Cc Validated Field with airSlate SignNow

Cc validated field

okay uh thanks very much for the introduction catherine um and and thank you very much uh to to you and the rest of the organizing committee for the invitation to come and speak to you today um as catherine said i'm going to speak to you today about thalidomide analogues um this is a incredibly interesting class of molecules that have a history that goes back all the way to the 1950s um there were aspects of the history that have uh were incredibly dark periods for drug discovery but recently there's been a number of advances and these molecules have really become the prototypes for a new approach to drug discovery called molecular glu which has incredible potential for human health and i hope you'll be able to get that conveyed by the talk today and just by quickly by way of uh disclosing my personal affiliations and and uh doing some upfront acknowledgments um my uh affiliation history is shown here um from 2007 until earlier this year i was at celgene bms a lot of the work i'm going to describe today as part of really what was a great privilege that mine was done during my period there i'm also going to give a overview of the field and this will all be uh literature uh discussions um and then since earlier this year i've been working at a startup that still is still too early to talk about we're in stealth mode but uh um i have a equity position in both of those uh organizations um and obviously um my employment has has ended with sergeant bms so i'm not here as a representative today so just by way of a quick uh overview of uh targeted protein degradation and i'm sure this is uh completely redundant for many of the people that are on the call today um what we're talking about here are specifically the types of drugs that combine to a piece of cellular machinery called an e3 ubiquitin ligase and that's shown by this bundle of objects here so these can be multi-component systems but the overall action is one where target proteins are recruited into this complex and they're a protein called ubiquitin is transferred onto the target protein and you have a ubiquitin chain formed onto the target and that targets this protein for destruction by the proteosome so in the instance that you have a protein that has a clear link to disease what you could do using this technology is use a drug to have this protein ubiquitinated and destroyed by machinery that would not normally act on that target protein so we're talking about a gain of function we're talking about redirecting the cellular machinery and we're doing this for the purposes of resolving disease and another feature of this system is shown here you can see that this is a catalytic process and it can cycle around so a single drug molecule can lead to the destruction of multiple proteins really there are two um distinct types of strategy for targeted protein degradation that have become clear in the literature over the last few years you have the molecular glue and then you have the heterobi functional type and and the difference um really comes from um the uh origin of these two approaches so um if you if you look on the left um you have uh molecular glue um this term was first coined by ningxiang in this work tan ital in 2007 and they were doing some work on a plant hormone signal uh signaling molecules and they work in a very similar way um to the way that some of the drugs i'm going to talk about today they work they're very small molecules they bind to the e3 ligase and that changes the surface of that interaction just enough to make a target protein be recruited and then destroyed on the other hand you have this heteroby functional approach such as protax and this was a a concept that was first published in 2001 by a team led by craig cruz and ray de shea and what they saw and what they developed were this bifunctional molecules here whereby you have a ligand that binds to the target protein you have a ligand that binds to the e3 ligase you link those two to things together and in the same way you can scaffold a protein into this complex and cause destruction and the distinction here um is you know if you work in the field like i think it is worth making because there are differences in the pharmacology there are differences in the strategy by which to approach these things um you know and there's also different target proteins that you can go after with these approaches so um and i'll go into some of the details of this during the talk but your dependence on a ligand binding site for the hetero by functional approach is still there whereas and what i'll describe today is uh that for the molecular glue system you have a much reduced or in some cases no dependence at all on ligand binding sites and that is really a completely different scenario to that typically encountered um in uh conventional drug discovery and so today the talk is going to focus mostly on on the molecular glu side and obviously i think as part of this seminar series you have craig cruz coming up and i'm i'm sure he'll um give a fantastic view of the hetero by functional approach and and so just to to talk about that a little bit more um i've the the general features of the conventional therapeutics that were available to drug discovery researchers until you know a few years ago and these uh two types of therapeutics still dominate the clinical landscape and so on the left what you have is a small molecule so these are small synthetic chemicals they can be formulated into pills that you can swallow and they can be available by the organ of administration and on the right you have biologics such as antibodies typically delivered by injection and what you can see is the therapeutics are shown in grey so you can see the huge difference in size and nature of these approaches and uh they both have their limitations so uh small molecules they can act in or outside a cell but they're limited by size and properties and they're very dependent upon the presence of ligand binding pockets you know and there's been you know some heroic efforts in drug discovery to target proteins with very minimal binding pockets um and these things and people have done incredibly well are pushing the boundaries of that but in general you still have you can see in this protein here you have this nice binding pocket in which is more molecule combined now your biologics your antibodies because they have this large protein surface they don't have that same requirement so you can target different types of of molecule but another limitation for especially small molecules is the the binding site the place where these bind must have biological consequences they must act in a way that causes resolution or at least a clinically impactful effect so um and and protein degradation really offers a solution to that because uh if you can bind a protein anyway you can destroy the entire protein so that that's much less of a dependency but the molecular glue approach as well bypasses this uh requirement for a binding site and so it really does it by leveraging this uh the larger surfaces that are available from making a ternary complex with knee 3 ligase and and that gives you a whole different type and a whole different number of proteins that you can target now so the molecular glues um obviously have tremendous power and potential and um the way that the validation became available for these was obviously um it would have been fantastic if it had been you know a set of scientists uh you know conceptualizing and uh and driving these things um you know from scratch but um as with many things it was actually a number of serendipitous discoveries and in the case of really i think the most uh um well-studied class of molecular glues the thylamide analogues um you know unfortunately a lot of the lessons were very harsh ones for society and drug discovery so thalidomide was first synthesized in the 1950s by a company kemi gruningthal in germany and you can see solidamide is shown in the bottom corner and it's an incredibly small and simple molecule unfortunately the safety studies that were done back in that era were incredibly limited and so the drug was used in situations that it should not have been used and that included being prescribed to pregnant women and that led unfortunately to an epidemic of birth defects so this was a global phenomenon it's likely that there were tens of thousands of babies affected the us was largely spared due to the efforts of francis oldham kelsey at the fledgling fda and then in the early 60s that the link was made and thalidomide was withdrawn um however it was discovered that it actually has beneficial uses in treating a complication of leprosy not in its enl and it also had beneficial effects that were discovered later in myeloma and so clinical approvals for those molecules followed but for those indications followed but um already at this point um the story of thalidomide and the geopolitical consequences of the decisions and the impact on drug discovery and then the the turn upon into a into beneficial therapeutics um was written up in a book called dark remedy um and it's a i think it's a fantastic history of uh this class of molecules but it was published in 2001 where the state of knowledge was exactly um as is shown uh on this slide so um uh i actually think it was later now that the the real action has uh started in this field because um it turned out that the mechanism of action of these drugs is through molecular glue and that has now brought them to the to the cutting edge of drug discovery and that was work that started in 2010 with some work done by hiroshi handa who identified cerebellum as the primary target and from there substrate proteins were identified ceremonies structural biology was performed new substrates were identified these molecules were repurposed into heteroby functional molecules novel substrates with novel chemical matter were described proving that you could now repurpose this ligase into new areas additional zinc fingers and neo-substrates and uh additional zinc fingernail substrates were described and structures were solved um salphor was identified as a as a plausible driver of trautogenicity i'm going to talk to you a bit about that today um there were some other molecular details added and the last piece of the story i'm going to talk to you today is is just visiting the south or ternary complex structure which loops all the way back to the to the earliest actions of this molecule so you can see that it had an incredibly difficult start and really as drug discovery has evolved this class of molecules has evolved um as well and uh additional findings have now led to this uh having uh being an area of uh incredible potential for human health um and i'm gonna talk to you about this last period 2010 onwards today and there's a few uh other things i think are interesting about this class of molecules clearly they've transitioned from completely phenotypic to now very well molecularly defined they're great case studies for that but really it's as molecular glues are these things are the most interesting case study to me so um thalidomide it binds to the crl for crbn e3 ubiquitin ligase so uh casting back to the cartoons shown before the link was made by takumito and hiroshi handa and then and the t uh and that team who used uh salidomide analogs bound to beads and they were able to pull these through extracts and they were able to identify these bands here for cerebellum ddb1 and previous work by ningxiang's team had placed that into the crl4 e3 ubiquitin ligase so this was a molecular structure that was understood um with the exception that nobody knew how cerebellum was fi was folded and where it was bound so we we had a good identity understanding of the underlying machinery but we lacked that detailed understanding of of the specific features related to cerebellum now at the time there was no knowledge of the neo-substrate mechanism of action and the identity of those substrates but that changed in 2014 with publications from ben ebert's lab bill kalin's lab and a team at celgene and what these teams reported was the thalidomide analogues so that's thalidomide and then the other approved drugs lenolidomide and pomalidomide trigger the degradation of the icarus family of zinc finger proteins so an example of a zinc finger domain is shown over here on the right it's incredibly small domain this binds into the groove of dna and helps alter transcription and protein expression but there's no ligand binding sites on this domain and these proteins typically take the form of tandem arrays of this domain so we really don't have many opportunities for intervening with this drug class with this protein class and there are a lot of members of this protein class so it was shown that these are robustly degraded by pharmacologically relevant concentrations of lennolidomide you can see it here so here you have your protein this is iolus and icarus uh two members of the the protein family and one micromolar you have uh removed all of the protein and so that's an incredible finding so you're seeing uh protein degradation you're seeing it in an an undruggable transcription factor protein class and it was shown that these that are key mediators of the anti-myeloma activity as well in in numerous subsequent works a second class of substrate proteins was described not long after in this publication by the ebert lab and what you can see here is that ck1 alpha is a substrate for lenolidomide so ck1 alpha is a protein that is completely unrelated to the icarus family of zinc fingers so it's a it's a protein kinase you can see um you have an atp binding pocket here this has a complete different function completely different fold completely different sequence and you see robust degradation in the presence of len but the other fascinating thing about this work um is that you do not see um degradation by these other analogs and some of them are incredibly close uh analog so what you're seeing now is a chemical differentiation in substrate preference and that was obviously a huge value now the structural biology at that point became i think very very important and this was a a fantastically fun project to work on as well um and there were two teams that published the structure uh in 2014 as well so there was my team at celgene and then there was uh uh eric fisher working in the time a lab in switzerland um who published um the structure as well and what both of these teams found is that you have this glutaramid ring so this is a conserved group amongst the clinical compounds it's bound in a shallow hydrophobic pocket on the surface of cerebellum and then you have anot another part of the molecule displayed on the surface so so this was um incredibly um interesting information for me um and what you can see is um um where you have sites of chemical differentiation you can imagine that this is where you're now able to select different substrates and bring them in you can also see that if you look at the whole complex that you have this sort of adap array where you have the adapter protein ddb1 you have cerebellum bound at the one end of that and then at the very far end of it you have the drug binding pockets oriented exactly where you would want to recruit a protein to have ubiquitin transferred to it so that was a very plausible finding too um but there were some other things about the the binding pocket that i think were quite startling at the time one of them is shown here so what you're seeing here is the surface of the thalidomide binding domain of cerebellum and it's colored by sequence conservation with the species listed above it so in red you have a hundred percent sequence conservation and so you can see that the monkey protein is completely conserved the mouse protein has these two changes and if you go to fruit fly or to soybean you can see that the sequence conservation gets much worse for the most of the protein but the thalidomide binding pocket is to 100 conserve so we're talking about a pocket of extreme functional constraint and you know the one thing i think we can be confident in is that this pocket didn't evolve for the binding of thalidomide so um clearly there's a a there's a really major uh and important unanswered question um in in in this field you know there have been a number of natural substrates implicated now but i don't think anyone has really nailed the you know put the nail in the coffin and found the the ligand that binds in this pocket and i i would love to see that question answered so another thing that we published back in 2014 um was this finding so um it's that if you take a myeloma cell line you can see you have sensitivity to pomolidamide you take the cerebellum away you lose sensitivity completely you can rescue that with the human protein but if you put the mouse cerebrum protein into that line you have no effect of pulmonary so so we learned that the sensitivity um to this class of molecules required the human sequence and the sequence changes in the cerebellum protein were underlying the species of resistance um that had been known in the class for a really long time and that was actually something that that probably did um you know that contributed to some of the bad decisions that were made in the early development of this class of molecules in the mid 20th century um the the fact that with these uh sequence changes and we proposed these two um as candidates for the the uh the drivers of that uh resistance and with that you can you cannot model the activity of the neo-substrate action of this class of molecules they're there and you can see how how startling that effect is in this graph um that was uh confirmed to be the case uh for uh ck1 alpha and icarus in in the work uh co-published with ben eber in uh 2015 and you can see that again the mouse protein does not confer uh degradation of cko and alpha human cerebellum does and if you change one of the mouse residues with an incredibly conservative change you can see it here very subtle change uh i to a v you can now rescue the degradation effect and uh so what we proposed at the time was that this incredibly subtle change here was probably uh interfering with what would turn out to be a hydrogen bond donor to the substrate protein so so we were confident from the initial structures even though we didn't have a ternary complex structure yet we were confident that because the thalidomide analogs were so small they didn't have the molecular weight they didn't have the size that would have been necessary to simultaneously engage two different proteins without the use of the protein surface and also um because of that you know we anticipated the glue action and uh uh expected that uh there would be uh critical elements on the protein surface and this species difference was really highlighting that um so um and it was and it was that sort of strict requirement on the protein surface that really you know took us down this path in in contrast to as i said before the heteroby functional approach um but these molecules were incredibly useful as ligand-binding uh moieties four hedgehog by functional molecules and there were two publications in 2015 describing that application one from jay bradner team and one from craig cruz and you can see that with these molecules you can see the solidimide analog on one end you have your bid4 binding group on the other you can confer degradation via this heteroby functional strategy you can convert robust degradation where otherwise you would not have it so now with the molecular tools becoming available there was a lot of discussion in the field regarding whether you would ever be able to have a target-based drug discovery that operates in a molecular glue system and so i think it was important to go in and start using the most modern tools to rationalize some of the properties of the molecules that were around and uh one case study that we published back in 2017 i'm going to discuss today is on this molecule ibidomide and so this is a a clinical compound in development by cell gene bms and this molecule is interesting um in that it has uh more potent uh binding to cerebellum than len upon so you can see that from this 2017 work here and you can see that it has a similar uh rank order increase in icarus degradation rate so i to me this was very interesting and it showed that um things that you can measure using modern molecular tools um could uh result in uh increased uh potency and of course these are things that you can now start to tackle with a rational manner and you could even start to think about applying structure-based design principles and so to see if there was a structural rationale for this difference in potency we solved the structure of vibrating wide bound to cerebellum and if you compare this to the structure with len um you can see you have your uh len here and you can see you have this extended ligand group on uh ibudimide that wraps around the surface of cerebellum it makes more interactions with cerebellum and so now you can rationalize the increased potency as being due to increased interactions with cerebellum and that causes increased degradation of the substrate so this is one path through which molecules of this class can be optimized now of course what we're talking about here is a much more complex system than the conventional drug discovery examples i talked about at the beginning where all you need is your therapeutic and your target protein we're clearly talking about repurposing machinery in the cell and that is uh something that brings all the benefits i've talked about it also brings some other risks and one of those risks is the risk of resistance because you're dependent on many many more components um for your activity in a cell and so to look for the components there were uh two teams um that published works in uh 2018 where they used uh crispr libraries to identify additional uh proteins that that that uh cerebral modulators and uh the thylamide analogs we've been talking about were dependent upon in addition to the ones that were known such as cerebral and i'm going to talk to you so that this is work from a celgene team led by gong liu and a team led by ben ebert and you can see here this is an example of a finding from this uh lewitowel publication and you can see that uh knockdown of this e2 so this is a protein involved in the transfer of ubiquitin to substrate to the target protein causes a difference in sensitivity in icarus degradation and in following up on that this is an in vitro reconstituted system this is work done by mary matasquiela's team at the celgiano bms and what you can see is uh this so this is a fully reconstituted system every individual protein is purified and what you can see is that you have this e2 protein ebe2g1 and you have another one ub2d3 and you have a priming event whereby a single ubiquitin is transferred so this is your parental icarus protein here this is icarus with the addition of a ubiquitin you have a priming event whereby you have the transfer of a single ubiquitin and then you need the second e2 to extend the chain so this is a priming and extension event that had been described in other ubiquitin ligase systems and you know very interestingly this is also the same uh sequential priming and elongation was found to be true for other ligand and uh neo-substrate systems um in this case gspt-1 and uh and that is a segue into a description of gspt-1 um so um so improving on the existing clinical activities the rationalization of uh ancient phenomena um were incredibly interesting exercises um and and i think are fun and hopefully valuable but really what you want to be doing with this type of system is getting into new space finding new substrate proteins and new clinical activities and i'm going to talk to you about one published example today and that is cc885 that's this molecule shown on the left and so this was work again done at the cell gene organization we published this in nature in 2016. so you have this molecule cc85 it has this core that is incre that is very similar to the other clinical compounds and then you have this extension um here and can just despite the very similar um structure of this molecule what you find when you put this into uh phenotypic assays is incred a very very different profile of activity so you can see um if you look this is each of these tiny little and probably unreadable um uh uh lines here is a tumor cell line and then across the top you have the iac 50 for cca5 you have the one micromolar mark here you have one nanomolar here so what you see is that there is a very broad spectrum effect of uh antiproliferative activities from this molecule so that's very very different to the clinical molecules and so uh what we were able to show is that this activity uh uh is uh both very potent but also very very cerebral independently so you can see here you have three several on knockout lines and you can see the activity of ccaa5 is removed in those so so the structure of the ligand the activity it was a cerebellum dependent event um and obviously at the time we became very interested in knowing what the substrate was um this was work that was uh done in collaboration with hiroshi hand and takumito again and uh what you can see is that they were able to pull down cerebellum ddb1 and in a drug-dependent fashion they were able to identify this band here and that band corresponded to a protein called gsptm1 so this is a translation termination factor it's a protein critical for cell division and you can see that this protein is rapidly depleted upon the addition of tc885 so you have uh down to a nanomolar of cca85 is able to clear gspt-1 in just four hours and again with the uh reconstituted uh in vitro ubiquitination system uh mary was able to show that with all of the components uh present you have very robust ubiquitin transfer to gspt-1 and you take any of the other components away and you uh you you remove that so we had a very strong feeling that that we knew what was going on here but obviously the piece of information that we were missing um was the structural aspect you know how was this molecule able to recruit that but first we wanted to tie this uh observation of ubiquitination and pull down and depletion of gspt-1 to the biology so we showed that uh gspt-1 degradation is necessary and sufficient to recapitulate the effects we'd seen in cell line so this was again work done by gang lu and his team and what you can see is that uh you have the the potent uh anti-proliferative effects in this cell line of cca85 you overexpress wild-type gspt-1 and you shift that and then you introduce this degradation resistant mutant and you completely remove um the sensitivity in the cell line to the molecule so so we knew that gspt-1 degradation was necessary for the anti-proliferative activity and then uh it was also shown that a knock down of gspt-1 using sh approach uh causes uh uh anti-proliferative effects so so it looks like it would be sufficient as well so then um on to the structural biology and this was something uh that we were um there was uh you know these things they they can look obvious in in retrospect but this was a project that was approached with some trepidation at the time because what we were talking about is a catalytic phenomenon and it wasn't a very potent effect it's a multi-protein complex that would need to be solved and we just didn't know if there would be enough affinity to drive our complex formation um in in order for us to get uh structural studies completed and now of course there's been several structures and and it's all uh it's much smoother sailing but uh but that wasn't the case at the time and so it was with great delight um that we're able to very quickly get to this uh a negative stain uh image shown here whereby you have cerebellum ddb1 and then you can see in a drug dependent fashion we were able to see this protein gspt-1 stuck on the end of the complex and this was work that was done in collaboration with the lander lab at scripps and that gave us the confidence to go into crystal trials um and and that confidence was certainly needed because uh this didn't fall out of the tree it's always very very painful to jump straight to the result um when um you know a lot of the uh a lot of the doing of these sorts of projects is slogging it out and getting it to work um but uh i will cut straight to the result and show that the the structure looks very much like the negative stain and you can see um that you have gspt-1 bound on the surface of this complex and if you looked you know very very closely you would see a tiny hint of cca5 bridging the complex there and for the first time we were able to have a really detailed look at uh a ternary complex um that was scaffolded by this uh synthetic ligand cc885 so and uh so this was a a a great joy to look and analyze it very much like we were anticipating we we saw interactions between the substrate protein and the ligand cta5 we also saw this direct protein protein interaction that we've been anticipating and that included the hydrogen bond to the residue adjacent to the species resistant uh side chain that we predicted in the 2015 nature paper so you can see the two uh side chains that are linked to rodent resistance you can see this e37 here you can see that uh why that one might mediate resistance to gsp t1 degradation it's critical in in recruiting this molecule cca85 you see these two bonds here and then you can see this v 380a and it's tightly packed against this protein interface you can see that even the very conservative change to an isoleucine would be enough to interfere with that especially at the sensitive location where you have this hydrogen bond now so we observed this protein protein interaction we observed the interaction with the ligand we obviously wanted to carefully validate the observations seen in the structure and the way that we did that was to do a surface scanning mutational analysis of cerebron so this was done in a uh immunoprecipitation format so you take your cerebellum you go through and systematically mutate side chains to alanines and um this is the thalidomide binding pocket in the middle then you do your immunoprecipitation and you look for residues that were critical for the formation of the ternary complex and then these are mapped with color onto the cerebellum surface and you see the fingerprint for the interactions with gsp1 and that for us uh perfectly corresponded to the interactions we observed in the crystal structure so we were confident that we were seeing a real phenomenon um in the crystal structure uh now that this system was up and running uh we went in a forward-looking fashion and just asked whether um the same uh dependency of interactions was found for other substrates in this case i'm showing icarus and although there are differences you can see there are differences here and here you can see that this core region is conserved and this is where the loop uh this uh loop feature binds so of course the question now is if the dependence on the surface is similar between these two different uh proteins perhaps the molecular feature involved is uh similar between these unrelated proteins too um so this is a zoom in on this protein protein feature from the substrate protein and so you have this hairpin um with a glycine and a tightly packed part of the pocket you have these three hydrogen bonds that come from the backbone of the feature and that's a critical feature here because the side chains can change between substrate proteins in this example but it's the backbone um that is critical which means it's it's a it's the shape of that feature that is important but not necessarily the chemical identity of the side chains and so i took this feature and was able to model it onto icarus there was no structure of icarus available at the time but we were able to predict one and able to see the icarus does indeed have exactly the same molecular feature and it has a glycemic exactly the same position but none of the other side chains were conserved with gspt-1 so so this was a phenomenal finding for us a completely surprising finding um in because you know we could have anticipated solving multiple multiple structures and doing compare and contrast exercises but it turned out that from solving the very first structure we'd actually defined a degron and so the degrom is the common molecular feature in a substrate protein that enables it to be recruited to an e3 ligase we were able to define the degree and actually it was it was a unique degree it's one that was uh driven by structure and not by primary sequence so so this was an incredible finding it was a it was a a very surprising one and one that was uh um and completely unanticipated for us so i was i'm just gonna try and uh show this uh uh finding in uh video form hopefully everyone can see it i'm sure the experience might be different for every individual based on bandwidth and other things but we'll have a go so so we have the complex now what you have here is your ligand bound and then creating a surface this is a the gain of function neomorphic surface on this complex and that surface then is able to recruit these neo-substrate proteins this is gspt-1 so you can see the fold it's adopted you see this hairpin and the protein protein interaction you can see a structure from the toma lab on ck1 alpha i'll talk about that a bit in a moment completely different fold the same hairpin feature and then switching to icarus initially we had the models and then again the timi lab have confirmed those with structural biology studies on zinc fingers and you can see the same hairpin feature talking against and so if you take those three proteins you can see they have nothing in common nothing in their fold nothing in the sequence but they have the same molecular feature uh here [Music] and so just to show that in uh just to summarize that in 2d form if for anyone that didn't manage to get the video um to work um what we can see is that you have different uh you have this common e3 ligase system with cerebellum you have different ligands and with those different ligands you're able to recruit different substrate proteins these substrate proteins don't have anything in common in terms of fold or sequence but they do all have this same loop feature that binds up against the ligand and so and to show you that uh sequence alignment and to show how surprising this is um you can see the glycine is aligned the glycine is found it's found at this position turn the laser pointer back on and this is where the glycine occurs here you can see that there's no sequence identity on them and obviously that's a great joy for a structural biologist because it's great to discover something that wouldn't have been accessible by primary sequence alignment so with that knowledge um you can go forward-looking and start to try and identify proteins that have the same degrom feature and one of the things that certainly me and my team we were keen to do at the time is to use this not just in a uh the way of trying to find new activities but also to try and explain some of the oldest clinical activities associated with these molecules and i'm going to talk to you quickly about an example of one of those uh today's uh this is cell four and you'll see in a moment why we were so interested um in trying to uh uh find this protein so so this is a zinc finger protein protein so you can see the zinc fingers shown by boxes here you see this array and you can see um where there's a star there's this degrom feature there's this loop feature containing a glycine and uh uh sure enough you can go in and then again this is using the in vitro ubiquitination system you can see that ubiquitin is transferred to the sink finger here then number two a little bit to number four but predominantly number two and that this leads to robust uh degradation of self or in a cerebral independent fashion and at concentrations that are pharmacologically clinically relevant okay and that's critical for all of these things because uh um you know they obviously to have a plausible tie into human biology it has to be at concentrations where um these things where the in vivo exposure is likely to be uh relevant and this is work that was published uh from my team in nature chemical biology in 18 and also by catherine donovan and eric fisher's uh crew um in e-life in 2018 as well so this was a concurrent finding fortunately the papers were very much in agreement with each other which is always a huge relief um and so this is why we were so interested in this protein in particular so mutations in cell 4 have been so these are loss of function mutations uh have been linked to a number of syndromes including aqua hero brain radial ray halt aurum ivec aquarino ocular syndromes okay these are syndromes that have a clinical presentation that overlaps with uh thalidomide embryopathy so closely that there have been uh uh clinical cases of misdiagnosis and that was uh the incredible finding uh for us what you can see is the title of a paper here from 2003 describing this phenomenon of misdiagnosis between the genetic syndromes and thalidomide treatment and so this is a very plausible human genetic link and indicates that it's the degradation of cell 4 that by by uh upon treatment with thalidomide that may underlie the the thalidomide uh teratogenicity crisis of the 1950s um it's also very very interesting that um you know anyone that could have read this paper back in 2003 if they'd been curious could have added some thalidomide to a cell uh and uh and watched uh southward be depleted that person could have bypassed the whole molecular glu thalidomide a targeted protein degradation set of work um and it's just uh it's interesting the route that sometimes you have to take to get to uh uh to these sorts of findings now obviously nobody should be using um validating this sort of work in in in human studies but what we looked for is a in vivo system that would enable us to look for a strong correlation um so uh here we have three systems here we have one that is sensitive to thalidomide embryopathy rabbit and we have two resistant um models we have a wild type mouse and a transgenic mouse carrying human cerebellon and what you can see if you look for the rabbit example is that rabbits degrade neo-substrate proteins in this case just using zfp 91 as an example so you see the loss of the stain upon salidomide treatment they also degrade cell 4. so that's great you know it's still plausible that cell 4 causes thalidomide embryopathy so wild-type mice on the other hand as we said because of those mutations they do not degrade neo-substrates and they do not degrade cell form now surprisingly the human humanized transgenic mass so this is the the mice that have the entire entire uh cerebellum human sequence uh incorporated into them you can rescue the degradation of neo substrates such as efp 91 so you can see that now solid mode is active in depleting cfp91 however you don't deplete cell4 um and so this gave us something of a control and a strong correlate correlative case uh linking cell for um to uh the imbibo embryopathic effects of thalidomide um and actually the explanation for that is because there's also mutations in mouse south for the block the activity and so just to to finish up our uh transition of uh this a system from a completely phenotypically uh and empirically defined uh biological phenomenon into one with all of the molecular tools that you would want um uh we moved on and sought to crystallize the cell for uh ternary complex and and we were able to do that and we published this earlier this year and so now you can see this is the molecular complex you have ddb1 cerebons and then the sink finger from southward bound on the surface so again you know we think this is the a plausible complex that underlies the 1950s uh tragic tragedy um you can go in and you can see the molecular details um so this is interesting from a historic perspective um but obviously it would be great if there was some uh use for this type of structure and that use obviously could be in uh in designing away from cell phone activity and in next generation drugs and with that view we did an analysis whereby we compared the binding mode of southward to icarus we saw this 15 degree rotation you can see that this corresponds to the shift in these uh in the hairpin the headpin is the same but there's this shift and uh we hypothesized that this was due to this steric interaction between the side chain histidine 417 and i371 um so we targeted those with mutations and in vitro ubiquitination you can see that you uh if you mutate this back to an alanine which is what you find in icarus you do indeed get much more robust ubiquitin transfer and on the other side of that interaction the isoleucine in a immunoprecipitation format you get much tighter tonery complex formation as well so it does look like salfor um is uh to some extent limited by the steric clash and it causes a rotation and that's great because now we have sequence differences and we have structural differences that can be used to try and uh generate the next generation of safer drugs okay so now we have a much more molecularly defined system we have degrons we have sequential e2s we have the identity of the components involved so this has really been a hq and we have a number of neo substrates and this is a very small list from what is available in the literature there's been a number of uh publications describing additional um uh substrate proteins um and uh and uh this is clearly gonna be a tremendous uh source of work uh for the future so uh and where does this leave us from a drug discovery perspective this is uh just a table from a review we wrote last year and you can just see that when it comes to approved drugs you're still in the targeted protein degradation space you still only have the thal line and pond um that have made it across that hurdle there's a number of other molecules in clinical development all of these are glues almost all of them are directed towards cerebellum though there is decaf 15 as well but now we're seeing the arrival of the heterobile functional molecules and so obviously it will be very exciting to see how this changes over the next few years but cereblon um is interesting as a as a model it's interesting as a case study uh for the glues um but now there are multiple other systems that have been described so the original coining of the glue term from this nature paper in 20 2007 based on the auxin analogues there's a cerebellum modulator that we've talked about today decaf15 and rbm39 degradation that also occurs with uh clinical stage molecules and then now we're seeing pre-clinical examples published including this work on uh direct repurposing of ddb1 towards substrates by small molecules and this was published by uh the teams shown down by here this year so this is a a cutting edge frontier in drug discovery and i'm sure there's going to be a lot of tremendous work in the future in this area oops i do not know what happened there just gonna let my computer think for a moment it's uh got the the spinning ah there we go and so just to wrap up um i'd like to acknowledge um a lot of incredibly talented and wonderful colleagues at celgene and bms there really are just far too many people to call out individually um there's a there's obviously a huge effort going on over there it's a great privilege to be able to contribute that during it was a great privilege to be able to contribute to that during my tenure there um we also had external collaborators uh benny bert hiroshi honda and gabe lander um and so um the individual uh contributors to the work i've talked about obviously that's all uh literature-based and so they are all captured in the authorship as well so uh so it should be very clear um who to speak to if you have additional questions so um thanks very much for your attention and uh and i'll take any questions uh that come up now thank you