of glucose around, you need to metabolize the substrates. PGC1 alpha-in-the-liver can also regulate triglyceride secretion. And what's interesting is that we know that PGC1 alpha-in-the-liver is really downstream of glucagon. So it's really a mediator of glucagon signaling in the liver during the fasted response. And interestingly, insulin can feedback negatively to control PGC1 alpha very quickly. So insulin shuts off PGC1 alpha very fast. So it's really a nice circuit to regulate the fasting response. But then PGC1 alpha kind of got a bad rap in the liver for a long time, in my opinion, because many people thought, well, overactive PGC1 alpha activity in the liver would lead to inappropriate gluconeogenesis. And that would exacerbate metabolic disease. But over and over again, others have shown that in metabolic disease, be it diabetes, NAFLD, cardiovascular disease, PGC1 alpha levels are low in the liver. And so we really kind of took this question and investigated this in the lab to see what these low PGC1 alpha levels in the liver were contributing to, if there were cause or consequence. And I just want to point out that it's not just low levels of PGC1 alpha in the liver or in other organs associated with metabolic disease, but there are also SNPs associated with metabolic disease in humans in PGC1 alpha. So I'm just showing one here that we've studied a bit in the lab where you have a glycine to serine amino acid change, a single amino acid change. And we showed in the lab that this leads to a decreased half-life of the protein. So it's normally very short-lived protein. This polymorphism is even more short-lived. And that's the one that's associated with the metabolic disease. So again, consistent with the literature, that lower PGC1 alpha in the liver is not necessarily a good thing, but why? So back to the original question of what PGC1 alpha is doing in the liver in terms of gluconeogenesis, we thought initially when we started to study this that maybe PGC1 alpha could inhibit insulin signaling, right? There was some link with insulin signaling we didn't know, but every time we overexpressed PGC1 alpha in liver cells in the liver, we would see enhancement of insulin signaling. And this was very confusing to us, which also meant in our head, well, if you're getting enhancement of insulin signaling by PGC1 alpha, would you get enhancement of signaling by glucagon, which again is paradoxical if you think about it, why would glucagon enhance insulin signaling? But that is actually what we would see many times. So we would take primary hepatocytes, we would treat them with insulin, you see phosphorylation of AKT goes up, and if you pretreat with glucagon, you have more phosphorylated AKT in response to insulin. Luckily, at the exact same time this was happening, we didn't have to argue it to the field. The group of Kirk Habegger also showed this in a very nice study in diabetes that was published in 2018, where he showed also in vitro and in vivo that if you pretreat or co-treat with glucagon with insulin, glucagon enhances the insulin signaling in the liver and improves glucose tolerance. And now, as many of you probably know, we have co-GLP1 glucagon agonists that are known to improve insulin signaling. So we thought it was kind of cool, but we wanted to know then the molecular mechanisms of how this was happening. So we dug a bit deeper and we wanted to see how glucagon was linked to insulin signaling. We can see that when we over, let's see if we can get this. Nope, okay. In the western blot and mRNA, when we treat with glucagon, we looked at the whole insulin signaling pathway, but what was very clear to us is that when you treat with glucagon, you increase the expression of both IRS1 and IRS2. Thanks, Vincent. Ooh, okay. So we, too bad you didn't give it before. You see protein and you see RNA. So glucagon can induce both, and the insulin receptor substrates bind to the insulin receptor that's activated and mediate the signaling downstream. So they're really the first step in the insulin signaling pathway. And many of the times you think of them kind of together, like IRS1 and 2, they're often written that way. But they actually, there's a lot of work that's been done to show that it's a bit nuanced. IRS1 can regulate lipogenesis more proficiently while IRS2 seems to inhibit gluconeogenesis more efficiently. What we found interesting is that when we knock down PGC1-alpha, it only affects the induction of IRS2 by glucagon. And so when, and this was consistent when we overexpressed PGC1-alpha, you can see that it also induces IRS2 and actually represses IRS1. And in a mouse that does not have PGC1-alpha in the liver, you have the opposite. You have overexpression of IRS1 and knockdown of IRS2. So you really have a shift in the balance. When glucagon and PGC1-alpha are active, you have a shift toward IRS2, which would suggest that insulin would work better to inhibit gluconeogenesis. So if you think about it, glucagon, which is promoting gluconeogenesis, I'm not saying it does not, but that's when insulin levels are low. So, but then what it's also doing at the same time is priming the liver for that initial insulin postprandial response that comes in so that the first thing the liver does is that it inhibits gluconeogenesis when it hits the receptors. So we tested this in terms of PGC1-alpha in primary hepatocytes. So this is inhibition of gluconeogenesis by insulin. So normally insulin can inhibit gluconeogenesis by about 40%. If you increase PGC1-alpha in the cells, you see that insulin works better to inhibit gluconeogenesis and that this is dependent on IRS2. In cells that don't have PGC1, the insulin is not able to inhibit gluconeogenesis. You can rescue that by increasing PGC1-alpha and that was again dependent on the expression of IRS2. So that was in vitro, we went in vivo. So you can imagine that when we overexpressed PGC1-alpha in the liver using viruses, we thought that we would have inappropriate gluconeogenesis in the fasting state and we did not. But what we did see is that when we did a pyruvate tolerance test where pyruvate in fasting mice will be converted into glucose through gluconeogenesis, you see glucose goes up. At this point, because glucose is going up, insulin is released. And during the time when insulin is high in the mice, you can see that mice that have high levels of PGC1-alpha in their livers are able to shut off the gluconeogenic response more efficiently. So in summary, we think that our study was able to add a little bit of a nuance to the whole glucagon insulin story and that glucagon and insulin are not always counter-regulatory opposing each other and that glucagon and PGC1-alpha help fine tune the hepatic glucose control during the fasted to fed transition. And then we can think about maybe relating that back to metabolic disease where we do see low levels of PGC1-alpha in the livers of humans and that these low levels might still be contributing to inappropriate glucose control that we see in these patients. So in the second vignette, I wanna add some more recent work that we've been doing and it's because we can't think of PGC1-alpha as one protein anymore. Fortunately or not, many groups including our own have discovered many other versions of PGC1-alpha. It's not a single co-activator anymore. It's about a family of 11, 13 proteins that are expressed in a tissue-specific manner. There are at least three separate promoters in the liver. There's a separate promoter for the brain. So it's really, there's a lot of proteins going on. We know a lot about canonical PGC1-alpha. We don't know a lot about these other isoforms. So we were lucky to be collaborating with Peter Matrakos at McGill University. He's a liver surgeon doing transplantation and so he gets samples of liver from people on a regular basis. And so we were able to get samples of liver with people with NAFLD, NASH, or cirrhosis. And we started to study just what are the transcripts that we see in terms of the different isoforms of PGC in this system. We saw what many people were seeing in the past, lower levels of canonical PGC1-alpha as the disease progressed. So these are normal controls and then you can see it goes lower. It's not like super striking. But what really was striking was that when you look at transcripts from the alternative promoter for PGC1-alpha, this is basal right there. So you can see that in people that have NAFLD, NASH, and cirrhosis, you get inappropriate or enhanced transcription from the alternative promoter. And so this is regular PCR that you're seeing over here. This is not quantitative really per se. We just wanted to see which isoforms we could see in the liver of humans. We can detect at least four. And you can see that in the NAFLD and NASH state, you start to see higher levels of this truncated version of PGC, this one right here. So at the time, we really didn't know what this little version of PGC1-alpha does. There had been a few studies before linking it to some aspects of mitochondrial function as well as exercise, physiology, and muscle. We wanted to know what it was doing in liver, so the first thing we did was did transcriptomics. And we were very surprised. If you look at this green circle here, this is untreated primary hepatocytes expressing either the canonical alpha-1 or the little one, alpha-4. Alpha-4 did nothing, absolutely nothing. It's the first time I've seen nothing change in transcriptomics. So we were, of course, a little bit disappointed, but we redid the experiment, and we did it this time with untreated and TNF-alpha treated and now we had an explosion of transcription. So it really, you needed some kind of biological signal, and in this case, it was an inflammatory signal. And you can see that the canonical version has a very big transcriptome. There's some overlap, and then there's a unique signature for alpha-4. In summary, we did what we normally do and everybody does is look for gene ontology and gene set enrichment analysis. I know you can't read any of this, but unsurprisingly, the red and the yellow where they overlap and the alpha-1 signature was really mitochondrial. It was really mitochondrial metabolism, nutrient metabolism, but the gray area was interesting to us. So genes that were very specific to alpha-4 in the presence of TNF were really focused around apoptosis and innate immunity. So the link between PGC-1-alpha and inflammation has been done previously, most notably by Christoph Hanschen looking in muscle cells, but we know that alpha-1 can down-regulate NF-kappa-B signaling, and we saw it very clearly in our primary hepatocyte system that overexpression of alpha-1 can shut off NF-kappa-B signaling. You can see there in the pink compared to the gray. When we looked at alpha-4, we thought maybe it did the same, but it actually doesn't do anything to NF-kappa-B signaling. So it wasn't through a canonical inhibition event of inflammation, but if you look at apoptosis, and we did this a number of different ways, but I just wanted to show one blot here. If you look at apoptosis, you can see that alpha-1 in the presence of TNF-alpha increased apoptosis, where alpha-4 almost completely abolished apoptosis. So here is an example of where they're doing completely opposite things. We then went in vivo to see if we can see the same thing, and not just in a culture system. So we took mice overexpressing PGC-1-alpha in the liver and treated with LPS, which is a bacterial saccharide that will induce inflammation similar to TNF-alpha. You can see wild-type mice, you get apoptosis, many markers of apoptosis this time. And then if you overexpress alpha-4, we eliminated that response. And then we also did the loss-of-function model. And the loss-of-function model's a bit difficult with PGC-1-alpha. You can't just knock out alpha-4 because this is a splice variant. So what we did was we knocked out the alternative promoter. So it's a few different versions of PGC-1-alpha, including alpha-4. But when we knock out that alternative promoter, you can see that apoptosis is enhanced in response to LPS. So how does this all fit together? What we think is happening is that, as I mentioned in the beginning, liver is really integrating a lot of signals all at the same time. You have the metabolic signal to boost cellular metabolism, but you also have inflammatory signals that it's seeing all the time. So what we think is alpha-1 and alpha-4 are working together to allow the cells to boost cellular metabolism, but also we have the inhibition of the inflammatory and the apoptotic program to protect these liver cells in this environment that's constantly changing. So in summary, we think that PGC-1-alpha is a protein that promotes really metabolic flexibility. This is not my hypothesis, I'm pretty sure. Many people work on it, think this way. But thinking about it as a really inducible protein that's induced when it's needed, and it allows rapid adaptation to metabolic challenges. In the liver particularly, which is our expertise, we've been able to show that it really fine-tunes hepatic metabolism. So it's not a master regulator, I would say, but it really helps to fine-tune the signals to boost capacity when it's needed, but also to protect against inflammatory damage. And that when PGC-1-alpha is low, say diabetes, metabolic disease, NASH, NAFLD, we have inefficient energy metabolism. So this happens when you have aging, diet can do this, genetic factors, and all of these things can come together to worsen the outcomes of metabolic disease. So I'm gonna thank everybody in my lab, particularly the trainees who did all the work that I showed you today, past and present, all our collaborators, and all the funding agencies. Thanks. Thank you. I'm gonna see if this question thing. I'll start again. I was glad to see that your R-alpha was at the center. Yes, I made sure. On your slide. Thank you very much. I was frantically changing the ions when I saw you. No, because in fact, we see, not studying PGC-1, but ERR, we see a phenocopy of what you see with the PGC. So I think it's the main guy there, for sure. Probably, for the mitochondrial aspect, for sure. And also for the apoptosis and inflammation. Oh, really? Okay, I haven't seen that. I can show you that. Yes. The other thing is that the isoform, do you think the isoform, the alpha-4, act as kind of a dominant negative toward PGC full length? Because you seem to see, like you say, totally different responses. So we've looked. We've looked at kind of the stereotypical PGC-1-alpha targets, and we don't see that. So you think it's independent? I do, but I don't wanna say it too loudly. Because I think that in some cases, it might. We just haven't found a good explanation for what we're seeing. It's kind of like what Dr. Campbell said. It doesn't always co-activate. It doesn't always repress. And in some situations, it might. But I think in general, it's not. If we overexpress both, alpha-1 still is able to do what it needs to do. Thank you. Hi, Nick Webster from San Diego. Great talk. So I had a couple of quick questions. You know, your data's really nicely fits with this whole metabolic flexibility that you mentioned. But if you do it in a more sort of chronic paradigm to sort of mimic hyperglucanemia, hyperinsulinemia, does that change what PGC-1-alpha's doing? Are you talking about the long one or the short one? Just the long one, just the end. The long one. So I mean, we've done the typical high-fat diet or NASH promoting diets. Is that what you mean? Yeah. Like where we've lost PGC-1. And we have done that, yeah. And so what we see chronically is that glucosomal stasis is not affected very much. And we only do it in the liver, I have to say. But definitely there's an inability of the liver to deal with oxidative stress because we see a chronic kind of accumulation of oxidative damage to the tissues when you reduce PGC-1-alpha. And then the degradation of PGC-1-alpha, I mean, it's turned over very quickly. And you know, a lot of the regulation in the liver is actually post-transcriptional. It's actually protein degradation. A lot of that is regulated. So are there conditions that will change the half-life of PGC-1-alpha? I would love to know. And that's what we're currently working on. And so we're focused on that SNP that I showed like very, very quickly, the data. Because that one's only regulated at the post-translational level. And so because it's a G to S mutation, we know, we now know that it's phosphorylated there. We don't know what phosphorylates it. So we, but we're trying to figure out what it does and in what situation that would occur. But it would be very interesting to know. Thanks. Do we still have time? Absolutely. First training question. Oh, jeez. Hello. Thanks. I just saw Ralph White III from University of Minnesota. I was wondering in regards to your studies with glucagon and the PGC1A axis. Is there any modulation occurring with the other PGC members such as PGC1 beta as well? I know that plays a role a little bit in co-activator type work. It is. We see less modulation with beta. Beta seems to be much more stable protein. It doesn't, its levels don't go up and down the way that we see with alpha. But I don't wanna speculate too much because it's kind of a really understudied version. It's not as misregulated in metabolic disease. It does seem to favor a little bit more of the lipid biology side, like lipid triglyceride. So there are some differences, but there's a lot of overlap between the two proteins. Thank you very much. Thanks. All right, our next speaker is Dr. Michael Garabedian. He is a professor of microbiology and urology at the NYU Grossman School of Medicine. His lab works on understanding how nuclear receptors regulate cell physiology and pathophysiology in areas such as prostate cancer, cardiovascular disease, and neuroendocrine function. Today he'll talk about the androgen receptor and co-regulator MED-19 and how they cooperatively function in prostate cells. Thank you. Thank you. And how they cooperatively function in prostate cells. So thank you for giving me the chance to talk about our work on MED-19, which is a mediator subunit that seems to engender hormone-independent growth of prostate cancer cells. So I have no disclosures. I just wanted to highlight right off the bat to Hannah Weber, an MD-PhD student that did a bulk of this work, and Rachel Roff, a technician in the lab, who are really the team that figured all this out. Okay, so I know a lot of you know this already, but I just wanted to get everyone at the same level. So we know the androgen receptor is a transcription factor, a nuclear hormone receptor that binds to DNA to activate a transcriptional program. In a normal prostate, it interacts with co-regulators to induce proliferation and then differentiation, which is this normal function. That function gets co-opted in cancer, where the proliferative response now, and the differentiation goes away, and the proliferative response dominates, and that can be targeted by anti-androgens or androgen deprivation therapy in early-stage prostate cancer. However, as many of you know, that cells become resistant to those drugs or to androgen deprivation therapy, leading to what's called castration-resistant prostate cancers. They're insensitive to most antagonists or androgen deprivation, but a lot of work has been done that they still rely on the androgen receptor activity. And that's because of a lot of reasons mechanistically, including AR amplification, so you just make more of the target, but the drug doesn't work so well because there's more of a target. You can get AR mutations that allow the antagonists to now work as agonists, and you get splice variants that allow for constitutively active androgen receptor that lop off the ligand binding domain and lose the target of the drug. But also, there's alterations in the expression of AR co-regulators. These are proteins that help activate the androgen receptor in cooperation with AR, and that's what we wanted to focus on in our lab. So we set up a very, almost a decade ago now, a very simple screen where we just wanted to ask if we knocked out every gene in the genome and assayed androgen receptor transcriptional activity, what factors would come out? And again, this was an agnostic, unbiased screen. We didn't care if they were direct interacting proteins or not. We just asked what happens if you delete a particular protein, does it affect AR activity? And from that, we got a list of genes, lots of different flavors and types, and as you can see, this is a simple transcriptional activation assay with an integrated luciferase reporter, and on the left, it's control. We have 100% activity. If you knock down androgen receptor, you see in the red box, you lose AR activity, and if you delete the rest of them, including MED-19, you lose some activity, and you're probably thinking, Michael, that's not very impressive, right? Why'd you pick that one? I know you like mediator, but really. This is why we picked it, because when we did another experiment where we took an androgen-independent cell line, LINCAP-ABL cells, and knocked down all of these same factors and asked what was the effect on proliferation, we got a much different result. We got a very potent growth inhibition compared to AR itself. We decided to try to figure out why this was such a great inhibitor of prostate cancer cell proliferation. What is MED-19? It's a component of the mediator complex, one of the subunits, and as many of you know, the mediator acts as a bridge between the enhancer and the promoter with transcriptional factors. It helps with the recruitment of Pol II and pre-initiation complex formation. It's involved in initiation of transcription, elongation, splicing, and also chromosomal looping. There are various domains, the head, middle, and tail, the kinase module, and MED-19 is associated with the middle domain. Really beautiful crown EM structures of the mediator have now been solved in yeast and humans, in complex with pre-initiation complex factors and Pol II, and here's just a structure. MED-19 is in this loop region in the middle, and it was shown by some biochemical studies and also through the crown EM structures that it's actually physically interacting with the CTD tail of RNA polymerase II, which is important for getting transcription up and running. So we knocked it down, as I mentioned, in LENCAP-ABL cells, so this is simply showing you a little more detail if we knock it down over time, and compared to CTRL-S RNA, we see a profound growth inhibition. If we take an AR negative cell, in this case PC3 cells, and knock it down, we don't really see much of an effect, so it's not a general toxic poison. We're not destroying the complex. It really seems to be quite specific for AR and for AR-expressing cells. We also looked to see whether it was modulated or overexpressed in prostate cancer, so here's some UNCA-type plots from cBioPortal, and it shows in primary prostate cancer, about half of the samples that were in this cohort show higher levels of MED-19. That correlated with worse outcome, and in metastatic prostate cancer, maybe 15%, 14% show elevated levels, and then in a separate study from another group, they showed that MED-19 by protein immunohistochemistry was low in benign prostate tissues, but higher in primary prostate cancer and largely nuclear. So that led to this hypothesis that really Hannah came up with, and this was actually her rotation project, and I thought it would, of course, never work. So what if you took an androgen-dependent cell, so you take an early-stage prostate cancer cell, in this case, LIMCAP cells, which need androgens to grow. If you put them in media that don't have androgens, they just don't grow very well. So we asked, okay, if you overexpress MED-19, would that lead to androgen-independent growth, and if so, is it affecting AR activity? So she made two different cell lines, pools of lentivial transduce cells with the flag tag MED-19, or just an empty vector, and asked what happened, and of course, I wouldn't be standing here telling you if it didn't work, and so lo and behold, she showed that in the absence, this is androgen-depleted media, this is a proliferation assay looking over time, that the control cells don't really grow if you don't have androgens in the media, but the MED-19 overexpressing cells now grow quite robustly in this 2D proliferation assay. We did it in 3D as well, and if you look in colony formation, you see that very few colonies can form in the absence of androgen in the media, but when you overexpress MED-19, you get robust colony formation. We did this in a couple other cell lines that were early stage, RWPE1s, and an AKT-transformed mouse stem cell line, and we got the same kind of phenotype. There was really no effect on androgen-dependent proliferation, and so we asked, does this actually affect castration-resistant prostate growth in a mouse model in vivo, and so we did xenografts using a castrated, in this case, castrated new mice, subcutaneous implantation of either the MED-19 LINCAP cells or controls, and you can see there's robust tumor growth under castration conditions. So what's the mechanism? How is it conferring this androgen-independent growth? So there's a couple, several different potential ideas. One could be you just get a lot more AR. I told you before that one of the resistance mechanisms is just upregulation of AR, and again, maybe there's splice variants that happen. I told you these mediator complexes are not just involved in transcription initiation, but also in splicing, and again, maybe it just bypasses AR altogether. It doesn't need AR, it just now has AR, but it doesn't really care, and they're driving a totally separate proliferative program, or maybe it's influencing AR activity. To cut a long story short, we tested all of these. There was no change in protein abundance of AR. There was no detectables, constitutively active splice variants, although we didn't look at all of them. We just looked at B7. And the cells remained AR-dependent. They actually were sensitive to enzalutamide and sensitive to AR knockdown. So that led us to conclude that it must be somehow increasing AR activity, or affecting AR activity. And so what's it doing in that case to affect AR transcriptional activity? And so I'm gonna show you a model, just to give you a little bit of context of how we think what's going on. So we think under low antigens, and the conditions low, mid-19, which is what we would suspect as LINCAP cells, we see there's some AR that's already bound, perhaps bound to co-repressors, mediator, very low acetylated and very low transcription of these target genes that are involved in proliferation, and there's really very little growth in the absence of any added androgens. But we think what's going on with mid-19 overexpression is that it's limiting, that you get more mid-19 incorporated into the mediator complex. That interacts with, we think, HATs to effectively induce acetylation. We think that this helps recruit AR to target genes, and with some mysterious factor here that we don't really know, that I'll tell you that we found out about through Hannah's work, can assemble a complex, induce cell proliferation, induce genes that are involved in cell proliferation, and increase antigen-dependent growth. So this is the model, and I'm gonna show you some data to support this model. So to test this model, Hannah did a number of studies under both antigen deprivation and also plus-antigen treatment. I'm only gonna talk about the studies in antigen deprivation, because I think that's the most relevant. So she did what we all do these days, RNA-seq, between the two to see what's different, ChIP-seq of AR, and mid-19, in this case, flagged the topic mid-19, and then, of course, H3K27 to look for activation targets. And so this is just the RNA-seq data, and surprisingly, not that many genes changed, only about 150, half went up and half went down. If you did ChIA analysis, asked what are the transcription factors associated with the genes, it's largely antigen receptors, so I guess that's good. And then, when she looked at the systromes, so here's AR occupancy, so it did shift a little bit, there's a lot of overlap, but there's some new sites that get occupied by the antigen receptors, there's some that are lost, that are controlled and not in the overexpressed cells, but importantly, all the mid-19 that we could find was always associated with, virtually all, was always associated with the antigen receptor, so it's going to places where AR is. So we wanted to ask, what are the genes that are really helping to drive this antigen-independent growth? And so one of the genes that we found that was linked previously to prostate cancer proliferation and one of the targets of mid-19 is under antigen deprivation is this gene called MAOA, which is monamine oxidase, and it's induced by mid-19 under low antigens and it's been linked to prostate cancer. So again, what is MAOA? Well, it's an enzyme that catalyzes deamination of amines, usually thought of in the CNS, where it affects things like dopamine, and the catalysis produces hydrogen peroxide and that is a source of ROS. It's also a known AR target gene, so already known that AR can bind and activate this target. And really beautiful series of papers from the late Lien Chung's lab showed that it's association with prostate cancer bone metastasis, and so this ROS seems to activate this HIF-1 VEGF axis to induce proliferation and angiogenesis. High MAOA is associated with worse clinical outcomes, and depletion by sRNA or inhibition by small molecules decreases tumorigenesis and bone metastasis in prostate cancer mouse models, and that's actually led to a clinical trial where there's an inhibitor of MAOA that's been used to treat castration-resistant prostate cancers. So we tested whether MAOA was important in our system by depleting it and then asking what's the effect on cell proliferation? So just to show you a little of the data before I show you the cell proliferation data, so here's what we see using kind of ChIP-PCR approaches. If you recruit AR, AR's recruited to the MAOA promoter in control cells, and that goes up in the MED-19 overexpressing cells. That's associated with MED-19, and that's associated with an increase in H3K27. And then importantly, if you look at the bottom here, the effect on proliferation, if you silence MAOA with sRNA, you get a profound decrease in the androgen-independent cellular proliferation. And this is a control that we use called KIF-11, which is a motor protein that screws up spindle-pole formation, and so cells that you knock this out, they just don't proliferate at all. So there's a clear association with MAOA and MED-19 independent growth. So what's the, so I hinted before, there may be, what are some other factors that may be interacting with or associating with to help drive this effect? And so Hannah did a very, she's smart, and she asked, okay, what's unique in the MED-19 AR ChIP-seq data that overlaps with MED-19 specifically? So when she looked at this small quadrant, she found that there's a transcription factor called ELK1 that was highly represented in that set. And so she asks, and so what's ELK1? It's a member of the transcription factor family. It's known to be an AR co-regulator, cooperates with AR and chromatin of your interactions with the AR ligand-independent domain, which is exciting for us, because this is all done in the absence of any AR ligands or low AR ligands. Its depletion alters target gene expression and reduces antigen-independent growth, and ELK1 expression correlates with prostate cancer recurrence. So Hannah did the following, again, these are ChIP-qPCR, where she looked at ELK1 recruitment in these MED-19 LINCAP cells, and found that at the MAOA promoter, she got ELK1 recruitment, a low level of it, but that was enhanced by, or increased in the MED-19 overexpressing cells. And then, importantly, if you knock down ELK1 and look at MAOA expression, it also goes down. And then if you look at proliferation, if you knock down ELK1, it also inhibits proliferation. So from that, we think that, again, when we overexpress MED-1, we think it's limiting, it then gets incorporated into the complex. That interacts, or helps recruit AR, along with ELK1, to promoters of certain target genes, like MAOA, to induce their expression. And in combination, probably not just MAOA, but several other targets, we think that that helps to induce antigen-independent growth. So just to summarize what I told you, we identified MED-19 in an unbiased screen for factors that, when we reduce them, when we deplete them, reduce AR transcriptional activity. So again, a very broad definition of a co-activator. We showed that depletion reduced the proliferation of AR-expressing prostate cancer cells. Overexpression promoted growth under low antigens, and tumor growth in castrated mice. And it promotes an alternative transcriptional approach. The AR occupancy increases, H3K27, at certain targets, in low antigens, including MAOA. And that we think that this, at least in part, is mediated through ELK1. So with that, I just wanted to thank, again, the people that did all this work. It was really Hannah Weber, just graduated, just got her MD a few weeks ago. Rachel Ruoff, a technician in the lab, there's Hannah here. And Karen Embrooker actually did the initial sRNA screen to identify MED-19. And the rest of the lab. And thank you for your attention. Hi, Nima Sharifi, Cleveland. Very nice talk, I really enjoyed it. I have a question about the specificity, the effect of the MED knockout. So if I remember correctly, you show that basically there's no effect on PC3 cell proliferation, there's no effect on antigen-dependent growth. But your second data slide suggested that getting rid of MED had a greater effect than getting rid of AR, which to me suggests it's doing something else as well. So can you comment on that, please? Yeah, I think you're right. I think there's probably other factors that are important in that MED-19 regulate. So, and I think that's the reason why you're getting this enhanced proliferation. But it's, so there's a lot of data out there with respect to knockouts of MED-19 now in a variety of other cell types. And in prostate cancer cells in particular, if you go to the DEDMAP project, you can see which ones are quite specific to knock down a MED-19, and they're largely things that have antigen receptor in them. And there's a really nice paper a few years ago, I forgot the group, that looked at MED-19, looked at all the mediator subunits and asked what cells. So they knocked out all the mediator components in mouse B cells, T cells, and stem cells, and MED-19 had no effect on their proliferation. But it did have a profound effect on the transcriptome in those cells, it did already seek on those. So I think it's a combination of when you get rid of it, it affects antigen-dependent signaling as well as antigen other factors. And unfortunately, we haven't done RNA-seq, I think, on those cells that we've knocked down MED-19, we've just looked at the overexpressing side, so I can't tell you what's antigen-dependent and independent at this time. But you're absolutely right. But that's what caught our eye, and we initially thought that, ah, this is doing more than just hitting AR, it's hitting AR plus a bunch of other stuff, and that's really why you get this profound inhibition of proliferation. I should say, I don't, in Hannah's paper, she knocked down all the mediator subunits in that LinCAP-ABL cell line, and MED-19 was still the winner compared to all the other ones, but there were other ones. MED-1 also reduces it. Some of the ones that are kind of scaffolding proteins, of course, reduce it. But it seemed to be the most profound, and whether that's just some idiosyncrasy of the LinCAP-ABL model or something more profound, we just don't know. Sure. Hi, Colin. That was a lovely talk. Thank you. What do you think about, were you surprised about the number of factors you got in your initial screens? Was it more than you thought you'd get or fewer? And I was waiting for you to tell me the CREB story that was in there, because you had CREB and CREB binding. Yeah, yeah, yeah, yeah. What's your thought? There's 1,000 other stories in there, and you guys are just grinding through them, or? Yeah, no, it's a great, great. So we, of course, had to prioritize. We got a lot of hits, of course, not surprisingly. And then we stratified, well, actually, we prioritized them against GR. So we found that, so things that affected both AR and GR, we put into one bin. Things that only affected AR, we put into the other bin. And so these are the things that only affected AR. And yeah, we'd love to do all the other ones, and we're, you know, it's always a resource issue. So yeah, but yeah, thank you. Interesting talk. In regards to your work in terms of looking, in terms of like the genes that correlate to AR and MED, I know you said majority go to AR. Do any of the genes that are outside of AR sort together? Do you find anything there? Yeah, that's a really good question. You know, sadly, I don't recall whether those also sorted out in a different way. So we focused mainly on the ones that were correlating with MED-19 and AR that were kind of unique, as we thought that those might be the most relevant ones. I mean, you could also argue, well, you've lost stuff too, right? There are things that are in the normal cells that you don't see AR being recruited to anymore. And what are those? So that may also be kind of getting the other side of the repressive side of things, you know, may also be that case. So there's still a lot to do, but I think, you know, I think the mechanism that Hanna put together is a plausible one for how this is working. It's certainly not the only. Certainly don't think MAOA is the whole story. We think that there's a lot of, you know, probably a lot of other targets that affect it. And just to get to more complexity, which I didn't bring up here, it'd be just because of time and bandwidth, but there's also two splice variants of MED-19. So, and ones that lops off the C-terminus. And we've shown, and we've made, basically, you know, if you have a hammer, you do the same thing. So we've expressed the splice variant in the norm, you know, and show that they have slightly different effects on cells. So it's even more complicated than what we make it out to be. But I think it's a, like I said, it's a plausible mechanism for how this may be inducing androgen-independent growth. And I think it's an early event in kind of the progress of progression to, you know, cancer and resistance. Absolutely, thank you. Michael, does overexpression of MED-19 change what complex is made up of mediator? Yeah, that's a great question. So the question was, does MED-19 overexpression affect the complex, mediator complex? We would love to know that. I can tell you we've done proteomics of, you know, we pulled out FLAG and then, you know, looked to see what came with it, and all the subunits come with it. A lot of other stuff comes with it. You know, the dirty laundry, of course, is that, frankly, when we overexpress it, it's a lot. It's not just nuclear, it's cytoplasmic, too. There are some papers out there that say other mediator components can work through cytoplasmic signaling, and we haven't done, I think, the definitive experiment yet of forcing it only to the nucleus or forcing it only to the cytoplasm to see if that makes it engender the pro-proliferative responses, so complexes are hard to, I mean, again, we did a very crude, it's very crude what we did. To do it much more definitively, I think, would take, it's kind of like beyond our resources at this point, but we do know that there are other components in there. It's not just working outside of the complex, so. Thank you very much. Thanks for your talk. Thank you. So it's my privilege to introduce the last speaker of our session, Professor Henriette Uhlenholt, earned her PhD in molecular biology from the Embol and Heidelberg, and did her postdoctoral training with Ron Evans in La Jolla. Professor Uhlenholt is the Chair for Metabolic Programming at the TUM School of Life Sciences, and is Director of the Institute of Diabetes and Endocrinology. Professor Uhlenholt studies fundamental mechanism of GR action in inflammation, where she's made numerous seminal discoveries on how GR regulates transcription. Welcome. Thank you. It's great to be here. And yeah, it's especially nice to meet everybody in person, in a room, so hopefully we can keep doing this. And here's your QR code. So about the pandemic, many of you in the clinic have probably realized that the potent anti-inflammatory and also lung regenerative actions of glucocorticoids were literally lifesavers during the past two years. But they, of course, come at a cost, so many patients are not too happy when they hear they are gonna be on glucocorticoids. So we are very interested in understanding how that is happening at the molecular level. So just as a brief recap, you're all aware that glucocorticoids bind to the glucocorticoid receptor, which then binds to glucocorticoid response elements at promoters and enhancers. And it does that in a very cell-type and signal-specific manner, because, as you're all familiar with, the glucocorticoid receptor depends on the presence of pioneering factor lineage-determining transcription factors to open up certain enhancers or promoters, which are then glucocorticoid responsive. So in order for any given cell to have an accessible GR binding site, the glucocorticoid receptor will then bind these open sites that are pre-specified or pre-determined by these lineage-determining transcription factors. It'll interact with other transcription factors present at these cis-regulatory elements, and it'll recruit a bunch of co-regulators, histone modifiers, et cetera, to then either activate or repress transcription. And that is the topic of this morning's session. So we have also chipped GR in a number of cell-type series, so each circle represents a cistrome, and it also does something very interesting in liver metabolism, but I'm not gonna talk about that today. It does something in circadian rhythms, et cetera. But for the purpose of today's talk, we're gonna look at its anti-inflammatory actions in primary mouse macrophages, where it cross-talks with AP1 and NF-kB transcription factors on promoters and enhancers of many immunomodulatory gene programs. So all of the data that I'm gonna show you today is from these primary mouse macrophages, and what we're gonna be looking at is a cell that has been activated with the LPS stimulus that was just mentioned to you earlier, and these cells are then co-treated with dexamethasone, usually for a period of three hours. And then the glucocorticoid receptor will bind a given binding site, recruit a bunch of different co-regulators, cross-talk with these inflammatory transcription factors, and then it will repress, potently, the expression of many inflammatory cytokines, chemokines, interleukins, TNF, et cetera, et cetera. It'll also induce and activate the transcription of classical GR target genes, GILs, KLFs, DAFs, FKBP5, et cetera, et cetera. And honestly, I'm not gonna really tell you how that is specified, because we have really no idea. Anybody who has one is very welcome to speak up. But just bear in mind for the next couple of slides that I'm always gonna show you examples for these two types of categories of GR, positive and negative target genes. So there's a number of scenarios out there on how GR might repress transcription, which include negative GREs, competition, tethering, composite elements, classical GREs, non-genomic actions, what have you. Again, I don't wanna go into that too much, but the point I wanna make here is that there's all kinds of different scenarios, and that even in the same cell type, there's gonna be different sets of target genes, and this regulation is really locus and target gene specific, right? You already also just heard that there's a number of enhancers per given gene, et cetera, et cetera, so that the complexity is daunting, to say the least. So that's what we started looking at and mapping out. So we have started defining these different categories of subsets of GR target genes. So if this represents a systrum here, then it comes in multiple different flavors. So I'm gonna give you a few examples of that. And since this is the co-regulator session, let me start out with our ChIP-MS experiment. So this is like a RIME experiment, where we pulled down the glucocorticoid receptor in these activated macrophages, and then mapped the functional categories of all of these GR interacting proteins. And as you can see from this plot here, the most prominent two kinds of complexes that we purified together with the known ones, like P300, NF-Kappa-B, the SRCs, et cetera, et cetera, are chromatin remodelers and histone methyl transferases. And we confirmed this by endogenous co-IPs, et cetera, et cetera. So the two complexes that we pulled down together with GR in mouse macrophages were the SETD1A compost complex of H3K4 methyl transferases, and the related trithorax complex, the MLL-BRG1-Switch-SNF complex. So these are kind of the two main co-regulators, in addition to GRP1, SRCs, et cetera, et cetera, that we pulled down in this big bunch of proteins that all came down in these IPs. So since the compost story is published, I just want to very briefly tell you that we have chipped the core compost components, we have profiled the histone methylation dynamics, as well as the histone H3K27 acetylation dynamics, and we have then divided them bioinformatically into all these different subsets. And I can tell you there is some genes that recruit the compost, some that don't, some that go up, some that go down, some that have methylation changes, some that don't, et cetera, et cetera. So all kinds of different scenarios that you can come up with. But bottom line, if you knock down the core component of this complex, you affect target gene regulation by EGR. Again, some genes go up, some genes go down, some go up even more, some genes go down even less, et cetera, et cetera. So it's very complex, but this compost complex is actually required for a subset of GR target genes. Again, not all of them, right? So it's just a selected group. So then the other part of the complexes that we pulled down was the SWITCH-SNF1. So we chipped BRG1 together with GR in these cells, and as you would expect, basically all of the GR bound sites and macrophages are also bound by the core component BRG1, right? That is not very surprising, and you can see that here. This is what you would expect, right, for a SWITCH-SNF chromatin remodeler. But when you look at what is happening in the absence and presence of GR ligand, you can see that GR actually recruits BRG1 to selected target sites so that there is enhanced or increased recruitment of BRG1 to chromatin in response to GR ligand. And I'm gonna give you an example here. So here's the KLF, for example, locus that is activated by GR. So when GR binds, it brings in BRG1 in response to ligand, and it's not there in the absence of glucocorticoids, while at a classically repressed GR target site, like the CCL2 locus, you can see that BRG1 is already present, and its occupancy doesn't really change with GR ligand. That makes sense because this gene is, of course, already activated by LPS alone, for example, right? So we have also profiled by ATAC-Seq the chromatin accessibility in response to LPS or LPS index, and this is what you can see here. And if you look at the dynamics here, so the increase in ATAC-Seq signal, you can again see that there's increased accessibility at these GR target sites where GR recruits BRG1, which are then transcriptionally activated in response to ligand. And I'm gonna show you the global plots here that nicely track, for example, with BRG1 recruitment, increased ATAC-Seq signal, and increased in H3K27 acetylation. So the purple signal here is always when something goes up, that's plotting the delta. So again, this is something that makes a lot of sense. At a classical target site that is activated by glucocorticoids, you have GR binding, you have increased in ATAC-Seq signal and increased in H3K27 acetylation, and then the gene goes up. But whereas at target sites that's typically repressed by glucocorticoids like the CCL2, there is really no changes in these ATAC-Seq signatures, for example, even though the corresponding gene and the H3K27 acetylation go down. So if we then knock down BRG1 by its iRNA, interestingly, this is again plotting the delta here, you kind of reverse the picture. So the genes that are usually repressed by GR are now de-repressed, right? So in the absence of BRG1, there's impaired repression of these inflammatory cytokines and chemokines, and there's impaired transcription activation of these classical GR target genes. But again, these are not all the GR target genes, it's a selected subset, right? Which you can then of course functionally annotate and look at in more detail, but bottom line is that you have a bunch of GR target genes that are affected by loss of BRG1, which include a lot of inflammatory cytokines and chemokines, and which interestingly, are not affected in the response to LPS, right? So even in the absence of BRG1, they're still induced. Now, if we want to explain that mechanistically, by ChIPqPCR, for example, we looked at cells that have been treated pharmacologically with a BRG1 inhibitor, so a small molecule that targets the active site of the BRG1 ATPase enzyme, and not surprisingly, if you inhibit BRG1 activity, you have of course reduced induction, so that's recapitulating the siRNA experiment of these positive GR target genes, and this is again showing you the impaired repression or the de-repression of these negative GR target genes. And this tracks, especially for the activated GR target sites, with GR and mediator occupancy, right? So again, not surprisingly, you inhibit switch SNF, the BRG1 ATPase, you're gonna have less GR bound to chromatin, you're gonna have less mediator recruitment, and that is of course why you have reduced activation of these target genes, right? So far, so good, but what about the repression? So we performed a couple of more ChIP-seq and ChIP-qPCR experiments, and I told you previously that the repression nicely tracks with the H3K27 acetylation, so in the presence of this BRG1 inhibitor, there is increased H3 acetylation, right, either at the total level or also at the H3K27 level. So the histone acetylation, which normally goes down when the gene is repressed, is now maintained, so it's not reduced anymore, and this goes together with a tendency towards reduced HDAC recruitment, but also we think it's not entirely explained by the presence of the HDAC per se, but really also by the HDAC activity. So it seems that the BRG1 complex, besides its role in chromatin remodeling, et cetera, in nucleosome positioning, is doing something to function as kind of a platform protein to bring in HDACs to then lead to the repression of a subset of GR target genes. And if we treat these macrophages with a classical HDAC inhibitor, like the SAHA, for example, then we recapitulate this effect, right, so again, not surprisingly, if you inhibit the HDACs, you're not gonna be able to repress your subset of GR target genes here. So essentially, what we believe is happening is that a gene like KLF or FKBP5, we have GR binding, and it depends on the presence of BRG1 chromosome remodeling, et cetera, to bring in mediator and to activate this gene, which is, of course, not happening when you don't have BRG1. Whereas the site is typically repressed by glucocorticoids, you have GR binding, and there the BRG1 seems to bring in the HDACs to then reduce histone acetylation and to transcriptionally repress this target gene, which is not happening when you don't have BRG1. So in this case, BRG1 seems to affect HDAC recruitment and activity. That's all we know at the moment, and again, we don't know why it's this particular subset, for example, that is being affected, as opposed to one other. So again, it's a subset of GR target genes that depends on this mechanism, and it's not all of the repressed macrophage targets. So just to sum this up, I told you that there's many different flavors and subcategories of these GR target genes, even in the very same cell type. In the presence of all of these same co-regulator proteins, some loci depend on Switch-SNF, some loci depend on the compass complex, some are tethered, others are not. Some make eRNAs, others don't. Some change their histone methylation patterns, some don't, et cetera, et cetera. Some of these functions are indeed overlapping, and some are specific, so it's complicated. The one thing that always seems to track with the transcription output, and that's a correlation, right, is the H3K27 acetylation, but again, that is simply a readout for enhancer activity at this point, really, right? So again, it's really an example of locus-specific and subset-specific mechanisms that we're seeing here, so I think, especially for those of us doing a lot of genomics and doing these global approaches, we really have to then dissect the individual loci and really look at our favorite target genes in a more detailed, painful, and thorough way. So with that, I would really like to thank you for your attention, and I would especially like to thank our collaborators from Matthias Mann's lab here, Michael Wierer, for a lot of the proteomics and the ChIP-MS, and I would especially like to acknowledge Franziska Greulich, who did a lot of the ChIP experiments in macrophages, and that is her, and of course, again, I'm very much looking forward to all of your questions and discussions. I'd like to thank all the funders, and if you have noticed, the Helmholtz Institute in Munich is so fascinated by this problem that they even incorporated this activation repression into our new logo. So with that, I'm looking forward to talking to you in person. Thank you. Nikki Partridge, New York University. I just wanted to know if you looked at any of the other, the regulatable HDACs, HDAC four, five, et cetera. You said HDAC one and three there, but I wonder about any of the others. We have tried looking at all of them. These ones showed the highest expression levels and showed an effect when we knocked them down by siRNA and also had antibodies that worked in CHIP. So I'm not excluding them, but these were the ones where we have data that made sense. The antibodies for HDAC four worked pretty well for CHIP. Yeah, but this one then didn't really do anything with respect to the siRNA, with respect to GR in the macrophage. Right, thanks. Again, in our hands in this time point, et cetera, et cetera. Very nice talk. Carolyn Cummins, University of Toronto. My question is related to the strong effect of DAX on those genes that are repressed. It looked like there wasn't a DAX-dependent regulation of BRG1. Did I miss that? So if you treat cells with dexamethasone, this will eventually downregulate GR itself as well as its co-regulator, so that again, what makes sense, there seems to be some kind of feedback mechanism. With this particular data, I think the BRG1 itself was okay, but the other subunit, the BRM, for example, is strongly downregulated. Okay, thank you. Hi, Trevor Archer, NIHS. A very nice talk. I was curious if you could tell us a little bit more about your mass spec experiment. I noticed that you had a lot of the subunits in there, and I wondered, because of course, as you know, the heterogeneity of the complex is key to its work and function. So can you say something about what subunits you saw or didn't see? Yeah, so in these ChIP-MS proteomics, the problem is, of course, that absence of evidence is never evidence of absence, right? So if we don't see something, it doesn't really mean much. For the switch SNIF, we did not, we saw a bunch of the BAF subunits that are known to interact with nuclear receptor, the 57, et cetera. We did not see all of them. BOG1 itself, actually, is rather hard to detect in this one. For the COMPAS complex, we saw all of the subunits. Yeah, I'm just, you can see him sitting right, like, why, because I was thinking, I don't, so it has to be something else than GR, because GR is one receptor, so it has to be something that all of these genes have in common, have different factors. So were you able to look at the individual and answers and see what factor correlates with what? We've tried motif analyses, right, of these ChIP-seq. The problem is a little bit that the subsets, bioinformatically, at some point get too small to get really good statistics, right? So there is some candidate motifs, et cetera, et cetera. If you look at your favorite genes, let's say. Yeah, so we haven't been able, I mean, you see IRFs there, et cetera, et cetera, but we haven't really seen anything that we were able to pinpoint yet. We're now trying some neural networks, machine learning, et cetera, to see if we can improve the predictions. Yeah, really, really great stuff, and I was struggling to think as well. So I love the numbers, so 37,000 ChIP-seq peaks for SMARCA4, no, BRG1, yeah, SMARCA4. Okay, so GR almost exclusively overlapped with that. But obviously, by definition, it doesn't. So if you mine those 37,000, what are the top nuclear receptor motifs coming up in that? I mean, is it, you know, I'm all about, we're all studying MIC, but we don't want to admit it. Are we all studying these SMARC complexes, and it's those that are the master regulators, but we just project our biology through our nuclear receptors? That's true, so I have to admit, we haven't really looked at the ones that don't overlap with GR, right? So if anything, your ChIP in macrophages typically comes up with PU1, CBP, AP1, and NF-kappa-P. Yeah, right. But yeah, it will surely be interesting to see. Thank you very much. So we'd just like to thank the audience for a vibrant discussion, the speakers for really wonderful talks about the depth, breadth, and complexity of transcriptional co-regulators and their function in biology. Thank you, everyone. Thank you.
Video Summary
The first video features Dr. Daniel Joelle and Dr. Denise Okafor introducing a session on co-regulators. The session includes four presentations discussing the roles and interactions of co-regulators with nuclear receptors, transcription factors, and modifying enzymes. The topics include the composition of transcriptional complexes, the regulatory roles of co-regulators in metabolism and inflammation, and their involvement in prostate cancer cell proliferation. The session is a hybrid format with both in-person and online participation. The first speaker, Dr. Murray Campbell, discusses the role of NCoR2 in prostate cancer progression, highlighting its interaction with transcription factors and its impact on gene expression. The second speaker, Dr. Jennifer Estal, focuses on the molecular changes involving diabetes and metabolic illnesses, specifically discussing PGC-1 alpha and its role in regulating various processes. Both presentations emphasize the importance and complexity of co-regulators in biological processes.<br /><br />In the second video, Professor Henriette Uhlenholt discusses the complexity of glucocorticoid receptor (GR) regulation of transcription in primary mouse macrophages. The study investigates the co-regulator complexes involved in GR action, with chromatin remodelers and histone methyl transferases being identified as the main complexes. The SETD1A complex of H3K4 methyl transferases and the MLL-BRG1-Switch-SNF complex were found to interact with GR. The SETD1A complex regulates a subset of GR target genes by affecting their transcription activation, while the MLL-BRG1-Switch-SNF complex recruits BRG1 to GR target sites for transcription repression. The presence of BRG1 in the complex facilitates the recruitment and activity of histone deacetylases, leading to transcriptional repression. Knockdown of BRG1 impairs the repression of GR target genes. The study highlights that different subsets of GR target genes rely on distinct co-regulator complexes for regulation, underscoring the complexity of GR regulation and the importance of careful analysis and investigation of target gene subsets.
Keywords
co-regulators
nuclear receptors
transcription factors
modifying enzymes
transcriptional complexes
metabolism
inflammation
prostate cancer cell proliferation
NCoR2
gene expression
diabetes
metabolic illnesses
EndoCareers
|
Contact Us
|
Privacy Policy
|
Terms of Use
CONNECT WITH US
© 2021 Copyright Endocrine Society. All rights reserved.
2055 L Street NW, Suite 600 | Washington, DC 20036
202.971.3636 | 888.363.6274
×