 It is a great pleasure to introduce Huda Zagvi for the second time. And for those of you who weren't here yesterday, let me just do my introduction very briefly. Huda is a professor at the Baylor College of Medicine. She's an investigator with HHMI, a member of the National Academy, and founding director of the Neurological Research Institute at Texas Children's Hospital. She, her work really has, begins with trying to understand disease and it's a fundamental molecular basis. And she's done spectacularly well on multiple fronts. In honor of that, she won the 2017 Breakthrough Prize in Life Sciences. And I think it's particularly apt that she is here today to give Steenbach lectures since Steenbach also was remarkable in his ability to identify molecular bases of Ricketts in particular and can't have a Steenbach lecture without showing Steenbach, who got his degree here in 1908 and was a professor in the first half of the last century. So it's a while ago now. But he's basically responsible for the founding of Worf with his own personal funds and we owe him a tremendous debt of gratitude. And with that, I would like to turn it over to Huda who will give her second lecture and I'm really excited to hear it. Thank you so much, Judith. I have to say as the second day is starting to widen down, I'm starting to feel sad because I actually had such a wonderful time here, meeting with faculty and getting to know... Is this better? Can you hear me better now? I'll turn it up. Okay, great. I hope, can you hear me now? Wonderful. So what I'm trying to say is I'm really sad that my trip is coming to an end because I thoroughly enjoyed it. I really loved every visit I had with the different faculty and just getting to learn about medicine and the great science that's happening here. So I feel very honored to be the Steambach lecturer and to have spent time with you and I appreciate all of you for making time for me. So before I tell you about the work, I would just like to give you a disclosure, particularly regarding my collaboration with UCB because as you'll hear about the work today, it's some of the work we've done. They have helped support and advance towards the development of therapeutics and other affiliations. There are really no financial monies that come from my own, but we heard the work yesterday. We collaborate with them and then I have a couple of advisory board responsibilities. So the talk today will tell you about my journey with the rare disease, but I hope that it will sort of show you how when we study some rare diseases, we might have a better handle on them than the more common diseases. Of course, the most common neurodegenerative diseases are Alzheimer and Parkinson's disease, but for the majority of Alzheimer and Parkinson's, they're sporadic disorders. We don't really have a handle about the molecular basis. There are a few familial cases with maybe 5% of Alzheimer and Parkinson are genetically determined, but it's those rare diseases that are helping us understand the pathogenesis, and today I'm going to tell you about a really rare disorder that we study, and I hope by the end of the talk, you'll see that some of the lessons we learned from this disease might actually have implication for the broader and more common and less understood neurodegenerative diseases. So spinal cerebellar ataxia is a balanced disorder. It is dominantly inherited. It's extremely rare, much rarer than red syndrome, affects 1 in 100,000 people, and the people typically are healthy till they're about 30 or 40 years old when they start losing their balance, and they'll have balance problems as you see these two individuals in the picture using a walker to get around for about 10 to 15 years before they become wheelchair bound, and once they become wheelchair bound, additional parts of their nervous system function deteriorate, and that's the ability to swallow and speak easily, and the swallowing difficulty lead to choking and the current infection and unfortunately premature death. This is the project that I worked on because there was no way in 1985 to really focus your career on a sporadic disorder such as red syndrome. I mentioned to you yesterday my mentor said, find another project. Find a Mendelian project, and this is the Mendelian project. As you would look at this pedigree, you will see that there are multiple individuals, the filled circles and squares are affected people, and you'll see this is a family that lives in Houston about 50 miles north of Houston. They have this disease. They were about a couple hundred members that I was able to identify, examine, and or obtain history about, and what you'll see clearly that this family, as the disease progresses, the onset also happens to be earlier and earlier. So these individuals had disease onset in their 70s, the next generation their 50s. Most of the people I examined are from this generation down with onset in their 40s or 30s, and those in the last generation, they're beginning to have onset as children, four-year-old, 10-year-old, 17-year-old. And this phenomenon of anticipation, it's called as such because it's almost like a clinical anticipation, that when you know your father has the disease, you start looking for symptoms and recognizing the symptoms early or the physician might anticipate it early. And that may have a little bit to do with it. You might notice if you're a little bit uncoordinated when the family history is positive, but I think what you need to know is that the death is much earlier. So the person who had disease in their 70s died when they were 85 or 90s, that's practically a normal lifespan, was those children, they were all dead by 20 years of age or younger. So it's the age of onset as well as the age of death. And, you know, my lab has mapped the gene to chromosome 6, and we were trying to walk closer and closer. We got it down to a million base pairs. And we collaborated with Harry Orr at the University of Minnesota, who also had a family, similarly had multiple affected individuals. And he and I were walking towards the gene and looking to find the gene. And one day I hear a seminar at Baylor College of Medicine by Tom Kasky, describing how they discovered the mutational basis of myotonic dystrophy. And as a neurologist, I knew exactly what happens in myotonic dystrophy. You have anticipation. The mother could be fine, but the baby will have disease at birth. And he found out that the repeats are smaller in the mother, but thousands expansion happens in the children. And immediately recognize this as possibly this is the phenomena. So while Harry and I were trying to find the gene in that million base pair, it immediately inspired us to look at CAG or CAG, any type of repeat expansion. So he and I decided to split the region in half and start looking in the region for a repeat expansion. The region was one megabase. We did not want to miss anything. So I said, why don't we make sure there's a 75 kilobase of overlap as we screen for these potential repeats in the DNA of the patients. And that we did and we discovered the mutation on the same day in both of our respective laboratory. And the mutation happened to be in the 75 kilobase region of overlap. So that was the culmination of a five years collaboration. And this year marked our 25th year collaboration. We started collaborating before the gene was discovered. And I was so thrilled that we discovered it together. So the CAG repeat that's expanded can be typically 30. It can be as little as six and as high as 39 in healthy individuals. But in patients with the attacks, yeah, it's usually 39 perfect repeats or more. And this is 39 repeats causes late onset disease where the 82 younger onset repeat. The repeat fell within the coding region of a protein that at the time when we did blast, there was nothing like it. So we had no homology to anything. First attacks your gene clone called it attacks in one for lack of a better name. And that's how our studies started. And this slide shows you now my CA1 family that I shared with you. And here the individuals that had their disease onset later 60s, 40s and so on. And you'll see they had a repeat size in the 40s. Where I hear the children and you'll see they had repeat size 70. And in this case 75. And here's one our youngest patient ever. She had 82 repeats and she had an onset at four years of age. So the longer the repeat, the younger the age of onset and the more severe the disease. At the time we cloned the CA1, there was only one other polyglutamine disease. The engine receptor and Huntington got cloned about the same time. And since then there are now nine polyglutamine diseases. The red triangle is where the polyglutamine tract is. And all you have to know is that there are proteins of various sizes, various functions. Quite different. Here's a calcium channel for a CA6 and a tata binding protein, transcriptional regulator, and so on and so forth. So they're very different proteins. And therefore people really focus on the polyglutamine tract. And many of them are called ATACSs because they all called ATACSs. So there are multiple ATACSs now caused by polyglutamine. But they're really different. They don't share anything amongst them, only the polyq expansion. So therefore there were many, many studies for a year that focus on the polyq expansion being the entity driving the toxicity. And today I'll share with you what we've learned and how fascinating it is. So today what I'd like to tell you is tell you about the molecular mechanism. How does a polyq expansion cause neuron degeneration? I also would like to share with you a new discovery we made about why is it that in this disease where the protein is widely expressed, it's in every cell of the brain, yet we see degeneration in the purkinje cells of the cerebellum as well as a few brain stem neurons and maybe some hippocampal cells. Why is it that we don't see it in other cell type? Similarly, for example, Huntington is widely expressed, but you don't see purkinje cell degeneration. So what drives the cell specific degeneration? I hope to share with you one story about that. And then now that we understand these two things, what can one do about modulating the course of the disease? And finally, to go back where we started, is something we're learning from SC1 going to be relevant to other neurodegenerative diseases. So focusing on the mechanism, when we found this gene, the first thing we wanted to know could the polyq expansion be causing loss of function of the protein? So the simplest thing to do is to delete the gene in the mouse, and the mouse has normally two Q. When we deleted the gene of the mouse, the mouse did not have any of the attacks of phenotypes that we see in the human patients. They had slight learning and memory deficit, but no SC1 phenotype. The next thing we did is we put a knock in and expanded repeat within the endogenous SC1 locus. And when we put 154 glutamines into the CAG repeat within that locus, replace the two Qs CAG, we're expressing it from one allele in the right spatial temporal expression. We now reproduced all the features of the human disease. So you might say, well, she's lucky. Every time she makes a mouse model, like with the rat and others, she gets a mouse that reproduces all the features of the disease. Not so quick. I'll have to tell you, our first mouse was 82 repeat, because that was the largest repeat we observed in a human patient, if you recall, our age of onset curve. And the mouse did not have any of the human features. And we were frustrated after two years of waiting for that mouse. And the reason is it did, if you homozygose the allele, and wait two years, you'll get some coordination problem. But which postdoc is going to work on a mouse that has a phenotype that begins at two years of age? Well, when we reflected on it, as I told you, the 82 repeat caused disease in a four-year-old individual, and the mice only live two-year-olds. So the time that the neuron exposed to this toxic protein is a factor. So all we did is we took the human family's data and extended the curve to see, when might we see a phenotype within a couple of months? And that's how we came with the 154 repeat. So I just wanted to share that. Not every mouse we made is immediately great. We have to think about it. Think about the biology, and we finally get it. So now that we had a good mouse model, we begin to investigate the mechanism. And one of the earlier studies on the mechanism was, attacks on what we knew nothing about. So one simple thing we asked, who does it interact with? And this is work we've done in cells initially, and using yeast to hybrid, and using both of those studies, one protein that we identified is called capicua homolog. Capicua was initially described in the fruit flies, and I'll tell you a little bit more about it. And you can see here, when we did this tandem affinity purification in cells, we purified attacks in one. Attacks on one was attacked protein. We identified two isoform of capicua, along and short. And the other protein that came was 1433 and HSP70. So we found that in fruit flies, capicua modulated the toxicity of attacks in one. And we found that it interacted with a domain within attacks in one that eventually we discovered actually has homology. By the time we began these studies, we now learned there's an attacks in one parallel that's now called attacks in one like. And the two proteins only share this, what's called attacks in one homology domain, because that's the only domain they share. And this is the protein, a domain that interacted with capicua, where there is a small region here within the capicua protein that's conserved among the two splice isoform that interacted with attacks in one. So what is capicua? It's an effector of the RASMAP kinase signaling pathway. It was discovered, as I mentioned in Drosophila, loss of function of capicua caused embryo patterning abnormalities, wing vein abnormalities, and some intestinal stem cell proliferation defect. Essentially, it was shown that capicua is a repressor and it is inhibited by the EGFR receptors. And upon activation of these receptors, basically you inhibit capicua. When it's not inhibited, capicua will be phosphorylated and... Sorry, when it is inhibited, it will be phosphorylated and degraded. Whereas when it's not inhibited, it acts as a core repressor and attacks in one is in a complex. We went into the cerebellum and we found there are any complex together. So what did we learn happens in the polyq expanded attacks in one? What we learned is that that normal function, the attacks in one capicua complex, is now exaggerated in SC1. When you have a polyq expansion, you have increased repression on some target genes. So when you have this kind of result, you wonder what happens if we reduce the levels of capicua by 50%. And when we did that, what we discovered that reducing capicua levels by 50% rescued the Purkinje cell degeneration. So this is a Purkinje cell count in wild type animal in the capicua heads. And in the SC1 knock-in mice, you see a reduction here of the Purkinje cells degeneration, as you can see on the skull bind and staining. But when you reduce capicua levels, you rescue some of that phenotype. You rescue the pathology. We rescued the attacks here. All the behavior coordination deficits were released. Because capicua is a repressor, we can now look at what happens to genes that may be altered in SC1. And here what you see is a handful of examples where there are some genes. The black is the wild type level. You see, because capicua is a repressor, when you knock out capicua, these genes will go up as you would predict from a typical repressor. And what you see in the SC1 knock-in mice, these same genes are hyper-repressed. And when we take away one copy of capicua, we now correct that hyper-repression to where it's back to normal. So this suggested to us, this gain function of the complex leading to hyper-repression, repression got rescued by decreasing capicua. And rescued the phenotype. But what was also interesting, we found that some of the genes that normally repressed by capicua, they were now derepressed in the knock-in animals and they actually got worse about taking on one copy of capicua. So what this told us, there's also a loss of function of this complex on some targets. So for some target, there was hyper-repression. And so some target, there was some loss of function. Because all of the phenotypes, the Purkinje cell degeneration, the ataxia were rescued, we believe it's the relief of the hyper-repression that attracts you. But we could never be sure whether this loss of function of capicua is contributing anything to the phenotype. So with this, we started new studies and this is now, I sort of gave you the old background to give you some of the more recent work. What we decided to do with these new studies is first ask the question, we know now that the gain of function of ataxia in one causes disease, causes ataxia. But does loss of function of capicua or ataxia in one and its parallel contribute to pathogenesis? And this is work done in different brain regions in the cerebellum by Maxime Rousseau and Sean Liu and in the cortex by Sean and in the hypothalamus by Truman 10. So, and this is just to show you the brain regions where we deleted capicua and or we deleted the two paralogs of ataxia in one, ataxia in one and ataxia in one like. Now you might ask me, why not look at the constitutive knockout? Well, constitutive knockout of either capicua or the ataxia in one paralogs is lethal. Embryonic lethal and they have, we published on that, they have lung defect, they have many other defects. So it's an essential gene. So we decided to focus on the brain, take it out from the whole brain, the animals are very sick, they don't eat well, they died immediately after weaning. So then we focus on the forebrain, the hypothalamus and the hindbrain where we hit it in the cerebellum and I'm going to show you these data to tell you whether there's lots of function contributions. So when we took it out from the hindbrain, what you'll see here is deletion of either ataxia in one like or capicua. In this case, we're deleting ataxia in one and attacks in one like and showing you the low protein levels. One thing you'll notice that by simply deleting ataxia in one and one like, you significantly reduce capicua levels by 50% because they stabilize it in the complex. So this is another thing it tells us how dependent they are on each other. And when we delete capicua, on the other hand, the stability of the ataxia is not affected, but you can see here major loss of capicua. So now we have our animal models, what happens? Normal survival. Normal activity. No coordination problem. Perkinjee cells aging the animals to over a year. Perfectly normal. So loss of the attacks in one one like paralogue together from the cerebellum is not contributing to perkinjee cell degeneration. Same loss of capicua does not contribute anything to perkinjee cell degeneration. How about the forebrain? I'm going to summarize the forebrain data in a cartoon. If you knock capicua from the forebrain, what you cause is learning and memory problems. If you knock it out from the hypothalamus, you cause social behavior problems. The mask doesn't interact with other mice as well. So when we saw hyperactivity and learning and memory problems, we wondered if loss of capicua in humans will cause disease. So we began looking and searching for patients who might have capicua mutations that may cause loss of function. And indeed, we found them. Here are some sampling of these patients with intellectual disability, autism and seizures. Here's intellectual disability and seizures. More of the same autism. These are different patients. This parent has probably germline mosaicism. This is what you see. All of these are de novo, but this one had germline mosaicism. Since then, we identified many more patients. They're even more than in this slides. And you'll see most of these are either in conserved amino acids or causing frame shifting, loss of function alleles. And these are the typical phenotypes. And one of the patients had ALL. One of the nine patients had leukemia. And whenever we observe something, as I shared with you yesterday, you do the mouse study. It tells you go look in a human. You find a human with a disease. Then the human teaches you something more. Two things we learned here. We learned that there's seizures in five of the nine patients. And we learned that there is leukemia. Just recently, we discovered that the seizures happen from loss of the gene and inhibitory neurons. Because in the patients, it's haplosefficiency through every brain cells. We did our knockout only in excitatory cells in glia. So we recently knocked out the gene in inhibitory neurons and the animals had seizures by EEG at least. So we think this will finally reproduce all the phenotype. But we also were interested in this phenotype. It could be by chance that this patient with haplosefficiency had leukemia or it could be actually due to the loss of capicua. So we decided to test that hypothesis by now deleting capicua from adult animals after they are mature, use a tamoxifen-cree and see what happens. And when we deleted capicua in adult animals, you'll see 100% penetrant phenotype of either T-cell lymphoblastic leukemia or lymphoma. As you can see here, the post-tumin really saw these big thymuses. And you'll see here in the knockout, the lymphoma in these animals. So single gene loss goes leukemia and lymphoma. And we didn't have to sensitize the mouse model to anything. And we identified some of the mechanism by which this is happening, the molecular targets of capicua that are driving this oncogenesis. The same is true for attacks and one attacks and one likes. So deletion of attacks and one attacks and one likes in adulthood. Also the reason we did it in adulthood because I told you this is embryonic lethal. So probably what happened with the patient, she had one allele loss and probably in her bone marrow she had a second hit and unfortunately got her leukemia. So what we learned from these studies is the complex has important functions during development. The forebrain knockout has neurobehavioral defect. And this complex is dispensable in cerebellar development and that loss of function of the complex doesn't contribute to SCA1. But haploinsufficiency of capicua causes human disease, neurobehavioral field type of human and drastic reduction of the complex in adulthood can cause cancer. So this is what we learned about the normal function. Now what causes the cell specific degeneration? And for this we got we returned to capicua. I mentioned to you that attacks and one capicua interact and we map the interaction domain to the AXH domain on attacks and one in this region on capicua that's a very small peptide portion of capicua. And we collaborated with Jijun Song to crystallize the two proteins these two domains together. And he was able to do that and was able to pinpoint one amino acid on capicua and two amino acids on attacks and one that are very critical for the interaction. So we got it down to one amino acid of capicua. If you disrupt that amino acid of capicua you wipe out their interaction. We went back to the full length proteins and tested it in the context of the full protein and lo and behold they lose the interaction and two amino acids on attacks and one. So now that I told you gain a function of the attacks and one capicua complex drive disease we can ask the question does capicua play any role? Is it the only player in driving disease through a gain of function in Purkinje cells or are there other gain of function interactions in the Purkinje cell? So to answer this you can disrupt the attacks and one capicua interaction and this we did in collaboration with Harry Orr and Tyler Trumperlin who is a graduate student in his lab at the time created transgenic mice because they've already shown in their lab if they express mutant attacks on one in the Purkinje cells they get massive degeneration and we knew it's a similar mechanism that involves capicua based on the repressed gene. So the question now what happened if we mutate the amino acids on attacks and one that disrupt its interaction with capicua and now express polyq expanded attacks and one with these two mutations and to do this what you have to know is that you really disrupted that interaction and what you see here is that if we this is the input but if we so the input shows you that you express both of them equally express attacks in one whether this is the mutant in the wild type this is the sorry both are mutant with 82 glutamine so long glutamine track but one has the amino acid interaction mutation and now if you IP capicua what you'll see is you can bring down the attacks in one with just the expansion but not the attacks in one that has the mutation that disrupts its interaction so we we can show and this is work by Vitaly Bondar in our lab who was at the time a graduate student and you'll see here you pretty much wipe out that interaction although the amount of the protein and the RNA both of those make equal amount of protein equal amount of RNA so it's not because we're expressing less of the gene it's really simply because we cannot interact so what happened to the phenotype and when we measure the coordination on the rotating rod and we do this over time all the way to almost a year and here shown up to 20 weeks what you'll see that the controls stay long on the rotating rod the attacks in one polyq expansion animals they stay very less and deteriorate with time but the polyq expanded that cannot interact with capicua is no longer toxic it's right here that's the wild type and that's the expanded protein so they behave very similar to wild type and if we look at the Perkinje cell pathology you can see massive Perkinje cell degeneration but here we don't see any degeneration so what conclude what we concluded from the studies is that capicua does drive the Perkinje cell degeneration and that's important because now we know at least the secret at least for one protein attacks in one we know that if you have a partner that drives that and if you disrupt that partner you pretty much avoid the degeneration so this tells us perhaps one way you can get cell specific degeneration is by who is your protein talking to who is it interacting with and how important is that interaction the other important conclusion from that we can say that the polyq expanded attacks in one causes disease through a protein-based mechanism because the RNA is still being expressed and still highly abundant so it's not the CUG repeat within the RNA it's really or it's not Iran translation of that RNA it's really the protein and we have other data that also show that I'll mention them later and what we also learned based on everything I've shared we do so far so when we do the gene expression analyses the polyq expanded attacks in one show as many capicula target down-regulated when we do it when it cannot interact all of those genes were corrected so it tells us at least through an enhanced function of attacks in one so as we were studying these and doing all these experiment and actually the first experiment would show that's the hyper repression enhanced function it started to tell us well if it's enhanced function of attacks in one maybe if we can understand attacks in one regulation and find ways to lower attacks in one level that could be a strategy to help the disease and I'm going to share with you two stories of how we did that the first story was done by at the time he was a postdoc in the lab he now just took a faculty position at Columbia is finding ways to find factors that regulate attacks in one at the post transcriptional levels what regulates the RNA of attacks in one and what Alessandro discovered is that attacks in one which has a very long three-prime untranslated region over seven and a half KBs has binding sites for pomellium one and if you knock out pomellium one in cells you increase attacks in one mRNA and protein so this was in cells which was fine and nice but I felt that for us to make this believable we have to go in vivo and to do this we asked high fan Lynette Yale to share with us his pomellium one knockout mice and he was generous to do so and you'll see here that in the pomellium one heterozygous mice you have 50% of the pomellium levels and in the knockout all pomellium levels are gone and if you now watch what happens to endogenous attacks in one we're no longer speaking about polyq expanded attacks in one this is just an animal that lacks pomellium what you'll notice in the het you'll see some increase of attacks in one and in the nulls you see more increase and we look at this in the cortex in the cerebellum and the results are similar so here's the quantification where you see now this is the normal levels of attacks in one you see the increase here and you see more increase in the knockout so if you and what did the animals do the animals had terrible attacks here they had other problems but they clearly had attacks here and percigisal degeneration so we figured if reducing pomellium by 50% which lead to 30% increase in attacks in one you get attacks here and percigisal degeneration could this be really because of attacks in one because pomellium will have many many other targets and the only way to prove it's really attacks in one is if you normalize the attacks in one level and ask does the behavior rescue and does the degeneration rescue and this is what we found out we bred our pomellium heterozygous mice to our attacks in one loss of function heterozygous mice so you see here pomellium heterozygosity you can appreciate attacks in one levels increase you see attacks in one haploinsufficiency heterozygosity gives you 50% attacks in one when you combine the two together we go back to normalizing attacks in one and what you see here here's the wild type animals here's the pomellium head and here's the double head and the attacks in one head have also healthy percigis cells you see the double head total rescue of the percigisal degeneration so this really told us that at least the percigisal degeneration and the attacks here that were rescued by normalizing attacks in one were driven by a 30% increase wild type attacks in one we did not have to mutate attacks in one now one thing I want you to pay attention this degree of degeneration with 30% increase of attacks in one manifested at 10 weeks of age I want you to see how severe this is because later I'm going to show you in the polyq expansion it's actually milder so this is a much more severe disease than we see in the SC1 knock-in mice so with this we concluded that maybe mutations in pomellium one could cause attacks, human attacks and here again we spend some time looking for potential patients and studying their mutations and we found that in fact there are two types of mutations in pomellium one there's some mutations that affect their RNA binding protein function or it's a deletion of the gene so it's a true loss of function and those patients with haploinsufficiency had a severe syndrome that we labeled pomellium one associated developmental disability attacks and seizures but what was interesting we found in one family with a dominantly inherited late on sex attacks a much milder mutation that doesn't exactly interfere as much with the function of the protein based on us measuring its targets and attacks in one so this we called pomelliorated cerebellar ataxia so I think by now you can appreciate that if 30% change in attacks in one upward change or a 50% reduction in pomellium one can cause severe ataxia whereas a milder mutation probably 20 to 30% inhibition of pomellior function will give you later onset ataxia and this is really I would say this was the most surprising thing to me about the protein levels where it's now a healthy attacks in one that you didn't have to mutate to see disease but it affirmed to us that the levels of attacks in one matters and with that we decided okay now we want to use this information to see how we can modulate the course of a CA1 I mentioned to you before that Harry's lab had studied many transgenic animals where he expresses polyq attacks in one in percini cells and you see the nice degeneration here in this animal model but Harry identified a particular phosphorylation site to the Harry's lab at serine 776 whereby they showed if you mutate that serine to an alanine even in the context of the polyq expansion and even in the context of similar RNA levels the protein is no longer toxic so we were interested to know what's different between these two proteins and what we discovered is that this serine is bound by protein called 1433 multiple isoforms but particularly we focused on 1433 epsilon because we had access to a knockout mouse and we knew that there's less function redundancy since the total null is lethal we could study the heads of this mouse and through cell-based studies what we learned is that when 1433 binds attacks in one it retards its degradation and what we also learned is that when the polyq tract is expanded the binding of attacks in one to 1433 is enhanced so this gave us a clue why we might be seeing toxicity from attacks in one when you have a polyq expansion you're basically expanding the protein increasing the binding to 1433 increasing the accumulation of attacks in one and that's leading to disease so with this we rationalize if we reduce 1433 levels we should rescue some pathology and indeed we found if we reduced 1433 by 50 percent now we can reduce attacks in one by 20 percent and we looked at both total 2q attacks in one and 154q attacks in one and both of them are reduced as you can see by these western brats and this is 1433 reduced by 50 percent behaviorally again we rescued the attacks here and you can see the pathology you can see the 20 percent reduction of attacks in one the purkinje cell degeneration is now rescued but what I want you to notice this is the kind of degeneration we see in the SA1 mice at 32 weeks now you recall with 30 percent increase in wild type attacks in one you saw a lot more degeneration than this it was much more massive which tells me that the polyq expansion is probably causing an increase in the 10 to 20 percent range because it's much milder than the expansion we saw from the wild type attacks in one but this now gave us an entry to try to find therapeutics with the idea that if 20 percent reduction in attacks in one can rescue the phenotype of attacks here maybe if we can find regulators that regulate its level that will give us insight into its biology but it also might give us additional candidate genes for attacks here pomelio is one example at the transcriptional level and hopefully therapeutic entry points for a SA1 and we decided to focus on things that change the protein level so this is going to be now post-translational and this is work that was done by a postdoc in the lab Jihay Park in collaboration with Ismail Aramahi who was at the time postdoc in Juan Botas' lab they both collaborated to find regulators of attacks in one so what Jihay did is she created a cell line with a reporter for protein levels of attacks in one and basically what you see here is she created a transgenic line that expresses DS red then there is an internal ribosomal entry site where from the same plasmid now attacks in one fused to GFP is being expressed so the cells will make both proteins and this way we can monitor the ratio and what we're looking for typically when because they're from the same promoter you should have similar levels but if we find something that decreases attacks in one we're going to see more red or something that increases attacks in one we'll see more green and this will allow us to screen for things that actually affect the level of attacks in one rather than just kill the cells or just affect the promoter driving this transgene in parallel to this the Botas lab screen the same genes that we were screening in cells in fruit flies and in their case they expressed mutant attacks in one and the fruit flies had ID generation so what we ask is we want to find things that work in both systems because the idea if it works in both systems it's most likely to be robust and trustworthy and we've done SIRNA screens without short hairpin screens and I'll show you here some example of the data here's an example for the SIRNA Kynome screen you'll see that there were 10 genes shared between the fruit fly screen and the mouse screen and here's a short hairpin screen pooled screen and you'll see again some shared genes here are some of these shared genes MSK1, MSK2 are among them PKA is amongst them but here's the bigger picture many of these genes work within one pathway so MSK is at the bottom of that pathway so when you find 10 genes from two systems totally disparate species and they work in the same pathway that's to me what makes the data trustworthy now how does this pathway impact attacks in one stability well we got maybe a little bit lucky here in that we did mass spec and asked do any of the kinases we identified in this screen phosphorylate attacks in one and MSK1 the one in that pathway was the primary kinase of phosphorylated serine 776 as you can see here in the red and PKA1 was the second kinase so discovering we also found RSK but the majority of the phosphorylation happened with MSK1 and we've done a lot of study to prove that MSK1 can impact the level of attacks in one only if it can be phosphorylated we then went in vivo there are two paralogs to MSK1 MSK2 and when we take away one copy of each you see the animals are healthy when we take now attacks in one expressing mice with Purkinje cell degeneration if you have no modifier here or one copy is maybe slight reduction in degeneration but when you take one copy of each we see a very nice rescue of the degeneration as well as a rescue of the ataxia so what this told us is MSK1 then is a wonderful target to explore potentially for lowering attacks in one levels and for this one can test MSK1 inhibitors and we chose to do that on eye neurons from patients with SCA1 we also generated now a mass I mentioned to you how to use transgenic mice now we've made knock-in mice that have an alanine mutation on the polyq expanded allele and I don't have time to show the data it's new work but at least we do rescue the cerebellar phenotype as well so what happens if you add a tool compound to IPSC derived neurons you see that you can nicely inhibit the phosphorylation of MSK1 which is required for its activity but we also lower the level of human attacks in one in human patients so with this background you could see now why this makes it worthwhile developing an inhibitor to MSK1 and see if we can use that hopefully to help SCA1 patients one thought we had is that so we know MSK1 regulates attacks on one but we know other targets must also regulate it and we are aware that these are chronic diseases and if you want to inhibit a pathway in a disease for 40 years you better be sure that's safe we do know that the MSK1-MSK2 double heterozygous knockout are healthy MSK1 knockout are not terribly sick but still that's a lot so what we rationalize that perhaps if we can partially inhibit one or two targets that may be a good thing and recall I mentioned to you that PKA1 also phosphorylated attacks on one so here we use tool-compant to inhibit either PKA or MSK1 or the two together and you will see that the two together has a much more potent effect and hopefully that would reduce the toxicity and since then we've broadened our screen to about 7800 genes what we call the sedragable genomes Steve Ellidge helped make these libraries with short hairpins and in this case we do now the screening where we infect the different libraries and sort the cells and sequence them and decode which short hairpin RNAs are enriched where attacks on one has been lowered and then we put these through the fly studies and through human eye neurons and now we're at this stage we're testing some of the candidates in vivo and the animal model our idea would be if we can find two to three targets that we can inhibit cooperatively we can start with MSK1 to develop a drug but one could maybe do two or three drugs so from the SCA1 studies then I hope I convinced you that we learned that the molecular mechanism at least driving the disease in the cerebellum is enhanced function of attacks on one capicua this what contributes to cerebellar pathology and it is capicua is required for Perkin G cell pathology I don't have time to show the data and still work in progress it is not what's driving the disease in the brainstem why do I know this? because when we lower capicua levels we slightly lengthen the half life the life of the animals we don't cure them so there are other drivers and we're now pursuing these other interactors and regarding what can we do about the course of the disease we hope that if we can partially lowering attacks on one using therapeutics that either target the modulator or attacks on one or both we can perhaps help this disease so this is what we've learned from SCA1 and I hope that other investigators working on other degenerative diseases will think about really exploring the normal function of the proteins their partners would drive the cell-specific degeneration but what we also learned that which I'm sure you appreciate that protein levels matter I showed you that polyq expansion stabilizes the protein but we know now that attacks on one levels by themselves the higher the worse is the disease and this really made us think about other disease drivers we know that APP duplication can cause dementia we know people with Down syndrome they have an extra copy of chromosome 21 where APP is localized and these people have dementia at a younger age in their 40s and they have the same pathology that actually you see in sporadic disease so somehow we know APP is a driver and there are a couple of individuals that have the trisomy minus the APP locus and they don't have Alzheimer's so we do know APP is actually the driver we know that levels and isoform splice changes in tau can drive neurodegeneration and also we know in synuclein a driver of Parkinson's disease having an extra copy of the wild type protein you don't have to mutate the protein you can have degeneration so knowing that all these proteins can drive degeneration without even mutating them we decided to look at modulators of these proteins as a therapeutic entry point for these diseases and we figured at worst we learned something about the biology of what regulates these proteins and at best we might find targets that might help in these diseases so here again we did the same strategy collaborating with the BOTAS lab where they did all the screening of flies and we did it in human cells and I just want to show you some of the cool things we do with the flies these are the tau flies and you can see how they have a problem in that they are slower in climbing so we use that when you have a modifier we not only look at the fly eye but we try to see if the modifier now will allow these flies to climb much faster so this allows sensitive quantification and when we did that I guess I sorry about this that's okay one of the candidates we identified is a kinase that we discovered using mass spec it phosphorylated tau at serine 356 now tau has been studied extensively many kinases have been identified that phosphorylated this kinase was not known to regulate it so this is the beauty of unbiased screen we're new to the field we just did it in a blinded way and it came about and what you see here when we mapped it to serine 356 if we use a wild type NOAC you'll see increased phosphorylation but if you mutate NOAC the protein is no longer phosphorylated but it has no effect on other phosphorylation sites on tau and then we explore the mechanism by which NOAC retards the turnover what does it do at tau and what you see here if you have wild type tau and NOAC you'll see the half-life of protein this is a doxycycline-inducible protein and you see when you don't have NOAC it turns over by 48 hours and 60 hours almost gone if you have NOAC it now the clearance is lower however if the serine is mutated NOAC no longer has an effect and this is shown here in the graphs where if it's an alanine mutant it has a much shorter half-life whereas if it has NOAC that glences the halftop from the wild type protein so with this we decided to test that in animal models and I'm just going to summarize the data when we bred these mice lacking one copy of NOAC one to mice that express toxic tau protein these animals have degeneration we found we decreased the phosphorylation and tau levels in these mice we restored learning and memory restored synaptic plasticity this is in vivo long-term potentiation study you see the tau animals have decreased LTP over several days and all the others are controls which look inseparable so the NOAC tau double mutants are inseparable we reduced the pathology and we improved survival by months so this gave us our another target that's now being pursued to potentially see if this would be helpful to reduce tau toxicity in people at risk and people perhaps with high levels of tau in their CSF and eventually the screen were done on APP and tau by Geon Kim and Ismail and Ramahi that they identified multiple modulators and just want to show you here an example you recall I told you when you find modulators in a pathway you gain confidence here's an example in for multiple enzyme in a sumo conjugating pathway which upon knockdown of these genes you'll see a reduction of tau levels this is the control vinculin here for a control this is the negative targeting short hairpin RNAs and this is UB2 one targeting you see reduction of tau AOS1 also reduction here's the quantification so basically many of these the black bars are the negative controls and all of these within the same pathway shows you that actually we are pulling genes at work in a pathway that regulate tau levels so what do you do when you have so many candidates you pick some of the ones that you think may be safe to modulate and you now want to test them in vivo we can't do everything like we did for NOAC bring the knockout mice do genetic interactions we began now to use AAV viruses at P0 to monitor the levels of tau by knocking down these genes in vivo and this is some example of the data trim 28 is has multiple functions but one of its function is a ligase activity and you can see here how knockdown of this gene can reduce tau levels this is on negative control and this is the trim 28 knockdown and here's some example of additional genes where we can see this is targeting tau directly in the mice but here's some other short hairpin RNAs targeting some of the other candidates and we see at least 20% reduction and that's really what we're aiming for we don't need much more than that so what we think is that these genetic streams are going to be extremely helpful not only to give us maybe insight into the regulators of tau or app or any of those proteins we have discovered NOAC1, trim 28 there are others in the works but what they will also give us is candidate genetic risk factors you can envision that inhibition of NOAC will reduce tau levels that might suggest that a mutation that cause constitutive increase activity of the enzyme might increase the risk of someone for Alzheimer or if we find something that normally will destabilize NOAC and when you inhibit it now tau levels will go up this will tell you that lots of function of those genes will be genetic risk factors so I think the beauty of this approach as we go through these modifiers of tau levels in either direction it can help others who are doing a lot of sequencing and Alzheimer patients identify risk factors because you sequence people with Alzheimer disease it's not usually like autism, a trio where you say here's a de novo mutation it's easy to pinpoint you're going to see variations they may be in the parent and the parent will not have symptoms but at least if you see some variation and you know there's some biological function in some of these genes that they may regulate tau levels or APP levels then that gets to be moved up on the priority list and my last slide is to show you it wasn't really by design but this is biology here are more proteins that their protein levels matter so we talked about attacks in one where I showed you the 30% increase some increase causes a C1 and hopefully I convinced you that through pomilio it's also contributing and pomilio one of course we know that the milder mutation cause late onset but haploinsufficiency, severe onset and here's sine nucleon the duplications or triplications have different severity this is a triplication they have the most severe form and then APP of course we know extra copies cause early onset disease so I think it's really important to pay attention to protein levels one thing I would say many of us used to think enzymes if you lack one allele it's probably okay you still have plenty enzyme activity well I don't think this is we can say this safely anymore because in Parkinson's disease GBA haploinsufficiency which you know loss of GBA cause lysosomal storage disease but loss of one allele and the parents of these children is proving to cause increased risk about a third of these individuals will develop Parkinson's disease because of some subtle effect on lysosomal function so I really learned from all the studies you learn about pretty much all but one protein atonal is transcription factor I work on but all the other proteins I work on you've seen now how every one of them is those are sensitive and I don't have the data that's another hour but it turns out atonal haploinsufficiency causes loss of inner ear, hair, cell and deafness so that's another dose of sensitive gene so I think that we really learn to respect that we need to pay attention small changes have big effects and I hope these data would convince you I mentioned the people who've done the work have we've gone along I think the one person I didn't mention is Christian Lasagna-Reeves who worked on the NOVA QAN story but all the others and of course my collaboration with Harry which has been really very exciting and precious and similarly with Juan Bota through all the nice fly work he's done and our computational biology collaborators led by Zhendong Lu and we have many clinical collaborators who helped us with the PoMone clinical study including Kim Boycott who led the study on the dominant ataxia and of course I'm grateful to the funders NIH, IDDRC ataxia foundation and various consortia including a Belford foundation and the Piccour foundation thank you very much okay time for questions yes Anita that's a good question by specificity mean if you inhibit NOVA QAN how do you know you're not affecting something else right one thing we propose and we did a lot of study those I don't want to say boring but I don't think there's studies you want to hear about what we did is we brought and we we proposed as we're working to develop the pharmaceutical industry develop inhibitors we said we don't have to have total inhibition of the enzyme partial so let's look at the partial inhibition so at least we look to the best of our ability in two areas humans and mice what we found in humans there are many individuals who are healthy who are have lots of function alleles of NOVA QAN and their adults and they have no problems these are people who are in the exact database there's several of them and if you look at the exact database NOVA QAN is called mutation tolerant in the sense they they saw the expected amount of loss of function mutations the expected amount of mis-sense mutations throughout the gene so that gives us confidence that at least the human being who lack one copy of this you know it helps it so maybe if you inhibit it by 50 percent as best as we can tell that's okay in the mouse we did additional studies on the heterozygous mice basically putting them through every potential assay you can do cardiovascular put them on high fat diets stress them wound here etc we did all that and I think that this is what gives us hope that that partial inhibition would be okay I would go further and I would say I would be thrilled if I can find another target and where the two conversion attacks on one and now I don't even have to inhibit NOAC on 50 percent maybe 30 percent just like we titrate blood pressure medications we should really start thinking titration combination I know it's not the norms nobody want to hear about this neither the pharmaceutical companies nor the FDA but I think if we're really going to be serious of trading these diseases in a preventative way rather than once the symptoms set in which means I'm going to take a healthy individual who we know has the SCI1 mutation who hasn't manifested any symptoms they're 30 years old and tell them for the next 70 years you're going to be on medicine I much rather give them a combination and milder dosage of something chronically than any one system because the odds of these two systems impinging on many other proteins is now reduced so this is my vision I'm not saying my vision will be appreciated by everyone in the pharmaceutical maybe in regulatory agencies but I think it's beginning to start this conversation and I hope that data will convince people down the line that this is a good way to go speak up after I'll repeat the question So, before co-factors in a taxi for example where you're on between the peaks but did you mention any of those there was work by Tijac back in the 90s which are people's sequestration model from that do you see a role for those in your so that's a really good question I think I appreciate the question which is getting back at could it be that the polyq expansion is really titrating factors some transcription factors and it's maybe in other disorders remember a lot of studies were done in the past using either overexpression systems or sometimes using systems that may be heterologous or not the full-length protein all I can speak to is the SCA1 model where we've done it in the context of the full-length protein and what we've learned is that you can have all the cues you want but if you now change one amino acid serine 776 or change the interaction with capicua you're not going to have disease so all what if this polyq expansion really plays a role in titrating some things maybe we don't see that in the lifespan in the mouse maybe in a human it's still in action but at least we know that the pathology even in the face of massive expression when Harry did it with the Purkinje cells but full context of the protein we didn't see the pathology anymore so I tend to think it's really more I think this is what's happening I think the polyq expansion it's changing the conformation of the protein enhancing some binding activities perhaps retarding slightly its clearance and eventually the disease is via the gain of function of that and I think the androgen receptor supports this because many elegant studies were done by people studying spinal bulb or muscular atrophy where the polyq expansion is within the androgen receptor which show it's really about the normal function of the protein because if you now take away the androgen the ligand you no longer have the toxicity so it's not that the protein itself is doing something or I can have it it's really when it binds its ligands goes into the nucleus and does what it has to do and we know it's a gain of function because people who lack the receptor they have a different disease they have testicular feminization syndrome where their sterile have female-like features in they don't have motor neuron degeneration in the polyq expansion you have partial loss of function just like we saw some of the capicua target but that doesn't seem to be what's contributing to the disease and you had a comment oh yes, sorry the comment was it could mean that there are patients cancer patients with solitumes or a given vortishment and there are also one side effects is ataxia it could be so vortishment there's an approachism inhibitor right and I wondered whether the mechanism behind it which isn't well characterized at all in terms of the ataxia could be a subtle increase in some of the ataxins I mean many of these ataxins are abundant in Purkinje cells so it could be that you might increase two to three of these ataxins just a little bit because many of them are proteins that are degraded via the proteasome pathway that is true and then when you stop the treatment they might improve that could be possible we did show that ataxin was degraded via the proteasome pathway so you saw this TALL, cancer in this one patient is that the only cancer or are there other kinds of cancers? thank you for asking a question the question is do we see other cancers so we did not because we only looked at children and these children with developmental disability but capicua is now pretty much known to be loss of function it is a tumor suppressor gene loss of function can cause a variety of cancers for sure oligodendrogliomas if you lose capicua I thought you have to have nerve neural tumors with this you talk about nerves and then you go into the blood so I think there are people that have mutations that have loss of function mutation and they have additional cancers definitely thank you I feel better life makes sense so the question is is there retinal degeneration in SCA1 or any of these models not in SCA1 in SCA7 we published on that another repeat expansion disorder that disease causes retinal degeneration and attacks yeah but SCA1 neither the humans nor the mice have retinal degeneration there was one question in the back so I can speak to attacks in one so the question is can the mouse help us model this dose of sensitivity I can only speak to attacks in one because as you recall it was the 30% increase in the mouse endogenous attacks in one that led to purkinje cell degeneration so clearly here the mouse was sensitive to the dosage of this protein and we also know from the SCA1 situation that when we reduce the polyq expanded attacks in one by 20% we were able to rescue the attacks yeah so I think it's safe to say for SCA1 for this attacks in one protein somehow it's the mouse that revealed to us the dose of sensitivity for attacks in one and also for pomegranate one I couldn't say the same for many other protein I can tell you for capicua haplitis efficiency of capicua causes increase in activity which the humans have hyperactive but it didn't cause the learning and memory we only saw the learning and memory in the mouse when both alleles of capicua were deleted in the forebrain so for some proteins it's easy to see for others it's not you may have to lose both alleles to learn that this is an important protein then you discover your human counterpart to be you know haplitis efficient so can I ask one last question so it's in trying to move forward with tau and its regulation what's the simplest and quickest model to test that in that has biological relevance I mean we use a model that was developed by Virginia Lee that she expresses one of the human disease causing tau mutation at a proline 301 position this is called the p19 line the reason we use this mouse is because it starts manifesting pathology and behavior problem after six months of age you need to age it to six to nine months of age but that's better than the other model that takes much longer I was thinking I mean I want to do it in flies personally oh flies there are one botus has flies that overexpress tau and they have neuronal degeneration and it was the flies did discover nook once yeah and we discovered that it lowers protein if you inhibit nook one you inhabit protein levels of tau in those flies is reduced but not the RNA so it's really so you can move forward with that yeah so we do it in parallel but nobody's gonna want to you know follow up further for drug development until you really nail a mechanism and do it in a mammal so yes everything we do is for cells and flies super any other burning or not so burning smoldering questions yeah right that's really an interesting question I don't have a very direct clear mechanism that downstream what I can tell you there are quite a few gene expression changes downstream of the capicula attacks in one complex that are altered there are also some genes that are critical for calcium homeostasis that are altered and some genes are critical for neurotransmission respond to you know neurotransmission that are altered so I cannot tell you if any one of those all of those once we see that there's so many genes changing we don't go by one gene and say this is the gene that's really driving it we think it's the combination and we know that when you prohibit the interaction with capicula all of these correct so we we think these are the downstream mediators we have a candidate there there we just published those but I don't know that any one of them is a major driver it's hard for me to envision one single driver now that's that's an interesting question the question is do we know the pattern of degeneration I can tell you in humans what you see is some abnormal axons with torpedoes bodies called torpedo bodies which contain we looked at them by em they contain mitochondria and some other cellular material but for sure there's mitochondria trapped in the axons that you know so probably can't move along the axons as well so we know the axons are altered and there's some axonal degeneration in the mice we've only focused on looking at the dendrites we see loss of dendrites quite early in the process and as the disease progresses you lose the cell body but what I cannot tell you if the axon happens before we have not looked as deeply in the mouse so if you tether a taxi as to other transcription factors does it turn into is it a co-repressor we don't we don't know per se if it is a co-repressor with other factors we know for sure with capicula is a co-repressor although we do see some genes going down when you knock out capicula so I cannot be 100% sure it's purely that but we have not identified ROR alpha has been shown to interact with attacks in one but we don't know for sure really how much those contribute to the disease since now we know that at least in the pregnancy cell capicula is sufficient we have not pursued this any further in the brainstem we don't know we're just beginning to IP the partners in the brainstem to see if we can find a key one driving the disease any smoldering questions let us thank Huda again for a wonderful second lecture as well