 Today we have the pleasure to have Bruno Correia, assistant professor at the laboratory of protein design and immuno engineering of the E.P.F.L. here in Lausanne. Bruno holds a bachelor in chemistry obtained in Portugal and in 2010 he completed his Ph.D. in computational biology. After having spent some years at the Instituto Colbenkian Ciencia in Oevas, sorry, and at the University of Washington in Seattle in the U.S., where he was also a research assistant until 2011, and over there he developed computational flexible backbone design methodologies for immunogen design. And from 2011 to 2015, Bruno worked on the development of wall proteome small molecule fragment based screening methods in the Scripps Research Institute in La Jolla in the U.S. And from 2015 he is assistant professor in bioengineering here at the E.P.F.L. And as well as a group leader at the S.I.B., the Swiss Institute of Bioinformatics. Computational and experimental techniques to manipulate protein structure and function hold the potential to make transformative contribution in the most diverse scientific domains. Bruno's laboratory develops and applies protein computational methodology to design novel functional and therapeutic proteins. And the assessment of the computational work is carried by the experimental arm of the laboratory where the testing of the computationally generated hypothesis is performed. And today Bruno will tell us more about the computational design of the functional proteins for biomedicine. Thank you again Bruno for accepting our invitation and the floor is yours. Okay good. So thank you so much for the introduction. I guess it's a pleasure to be here, it's not really very far away from where I work, but somehow I don't get to come here very often. So let's see. So the first disclaimer that I want to do before I start my talk is that this is the third hour of lecture that I'm doing today. So if in the middle of the presentation I pass out and fell in the floor, you know this is to be expected, okay? So hopefully nothing is going to happen. So let's see. But let me tell you a little bit about the activities that we've been developing here at TPFL and when we started or when I started my own independent research group. So sort of like I gave it a broad title because we have a little bit of how should I say this, explorations going in different kinds of domains. But so today sort of like what we're going to hear about here is related to structure based computational protein design and I'm just going to give you a very, very brief introduction about what do I mean by structure based computational design and then I'll tell you about one of the main applications of the lab which essentially comes from a lot of the work that I've done in the past and that we're trying to push forward in trying to design computationally this what we call the PITOP focused immunogens for vaccine development. And then to finish up I'll tell you a little bit about the new branch of research that we have in the lab which is more related to designing proteins which can be used to control cellular activities. Okay, so computational protein design, right? So often when you think about proteins or most of us we actually think about them as sort of like single entities which arrive to us and there's not much we can do with them or they are what they are and we just do what we can with them. But if we think about proteins in fact what they have, they do have an evolutionary history which arises from the different organisms that they came from and the different conditions that they have and this essentially shaped their structure and function. So when we're talking about protein design the idea is in fact to forget about this evolutionary history and make these proteins do what we want rather than what they were evolved to do. So how do we do this? So in the end of the day in our lab so many people do this in very different ways but in our lab we like to use structures and in the end of the day structure is still or protein structure is still the best surrogate for the best way that we can understand protein function and therefore what we do is we use the structures or we make up our own structures and then apply computational methods and then we generate novel sequences that we then characterize experimentally. So this is the overall idea of what you're going to see. So not a lot of detail but you get the idea. The idea is that we have these structures and we're trying to find new sequences that have different functions into it. So now more in terms of the application of what we do so I think that for this kind of audience I don't really need to explain you why vaccines are important although a lot of people question these days but vaccines are clearly one of the most successful interventions in human health and it's likely one of the few interventions which actually was able to eradicate completely different kinds of pathogens. Now it is also true that there's still a number of pathogens for which we are lacking efficient vaccines so we're going to talk about some of these. Now talking about vaccines so there is one key ingredient which vaccines that work in fact need to elicit and this is antibodies so every time that we get a vaccine it's different for different vaccines but one of the most common features that you hear on vaccines that are eventually efficacious is that they are able to elicit neutralizing antibodies that can neutralize the pathogens that then eventually will get infected with. So what I'm giving you here is an image of a protein which is in the surface of the HIV virus and each one of these little codes that you see here is a neutralizing antibody so essentially the way that the structural biologist then looks at this looks at a structural segment which an antibody that is made by the human immune system can recognize and neutralize the virus so eventually this is almost like a map of the weak points of the virus so but of course that there's a number of challenges in terms of making better vaccines and more efficacious vaccines and some of the challenges are not just related to discovering neutralizing epitopes but but they are more focused in this idea of like how can we actually focus the immune response into making antibodies against these particular sites and so this is what people call immunofocusing but let me just drive you through this slide a little little slower so the idea is that these pathogens can escape the immune response by something that it's very clearly understood which is highly antigenic variability and what this means is that these pathogens are able to mutate and once they mutate the immune the immune system cannot recognize them anymore and therefore the antibodies are not neutralizing and we are not protected and then there's something which is a little a little bit less understood which is this what what people call immunodominance and generally immunodominance what this stands for is that for some reason there are pieces of structure in in in this proteins or in this in this pathogens where the immune system sees them a lot and makes antibodies against them a lot but it turns out that making antibodies against those particular regions of the molecule doesn't afford you any protection so now the problem with this is the following is that once your immune system is making antibodies and and it's it's binding to the pathogen it thinks that everything is okay but in the end of the day not everything is okay what because the kind of antibodies that you're making is in fact the wrong type of antibodies and so and this is what we what we call this idea of immunofocusing right where our idea is here to design proteins where we can actually modulate the response of the immune systems just so that the immune system sees epitopes that are more relevant than others that are more relevant so and just to finish up sort of like with this with this idea of this slide so some of these antibodies not only they're just completely useless some of them are actually arming to the system right so they don't protect you but in some instances they can actually provoke disease enhancement and so this is the case for instance for for for dengue for the infections okay so so with this now you understand what we're trying to do here in the big scope of things so let me tell you a little bit more of how we are going to tackle the problem still at the high level so the idea here is that by understanding what kind of immune responses vaccinated or infected individuals mount against particular pathogens and by immune responses I mean antibody responses and what kind of antibodies these individuals make and assuming that then we can pull out the antibodies or other people can pull out the antibodies and understand how do they neutralize the pathogens by recognizing different epitopes at the surface then we can use this information in order to make new immunogens and make better vaccines so that's essentially the idea and now I'll just give you a brief introduction about the pathogen that we are going to tackle on so most of the work we do in our lab is related to to RSV and RSV is a virus that you know it's not the most appealing virus to work but nevertheless it's still a very important virus in terms of the disease burden that it carries particularly in infants and people that are immunocompromised and still you know one important thing about it is that the the pharmaceutical and the scientific community has been searching for a vaccine for such a virus for more than you know now like 30 or 40 years and still doesn't have one so so this this makes it an interesting target for us and so just a brief history about it so it was discovered first in 1957 and one of the key events in the vaccine development efforts for this virus was a clinical trial that happened in the 60s where young children were vaccinated with a typical formulation for a vaccine which is a formula in inactivated viral preparation and what was observed in that in that study was that not only this was not protective as in fact it was disease enhancing and there was a number of kids that actually died in this in this clinical trial so this has been sort of like been hurdle in the field to develop vaccines that are based on the pathogen so in 1998 what we have is one of the first monoclonal antibodies approved for any disease in fact but in this case was to treat RSVF and this was to treat RSVF prophylactically and then what we start seeing in the 2010s and until now on is that we have been flooded with structural information about the viral proteins and the way that antibodies neutralize this viruses okay so and from this information what we can map is is a number of interesting things that are that are important for vaccine engineering so here what I'm showing you is a structure a structural representation of the F protein and what you see in these different letters is epitopes that have been mapped to be recognized by different antibody and neutralizing antibodies and so basically what this gives us is an idea so okay I understand that in this red segment of the protein there are neutralizing antibodies that can bind and neutralize this this pathogen the same thing for this segment the same thing for this segment and so on so what do we do in the lab with this information so in the end of the day what we do we use computational model modeling in order to design proteins that mimic the segments and then we characterize them and essentially what these immunogens look like if we put it in a structural perspective they look just like this so if you see you can recognize this epitope that is presented here this epitope is presented there and ultimately the idea is to make a number of different synthetic epitopes that then we can put together and and use it as a vaccine okay so to give you even a better picture of exactly what's happening in this process generally as you've noticed by now we use structures to do a lot of these things so here I'm giving you an example of a simple structure where we designed immunogens for vaccine development for this particular for this particular epitope and here I'm talking I'm giving you an example an HIV epitope but it's not really important but the ideas that we always start with a structure of neutralizing antibodies which recognize a structural segment and then what we do is we use computational approaches in order to search for proteins that have the same kind of motifs bearing in them and then we design their sequences and I'll tell you a little bit more about how we're going to design those sequences so once we then have these molecules that are essentially different but they have in common one of the different epitopes that we're interested we can then use or make these proteins in the laboratory and study them using biochemistry and biophysics and I'll show you a number of different pieces of data today about how we study these proteins eventually we'll push these things into animal immunizations and the last stage is to understand once we immunize animals with this with these proteins what kind of immune responses they they in fact elicit and the questions that we asked this immune responses is are they able to neutralize the virus what is the potency the breadth and so one thing that you shouldn't forget about this talk or at least about the first part of this talk is that the aim here is to design proteins that we elicit epitope specific neutralizing antibodies and that's what I'm going to be telling you about the most so and of course that we've had a number of important contributions from the immunology community just by being able to profile B cells and pull down large quantities of antibodies which bind to different sites and you'll see how we're going to use this structure in the in the or this information in the next slide so but let's let's use this oops let's use this molecule essentially as a map and the different parts of my of my talk will we'll touch in the different sites that we've been doing work and I'll try not to be too extensive but just to give you an idea of the methods that we use in order to do this so the first one that I worked still as a PhD student was what we call the site to and essentially what what we did here was to do exactly what I just told you and we use this structural information we designed scaffolds that mimic this this sites that were being recognized by the antibodies and then we were able to elicit or sorry to immunize macaques and see that as we were immunizing the macaques further and further we could have more animals developing neutralization activity so in the end of the day what was happening here was that this synthetic molecules that were designed in a computer in fact were able to work as vaccines now I think to give you a better perspective because I keep telling you here well we use computers to do this and to do that I think the best thing is always to show what is actually happening in the computer so the way that we actually designed these molecules so this is a method that is called fold from loops which I started developing also a few years ago and the idea is that using this method we can fold proteins with different conformations but also design sequences that then can stabilize this protein so what you've seen here is that we started from an extended chain until that the protein structure will collapse into a folded structure and then once it collapses into a folded structure we then start optimizing the sequence and that's what you saw in the step before when we when we're changing the sequences of amino acids so by doing a run like this you can then generate thousands and thousands of different sequences and of course that then this is when when the fun begins is when it's time to select the different sequences that you're going to to characterize in the lab so of course you cannot have 20,000 sequences that you're going to characterize in the lab you have to use a different number of criteria in order to select the sequences that then you're going to spend time on okay so but just to put things in perspective after this paper sort of like what we've learned is that and what we started being able to compare is a very simple simple question so and the question that that underlies this slide is as simple as this so how well do we do using our small synthetic immunogens that only have one single epitope of the virus compared if we do use the full protein from from the virus which has obviously a different number of epitopes probably up to hundreds of epitopes that can be neutralizing and so what you see here is that in this protein you have all this surface area that can be recognized by the immune system and here this is the site that we that we mimicked on our on our small synthetic epitope and so what we what you see here is essentially on the on the right side where with our little scaffolds we can elicit as much neutralization activity in in average in groups of non-human primates as the post fusion molecule that that is part of the RSVF virus so so and just to give you a little bit more of a piece of information this surface protein that exists on the RSV virus can exist both in a pre-fusion state as well as in a post fusion state but clearly the pre-fusion state is way more immunogenic and makes much more neutralizing antibodies than the post fusion state okay so so that's that was our work until I got to the PFL for for site two but then we started looking at different sites so we start talking we start doing work in and in site four and so in fact what we can start doing is also start categorizing how complex structurally this these epitopes are so you can see that the first one that we tackled at least in the scale that we have right now was likely one of the most simple epitopes that we could have tackled to start with so now we're going to move to the next one which is one on F and I'm just going to be relatively brief about it so there's a number of challenges related to this epitope but in the end of the day all that we're trying to build is a protein that is displaying this segment that you see here in red so and so we applied the methods that I've described you before and eventually we created a different number of sequences which we characterized and so to guide you relatively fast through this data so some of the things that we look at is if the proteins are well folded so what you're seeing here is circular dichroism spectroscopy and generally here we're looking for for the secondary structure signatures of the protein so here we were expecting a protein that would be mostly beta a beta containing structure protein and so that's what we see we also see that this protein even when you raise it up to high concentration is maintains maintains it its structure and also what we see is that this protein behaves as a monomer in solution we then also took collective NMR data for this for the protein and we saw that biochemically was well behaved so that was all good and fine but eventually we do these things in the computer and we come up with these sequences but generally the sequences are relatively suboptimal and so and I'll tell you what I what I mean by suboptimal there's often they are suboptimal in terms of sequence or sorry in terms of stability or in terms of the binding affinities that they have so once we have the suboptimal sequences what we actually do is we use this use display system in order to evolve the sequences and to further enlarge the sequence space that we sample in order to find clones that have the best the best properties and so essentially what you're seeing here on the y-axis is we have each each of these dots is a different clone of this protein so a different variant which might have one or more mutations we don't know yet until we actually sequence them but then we select the populations of cells that have improved binding and also improved display and the binding is the surrogate for the function of the protein display is eventually the surrogate for the folding of the protein and so and eventually then you do this for several different rounds and you get you get your proteins that bind the best of course that we can also do different in different ways and this is the result that I'm that I am going to show you next and we can do this in a single sort kind of idea and then use next next generation sequencing in order to understand the populations of sequences that exist in in this quadrant versus the population of sequences that exist in this quadrant and essentially what this gives you is if you have a sequence that is highly overpopulated in this quadrant it means that this sequence binds better than sequences that appear a lot in this quadrant so it's relatively simple but what you can then do is to actually map the mutations that you that you are observing so let's see so here we did this particular kind of experiment and then we mapped which kind of mutants we were seeing and how often they were being shown and from this data what we could see is that for instance mutants in position 20 it really doesn't matter which mutants are but if you mutate this this protein from is native aspartate or glutamate in this case to any other thing this would seem to make a protein that would bind best and eventually we end up doing the experiments and this is exactly what we see so we started from a computational design that would bind to the antibody of interest 80 micromolar and then we could with only two mutations we could actually bring it down as low as 16 nanomolar which is now the affinity that is the target is very close to the target between the antibody and the native immunogen okay so that's what we did for 101th so and this is still ongoing work but then of course that one of the epitopes that we were more interested for a number of different reasons mostly because the structural complexity was also very high and in terms of the methodology developments that it would force us what would also be more more demanding so this is the d25 epitope and so here the reason why I say that this structural complexity is higher is because rather than having one single segment of structure in fact if you want to mimic this epitope you have to put two sink two different segments of structure which have to be in the right three-dimensional orientation so to do this we actually had to come up with with a different method than than what we're using before or at least in addition to the method and the idea here is that while in the previous methods what I was telling you is that if we need to find proteins that have this structural motifs we go to databases and search for them where this method our idea is that we don't go to any database we just assemble them in the computer and then we design them and we see what comes out of it and so and how do we do this right so overall the idea is that we start by doing very simply laying out the 2d topologies that could be possible for a particular protein fold so let me just give you the simplest example that I have here so here what you see is the loop and the helix of the epitope which are represented here and you can see them here in the same color and what we're trying to do in this protein is to put two helices in the back that support this this structure now of course that there's a number of different ways there's a number of different topologies that that you can design or assemble with this with these four different elements of secondary structure so for this in fact we we we leveraged one scheme that allows us to just very quickly enumerate all the possible topologies based on on on a string enumeration system so we use strings and from the strings we can then just make all the possible combinations and connectivities and then we can reconstruct from the strings the topologies that we want so but once we have that that information then it's time to actually go from the 2d space to the 3d space and so and this is what what we do around here so then we have to to start working with a structure of the epitope and we locate the structure of the epitope on a three-dimensional space on the on the y-axis and then you start placing different secondary structure elements which were determined by our 2d topologies once we have this then we can drive spatial constraints and we can use the same methods that I've told you before in order to do folding in design of the proteins so and that's that's that's essentially you know an overview of the new methods and the new ways that we're following in order to design proteins so here let me tell you about one example that we have for a protein which we designed four helices in the back here is the epitope and another one there on the side in order to support the epitope itself so again we designed a number of different sequences in this case we designed around 10 and two of them were expressed and purified so we could actually study them biochemically using size exclusion chromatography and light scatter we could also see that the proteins even when you raise the temperature up to 90 they were still well-folded and stable and more important than that they did recognize or so they had a binding signal to their antibody of interest and here is just a mutant to confirm to confirm that this binding is specific so of course that we were still in the range of one to two micromolar which is extremely low for what we wanted we were aiming for something like 125 picomolar so a little far so in and eventually we turned into yeast display system again and eventually we were able to with a with maybe around five to ten mutations to bring it down from one point two micromolar to around a hundred and forty nanomolar which is now better than what we had before we also found out that during this this yeast display experiments that we didn't need such a long protein but we could actually live with a smaller protein so we could deplete one of the modules of the protein and that that then this became our our leading scaffold okay so but now taking a little bit of a break about all this kind of modelings that I that I was telling you and regarding the strategy that we're following so obviously for for this strategy we mostly rely on one structure and this is the structure that I'm showing you and this is true for most of the epitopes that we work with but as we know the responses or the immune responses they're polyclonal they're not really monoclonal so when you have a response to a particular site of a protein you'll have multiple antibodies against it so together with that you know there was also another thing that it was bothering us and what was bothering us was that when we would look at the several structures that were available in the PDB you could see that the epitope didn't always have the same conformation so this is this is a little bit of a problem right because we didn't really know if the conformation that we were working was was valid or not so this is when we turned this is when we turn to this study well let's see don't send I guess it always happens when it shouldn't right okay well good so so I was telling you about this idea that you know immune responses are really mostly polyclonal and not monoclonal and generally all that we're doing is is really based on one single monoclonal antibody so here this is when we turned for this to this publication came in coming from actually a company called edimabs where a large large panel of human monoclonal antibodies that are neutralizing were isolated and so another question was was was very simple right you know so we're using one single antibody as as a template now if we use if we if we check if our scaffold will bind to a panel of this T25 epitope specific monoclonal antibodies how is it gonna look and I think the answer is is just right here so if we compare our latest version of our of our scaffold compared to our initial versions we can see across a different number of human neutralizing antibodies we can see that the binding affinities are definitely getting much closer to the reference binding affinity and so essentially what this says is that not only biochemically but also we use this information as a surrogate for the structure that we're mimicking eventually what we're saying what we're seeing here is that we are mimicking closer the epitope that is being presented on the RSVF protein which is in the end of the day our main aim and so so this is just some of the deepest biochemical and immunological characterization that we're doing with that with our scaffolds so so yes okay so in terms of how we stand in terms of the epitopes that we have covered right now we really we have synthetic immunogens for three of the different sites but as I told you you know the idea is that we not only want to do this computational modeling and design but we also want to find out what these things are worth as vaccines right so and this is the time where in our lab we turn into okay well let's inject them into mice and see what what we can do okay so a few things that we that we've learned about it and I'm gonna be really really fast about this it's just really two slides so so eventually we had to optimize the conditions on how we actually have to inject this this antigens into into into the animals and we had to find out which one was the best the best adjuvant eventually we did this and here I'm showing you most of the data for the site two which is a scaffold which we call FFL01 but I guess the most exciting result is essentially that we do have neutralization activity this had been seen before in non-human primates but we had never been able to see it in mice and of course that there's a number of different implications with this but it's good that we can see it in mice because now we can work in mice and we don't have to use macaques and that's that's always a good thing but so you can see here that you know above this sort of like protective threshold we can get a few animals that really set the standards and we can do our further development for for next generation immunogens starting from here so this is the data that we have for 401 and for site two but also we also have immunogens which are the D25 immunogens which may make a different site in this case site zero and also what you can see is that the same thing is what you observe so you don't have very strong neutralization activity but in fact it is there and of course you can see and this is something that I also like to bring up is that if you use the native protein then the neutralization that that you achieve is much higher and this to some extent is to be expected once that you have the native protein in fact you have many more neutralizing epitopes so it's it's more likely that you that you will elicit neutralization activity but so with our simplified immunogens we can we can definitely see some of this of this starting neutralization occurring so of course that the idea is to put these things together and so just to sort of like wrap up this this part of the talk I'm just going to summarize what what I showed you in the last slides where the idea is that we start by leveraging structural information for neutralizing epitopes and we we create synthetic immunogens which mostly mimic the structure of the epitopes once we have this we use in vitro evolution to improve their their biochemical features and we characterize them and then we study them in animal immunizations and of course that here then starts a whole other field of things or or number of questions which I haven't talked too much about today but the idea is to figure out if we in fact induce neutralization activity okay so I'm going to go very quickly over this so ultimately I've showed you a number of different aspects related to how we design and the novel computational approaches that we're using to design the novel functional proteins in this case we're using immunogens but this could be actually used to design any kind of other functions into proteins some of our initial d25 epitope scaffolds in fact are able to elicit neutralization activity and the idea now is to keep doing these efforts for other non-neutralization epitopes and eventually the aim is to immunize with all these epitopes together and see if we can bring this neutralization activity up okay so that's it for vaccines for now and I think I still have sort of like five or ten minutes kind of so I will just tell you a little bit briefly about a completely different subject that that we work in the lab and that it's also of course related to protein engineering but in this case is a little bit of a more broad application where we're trying to develop proteins where we can tune their activity using small molecules and of course that the most the biggest realm of applications that we see here is in in applications related to synthetic biology okay so the basic principle principle here when we have this extra dimer forms this will regulate some kind of cellular activity and so if a is interacting with be eventually this the output that we're going to have is an active cell if we use a small molecule a that that can break the interaction between a and b so this will eventually turn off some sort of like signaling pathway which will render the inactive cell for a particular function so this is at the very general perspective of things but the principle is simple and the idea is that we're building on and off switches by the action of small chemicals mostly of them drugs and what we call this is the Sims or the chemically induced monomers so let me just give you sort of like a very you know straightforward application for this so let's say that you know right now a lot of people are excited about this idea of regulating activities of T cells and one of the concepts that arise lately is the car T cells but they have a number of different limitations and some of these these things that we're working on could it could potentially be applied here where the idea is that once you have the receptor in this car T cell it is only active when there's a all this activation domains and the recognition domains and the cancer cells come together but in case you can use a drug in order to detach this activation domain then this this car cell is no longer active so that's kind of it for for the principle but so but how do we really want to achieve this so so the way that we're planning to achieve this or that we are on our way to achieve this is to use protein design in order to design this this different proteins which we can then dissociate using small molecules so we started with some initial information from available in protein in the protein data bank where we have a segment of protein structure although in here it is a peptide and it has been crystallized as a peptide where we can displace or we can compete for the same binding site using a drug okay so in terms of the computational protocol very very briefly the way that we do this we first start I'm searching for protein scaffolds which we can place this functional segment we check if the scaffolds can bind to the binding partner which in this case is shown in in gray and then we put the side chains that are responsible by the interaction we design the interface and we design and we select the designs and eventually what you end up after this this computational protocol is with a number of different proteins which all resemble the structural the functional motif that you have here okay and so and then of course we go to the lab we make these proteins and so let me just tell you a few results that we got recently so the first the first question that we always have to ask is you know do our proteins bind to the target that they're supposed to bind so here you go so this is surface Plasmon resonance data and what you can see is that once you have these proteins interacting with each other you can see that they have a higher signal and once you you you determine the affinity which would divine to each other it's around 450 pack picomolar which it's pretty high affinity then another types of experiments that you can do is you can mix the two proteins together and see if they form the dimer that you want but then you can also mix them in the presence of the drug right and so here I'm showing you very classical chromatograms of size exclusion chromatography and so this is a peak that corresponds to the proteins in the absence of the drug meaning that the proteins do elute at the higher molecular weight and they form a dimer or once you actually do the same analysis in the presence of the drug what you see is two distant two different peaks which each one elutes at their own weight meaning that the dimer is broken down so this is more of an equilibrium experiment but here you can also do it kinetically right so so what what we're doing here is we have the two proteins binding together and this is the blue signal but in fact once you inject the drug you can see that the interaction just essentially goes away ultimately the way that our our interaction our our dimer dimeric system looks like and this is a computational model this is essentially what we what we what we achieve is a dimer that binds to this protein target which is in this case is bclxl and that we can displace with the drug that is available now of course until now I only told you about computational models but often computational models are what they are sometimes they're right and sometimes they're wrong so when we do protein design one of the things that we that we are required is to solve crystal structures to make sure that the models we generate in the computer in fact are accurate so what I'm showing you here is a superposition between the computational model so you can see that this one is in cyan and you can't quite see very well but the dark one is the bclxl and a crystal structure which was determined so one of them is a computational dream the other one is the reality and once we actually superimpose these two crystal structures you can see that our prediction matches very closely to the real crystal structure okay and I think with this I'm just going to close and tell you a little bit about the future perspectives for this sim so for now we have multiple systems that we're working on and our idea is to expand this design strategies to to achieve both on and off switches and then to apply this particularly to modulate functions in cells being those CAR T cells or just simply protein cellular localization as being that one of the key strategies that cells use to modulate their function so okay so with this I this brings me to the most important slide in the presentation which in the end of the day is the thank you slide for all the work that people have done in the lab and particularly the team of postdocs and PhD students and the staff that has helped us that that own this work and essentially this is this is we I can only be presenting this here today because they gave their hard work into into this project and of course also the the funding that we have been able to obtain without that wouldn't be possible and so with this I'll be happy to take questions if you have