SLC12A6

Transcription

Well it is my distinct pleasure to introduce Rick Lifton who is a native of our nation’s capital and then went on to Dartmouth University where he got his Bachelors degree, moved transcontinentally to Stanford University to get his M.D. and Ph.D. degree in Biochemistry and then move back transcontinentally to Boston where at Brigham and Woman’s Hospital he received his house staff training and stayed on as a faculty member at Harvard before being lured away to Yale University in 1993. There he has risen through the ranks and is currently the professor and chairmen of the Department of Genetics at Yale University and Rick has done a wonderful job at using very powerful tools of genetics to understand renal disease and, as I think, taken one of the hypothesis of Arthur Gieten about the kidney being what controls the amount of fluid in that balloon and helps control blood pressure. Rick has taken that and has actually shown that hypothesis to be true by demonstrating several human diseases, uh, human examples in which mutation in various transporters or channels and regulators in the kidney are important contributors to changes in blood pressure so it’s a great deal of pleasure that I introduce Rick Lifton [Lifton] Thanks Walter, it’s a real pleasure to be here and have the opportunity to speak today. What I thought I would do in the context of this symposium is discuss the natural alliance of genetics and physiology, because I think genetic tools are spectacular at giving us new insights into physiologic pathways into rate limiting steps that we might have had a very difficult time sorting out from physiologic analysis alone. So I wanted to develop that theme a little bit in the course of today’s discussion. So, the notion is that when we don’t really know who’s the driver and who’s the passenger, genetics is a spectacular way of making that decision. Mutations can tell you this by specific mutations can identify key nodes and rate determining steps in complex pathways and of course it shouldn’t escape our attention that these targets that are identified might might be manipulated for health benefit in the general population. So I came in to this from a background in Drosophila genetics, and when a generation ago we started trying to understand the basic drivers of development in the fruit fly, we took a simple genetic approach of doing eonu mutagenesis and looking for mutations that had drastic effects on the phenotypes we were interested in. And this is an example of such a eonu induced mutation that results in single gene mutation that results in not a subtle effect on body size, but instead, a three-fold effect on the size of the larva and dramatic effects on the size of the adult. And we are interested in looking for similar effects in the human, so we might identify, not places were we might be able to tweak the pieces a little bit for health benefit but places where we might be able to have spectacularly large effects that would be important clinically and I’ll give you one example of this. So this is the in vivo equivalent of the Bruce Willis character from the movie Unbreakable. So this is a fellow who suffered a motor vehicle accident in Bridgeport, Connecticut on a Saturday night. He was taken to a local emergency room and evaluated. He had routine cervical spine films done to rule out an occult fracture and the internist who saw him was a resident on call and he said, “You know, you don’t have any fractures but I’m very concerned about you because you have the densed bones I’ve ever seen”. And he was concerned that he might have multiple myeloma and he referred him up to Yale to see Carl and Sonia in the endocrine section. Carl measured his bone density and he found that he, indeed, had the highest bone density on the planet. His lumbosacral spine z-score, or the number of standard deviations removed from the mean and standard deviation of the population, was eight standard deviations out, and if you could find a z table that goes out that far, it's one with many, many, many zeros on the end of it. But it turned out when we evaluated this fellow’s family, he did have the highest bone density on the planet but not by that much. Half of his family members had bone densities that rivaled his, whereas, the other half were just like anyone else in this room. They were stone cold normal. And this demonstrated that this was a simple autosomal dominate trait transmitted through his family, that we were able to map and ultimately identify the underlying gene product. And the mutation in this family turned out to be a single amino acid substitution in LDL receptor related protein 5, a gene that through Drosophila genetics had been implicated in the whence signaling pathway as a coreceptor with frizzeled The mutation in this patient, and family, was a single amino acid substitution at a position that had been completely conserved in LRP6 orthologs from fruit flies to humans. And when we did biochemistry on this system we demonstrated the mechanism by which this mutation imparts its effect. It turns out that the Dkk and the sclerostine are normal antagonists of whence signaling and the antagonized whence signaling by binding to LRP5 and inducing its clearance from the cells surface through endocytosis through clathrin coated pits and that in the presence of these single amino acid substitutions these patients have sustained whence signaling. Now the beauty of this, I think, is if I’d come to you and I said “You know a lot about physiology, what is your guess as to what the phenotype of these patients is going to be if they have constituent gain of function in the whence signaling pathway?” Those of you who work in development might have said, “well, gosh, we know that in Xenopus we can get two headed tadpoles from gain of function mutation in whence signaling”. Those of you who work in cancer might say, “we know a lot of epithelial cancers that have increased activity of the whence signaling pathway, so you might be surprised to find that the only clinical complaint these patients have is that they sink when they try to swim, and other than that, they have extraordinarily high bone density and they are protected from the development of fractures,” these bones have extraordinary integrity and suggests that this actually might be an attractable target for therapeutic intervention. Interestingly, with, in this internet age, once this paper was published, a number of patients self-identified themselves to us. One interesting example was a physician in Alabama called me and said, “I think I might have this trait” and I said, “why do you think that?” and he says “well I’m 80 years old, I’ve developed arthritis in my hip, and they took me to the operating room for a hip replacement. I woke up after the surgery, I had this big scar on my leg, I was in a little bit of pain, my physician came in and I said ‘ how’d the operation go?’ He said, ‘we were unable to replace your hip’ and I said, ‘Why was that?’ and he said, ‘because you broke every drill bit in the operating room.’” So we have had a number of examples like this and because of an understanding the biochemistry of this rare family there are now monoclonal antibodies to Dkk and sclerostine that are go, in human trials trials that are having quite interesting effects not just to prevent further bone loss but to promote bone anabolism and formation of bone. So this is one example of the kinds of things you can do with combining genetics and providing insight into basic physiology. As Walter indicated, one of our major interests over the years has been the understanding of the pathogenesis of hypertension, which again, is a physiologic conundrum. It effects a billion people world-wide and a major public health problem which again, is a physiologic conundrum. It effects a billion people world-wide and a major public health problem problem because of it’s contribution to death from stroke and myocardial infarction and it’s very frustrating that we can’t, we have a hard time deciding even which organ is the primary cause of this trait, and as Walter indicated, this goes back to the physiologic analysis of the trait and this is Arthur Gieten’s working model for the regulation regulation of arterial blood pressure, devised at a time when we knew perhaps a third of the components that we we recognize now and I’ll think you can see that this is a very complicated tree and as a consequence this looked to us like an ideal place to use genetic approaches to try to understand the trait, and so the notion was could we find single mutations that would disrupt the overall behavior of this entire trait and drive blood pressure to the extreme high end of the distribution, or instead, the extreme low end of the distribution, and of course we knew to the extreme high end of the distribution, or instead, the extreme low end of the distribution, and of course we knew we were onto a good idea because so many people in the field told me when we were starting out on this that this was hopeless, and would never work because can’t you see that this is such a complex tree that if you place any lesion here there are so many ways you can compensate that you will never find single lesions that have large effects. Well fortunately, are so many ways you can compensate that you will never find single lesions that have large effects. Well fortunately, that turned out to be incorrect, that we have identified, now, mutations in ten genes that will drive blood pressure that turned out to be incorrect, that we have identified, now, mutations in ten genes that will drive blood pressure to the extreme high end of the distribution and another ten that will drive blood pressure to the extreme low end of the distribution in the population, and these genes are not littered randomly across the physiologic of the distribution in the population, and these genes are not littered randomly across the physiologic and the simple bottom line is, all of the mutations that increase net salt reabsorption by the kidney, raise blood pressure dramatically and mutations that impair salt reabsorption by the kidney dramatically lower blood pressure. There are diverse effects on handling of calcium, magnesium and potassium, but if you know the vector of what’s happening to sodium and chloride, you know what is happening to blood pressure in humans. So these are all single gene Mendelian traits that have been identified by positional cloning and I’ll give you a few examples of these. So this is a diagram of a nephron and some of the salt transport pathways are indicated and the overall activity of the system is regulated by the renin angiotensin system which through angiotensin 2 causing increased secretion of L Dosterone through its receptor, the mineralocorticoid receptor, a member of the nuclear hormone receptor family, of L Dosterone through its receptor, the mineralocorticoid receptor, a member of the nuclear hormone receptor family, a fellow in the laboratory, was a single amino acid substitution in the ligand binding domain of the mineralocorticoid receptor that caused a syndrome of severe hypertension that caused profound exacerbation during pregnancy and the single amino acid that caused a syndrome of severe hypertension that caused profound exacerbation during pregnancy and the single amino acid substitution was in the ligand binding domain and by combining structural biology and cell biology and biochemistry, David was able to demonstrate that this single amino acid makes a new Van der Waals interaction across the ligand binding domain between helix 5 and helix 3 and this new Van der Waals interaction eliminates the requirement for the normal interaction between the 21 hydroxyl group of L Dosterone and helix 3. So in the presence of this interaction, you no longer need this interaction and the consequence of this is that steroids, such as progesterone, of this interaction, you no longer need this interaction and the consequence of this is that steroids, such as progesterone, which normally bind but fail to active the receptor are now potent ligands for the receptor, and so women when they become pregnant who harbor this mutation, progesterone levels go up 100-fold during pregnancy when they become pregnant who harbor this mutation, progesterone levels go up 100-fold during pregnancy and they develop rip-roaring pregnancy induced hypertension, and all women with this mutation have required delivery at about week 30 because of uncontrollable hypertension as a consequence. Just downstream of the the mineralocorticoid receptor is one of it’s major targets, the epithelial sodium channel and the Mendelian form of hypertension called Liddle syndrome is caused by mutations that cause constitutive activity of the epithelial sodium channel. So the channel is normally cleared from the cell surface by endocytosis through clathrin coated pits. There’s a typical NPXY domain in the carboxy termini of the epithelial sodium channel that permit it’s clearance and mutations in patients that cause either the complete loss of the carboxy terminus or mutate one of four amino acids lead to sustained expression of the epithelial cell, epithelial sodium channel. At the cell surface, they remain active and enter into the same final common pathway of increased salt reabsorption, leading to expanded plasma volume initially, increased cardiac output and elevated blood pressure. So, starting with a blank slate, we now have a host of mutations that do this. On the down side of blood pressure, there are a variety of syndromes that cause life threatening, typically fatal, hypotension in the neonatal period that all result from impaired salt reabsorption. The most potent place to induce this trait is in the thick thick ascending limb of Henle where about 30% of the filtered load of salt is reabsorbed and we’ve identified four genes that when mutated, cause this trait called Bartter’s syndrome. If you have mutations in the target for furosemides, the sodium potassium 2 chloride cotransporter, you have a severe salt wasting phenotype associated with the activation of the renin angiotensin system, induction of hypokalemia and metabolic alkalosis, but you can’t supplant this missing channel adequately and these children frequently die from the severity of their salt wasting, but we can get similar phenotypes from mutations in a chloride channel that the genetics suggested in subsequent physiology has proven mediates the exit of chloride, so chloride that comes in across the apical lumen has to get across the basolateral surface to the interstitial space and loss of function in this specific chloride channel results the basolateral surface to the interstitial space and loss of function in this specific chloride channel results in a phenotype that is, clinically, almost indistinguishable from that induced by the loss of the entry step, but more interestingly from a therapeutic perspective, we can get a similar phenotype from loss of function mutations in a potassium channel, ROMK, first described by Gerhard Giebish and cloned initially by Steve Hebert, and loss of function mutations in ROMK give a very similar phenotype with one very intriguing difference. These patients all have marked hypokalemia, typically with potassiums between 1.1 and 2. These patients never developed that level of hypokalemia, and in fact, their potassium levels tend to hover between 3 and 3.5 and typically don’t require potassium replacement and the reason for that, is this same potassium channel is used in the distal nephron for net potassium secretion and as a consequence, this is become a very attractive target for development of novel antihypertensive diuretic agents. There are several aspects to this that are interesting. First, they never, these patients with this never develop the same kinds of hypokalemia, which is attractive. Second, this target up regulates in response to treatment, so those of you who are clinically active, recognize that when we give furosemide to a patient, the first dose the get a brisk response, the second dose a little bit less response and then they develop a threshold you have to keep increasing the dose to get sustained salt wasting. The reason for this, is the target is auto regulated and if you inhibit it, it’s gene expression goes up dramatically. That doesn’t happen with ROMK and so for this reason this actually might turn out to be a good target for anti-hypertensive agent, whereas this has never turned out to be a particularly attractive target for hypertension and is been more confined to the treatment of congestive heart failure and these agents are underdevelopment in the pharmaceutical industry as well. So, I’m frequently asked what are we going to do when we run out of these rare Mendelian traits, and I’m always reassured we will find more and we published a new one just last month that I’ll tell you about briefly, because we will find more and we published a new one just last month that I’ll tell you about briefly, because it gets us into a new territory. So Uta Shaw is a physician that came to the laboratory, and I tell every physician who comes to the lab that there are new traits out there they should try to recognize. So she went through our database of a thousand patients with hypokalemia and metabolic alkalosis and identified four families, five patients from four families and metabolic alkalosis and identified four families, five patients from four families who had the phenotype of Gitelman syndrome which features hypokalemia, metabolic alkalosis, low magnesium levels and low urinary calcium levels that’s attributed to loss of function mutations in the thiazide sensitive sodium chloride cotransporter, but they had an addition to that trait, profound neurologic distorters featuring seizures, deafness, ataxia, severe mental retardation and lower extremity weakness. So this was a new syndrome which we’ve called SeSAME to reflect all of the diverse clinical features, and she posited that this might be caused by a mutation in a single gene and, you know, quite remarkably with the technology we have now, in the span of two weeks went from identification of these patients, to identification of the underlying gene and what she found is that all of these families have, are linked to the same chromosome segment and they all have and what she found is that all of these families have, are linked to the same chromosome segment and they all have of the distal convoluted tubule of the nephron and is also expressed in astrocytes in the brain and we think the mechanism by which these mutations impart the diverse clinical features goes as follows. In the kidney, it turns out that the distal convoluted tubule on a per cell basis is the most active cell in salt reabsorption and as a consequence, the whole process is driven by the sodium pump, extruding potassium and bringing in, extruding sodium and bringing in potassium and because this is being done so briskly, in this cell type, salt reabsorption becomes highly dependent and bringing in potassium and because this is being done so briskly, in this cell type, salt reabsorption becomes highly dependent on the ability to recycle the potassium that comes into the cell back out across the basolateral surface. If your, and this turns out to be mediated by, at least in part, and dependent upon KCNJ10’s gene product cure 4.1. It’s believed it may also, it’s likely also an active heterodimer with cure 5.1, however, we have not found mutations in cure 5.1 in any of these families. So, we think we can explain the phenotype that looks like Gitelman’s syndrome because you’re stopping salt transport across the epithelium in this cell in a similar way to the, your stopping salt reabsorption through the thiazide sensitive sodium chloride cotransporter. We think the seizures and other neurologic phenotypes come from the absence of spatial buffering by astrocytes. Turns out, an astrocyte in the brain, KCJN10 sets the resting potential and is in close, in close apposition to active synapses and that KCNJ10 is responsible for clearing potassium from the synapse in response to neuronal activity and that KCNJ10 is responsible for clearing potassium from the synapse in response to neuronal activity and the absence of this spatial buffering, this promotes increased activity and results in seizures and the other neurologic phenotypes. So this is the first example of a potassium channel acting on the basolateral surface that is impairing salt reabsorption in the kidney. In the rest of my time, I want to talk about a project we have been working on for a number of years that continues to interest us and has actually turned us into more physiologists that I’d care to admit. So this starts off with a classic paradox though, and has actually turned us into more physiologists that I’d care to admit. So this starts off with a classic paradox though, in physiology and it’s a paradox in L Dosterone action. The kidney sees L Dosterone in two physiologic states. in physiology and it’s a paradox in L Dosterone action. The kidney sees L Dosterone in two physiologic states. First, if you become dehydrated, you activate the renin angiotension system and this, angiotension 2, causes the adrenal glomerulosa First, if you become dehydrated, you activate the renin angiotension system and this, angiotension 2, causes the adrenal glomerulosa glomerulosa to increase it’s secretion of L Dosterone and that promotes salt reabsorption. On the other hand, we know that potassium hyperkalemia is a direct secretagogue for L Dosterone and causes the kidney to maximize K secretion, and this raises the question when the kidney sees L Dosterone, how does it know whether it ought to be maximizing salt reabsorption or maximizing K secretion? One of the great things about being at Yale is you have all of these great physiologists that you can go down the hall and ask this question to, and when I asked is you have all of these great physiologists that you can go down the hall and ask this question to, and when I asked Gerhard Giebish and Peter Aronson and Steve Hebert, they all said, well you know, human dynamics, everything kind of works out and that was not a very satisfying answer and it was particularly perplexing because of this disease, Pseudohypoaldosteronism type 2. So these patients have hypertension with hyperkalemia despite normal L Dosterone secretion and it looked to us like these patients behave as though hypertension with hyperkalemia despite normal L Dosterone secretion and it looked to us like these patients behave as though they are stuck in a physiologic state in which they can only do this, and can never do this. So this is what got us very interested in this disease and Rick Wilson, a really brilliant Ph.D. student took this problem on and he positionally cloned the genes that caused this disease and it turns out that they’re caused by a novel family of serine/threonine kinases, the WNK kinases and what Rick showed was that mutations in WNK 1 can cause this trait and the mutations that cause this in WNK1 are large deletions that take out a big chunk of the first intron of WNK 1 and cause increased expression of it’s gene product. Conversely, a paralog WNK 4 on chromosome 17 is not, does not have similar mutations but instead has missense mutations that cause the disease, and these missense mutations are clustered within four amino acids distal to the kinase domain of the protein suggesting that these are going to be some peculiar gain of function mutation, because these do not look like simple loss of function mutations. So we’ve had at this point, the typical paradox in human genetics, a genotype on one hand, a phenotype on the other and no idea what connected the genotype to the phenotype and so we have had to do a lot of work over the years to figure this out, but what has emerged from this is that and so we have had to do a lot of work over the years to figure this out, but what has emerged from this is that these WNK kinases are previously unrecognized layer of regulation on electrolyte homeostasis in the kidney, that are actually orchestrating the behavior these WNK kinases are previously unrecognized layer of regulation on electrolyte homeostasis in the kidney, that are actually orchestrating the behavior of the salt and potassium flux pathways in the kidney, and so Chris Colley, a very talented M.D., Ph.D. student in the laboratory, of the salt and potassium flux pathways in the kidney, and so Chris Colley, a very talented M.D., Ph.D. student in the laboratory, has really led this activity, and what he has shown is that WNK 4 normally in it’s basal state, inhibits the activity of the thiazide sensitive has really led this activity, and what he has shown is that WNK 4 normally in it’s basal state, inhibits the activity of the thiazide sensitive cotransporter, inhibits the activity of the epithelial sodium channel and inhibits the activity of the potassium channel, ROMK, and the single amino acid substitutions cotransporter, inhibits the activity of the epithelial sodium channel and inhibits the activity of the potassium channel, ROMK, and the single amino acid substitutions in WNK 4 that cause disease in humans, have divergent effects on these different pathways. They eliminate inhibition of the thiazide sensitive cotransporter, they eliminate inhibition of the epithilal sodium channel but they augment inhibition of the potassium channel, ROMK, sensitive cotransporter, they eliminate inhibition of the epithilal sodium channel but they augment inhibition of the potassium channel, ROMK, and they further augment opening of the paracellular pathway for chloride, and the net effect of this is to maximize salt reabsorption and to minimize potassium secretion. So we think we can explain the physiology that we see in patients on the basis of these observations. So we think we can explain the physiology that we see in patients on the basis of these observations. This got us to thinking, that well if this is true, perhaps there’s a third state of this protein that would actually mediate the converse to maximize K secretion without maximizing potassium secretion and a very talented undergraduate at Yale, Aaron Ring, who is now a M.D., Ph.D at Stanford, demonstrated that this is, in fact, the case, that there is a WNK 4 can be phosphorylated in it’s carboxy terminus by the kinase SGK and this specific phosphorylation has a different effect from the mutations that cause PHA2 and these alleviate inhibition of the epithelial sodium channel which provides the driving force, which in conjunction with loss of inhibition of ROMK, promotes K secretion and so this suggests that there are three states for this protein, one that a basal state, one that promotes salt reabsorption without K secretion and then the converse, and most recently in collaboration with Gerardo Gomboa, we’ve tied this as we’ve anticipated the outset to angiotension 2 signaling. That angiotension 2 is acting through WNK 4 to achieve this state suggesting that the mutations mutations that cause disease in humans are a phenocopy of the state induced by angiotension 2, but I have to be mysterious about this because we really don’t understand what is going on at the site that is mutated in patients with PHA2 and why this is a phenocopy phenocopy of what occurs with angiotension 2. So if this all really occurs, it makes a very specific prediction, that if we were to make mice that had increased activity of the wild-type WNK 4 it ought to lower blood pressure and lower potassium and if we do the opposite if we make mice that had a copy of the PHA2 mutant on WNK4, it ought to raise blood pressure and raise potassium levels, and this turns out to be the case. Maria Lalioti made these BAC transgenic mice gene is expressed under it’s endogenous physiologic promoter, the mice with an additional copy of the wild-type WNK 4, have lower blood pressure then their wild-type litter mates. Mice with the PHA2 mutation have elevated blood pressure and this is a pair of mice with the different copies of the gene that were pair raised showing the differences in blood pressure. The effects on potassium are equally dramatic, most impressively, the, if you put these mice, the PHA2 mice on a high potassium diet the mice all die within a day, and if you back off on the K that you put in the diet, you can have mice walking around with potassiums at 8.5. All of this, interestingly, is corrected by giving them thiazide diuretics and the mechanism for this is interesting and mysterious. We think that the major mechanism of this mutation is to augment activity of the thiazide sensitive sodium chloride cotransporter, and when you look at these mice the, compared to wild-type this is staining of the distal convoluted tubule and staining for the thiazide sensitive cotransporter, this is low power and high power. The mice with the additional copy of the wild-type gene have actually, are hypoplastic for the DCT, whereas, there is dramatic hyperplasia of the DCT in mice with just a single amino acid substitution distinguishing this from this and the mechanism of this is not entirely clear. So there are four of these WNK kinases in the human genome, two of them are mutated in these patients with hypertension and there are two more and we are interested in what they are doing, and Jesse Rinehart in the lab has led this effort, and what he found was that WNK 3 has the interesting property that it is an activator in Xenopus oocyte expression systems of all of the sodium chloride cotransproters that mediate physiologic chloride entry into cell using the favorable gradient for sodium and on the other hand, they are inhibitors of all of the casial cotransporters which mediate chloride efflux out of cells using the favorable gradient for potassium. So this, of course, raised the question, why would you want a single protein to be regulating these two alternatives in reciprocal ways, and we, of course, immediately thought of two possibilities. One is a cell volume control in response to osmotic stress. So cells are routinely exposed to different osmotic states and they have to respond acutely in order to avoid either turning, becoming pyknotic or exploding due to water flux across the membrane, and they, it’s known from years of prior work that they do this by acutely modulating their intracellular chloride level and they do this by modulating the balance between chloride entry and chloride exit, and has been known from pharmacoulogic studies that the, that this is mediated through phosphorylation of these proteins. In the phosphorylated state, the sodium potassium, the entry step for chloride is active, the exit step is inactive, and in contrast, if you dephosphorylate these proteins the entry step is inactive and the exit step is active, reducing intracellular chloride. There’s another state in which this is relevant, and this is in the neuronal response to GABA. As you know, GABA opens a chloride channel and as a consequence, the resting intracellular chloride level determines the response to GABA and it’s been known for a number of years that in neonates, ranging from rodents to humans, GABA is excitatory because the resting intracellular chloride level is high in the neonatal period and so GABA opens a chloride channel, chloride runs out and this makes GABA an excitatory transmitter, however, in most nuclei, in most of the brain in the adult, intracellular chloride level is low because the balance is shifted to marked increase in K-CL cotransport. So this got us interested in this problem and Jesse decided to pursue this in more detail, and so he has pursued this by developing and getting up and running, quantitative phosphoproteomics in the laboratory and the way he approached this was to make stable lines that had KCC3, one of the K-CL cotransporters, that he introduced at a defined site in the HEK cell genome, using the FLP recombinase system, and this K-CL cotransporter is under TET regulation so he can turn it on and off. It has a Myc Epitope Tag on it, so he turns on the protein, expresses it, purifies it by immunoprecipitation, cuts it out of a gel, does tryptic digestion and then puts it through tandem mass spectrometry, in order to identify sites that are phosphorylated. This technology has improved dramatically with the recognition that we can quantitatively enrich phosphopeptides on titanium dioxide columns, and this enabled Jesse to identify five phosphorylation sites on the K-Cl cotransporters shown here. This then led to the question, if you acutely expose these cells to hypotonic conditions, do any of these sites quantitatively change? In order to do this he used another new proteomic technology called SILAC which uses stable heavy isotopes of lysine and arginine to differentiately label proteins in different conditions. So you grow your cells in either normal media or these heavy stable isotopes, expose these to hypotonic conditions, these to isotonic conditions, lyse the cells, mix them all together and repeat the mass spectrometry, and you now have a beautifully experimentally controlled experiment where you can quantitate the level of these and compare them directly to one another, and compared to the nonphosphorylated proteins, the phosphorylated sites, three of them showed no change in response to the hypotonic conditions and two of them showed about a 33% reduction in phosphorylation and these two sites turn out to be interesting, because when you mutate these sites individually or together to alanine you can produce constitutively active KCC3 that is no longer dependant upon exposure to hypotonic conditions for activity. So this shows that this has worked in collaboration with, before bush and shows no activity, so these are individual wells and rubidium labeling in response to hypo, in response to brief exposure to rubidium. So there’s virtually no activity in the wild-type, the single amino acid substitution at one of these sites results in about a five-fold fold increase in activity. That’s equivalent to what we see with the wild-type cotransporter in hypotonic conditions. The second site individually has an equivalent response but you combine these two sites together and you get twenty-fold induction activity constitutive of activity that’s quite dramatic. There’s a, just as a control, one of the sites that didn’t change, if you mutate that to alanine, has no effect. So, these two sites that are, have this effect, turn out to be completely conserved among all vertebrate KCC3s and also completely conserved among all of the other KCCs 1, 2 and 4. Interestingly, the phosphorylation sites are sites, for no known kinase indicating that there’s work to be done to identify the specific kinases that regulate these sites. Jesse has tried to identify the phosphatases in vitro, phosphatases PP1 and PP2A and in vitro, are capable of dephosphorylating in the presence of calyculin A. You can antagonize the activation in hypotonic conditions with the wild-type cotransporters. The 991a mutation is now insensitive to calyculin A, suggesting it is the major site that is regulated by this phosphatase, but there are still a number of questions to sort out. We don’t know whether phosphorylation, dephosphorylation at one site promotes the dephosphorylation directly at the second. We don’t know whether the dephosphorylation events are stochastic or not. So does this really happen in vivo? Jesse has gone on to study this in human red blood cells, and in human red blood cells, using phospho specific antibodies, if you expose these cells to hypotonic conditions for five minutes, you see dephosphorylation at these two regulatory sites just like you see in the HEK cells, telling you that this is, in fact, occurring in red cells as well. Another nice technique that has come online with proteomics is the ability to not just to say yes, it's phosphorylated and no it’s not, but to determine the precise stoichiometry using multiple reaction monitoring using synthetic peptides that are heavy labeled so you can direct, you can correct for differences in ionization efficiency. So this enabled the Jesse first to ask, does phosphorylation regulate whether KCCs get to the membrane or not, and it turns out that it doesn’t. That if you bianttenuate cell surface proteins and purify them, you can see by the antibody staining that they’re already highly phosphorylated at these sites, but by MRM, Jesse has been able to determine the absolute stoichiometry of phosphorylation and that in both HEK cells and red cells, the 991 position is at least 90% phosphorylated in both, and the 1048 position is phosphorylated 40-65% stoichiometry at both of these sites. So, finally, in the brain, I, we posited that this, that these sites ought to be phosphorylated in KCC2 in the neonatal brain and dephos, and when K-Cl cotransport is low and dephosphorylated in the adult brain when K-CL cotransport is high, if this model is correct, and it turns out that it is, in fact, the case, that both of the sites, this just shows the Y and Z plots for one of the sites in mouse KCC2. Both of these sites are phosphorylated and in the neonatal mouse brain, we can’t do cell surface purification like we can with cells in culture for mouse brain, but overall there is about 20% stoichiometry of phosphorylation at the 906 site in neonatal mouse brain and this diminishes to undetectable levels by day 21 and persists into the adult. This work is unpublished but is coming out soon in Cell. So, finally we are interested in trying to tie this to the WNK kinase pathway, and other kinases, because we don’t know what kinases are responsible for the phosphorylation at these sites and so we’ve begun doing a large screen of RNAi of kinases to see which kinases will knock down phosphorylation at these sites. So far, we’ve completed analysis for six good candidates, spak and OSR1, as well as all four of the WNK kinases in HEK cells and so it’s quite simple, you expose the cells to RNAi, grow them for awhile, then induce expression of the protein, briefly and look for phosphorylation with our phospho specific antibodies, and to cut to the chase, the only one of these kinases who’s knocked down, suffices to reduce phosphorylation at these sites is WNK 1, but further work is required to determine how this works. We don’t think it’s likely that WNK 1 is actually the direct kinase that phosphorylates these sites that we’ve identified. So a lot of work remains to be done. So what I have tried to show you this morning is how taking relatively unbiased genetic approaches can identify new layers of physiologic regulation that we haven’t previously recognized and what I want to leave you with is the recognition that our genetic tools continue to just get better and better. So, I’ve chaired the oversite of the large scale sequencing program for NIH and when we started sequencing the human genome in 1998, it cost about 100 bucks to sequence about a 1,000 bases and we have beat the heck out of Moore’s law over the last ten years, because the cost of DNA sequencing has come down 14,000-fold over the last ten years, and we can now sequence all of the exons in the human genome for about $5,000 and that number is going to come down, probably, another order of magnitude over the next several years, and when we started talking about a $1,000 genome it seemed like a fantasy, but I think it’s actually going to become a reality much sooner than we might have guessed even a year ago, and as a consequence, the relevance of these kinds of genetic approaches to complex physiology, I think, are going to get better and better, and human as a model system, I think, is only going to get more attractable as we go forward, and I think underscores the ability of thinking about complex problems in human. So I’ll stop there and thanks very much for your attention. [Applause] [Walter] So Rick, thank you very much for an illuminating presentation, do you have some questions? [audience] Going back to the first of the slides, you suggest that somehow (inaudible...) close the door and you stop something from happening. (inaudible...) but i would like to make a little bit of an exception, (inaudible...) [Rick] So, the point is, if you take one thing out, it doesn't necessarily prove it's a rate limiting step, because there may be many different pathways that would do the same thing. And that's a great point. So, if i had merely shown you one gene that when mutated, caused hypertension, i would have a very hard time convincing any one in this room that salt reaborption in the kidney was an important rate limiting step. However, i didn't show you one, I showed you in the summary slide that taking an unbiased approach and looking for families all over the world, we have identified 10 Mendelian forms of high blood pressure, 10 Mendelian forms of low blood pressure, and we've identified the genes responsible in each of those, and they haven't been scattered randomly across the physiologic landscape. They've instead, focused on a single final common pathway that regulates how the kidney handles salt, and so that's how I have come to the conclusion that this actually is one of the major pathways that sets long term blood pressure in humans. But I think it's very important, you're point is a very important one and I accept it. It also, however, is worth pointing out, that reguardless, some of the targets that we identify that have large effects are actually attractable therapeudic targets, and are worth pursuing. [audience] What percent of hypertension accounted for by the gene you've identified? [Rick] So the question is, what fraction of hypertension is accounted for by the genes we've identified and this is a wonderful question. Because hypertension affects a billion people worldwide and these rare Mendelian forms affect a tiny fraction of the total population, but to address this point as whether there may be more common variations in these genes that affect blood pressure in the general population, Last year we've completed a study, resequencing some of these genes. In 3,000 members of the framingham heart study. This is a study that has longitudinal blood pressure followup over a 60 year period, and what we found was, we intentionally picked diseases where recessive loss of function mutations caused severe lower blood pressure in these rare patients, and so what we found was that at 2% of the Framingham population is heterozygous for these mutations, and these mutations acually have quite large effects on blood pressue, about 10mmHg at age 60, which is about what we would get from use of a single anti-hypertensive agent, and that in fact, is sufficient to confer a 60% reduction in the diagnosis of hypertension among the carriers, and further is sufficient to what appears out to age 80 there is not a single cardiovascular death and not a single stroke among the mutations carriers in the Framingham cohorts, and so there are about 100 million people worldwide who are carrying just those heterozygous mutations in just those 3 genes, and as the cost of DNA sequencing gets lower, and lower, I think we might be able to address these more broadly. [Audience] What’s your opinion on why there were no mutations identified in genes that we think regulate vascular resistance? [Rick] Yes. So, I think that’s a great question. Why don’t we see genes that affect vascular resistance, and I think there are two compelling alternatives. The first, is really simple. Genetics can's see the effects of mutations if they're lethal. So, if these mutations caused embryonic lethality, we would never see them as, in human genetics, and so I think that's a serious consideration. The other possiblity, of course, is the, the reason is that if you had primary effects on the vasculature that did not include the kidney, the kidney would simply adjust the, as we know it does in vivo, would simply adjust it's salt handling to return blood pressure to normal, and there have been quite elegant physiologic experiments that have done just this. If you cause an increase in systemic vascular resistance by reducing profusion, for example, to the extremities without effecting the profusion of the kidney, you don't change blood pressure in the long term, despite the fact you have had a dramatic increase in systemic vascular resistance, initially, the kidney adjusts intravascular volume, and these are experiments that Gieten and Hall did 30 years ago, and is quite convincing that the vasculature alone, without the participation of the kidney, is insufficient. On the other hand, if you tie off the renal artery or just impair the renal artery, you can enter in this same final common pathway by activating the renin angiotension system but impairing the kidney ability to respond to it by changing volume and there are humans who have renal artery stenosis who have hypertension on this basis. [Walter] So if we have no futhur questions, Rick i would like to thank you for your presentation, Thank You [Rick] Thanks very much [Walter] And know i would like to introduce to you George Dubyak. [George] So, our next speaker is Dr. Amira Klip from the hospital for sick children and the University of Toronto. Dr. Klip's trajectory is relatively simple, Mexico, to Switzer, to Canada to Switzerland and then back to Canada. Throughout all that time, though, she's really focused on a very fundamental problem, that has a lot of both physiological and pathological significance. How does insulin regulate the glucose, the transport of glucose in the skeletal muscle which, of course, everyone knows it's the largest depo for glucose. So this is alot of both physiological and pathological significance. But, but again, as most of you know, the biology behind this is a fundamental area of cell biology, exocytosis, endocytosis, putting a transporter into the membrane and retreiving it, and this has been a problem that cell biologists, endocrinologists and physiologist's have been studying for a long time, but I think, Dr. Klips work has really provided some really fundamental insights into this. Amira. [Amira] Thank you, and I would like to start by congratulating Walter and the department on their new initiation and I'd like to thank you all also, for the oppurtunity to share our work here. I think, just like George said, we are going to move from hypertension from it's ominous prevalence to a disease that, also, faces us, very much, as a population and as individuals, and that is diabetes, but i will, type 2 diabetes, in particular, but I will speak much more about how we understand insulin in action, and I will be happy, over question period, to tell you where things go wrong. The title I gave to this presentation is 'Bifurcations In Insulin Action' and if you bear with me through the talk, you'll see why this is not a linear pathway and what we know so far, but what we are going to try to do is get us from the insulin receptors to a molecule called GLUT4, which is the muscle specific, muscle in fat, specific glucose transporter that is in charge mainly bringing in glucose during a meal, but also, in bringing in glucose when we need to exercise, and regrettably, I wont be able to talk to you about exercise today. So you already have, the where there translocation, and the reason for this, is that, again, as George said, this is the molecule that we've known for twenty five years does the job, but we've known much less on how it does the job bringing in glucose, but it involves posttranslational cycling. I'll start from this side of the slide, GLUT 4 being this red squiggle. It is known that it populates a diversity of still poorly identified compartments. It's not purely the normal constitutive recycling compartment, it segregates a way in to specialized regions, specialized the proteins in those regions, and it cycles, and again, as George said, it rapidly gets in endocytosed and very slowly gets externalized, and it's this balance of endo and exocytosis that insulin and exercise dependent cues and hypoxia have utilized, to change acutely the levels of glucose transporters on the membranes so that you can acutely, again, bring in more glucose, and glucose is the main initial source, of course, of energy in to our cells. The glucose that gets internalized in response to insulin, it's fate is not so much to give us instant energy, but to be stored as glycogen. This is a prediction of how GLUT4 is folded because it is based on other transporters of the same family, but the point to make is that it is NC terminal side is facing the cytosol, and if this molecule is cycling, it will depend on it's regions that face the outside to really tag it when it's outside, and when those regions present very few opportunities for epitope tagging, and so as others have resorted to actually label those exofacial regions, and I'll show you in a moment, how that has helped our work. So, in addition to the transporter having very few regions to reach from the outside of the cell, skeletal muscle, which has mentioned, is the tissue that largely takes up glucose during a meal, is extremely complex and if you wanted to study it with cell biology tools or biochemical tools, where in all of these different cell types is GLUT4 in a subcellular level, you would have a really hard time to get to it. So instead, and perhaps a little bit as a cop out, and as a cell biologist, what we use is the muscle cell line that differentiates in culture from myoblast to myotubes, and that's what muscle cells do normally during muscle regeneration, or musclegenesis, and these cells are shown here basically, they look simply as your growing monocellular cells. They will form myotubes in other slides you will see, but we've stabley transfected into these cells, the GLUT4 molecule I was mentioning to you, in which is now an inserted myc epitope that will just facilitate it's tagging both biochemically or for immunofluorescence or electron microscopy. So now we have these muscle cells, they, I should say, have all the capacity to respond to insulin in terms of all the singling of elements, we will review them today, and now they are expressing this glucose transporter that will allow us to tag and ask where is it in time, and we've recently incorporated the same transporter, stably, in transgenic mouse in muscle only. So, we hope that eventually we will be able to go back to the mature tissue and be able to move away from the cell lines. But for today, let's focus on what these muscles look like. The myoblasts that convert to myotubes. The assay I've been mentioning to you that we used to detect the arrival of the transporter at the membrane consists of containing the myc epitope, and let me show you two examples, both of these are fluorescence microscopy. Here we have a few myoblasts, so magnification of this kind of field, and without metabolizing them, we are detecting GLUT 4 on the surface. And you can barely see anything on the surface when these cells are not stimulated. If you stimulate the cells with insulin, and these images are taken to ten minutes after given isulin, but within a minute and a half you can begin to detect GLUT4 on the surface. So it's a very robust response, it's very gratifying to detect it, you can quantify it. And this is another way in which we look at it, and we like to do it because it allows us to look inside the cell and ask, not just when did the transporters arrive but where were they coming from. So in this case, we have lifted out the myoblasts and now they look like your any cell in culture when you're passaging it and they go spherical for half an hour to an hour. When they are spherical in these cells, we have previosly transfected them trangently in this case with a plasma membrane marker, and this case, the GFP tag, k russ molecule, then you see the plasma membrane of the cells and the cells are permeabilized, these were not, these are permeabilized, and in red we were detecting where GLUT4 is, and so two things i want to point out to you. First, is that, when the cell is not challenged with insulin, GLUT4 is in a nugget that happens to be a perinuclear region. When the cell is stimulated with insulin, GLUT4 first disperses thoughout the cytosol into vesicles and more importantly, it fuses with the membrane, and you can again, score this insertion now by the colloquialization of the plasma membrane. So these assays are useful to us because they will allow us to not test just physiologically what makes this transition to happen, but also, pathologically, at what stage is GLUT4 halted when it doesn't get there, when it doesn't get to the membrane. But lets start with the signals that get GLUT4 to the surface, and I think it's been known for about, almost 20 years, that the insulin receptor, of course, is a tyrosine kinase in itself that has a target and typically everyone always talks about two targets, the insulin receptor substrates 1 and 2, very homologous molecules. They bind to the receptors to get tyrosine phosphorylated on mulitple sites, and I'm sure you all know this. This attracts PI 3 kinase class 1, PI 3 kinase, which then exerts a series of signals that we will go over later on. So one of the first things we wanted to know is, given that both IRS 1 and IRS 2 are present in muscle cells, they both get phosphorylated, do they do equal jobs? And to approach this, we took advantage of the fact that the muscle cells are very easy to transfect with siRNA oligonucleotides, and in this way, we can knock out specific genes. You can appreciate here that we can knock out specifically IRS 1 this is siRNA IRS 1, IRS1 blot it goes down, or we can selectively knock out IRS 2 and then you can see the IRS 2 levels going down. In the graphs below, we have used either IRS1 to knock down cells or IRS 2 to knock down cells, to inquire, what is the response of GLUT4 to insulin, and let's just focus for simplicity, at what happens at 5 nanomolar insulin. What we are showing you here is the normal response of the cells, this is the GLUT4 value of translocation when the cells have IRS 1 and this is what happens when cells don't have IRS 1. You really impair significantly the ability of the muscle cells to bring in glucose transporters to the surface. This was done using the assays I told you before, if the cells have diminished IRS 1. But of the same concentration of insulin, cells that have diminished levels of IRS 2 couldn't care less. And through a series of analysis of other outcomes of insulin, we now can confirm that insulin works through IRS 1 to lead for GLUT4 translocation and not through IRS 2. This is interesting because when we did analyze other outcomes, not just GLUT4 translocation, we found that the mitogenic response to insulin is via IRS 2. So right at the first step, there is differentiation of function. There's another important thing about IRS 1, of course, and that's if this molecule, but IRS 2, loves to get phosphorylated by a series of serenine and threonine kinases and all of those have one thing in common. They act on many sides, but they are going to lower the ability of the insulin receptor to tyrosine and phosphorylate IRS 1. So IRS 1 is not only the first molecule that recieves the insulin signals that we will be required to lead to you to glucose translocation, but right there saturated fatty acids, cytokines and other conditions that may activate any of these kinases, is shown here, has the ability to stop things at the core. Now today we are going to move downstream of IRS 1 and see what are the other signals there for that emanate of IRS 1 and PI 3 kinase. So, I have mentioned to you that IRS 1 binds PI 3 kinase and I want to elaborate on that. There are, of course, numerous studies in the literature, but I want to convince you that there are two important branches of insulin signaling downstream of PI 3 kinase, and I'm going to paint them all on this graph now, this slide now, and we are going to go one step at a time to see how this was demonstrated. So, I'm going to try to convince you that does PI 3 kinase Rac, Rac 1, is GTP loaded. That this among other things, but presumably because of the change in activity of actin remodeling is goin to change the dynamics of where molecules are in the cell. I'd like to show you another player in this signaling, and the molecule that binds to remodeled actin, and that's called actinin 4, and on the other side of this story, I'm going to show you that dousium of PI 3 kinase we have the classical activation of Akt through PDK 1 and 2, and so we are going to talk about two different modules, modules that are downstream of Rac and moduling involving Akt. There's been alot of finding, very exciting findings, of what comes downstream of Akt, and just for anybody to ask me, if atypical PKC's, which are also activated by insulin and some cells participating in this pathway, i can tell you that, although the jury is still out, the majority of the evidence suggests that this might be, if anything, just an outlier and not leading to GLUT4 translocation, but instead what we know is that this Akt molecule or PKB, phosphorylates on four very important sites on the protein called AS160, it stands for Akt substrate of 160,000 molecular weight. And AS160 is the gap Rab, and today I'll tell you how it works on what we think is Rab 8 to perhaps interact with a motor protein, and this motor protein interacts also with actinin 4. So we think that not only we are identifying divergences but also to convergent in how all these steps lead to GLUT4 translocation. So now I need to prove to you that at least some of these step, indeed, worked as advertised. First of all, Rac activatation. The assay shown here for Rac activation is one that many of you might have used. It consists of pulling down active molecules of Rac, only Rac molecules that contain GTP by verge of their binding to a peptide that has affinity for GTP bound Rac, and so any time you will see by this Rac pulled down assay a positive signal, it says that Rac is GTP loaded. You can see here the GTP response to insulin at 5 minutes, but again, this can be seen at 1 minute already, in response to insulin 5 or 10 minutes you can see Rac activation. The other lanes are controls and I won't spend time describing them to you. We also have a postive control that indeed we are looking at a GTP bound form of Rac. So insulin rapidly activates Rac and why is this imporant? Well, you can test this in two ways. One of them is, you can transfect the cells tangently with either wild type Rac or dominant negative Rac. And then you can aquire what happens to GLUT4 translocation. Here, we were quantifying GLUT4 translocation and here I'm just showing you an image that is easier to follow. So if we are looking at this 4 images here, what happens to the surface GLUT4, these cells are not permeabilized, this is the basal conditions, these is the insulin stimulated cell. Normally, response, you can see in this cell that is there in both places, in both squares. There is a cell but you can't see the surface GLUT4 because it didnt really get increased with insulin, and that's because these cells are expressing dominant and negative Rac, and for those of you who care to see the contification of similar assays but in monolayers, you do not cause translocation of GLUT4 to the surface when the cells express dominant and negative Rac. It's interesting, because if you express wild type Rac you may, in fact, already potentiated beta-insuin response. So this suggests that Rac in some way is important for GLUT4 translocation. This is another way of doing it. If you knock down Rac at this level, of decrease of expression of siRNA, the normal response of GLUT4 going to the surface is also markedly reduced. I dont think I have it shown in this graph, but if Domininat and negative Rac, or siRac do under the same conditions is a preclude actin remodeling, and I'm sure all of you expected that because that's what Rac does, it reduces actin remodeling. In a slide or two, I will show you what that actin remodeling does, but what I first wanted to bring the question, how do we know that this either downstream of PI 3 kinase or, as we've been painting it, parallel to Akt activation. Well that it is downstream of PI 3 kinase was actually shown here when you look at Wortmannin and when you get Wortmannin, you markedly reduce but dont obliterate the ability of Rac to get GTP loaded, and so we think this credence to the fact that, at least to some degree, PI 3 kinases required for Rac activation. There are gaps for Rac that are pip 3 sensitive What about independance or dependance on Akt? Well, what you can see here is that for one thing, we are painting Rac separately from Akt, because when we knock down Rac, as it is shown in the last two lanes of all of these gels, we knock down all the known signaling from Rac, which are shown here for PAK and LIMK kinase, but we do not affect the insulin dependant phosphorylation of Akt. This is no insulin, this is insulin, and this change is the same as the control cells, yet these cells in the last two gels in each block do not have Rac. So we think that this tells us, that you do not Rac for Akt to get phosphorylated. We also have evidence for the converse, that I'm just going to tell you about. That is that we either knock down, sorry, either use dominant and negative Akt, or have used very effective Akt inhibitor and under those conditions Rac couldn't care less, and those are the reasons why for putting both of them as downstream of PI 3 kinase, is well known, of course that Akt is downstream fo PI 3 kinase, as well as, in a bifurcation pathway. So, let's continue with what's downstream of Rac in terms of leading us to GLUT4 translocation and I'm not going to focus on these two molecules but rather, on a consequence of these and other molecules like those, that is remodeling of actin. I've been talking about remodeling of actin, this is what it looks like. In all these micrographs, what you see is only the filament of actin. These are segments of actin myotubes. Myotubes are many hundreds of microns in length. This is a peice of a myotube, these are these multinucleated cells that the muscle cells like to make. Filamentous actin as stress fibers. Insulin stimulated cells form a mesh in dorsal region of the cell that is prevented by Wortmannin. This would be the same type of response but in myoblasts or in these rounded-up myoblasts where you can see the cortical actin forming hybridizations in response to insulin that are sensitive to Wortmannin. So, indeed, insulin causes actin remodeling and all of this, this remodeling is sensitive to Rac. If I had shown you the dominant-negative Rac, or siRac, you wouldn't have seen any of these manifestations, and as you can imagine, and this is very old work, if you use any of the usual sledge hammer type of approach to prevent actin dynamics, you prevent the gain in surface, GLUT4 at the surface. This are cells pretreated with cyto-Case D or latrunculin B and this is the visuals of it. This is again, surface GLUT 4 in an untreated cell, barely you can see any insulin response pretreatment with latrunculin B, GLUT4 doesnt get to the surface. So I think I've told you, you need Rac because actin remodeling, you need Rac because for glucose translocation, you need actin remodeling for glucose translocation. So what's the relationship between GLUT4 and actin? So we collaborated with Jen Hardwick and what we are showing here are images of the cortical mesh just beneath the membrane, and that's those big filaments throughout, and if you look at the 100nm scale micrograph here, you might be able to see the, depending on where you're sitting, black dots, and those black dots are gold labeled myc epitopes, gold labeled antibodies, gold labeling myc epitopes on GLUT4, and you probably can see that they are all positioned along the filaments. So we think that in some way, the actin mesh, just beneath the membrane is, at least physically, close to where the vesicles are, but of course we wanted to analyze this relationship between actin and GLUT4 in a more protein meaningful way, close to where the vesicles are, but of course we wanted to analyze this relationship between actin and GLUT4 in a more protein meaningful way, and we just heard from Rick about they're use of SILAC. So we use the SILAC approach, as well, not to look for phosphoproteins but to or phosphorylated sites, but to look for proteins that may associate with GLUT4 in insulin dependent fashion, and just like you heard before, and this was the work of (unknown) and Leonard Foster when they were in my lab. You can carry on parallel cultures labeled with leucine or with deuterated leucine, and you can indistinctly label, stimulate one with insulin or one with the other, of course you need to know which is which, and then you lyse your cultures. You make a single immunoprecipitate of GLUT4, in fact, we immunoprecipitate via that myc epitope, and you analyze your proteins by mass spectrometry, and the strategy to understand what's going on, is that any proteins that are going to have in this case, more deuterium than hydrogen will be proteins in which insulin has increased their association with GLUT4. Those that don't change, don't tell us anything. They might be background or they might just be proteins that don't change in their association with GLUT4 with insulin, and those in which we have less red signaling, in this case, less deuterium than hydrogen, are going to be proteins that insulin use to decrease their association with GLUT 4. So the next table is impossible to see, but I've highlighted the box at the top, which tells us the three proteins whose association was increased by insulin. They happen to be cytoskeletal proteins, but not actin, in fact, but proteins such as actinin 4, and we are going to follow up on what happens to acinin 4. Interestingly, the proteins that come off with with insulin include glycolytic enzymes and we've also followed that up for other studies that I will be happy to tell you later. So what is actinin 4? This protein that increases it's co-precipitation with GLUT4 in response to insulin. It's a form, sorry, it's a protein that forms homodimers, and in this case you can see that they have an actin binding domain and they have spectrin repeats. These ACTN4 dimer is known to mediate membrane to actin binding, so it is a link, a linker to membrane proteins, diverse membrane proteins, including some protein transporters, and so this was an interesting question for us to ask: Is it possible that it binds through it's actin binding domain to actin and through what we just demonstrated, that the co-precipitation with GLUT4 to GLUT4? And before going into analyzing this biochemically, we said, do we even care? And so what we did what we like to do best, which is let's knock down this protein, but before, I'm just going to show you the co-precipitation of GLUT4 and ACTN4, not just from the mass spec table I just showed you, that you couldn't see, but rather here through immunoprecipitation. So just look at the gels. What you can see here is that we immunoprecipitate via the myc epitope, from, just look at the myoblasts. The GLUT4, if the cells were not treated with insulin, there was no actinin coming down, but if the cells were pretreated insulin, there's actinin coming down with GLUT4. So, indeed, biochemically, you can GLUT4 and ACTN4 like to come down together, so what happens when you knock down ACTN4? When we knock down ACTN4 to this level, which is about 60-70% only of expression, knock down, we obliterate the gain in surface of GLUT4 that insulin would cause. So ACTN4 is an important protein for GLUT4 translocation. And I'm going to skip this image here because it says the same thing. So we wanted to know, where is it working? Is it working upstream of actin remodeling? Is it working on the Akt side of things? So first answer is that it is not working on the Akt side of signals, those signals that I had Those signals on the far end of the slide. Why? Because when you look at Akt phosphorylation, in response to insulin, and I apologize, this is now plus insulin minus insulin. So this Akt phosphorylation couldn't care less if ACTN4 is expressed or not. So this tells us that whatever ACTN4 is doing to block GLUT4 translocation doesn't involve Akt. And now we ask, is it upstream or downstream of actin remodeling and so what you can see here, red doesn't show up very well with the lights but we will leave it like that. What you can see here is these images on this side, these 6 images. In green we have where ACTN 4 is, in red we have actin, and in the merged ones we have perhaps you can see that when actin is not remodeled, when there is no insulin, the 2 proteins are not together. When actin remodels, actinin follows into the actin, but more importantly, in cells that do not have ACTN4, and that's why these are black and there's no green signal, actin continues to remodel. I hope you can make out the mesh of actin compared to the basal cells. So this tells us that ACTN4 must be downstream of actin remodeling because if you eliminate ACTN4 actin still remodels. And you can see that better in these images where we also ask the question, Where is actinin, visa vi, actin and, visa vi, GLUT4. So we are going to go through these slowly and I am just noticing how poor, perhaps, this is projecting things in terms of colors. (Ah, that's great, thank you, so this perhaps is the only time we may need something like this) (That's great, Thank you very much.) So, let's look at the images just under insulin in each case, for the interest of time. This is, these 3 panels here are in white, but it refers to the localization of ACTN4, actin, and GLUT4 in an insulin stimulated cell by immunofluorescence of regions where actin remodels. You can find ACTN4 and GLUT4, and therefore if you look at a merged image everything looks between yellow and white. What happens in cells that do not have ACTN4? Well, we now don't have a singal here, let's just follow the insulin, acin stil remodels, as I told you before, look at it compared to a cell that doesn't have insulin, actin remodels. But there is something very interesting here, unlike the normal cells where GLUT4 tracked with actin, in the insulin stimulated cells that don't have ACTN4, GLUT4 doesn't track with actin. This tells us that potentially, ACTN 4 allows GLUT4 to be at the same sites at the remodeled actin. (That's great, we can go on) So the model we are entertaining is that we have this signaling arm of Rac activation, downstream of PI 3 kinase, leading to actin remodeling, and in some way, actinin being downstream of actin remodeling, and I showed you actin and co-precipitates with GLUT4. So we imagine this arm is important for physical contact of glucose containing vesicles where actin is. So, I'm going to take us now to the other side of the question of the equation and ask, not so much what's upstream of Akt or what's Akt itself, but what I want to tell you about this molecule that you may have heard less of, which is AS160 and whatever is coming downstream of it. So what is AS160, I told you is this gap for rabs that is a substrate of Akt, that's why it is called, like I said, 'Akt substrate of 160,000 molecular weight' it gets phosphorylated by Akt, and being a gap protein, what leanhardin collaborators have hypothesized and several groups, including ours, have proven to be the case, is that, maybe this being a gap, what you need to do is shut it off so that now whatever rab is target can prevail in it's GTP bound form. So let me show you what happens if we express an AS160 mutant that cannot get phosphorylated. The hypothesis is that phophorylating AS160 is going to inhibit it's gap activity. I have to tell you, however, nobody's been able to detect the gap activity towards a full lenth substrate. So right now, this is all a little bit based on information when you use a small region of AS160 towards rabs. But let me introduce to you the AS160, the molecule. The molecule has all these red marked sides for phosphorylation by Akt. They all get phosphorylated in response to insulin and what we have hypothesize, and Leonard has hypothesized, as well, is that these phosphorylations some how may inhibit the gap domain, so that now you would have a rab target that would remain in the GTP bond, if the hypothesis is as if AS160 upon phosphorylation, can no longer to it's gap activity. So how do you prove this? Well, you can do this in two ways, one is you can express overexpress in AS160 molecule that cannot be phosphorylated, and in this way you would eliminate the ability of insuin signaling to shut off the gap. So, it would remain always leading rab towards a GTP bond form. That is, bears correct when you do the experiment. I'll show you it in a moment in the next slide. So in addition you it in a moment in the next slide. So in addition, the hypothesis that I keep saying, is that insuline by causing it's phosphorylation would inhibit it, and Rab-GTP would prevail. The mutant would not be able to be shut down because it wouldn't be phosphorylated, so you prevail in Rab-GTP. The other prediction is that if you do a double mutation, one in which you cannot phophorylate the protein, but actually the protein was not a gap anywaays, that under those conditions, the phosphorylation was not going to do anything. So, if you want to think about it, and I tell my group this, think of this protein as a 'hand brake'. If, what insulin needs to do is remove the hand brake, but if there was no hand brake to start with, then it doesn't matter whether it can or cannot phosphorylate it. So, let's look at the actual results, and this result tells you very basically, what happens when you can't phosphorylate AS160, but by overexpressing these mutants that we call AS160-4P, and in reality, it should be called 4A, because it has 4 alanines in the site that gets phosphorylated, but that's the name it's called. So if you overexpress this AS160-4P and if you only look at these panels here under 'surface myc', this is what happens in the basal state, this is a normal response, this is a normal response to insulin, no surface GLUT4, more surface GLUT4, and if you just look down below, the response is totally obliterated in cell that over express the mutant, AS160. So you need the phosphorylation of AS160 in order to, presumably, remove a brake and now be able to have GLUT4 translocation. So, so much for AS160, I am just going to give you a snapshot now of what is known about AS160 and it's possible targets. If AS160 is a gap for Rabs, how do you find out which of the many many Rabs in the cell are required, are it's targets? So there were a couple of studies that made some educated guesses. One is, as I have mentioned before, that the TBC domain, the gap activity domain of AS160, targets only a handful of Rabs. Those are 2A, 8A, 10 and 14. Now this is interesting, because 2A is in the ER and we probably can forget about it for the purpose of GLUT4 translocation, at least in terms of establishing a test. So 8A, 10 and 14 are interesting candidates. On the other hand, other investigators have isolated GLUT4 containing membranes from inside the cell, and have asked by mass spectrometry, what Rabs are present there? And this is the interesting thing, 2A, 4, 8A, 10, 11 and 14 show up, here underlined, our very same candidates. So then it's a no brainer where you are going to store it, and we decided to knock down, one at a time, Rab 8, 10, and 14 and we also asked, can we in fact, do a gain of function strategy, and I just noted last night that I did not bring the results of this experiment so I will just have to tell you. If you inhibit, if you express the P4 mutant, and the one I just told you about, if you express the mutant and translocation can no longer proceed because we assume you cannot keep the Rab in the GTP bound form, however, now you over express either wild type or constitutively active Rab 8A or 14, we could rescue with 8A and 14, but not with 10 the translocation. Im so sorry, I don't have the slide, the slide I do have here for you is what happens when you also try to ask the question, independently of AS160, if you knock down 8A or 10 or 14, what happens to GLUT4 translocation? So here you have evidence that we can knock down, selectively, 8, 10 or 14. These are controls of signaling, we wont spend time on those Let's look at what happens here below. All of these are cells that have no insulin, all of these are cells that recieved insuin, and the bars that did not amount a good response are those from cells that have diminished levels of Rab 8A or of Rab 14, and regrettably, together we didnt push things too much further. So this tells us that Rab 8A and Rab 14 independently and sufficiently bring down GLUT4 translocation when they are eliminated. So this brings us to just the final question in the presentation that I want to give you, and that is, if Rabs 8A and 14 are important, what do we know about them? I think I have shown you already that we inhibit GLUT4 translocation if we knock down Rab 8A and 14, just showed you that, and I mentioned to you without the data that the inhibition by AS160 mutant is rescued by Rabs 8A and 14. So, what do we know about them? We a going to focus on Rab 8A. So far, it's the only one we have analyzed and what I can tell you from the literature, just in the last two years, we found out that Rab 8A interacts with myosin 5B and this by yeast two hybrid system. That made it very interesting because myosin 5B is the motor and we thought if we know have a motor and we now have vesicles that bind to actin, perhaps, we're begining to close the loop of the signals and the motors, and that's of course what myosin 5B family looks like. These are processive myocin molecules. Myocin 5B, as I have mentioned to you, interact with Rab 8, in this case, Rab 8A. I'll just show you the last data slide. We tested the possiblity that myocin 5B, through it's interaction with Rab 8A, might be involved in this gain of surface GLUT4 by insulin by expressing a fragment, this fragment or this fragment, these are fragments of myocin 5 B. The fragment that is cotailed contains a region between amino acids 1221 and 1384, contains a region that binds Rab 8A. So this is a fragment that can still bind and endogenous Rab 8A. This other fragment no longer has the ability to bind Rab 8A, and as you can imagine, we did this with the intention to use this fragment to mop up endogenous Rab 8A and this fragment as a control, and if you just look at the graph here, I hope you can see the gain in surface GLUT4 caused by insulin was reduced in cells that express tail but not reduced in cells that express the other fragment. So, we think that this gives us, at least, evidence to follow up more on this possibility and I have to really say, this is really a model that guides our thinking, I don't think any of these have been fully confirmed and probably other ways are required for each step, but what I think from what we have done so far we think that insulin binds to it's receptor to activate through IRS1 to downstream, cascades one that includes Rac and actinin and ACTN4 with links, to translocation. Very interesting data in the literature, not from us, show that ACTN4 can interact with myosin 5, in fact, myocin 5B, which of course is the molecule that we are arriving at through this other cascade, in which I think I showed you that downstream of Akt there's a target protein, AS160, that is a Rab gap, that if you prevent it's phosphorylation, you probably prevent a particular Rab getting loaded with GTP. This remains to be measured but the target Rab that can be shown in vitro to be affected would be Rab 8A and a couple of others and we know from the work I just showed you that Rab 8 interacts with myocin 5, it's Rab 8A in this case, myocin 5B, and we think this interaction might be important for the translocation because when we express the peptide that interferes with this interaction, and I should say, was shown by flourescence the they interfere with that interaction, then we abrogate GLUT4 translocation. So this is, in a little bit of a big mouthful, what we think is underlined the movement of vesicles containing glucose transporters to the cell surface. I want to leave you all guys with one more word and that is What does this have to do with diabetes? Well, type 2 diabetes is preceeded by insulin resistance, and yes, you only develop the disease, as diabetes, with hyperglycemia as hallmark when the beta cells from the pancreas no longer resist the insulin resistance. But you can have insulin resistance for 10 years as part of the metabolic syndrome and that is going to be an absolutely necessary preamble to type 2 diabetes. What is it affecting? Type 2 diabetes, as far as we know today, in all the many cohorts of patients that been analyzed and up to 1 in 20 patients has type 2 diabetes in the Western world. No mutations have been identified in any of these molecules. All the way from the insulin receptor to GLUT4 translocation. Yes, we do have just like in the case of hypertension, selective monogenetic, few, few, few patient around the world that are affected by a few, and not even quite glucose in this pathway. So what is a large number of patients suffering from? These are not, at least not single gene mutations that are loss of function mutation in any of these particular molecules. Yet, the whole signaling pathway is toned down, and I'd be happy, over lunch, perhaps that's fitting, to discuss with you how nutrients tone down that signaling, but also, I think it was necessary toh ave this platform of targets to understand then how our nutrients, cytokines, stressors are going to challenge signaling, because ultimately, insulin resistance to type 2 diabetes is defined as the inability of GLUT4 to get to the membrane in response to the hormone. So I'm going to leave it there and I just want to, I'm just going to skip this, cell biology part and I'm going to thank the members of my group that have done the work I told you inparticular in today's work Ilana Talior worked on ACTN4, Tim Chiu and Alex Koshkina on Rac, Varinder Randhawa on AS160, and the rest of the group are very heavily involved in trying to understand insulin resistance. So thank you very much. [applause] [George] yes [audience] The microtubules are known to be involved in traffic in a variety of molecules, and also your model shows the myosin 5 and and (inaudible) translocating this GLUT4, and this myosin 4 is not to literally walk on microtubules. So you haven't mentioned the role of of these microtubules in translocating this GLUT4 to the cell membrane. And also the second part of my question is, do you think our the actin myocin, the actin cytoskeleton roles are limited to stabilizing this GLUT4 in this cell membrane, other than translocating this cytosolic GLUT4 to the cell membrane? [Amira] So, absolutely, and although this again is a diagram, this conforms exactly the molecule you're just describing, based on data I don't have time to show you today. I can show them to you later on where is GLUT4, why is the actin cytoskeleton needed, for to bring things or to keep them near the membrane? But if you follow this cartoon, this is based on work from different labs than our own. We believe that the vesicles are going up and down on microtubules, perhaps, in an idle way, and we think that myocin 5B is shown here as myocin 5A, 5B, we think that myocin 5B allows the switch of tracks, precisely as you were mentioning, from microtubules to actin. This is based on other things that are known in the literature about myocin 5B, and we think this is under the control of signaling through Rab 8A. So the signaling to Rab 8 may allow myosin 5 to really bind on the vesicles, and make them change tracks. You may need to prove this in a mechanistic way. We think that at the same time, in response to insulin, Rac is causing the actin remodeling near the surface. ACTN4 is in that actin mesh and we think this is going to help, perhaps, stabilize their position, or guide GLUT4 near the membrane, because the next step has to be to find productive snare complex formation between VAMP2, which we know is on the vesicle, and syntax N4 and snape 23 is on the membrane for final fusion. We think that, perhaps, signaling downstream of AS160, not through Rab 8, but perhaps through Rab 14, we don't know what might be involved in the fusion, This is based on work from other groups that find that AS160 is important as much for the fusion. Our work, that I can show you later shows that it is important, myocin is probably involved in the stabilization near the actin and that Rac is required for that retention. So, just like to describe, that's our way of thinking. [audience] Is there any evidence that any of the insulin promoting cytokines can target any of these upstream proteins, mostly focused on IRS proteins? [Amira] Absolutely, Im glad you asked because the literature will largely focus on IRS1 and you will almost find any study [Amira] Absolutely, Im glad you asked because the literature will largely focus on IRS1 and you will almost find any study looking at insulin resistance showing you that Akt is reduced by about 50% and almost any model, whether it's cytokine reduced insuline resistance, fatty acid reduced insulin resistance, and yet, if you deliberatly inhibit Akt only by half, which is what is found in all those studies, nothing happens to GLUT4 translocation. This arm is extremely resilient. So we have done a series of studies using cytokines, using saturated fatty acids, seromides, and what we found was that actually, Rac was exquisitely successible. Under conditions where IRS 1 bindind of PI 3 kinase is normal, phosphorylation of IRS1 is normal, we find Rac-GTP loading being quite susceptible. So we believe that if nothing else, it's an arm that is important to pay attention to and also it's an exciting arm because if it has a protein, it may interact directly with GLUT4, it might give us some ability to improve such interactions. But you're totally right. Everything included in the literature that looks like insulin resistance to the point of signaling says, IRS1 phosphorylation is abnormal, Akt phosphorylation is abnormal, this is never abnormal by more than 50% and that's inconsequential. [audience] And so, you obviously did a beautiful job at dissecting the GLUT4 translocation pathway but you alluded at the end to the observation that we haven't found many genes in the genome wise association studies that contribute to insulin resistance as a causation of diabetes, and I'm wondering how you are thinking about insulin resistance, whether you think it is contributing to diabetes or whether you think it's actually adaptive in response to the overfed stake. [Amira] Yes, so it's an excellent question, so if you bare with me for 2 minutes and you all are happy to stay for lunch. This is a current thinking, not really my particular view, but I do think there's a lot of truth in it. Obesity linked type 2 diabetes has a large component of cytokine assault to the muscle, and this whole pathway is very much cytokine sensitive. Where are those cytokines coming from? They are coming from macrophages largely, that are coming to remodel and then the adipose tissue and in the course of it, they can get activated into the M1 mode and they start to spill out inflammatory cytokines. So inflammatory cytokines and fatty acids together affect signaling so this has nothing to do with specific genes, it has to do with specific nutrients. It's over-nutrition, as you were mentioning, that's in the case of obesity induced insulin resistance. Now, obesity does have, and very beautiful work from Steve Riley and others have shown, as I'm sure you know, obesity does have those trapped with certain genes. Some very few ones are monogenetic, monogenes, and the other ones are, of course, more complex traits. So, our ability to consume energy, our ability to expend energy, through exercise, through fat oxidation, all of those are probably targets. So it's really fine tuning of metabolism that is probably altered in all these individuals that are going to have high levels of fatty acids and high levels of cytokines, and that's going to, as you said, secondarily, act on insulin resistance. So do we know of primary genes causing insulin resistance? No, not yet, in any case. Do we know is this secondary as in adaptation? The insulin resistance is happening as a result of the overfed state, it is a preamble to type 2 diabetes, but when the gene studies have been done to find genes that lead to suceptibility to type 2 diabetes, all of them point to beta cell targets, Logically, you are defining the disease as the moment when the beta cells fails. It doesnt say you need a preamble of insulin resistance. [audience] So this raises the question, if you could dramatically increase GLUT 4 translocation in patients, the typical obese diabetic, do you think that would be beneficial or do you think it would be maladaptive? [Amira] Well, in animal models, it has been shown that you can overcome insulin resistance caused by beta cell attack. So it's not a model of type 2 but it's for the same reason as in type 2, so if you use streptolysin or (unknown) etc.., in those mice that have that over expressed GLUT4, you prevent the rise in glycemia, and the reason for this is the beta cell gets involved because the insulin resistance exerts high demand. If you don't have that high demand, you aren't going to have that beta cell involvement. So, I would argue that this would be beneficial where there are alot of targets in a chemical trials right now, in clinical trials are of chemicals that increase fatty acid oxidation, as I'm showing you know. So, agonists because by increasing fatty acid oxidation, within limit, you are going to reduce a circulating levels of fatty acids. So, probably at any level there is an oppurtunity to intervene but it's more metabolic. [Walter] So with that we thank you Dr. Klip and think about that for lunch. [applause]