 My talk will have minimal mathematics in it, which may be a relief for a few of you. I'll mention a few things which might be more mathematically applied, investigated by not by us, but by others. One thing I'm delighted to participate in such a broad topic spanning from molecules to human health. And taking that at heart, I will present actually two different stories. I'll warn you these two stories have nothing to do with each other. So if you miss the transition, I'll try to make a warning when I make it. But there will be two different talks. I want to thank Boone for pointing out the distinction between these talks that we're going to be talking about interactions. And we're very much interested in regulatory protein interactions. And the reason I should say that these interaction networks end up putting order of the regulatory molecules in the middle is that they interact with everybody. It's not uncommon to find proteins that have 100 interaction partners. And the reason these interactions are so critical for these regulatory proteins is that the regulatory proteins lack identity. They lack specificity individually. They acquire their specificity through their interactions. And it is that very process of acquiring the specificity that we're interested in studying. The reason we use direct imaging is also because for us, the localization, where something happens, the responses to stimuli when it happens relative to other processes in the cell, those are the main part of the biology, if you will, of regulatory proteins, it is not sufficient to know who interacts with whom. We need to know what controls the interaction and where it happens in the cell. And we have a number of examples in the lab of studying these kinds of problems and finding out through the investigation of the interactions that we were completely ignorant of the whole area of biology in the process. So the two stories I'll be telling are one done by Huai Deng, a post-doctor fellow now at the University of Minnesota, whose expertise in Drosophila provided the opportunity of translating these imaging approaches, which we've done in living cell systems for many years, into live animals. And the other one, which is going to be our attempt at making some impact on human health, is in the area of developing drugs for rare cancer by Veronica Burns and Yunhui Cheng. My introduction, since I have two very different topics, I apologize, it's somewhat presumptuous. But bear with me, I'm going to start with asking a question, which to me has been puzzling. I think most of us in this room who have been at our area of investigation for some time are proud of the advances that we've made over the last five or 10 years. We think of things that happened 10 years ago. They were really hard. And now we can do things that are amazingly much more than we could 10 years ago. Now, if we compare with what we're doing in many basic sciences to what's happening in drug development, this is a plot of a number of drugs approved. Please apologize, a rather ancient slide. But the picture hasn't improved. Despite enormous effort as measured here by a number which is beyond my sort of comprehension in terms of financial impact, the success rate of drug development is woefully low. And the puzzle here is why is this? We are making such enormous advances in science. Why is this not translated into patient benefit? And this is a too big question for me to really answer, let alone to solve today. But I want to give some flavor of the reasons for this and perhaps where things might go. Now, this is sort of the standard drug development program, if you will. And this is not the scale. But nevertheless, the point here is that we're starting with identified lead compounds through some large-scale screen. And then through these individual steps, we screen out the ones that do not fit some criterion. And the reason this is not the scale is that this is, in fact, a much less efficient or much lower yield process than we know about. The ones that we actually have numbers for, the ones that involve US federal registration, it's an order of magnitude at every step. And it's at least as bad at the lower levels, which means that we're simply losing almost every hit along the way. And we have no solution for improving this. Well, the solution at the lower levels is, in fact, already solved. We can easily screen. We're efficient at going through large numbers of compounds through these steps. The problems happen here at the higher levels, where our ability to make any sensible prediction, any sensible analysis of what kinds of processes are involved is really limited. So one of my first stories is going to focus on the era of pharmacodynamics, that is to say, how animals respond to foreign compounds. And of course, the targeted effect is the one that we are striving for in terms of health and benefits. There are many non-target effects. And what I'm focusing on today is these induced response pathways, which is something that is initially rather surprising, that all of us, all of ourselves, have the ability to recognize foreign things. It's almost like an immune system at a molecular level. At the molecular level, what's amazing about these systems is that they're able to recognize the chemical characteristics of foreign compounds and respond in a purposeful manner, typically by inactivating, exporting, somehow getting rid of the foreign entity in order to protect the cell, because of course, things that come from the outside is generally bad for us. It's hazardous. However, drugs, of course, we have to get into the cell in order to have an effect. So we have to somehow understand and overcome these pathways, which are, again, one of the steps in this drug development enterprise, which are really failing. It's these later steps in this pathway, if you will, that are the reason why drug development, they become really expensive here, so we cannot afford being so ignorant about what's causing the failures. Now, from a biological point of view, there's a lot, and I don't expect you to read this list. It's basically an admission here that we're looking at a very small part of this picture. There are many biological mechanisms that protect us from the outside, being in our skin, the separation of cell from the environment. What's amazing about these things is that many of these things can discriminate newly synthesized compounds. Compounds come from chemical factories, things that have never been present during evolution. How did evolution develop the capacity to distinguish these kinds of compounds that we're very familiar with many kinds of biological stimulus response pathways? We can recognize much of normal biological stimuli because they have been present over eons, billions of years in some case, and have had all that time to fine tune the response. But how about the next drug, which has never been present during evolution? How can ourselves recognize it and respond in any purposeful manner, not to speak of the pesticide or what other compounds that we're exposed to? And this is the problem that we've been really interested in addressing, and because it's such an evolution development in my mind are two sides of the same coin. So we've studied this in a developmental context in Drosophila. The model system, and again, I emphasize that this is a tiny little part of this big picture, is this system which has been characterized for xenobiotic responses, xenobiotics being compounds that are artificial, which is based on a protein called KEEP-1, which acts as a sensor, and it is dedicated to control of this transcriptional factor, NRF-2, and under basal conditions to degrade it. And I'm going to admit right away that I'm putting up something of a straw man here with the purpose of knocking it down. There is much data to support this model, and I don't want to say that this is wrong merely incomplete. One thing that's amazing about this sensor KEEP-1 protein is that it's able to recognize this extreme variety of structural chemical diversity, which allows us to function as a mechanism for xenobiotic responses. Again, the canonical model is that this modification or structural change in KEEP-1 releases the NRF-2 transcription factor, which then now activates this wide variety of response pathways, which act at the level of protecting the cells. Now, there are some flies in the ointment, if you will, in that the KEEP-1 protein is also seen in the nucleus. And exactly what it's doing there at present or formerly was thought to be an export factor for NRF-2 protein. As you'll see, it does much more of interest. And one of the things that was intriguing for us and the reason that we got into this game is that the coupling between the stimulus and the response was seemingly mediated by a very tenuous link, this NRF-2 transcription factor, which would limit the ability of selective activation in response to particular classes of compounds and particular response subsets. And this is something that's observed. If we treat a cell or an animal with a particular compound, we don't get the identical response pathway. So therefore, there seemed to us to be something missing. As I mentioned, we're studying this problem in Drosophila. Drosophila has homologous proteins to these mammalian proteins that have been studied extensively. In Drosophila, these proteins do what we might think they do. They protect the fly from pesticides, et cetera. These are the pathways that pesticide makers concern themselves with. The reason we use the fly is that it has many nice advantages in studying what, to us, is rather a mysterious process. And it also allows us to ask, in hindsight somewhat, the question of where did these pathways come from? How was this ability to respond to something that is unknowable, or at least initially not evolutionarily present, developed? Again, we like using imaging. So the first question we ask is, where these major actors are? Both KEEP1 and CNCC, these are salivary gland cells. This is a nuclear stain with a DNA binding dye. Our nuclear, which initially was somewhat surprising. Remember, we usually think of the KEEP1 protein as being cytoplasmic sensor. At least this was the model. Just to make sure that our antibodies were specific, we also make transgenic flies in this case. And the transgenic gas breast proteins are clearly nuclear. One of the tools we like to use to look at interactions, and this will become relevant later in the talk, is that we like to look at how an interaction between two partners alters the specificity of their biology. And in this case, we use a tool that we developed some decade ago, which allows us to visualize selectively the protein-protein complex, whereas the individual subunits are invisible by fusing them to fragments of a fluorescent protein. Now, the fragments of the protein, just like half an enzyme, doesn't have any activity. When brought together, provide as a reporter a fluorescent signal, which we can now visualize, and see where the complex is. And here, perhaps no surprise, the complexes are present in the nucleus. Now, again, we're using this Drosophila system, because Drosophila imaging is really powerful. Now, those of you in the audience who use imaging know that the instrumentation, the power of optical visualization has grown by leaps and bounds. And this is really due to these enormous instruments that are very useful for many purposes. In our lab, we tend to take a rather more pedestrian approach of just adapting our system of visualization to the particular problem at hand. So in this case, we take advantage of something that is characteristic of fruit flies, particularly the salaried glands of fruit flies, the classical observation over what are called polytine chromosomes. What these are are basically strands of the genome, the entire genome, stretched out and copied in thousands or copied and aligned in parallel. And what this does for us is that it positions individual genes in thousands of copies in a physical location. Now, this is not to say that we couldn't do a similar experiment in a mammalian cell and seeing a spot correspond to an allele. But what this polytine chromosome system does is that due to classical work on the part of many people, individual genes have been mapped, relative to physical reference maps, on the chromatin so that we can, by simply looking at one of such spread, get a genome-wide map of where the proteins are bound. And in this case, we were really quite surprised that the pattern of binding here by these proteins which have this function, known function in protection from foreign compounds, corresponds to a very characteristic developmental program which is something called the ectosone response pathway. This is the pathway that the fly uses to conduct what's called metamorphosis, the remarkable process of the fly larvae to completely reorganize its own body structure. Now, to confirm that, in fact, these patterns of binding meant something biologically, we look at knockdowns, essentially, loss of function mutations in these proteins. And we find that this class of classical response genes is reduced in expression relative to wild-type flies, suggesting that indeed, they require both the KEEP-1 and CNCC proteins for their expression. This is not because we have compromised development, at least at the level of the salaried gland, because the late genes are expressed, and therefore the flying general is intact. Now, I'll have to do a little education here in terms of fly biology. So fruit flies develop from an embryo. And in response to these pulses of an endocrine hormone, a steroid hormone called ectosteroid ectisone, undergo specific transitions in development. And it was studying these. So I'm going to have to sort of cut out a big part of my talk here in the middle and say that we found that when we were depleting these regulators in the fly, we were finding that development was arrested. So it seemed like something was going wrong beyond the inability of the salaried glands to produce what are rather pedestrian glue proteins, things that make the larva stick to the surface where it's going to pupate. So we suspected that perhaps the production of this ectosteroid was being compromised. So we looked at a depletion experiment of these proteins in a particular tissue where the ectosteroid is produced. And we found that indeed the depletion of these proteins, in this case the CNCC protein, results in a delay and reduction of synthesis of this fly endocrine hormone. This is the time scale of fly development. These are, they ultimately form these pupae as you can see the pupil formation is significantly delayed in this context. This is because in this pro thoracic gland and endocrine gland in the fly, a pathway that leads from cholesterol just like we make steroids for our use in reproductive and other regulatory processes. There's a pathway encoded by a number of different proteins which convert cholesterol to this endocrine hormone. And we find that the genes that are encoding these P450s are again regulated by these proteins which we think of primarily as protective, as providing protection from the external impacts. So it appears that in this context these regulatory proteins are functioning to control the fly development. And again, a control we're showing that the gland itself is intact. So it's not that taking away these genes has caused development to hold, but rather in this case it's this bio transformation that is compromised in the absence of this regulator. So to actually look at this in a biological context we can look at the time at which a larva after hatching forms the pupae and look at what happens when we deprive the larvae of one of these regulatory proteins and show that there is a shift, a delay in time of hatching. We can rescue this fully by feeding the larva the endocrine hormone. So we can add the ectosteroid into the food and the larvae hatching is restored to a near wild type timing. So this indicates that it is the form production of this ectosteroid which is as regulated by these regulatory proteins that is essential for maintenance of the development. And rather to confirm that it wasn't that we had somehow compromised, somehow made non-functional larvae. These are the larvae that have been depleted and as you see they keep growing. In fact, they become larger than the wild type larvae because again they keep eating and not pupating and it is this absence of the ectosteroid which is essential for pupation. So summarizing this part, what we find is that these proteins which we know as regulators of xenobiotic responses function in two different tissues or in several tissues, both the central endocrine gland that is synthesizing the ectosone. This function is in response to a neuroendocrine hormone PTTH and via the distribution of ectosone in the circulation targets peripheral tissues where these proteins again function in concert to regulate the peripheral response to this endocrine hormone. So these data suggest that this regulatory network has existed or at least also exists as a function to control developmental progression and we think that it occurs because the development has to respond to the environment. It needs to be able to modulate the timing or the five development depending on external conditions. Now I mentioned that we like to look at the protein-protein interaction using this imaging but I haven't told you anything about what these complexes do and in fact we were surprised to find that the distribution of these complexes is distinct from the distributions of the individual protein. So here is the distribution of the complexes again visualized on one cell. The nice thing of this tool is that we can look at individual cells and variations among individual cells compared to the binding of the individual proteins and all of these are different methods. It should be obvious to you that the patterns are quite different even without the annotation of these bands. So we were curious what these genes were which selectively bind the complex. This is again one of these features of these regulatory protein interactions where it's the protein-protein interaction that alters the targeting. I'm gonna skip this in the interest of time simply to say that we can quantify these differences and show that it is not just a qualitative ability to detect but a shift in specificity which is occurring. This makes a more qualitative image again without trying to look at the individuals. The left half of this graph here shows the patterns of binding of the individual proteins by various different imaging strategies. The right half shows the patterns of binding of the complexes and again the patterns are quite different telling us that the protein-protein interaction re-targets the specificity of binding of these proteins. Now we look at particular genes as sort of a biological representative to this pathway. Some of these have developmental context, juvenile hormone, hydrolysis are regulating the stability of other developmental endocrine hormones. The de-keep one protein itself is a target for transcription regulation by this complex. Again, so the question for us was, well what is the biologic process that is reflected by this complex formation by these proteins? And we looked at the effects of drugs on these flies and found that there are several bands which are absent here in control flies. But when we see these flies, particular drugs, are now present telling us that in fact the binding of these proteins is reprogrammed in response to the drugs in a way that is similar to the other proteins. In a way that is similar to the reprogramming of the binding by the protein-protein interaction. So this turns out to be specific to particular drugs. Other compounds which are also regulating the keep one protein do not regulate binding at this particular loci, they regulate binding at other loci. So this appears to be a case where the binding of the drug, we think in the context of chromatin, now this is one thing we have yet to demonstrate that the drug actually acts on the DNA-bound complex, alters the binding specificity of the complex. We look at the functional consequences of this and again this upper group of genes are the ones which are directly bound by the complex and we measure the levels of transcripts of these genes and find that indeed phenobarbital, one of the drugs that induces this binding increases transcription of these genes and other drugs which do not induce binding at this gene do so to a much lesser extent. This is the control group of transcripts that are regulated by, that do not bind the complex. So these are the genes which we chose as representatives of the binding of the complex selectively. And again, we can also measure the protein products of these to say that they are significant. To look at the functional necessity of this binding insofar as does it alter, are these proteins in fact required for the transcriptional function? We look at the basal expression here in the absence of phenobarbital and we compare the genes where the proteins are bindings of complex and other proteins, other genes where the CNCC proteins appears to bind independently of the KIP1 protein where we don't detect KIP1 binding and in the context of phenobarbital both groups of genes are activated. The CNCC protein depletion of thereof causes a decrease in this phenobarbital activation of both groups. Now in contrast, when we do the same experiment with the KIP1 protein we find that again KIP1 is that the phenobarbital activates transcription but the KIP1 protein is required for activation of this group, the ones that are bound by the complex but not for activation of this group of genes where CNCC appears to function independently. So there seems to be a selective activation of different groups of proteins depending on this particular protein-protein partnership. We can do the gain-of-function experiment the same way in that by expressing KIP1 independently we can activate transcription and CNCC with some of these genes functions as a synergistic regulator such as here. So the combination has greater activity than the individual proteins whereas KIP1 acts as an antagonistic inhibitor, reducing transcription of these genes where CNCC functions independently of KIP1. So again, the protein-protein interaction at this specific locus seems to have a distinct effect depending on the binding. So to put this half of my talk in context this is the classical model which can very easily explain how a compound, a drug or some other external molecule functioning through a pathway activates a set of genes which in some ways help detoxify this compound. Now the challenge becomes, as I mentioned more difficult to explain when we start talking about many different compounds which have partially overlapping but distinct responses. So how does this system mediate this kind of response? And further as I mentioned, we find this group of developmental signals which regulate a very slightly overlapping set which also depends on these same mediators. So how can this sort of bottleneck if you will be tolerated? We say that it probably is not and in this case that it is the complex acting on chromatin which mediates through some, we imagine, allosteric effect. In other words, through binding different co-regulators the specificity of this complex is altered such that it regulates different sets of targets depending on the small molecule mediator. And the mystery remains to us that how this system has adapted to respond to essentially unknown inducers by ways of modulating specificity in a way that now allows it to target genes that are specific or effective for each of these targets. So we like to think that this system operates on many different levels and I just wanna thank the people participating in this part of the work and our collaborators who helped us with some reagents. Now for the second which will be a shorter half of my talk I wanna return to this problem of I have barely scratched the surface of the problem in the sense of I told you that I would address this pharmacodynamics problem but really all we've addressed is, well we have some hazy idea about where this might have come from evolutionarily which does very little to address the practical problem of well if we're going to develop a drug how do we approach the problem such that we're not going to get tangled up and tripped along this way. And this becomes especially critical in drug development programs that don't have the tens or hundreds of millions of dollars that are typically devoted to developing a drug these days. And that is the case for in fact the majority of diseases which are rather rare and attract very little interest on the part of pharmaceutical community. And the problem of course is that it is not just going back and forth and optimizing each of these steps one at a time but rather that a failure at the late stage here often leads to abandonment of the entire pro program and you read about the drug companies that shut down an entire division say cardiac diseases and you know that there are going to be no follow up on these projects. So what can be done? Well, this problem sort of has a similarity to me with the children's game of snakes and ladders. I'm not sure if you played this. It's one which is can be quite frustrating especially in the context where in this system at least there are very few ladders you keep going back and go starting over. So how can we if you will put back some ladders into the system and allow us to anticipate or perhaps even plan ahead for the challenges of dealing with these steps in the drug development which are rather poorly understood. So I'm good to tell a story of a project at my laboratory and I should tell you that I have very little background in drug development but there are many rare diseases adrenocortical carcinoma being one which there is essentially no drug development going on which means that this is still treated in the same way that it was treated 50 years ago with compounds that were approved in the 1950s and there's really no company has that is doing any development and the outcome is predictably unfortunate. There are many challenges to this and many such drug development problems. Diseases are often complex and despite the genomic genetic advances that are made the feedback, the turnaround of using that information for drug development is very slow. One key aspect of this cancer drug development problem is that the tumors in fact retain a very unique metabolism. They have characteristics that are akin to their progenitor cells which in this case is one that we take advantage of as an Achilles heel. In other words, we try to target the unique metabolic properties of these tumors in order to potentially treat them and we do this by using compounds that have been identified that have a very selective tissue or cell type specific toxicity. This is a very classical way of doing cancer drug development but I think it has lost perhaps some of its cachet. In our case, we take advantage of the fact that this kind of information is if not entirely publicly available. There are certainly many drug companies who discard drug leads which have very extensive characterization and here is sort of the key to our development plan is that we need to start effectively very close to the finish line. We need to start with compounds where we know that if we sort of get over the last few jumps we will actually have something to release into the clinic, we cannot afford to have the risk of losing our way along the way. So I'm going to, since time is short, going to sort of give you the answer and then give you some of the steps that we got here and some of the problems that we face. So we've identified mainly through literature investigation a compound discarded in the 1980s which has a selective adrenalic activity and this compound through some preclinical studies that we've conducted that are described in brief has been now in three phase patient trials and so far we know that the compound is well tolerated has low potency which is one challenge that we're facing in these preclinical trials. So briefly going over that kind of characterization many cancer drug development projects undergo so we use a xenograph model which is to say we take cancer cells, we put them in a mouse and we give the mouse a compound in this case this ATR 101 compound and this decrease in the growth of the xenograph isn't astounding but it is at least a clue that we have something that might have some benefit we weigh the xenograph at the end and very importantly and here's the critical aspect of trying to decide on which kind of compound would be appropriate. The patients that are treated for this disease are quite sick given that it is a late diagnosis is often at a late stage and therefore we have to choose a compound which has no adverse effects. This is not true of course for the majority of chemotherapies which are essentially toxins which you use at a sub toxic sub acute toxicity level but this compound was chosen selectively because it is so well tolerated here simply represented by the body weight of the mouse that is not decreasing. The compound can also be relatively effective in this case we're looking at the same kind of xenograph assay but we're starting administration early. Now this is not really a realistic cancer model cancer is usually treated at the time when the tumor is quite large but in this situation we can in fact suppress the tumor development almost entirely this might be perhaps representative of a situation of a adjuvant therapy after surgery or such. This compound acts by inducing apoptosis it induces the cells to kill themselves and this apoptotic stimulus has been the area of interest and you might wonder well we have these compounds in clinic what are we doing trying to understand what they do? Well the critical problem for us is that the compounds potency is not really adequate or at least is going to pose some difficulties for using the clinical. So given that we have no clue how this compound has this potential therapeutic benefit we have to go back and try to understand how it works in order to improve on it. Very briefly and without giving you much background we found that this compound causes cholesterol accumulation so this is a cholesterol stain where this compound the active compound causes accumulation of cholesterol in the membranes whereas a controlled compound lack in this dimethylamine is not causing the cholesterol accumulation this occurs in many cancer cell or many cancer cell lines of the adrenal and the adrenal is rather a unique sight in that the cholesterol metabolism of the adrenal that makes all of our corticosteroids the ones that regular blood pressure, et cetera has its own very dedicated cholesterol metabolism which is why we think that this cholesterol accumulation turns out to have such tissue specific effects. It is very rapid so this compound within a few minutes causes this accumulation of cholesterol which is in fact faster than the time at which the cells are starting to die losing ATP and increasing their caspase signaling suggesting that this is in fact the potential mechanism of toxicity. What makes it most compelling to us is that using this polysaccharide cholesterol chelating agent methyl beta cyclic dextrin we can remove the cholesterol effectively treat the cells with this compound but not cause the cholesterol accumulation which saves the cells effectively by adding this cholesterol chelating agent we can restore ATP to cells telling us that it is in fact the cholesterol accumulation which is essential for this ATP depletion in the absence of this chelating agent and conversely prevent that caspase activation which results in the cell death and indicating that in fact the way that compound works is by causing a excess cholesterol accumulation. How does it do it? I was pleased to see that in the afternoon we will hear one of the presentations of ABC transporters which are in fact the target that we've identified for this particular compound. The compound prevents cholesterol efflux so adrenocortical cells export excess cholesterol and in the presence of this drug the cholesterol accumulates with a time frame that is similar to the ATP depletion and this is in contract this is control compound which does not influence the cholesterol export and this group of ABC transporters is a complex one I won't try your patients or my lack of knowledge of them by describing them but there are literally dozens of them there are multiple substrates here that are affected in addition to the cholesterol efflux we're also finding that export of cortisol one of the products of steroid synthesis is inhibited in the presence of this compound and this inhibition seems to also be essential for the inhibition of viability so that we can mimic the effects of this compound so the compound effect here on ATP levels is quite substantial by combining multiple inhibitors of ABC transporters with different specificities we can somewhat mimic though we don't quite reach the efficiency of this compound indicating that this particular set of ABC transporters substrates in other words these particular inhibitors target different ABC transporters must be inhibited in order for this compound to have its cholesterol accumulating effect and therapeutic benefit what has this gained us so far? well, so far we only find that by adding to this compound known inhibitors of particular ABC transporters we can increase the potency so we can do what was mentioned earlier in the previous talk effectively combination therapies combination therapies have certainly an appeal the problem is that they are problematic in clinical trials and also, well, you have to get many companies to work together to underwrite such progress the other thing it gives us is that it tells us that there are specific limiting activities in other words if we can increase and find a compound which increase targeting specificity for some of these ABC transporters we may be able to do a better job with patients so at present, this is the model we basically have identified that cholesterol accumulation is a process that's driven by many inputs and outputs and in the presence of this drug candidate we cause a accumulation of cholesterol by blocking several of the pathways of export causing toxicity many people have participated in this project Veronica Burns and Yunhui Chen leading groups of students in the project a drug development program requires participation of people of a variety of different interests from expertise in drug development which is not me and again, the clinical trials are run by a startup company that we founded and we've had assistance from many clinicians which is something which again complements my expertise and has been essential for the project so I thank you all for your attention and happy to ask questions answer questions can you tell us what was the fundamental difference from what was known before what you guys did for this? well, what was known before is that there are for us for this particular compound for us what's valuable is that there are very large data sets of compounds that have been tested in animals in this case thousands of compounds these compounds typically come from projects such as heart disease drug development and often these projects come up with discarded compounds that have been administered to guinea pigs, dogs, monkeys animals which are relatively good models for humans I think what mathematics could do for us that isn't happening right now is that although we know the structures of these thousands of compounds that have been tested and we know something about the physiological outcome in this context for example we knew that the compounds had bioavailability they got into the animal they had bioactivity in the form of their cholesterol lowering effect and in fact in this case the operative word we knew something about the toxicity we knew something about how these compounds caused the adrenal to we knew that they caused a selective effect on the adrenals now relating the one to the other is something that is currently completely empirical in other words we don't we have no clue about what it is about the difference between these groups of compounds that make some of them adrenthoxic and others not I think that is a problem which there should be much more information about not obviously just for adrenotoxicity but questions in general in terms of drug distribution what causes a compound to accumulate in one tissue versus another again a problem for which there's really minimal information it's empirical we give a compound to an experimental animal and we find it in a target organ we say good this is a compound that could have benefited a priori we have no way of predicting those things so that's perhaps the the challenge and the difference that we made in this project of making sure that we understand roughly how the compound works in continuity with this question is there a maximum about the physiopathology of adrenocortical carcinoma a pathway a gene or several genes heredity, family, etc, etc, etc because if I understood correctly what you did is the very very old way just by Azar taking everything and the way now I think 10 years the companies if they have a pathway they have some enzymes or phosphorylation or something like that and then they choose the active site they try with many compounds just specifically to see-blay this site but yeah, you have something familiar so I've sort of glossed over many diseases like adrenocortical carcinoma are certainly not highly studied but that doesn't mean we don't know anything in fact, we have quite a bit we have tumor sequences from hundreds of patients which as some of you know we can identify candidate genes those candidate genes tend to be the usual suspects they are in pathways that influence genome integrity p53 they're signaling pathways beta-catenin, etc the success at targeting those pathways has been quite spotty in other words, we've known about pathways like RAS for 30 years we still don't have a drug we don't know how to target MEC well, those will certainly come but the other problem with many diseases such as adrenocortical carcinoma is that they are highly heterogeneous in other words, the molecular causes the clinical presentation of the disease varies quite broadly so although we may in a not too distant future be able to treat 1%, 5%, maybe 10% of the patients no seemingly reasonable time frame will we be able to treat a large majority of the patients by targeted mechanisms the other thing is here I mean, we can get past these but it is up here where all the failures happen so by taking the approach that we're taking we are somewhat saying these stumblings up here are so catastrophic that we have to focus on doing these instead of spending all of our time here I mean, it is nice to have a specific target but unless you have a way of dealing with the problems that will come up here you will never get to the finish line that is my argument I mean, we have that experience for example, I'm in a group that is looking for inhibitors for viral fission so the target was found they're potent, they're specific a viability becomes already a problem because they start to stick to everything inside the animal and then the pharmacodynamic is a mess but it's very, very hard to do that in the academic environment it's very expensive so there's not enough resources that come from a normal grant to do that, it's impossible there are big challenges now there are big challenges I wouldn't call it impossible no, but it doesn't have experimental problems I mean, I don't think that's an intellectual problem it may not be and I think I want to contrast the two halves of my talk there's a very different approach that we take when facing a problem of developing a therapy than we take when we try to address a problem having to do with mechanism in even an animal and I think that that difference in approach it doesn't solve the problem but it puts our emphasis on thinking about these problems early and trying to anticipate trying to... I wouldn't say that our solution in this case is generalizable I mean, I think every disease has its own peculiarities understanding that disease context I think contact with clinicians has been key for us in other words, talking with people who treat the patients to understand these are very sick people in order to develop a drug we cannot start with a compound that's toxic we have to sort of rethink of what a therapy might be so challenge is a bound but I would say that, yes, except for this part it doesn't have to be outrageously expensive but that requires sort of accepting compromises and taking shortcuts along the way I have a couple of questions about the beginning really more clarification I didn't quite understand with the KEEP-1-CNCC are they direct binders to all these antibiotics? are they feeding in upstream somewhere? and if they bind directly to other structural data? so KEEP-1 is a remarkable molecule and there are crystal structures of KEEP-1 it operates by many mechanisms one of the major one being that it contains cysteine residues that are hyper reactive those cysteine residues react with electrophiles compounds that are more reactive than most bioactive compounds so that's one big class of KEEP-1 reactive KEEP-1 activation mechanisms this model here of a changing conformation state of KEEP-1 has been somewhat corroborated by biophysical data the piece that we've been trying to work on is how do we get from this reaction to a target and that has been the thing that has had sort of a missing link which we think we've established I still think that the question of how KEEP-1 especially in the physiological environment in a cell let alone an animal is able to discriminate between molecules that are hazards than molecules that are mainly active intermediates of some intermediary metabolism it doesn't do a great job in some respects fumarate for example is a great activator KEEP-1 but it just operates at a sub so it's tuned to react with compounds in a very specific way it also binds polycyclic hydrocarbons and it does that with a very sort of relaxed specificity that again discriminates between steroids and things that otherwise might seem like it would activate high levels of glucocorticoids will activate KEEP-1 but not the levels that are present in most cells the second question for clarification is so the two proteins that are going to these puffs these sites in the polychine chromosomes maybe I didn't understand this correctly when we did the localization initially of each one they both localized to the disome ones and yet when you have them presumably they're forming a complex there ah that's presumably separately or what? because you said they redistribute when you indeed what is different about the individual binding of the proteins and when we see them together a big difference is we only see the redistribution when we overexpress the proteins or when we treat the animal with drugs so there's something which admittedly is a bit nebulizer between the proteins individually at their endogenous levels and the proteins overexpressed as a complex now that's something isn't the trick that we play this complementation assay it's not the fact that we've trapped the proteins on the chromatin because we can also overexpress them in ways which don't allow them to form the stable bimolecular complex so for us the overexpression is sort of a it's a it's a unfortunate but perhaps informative ah tool ah we think it reflects because it produces a similar redistribution or at least some aspects of redistribution can be reproduced by drugs we think it has some relevance relationship to that drug induced targeting but in general I think that the effects of overexpression should be considered in other words so to continue Mark's question to really know if the complex has a function and not just an overexpression artifact you probably need to do an interaction mutant I'm not saying that it's easy no no no it is easy so in fact all of our experiments we do this a lot and we often find overexpressed proteins have characteristics that are non biological that we can't hand wave such as here and say maybe this overexpression actually does something that is relevant most of the time when we find overexpressed protein doing something weird it's bad and indeed our prime strategy including this project is to show that a mutation in the protein prevents the signal we don't get fluorescence and hence that at a certain level represents a biologically relevant complex formation so what you need to see that now they do not the genes which are attached to are not expressed so in our case the mutation isn't quite as subtle as that in our case the mutation in this case is one that simply disrupts the complex so it's yes yes what you're asking is for us to show that re-targeting is something that depends on the specific interaction where all that we've shown is that disrupting the complex prevents us from seeing a signal and that we don't know partly because we don't really know what the signal is what the conformational whatever the mechanism is that converts this protein from one that is targeting one set of genes to a different set of genes thank you