 So Jacopo has started recording and I'm glad to present our webinar speaker, Talipe Keros. He has recently established a laboratory in George Tech and Emory University. Since two years, if I correctly remember, his team has been studying addressing problems of cell possibly of biopolymers, and their role and possible applications in bioengineering systems and biologic general. In today's talk, Talipe will tell us about his research and results related to liquid-liquid phase separation mechanism in cell systems and skin. All right, Talipe, I will concede the stage to you. Thank you, Roman. I really appreciate the invitation to speak to you guys. I'm excited to interact and show with you guys some of the work that we've been doing over the last actually 10 years now, the second liquid-liquid phase separation questions in both engineering and biological systems. And the focus of my work and my lab is really self-assembly of intrinsic blood sort of proteins. And so as you know, proteins come in many different flavors, but we really focus on proteins that are unable to assume a defined trimational structure. So if you run multiple simulations, they really do not establish the same trimational structure. And we're interested in those proteins because they really excel at undergoing a transition from a solute state to an insolute state in a similar responsive manner. And because we have full access to molecular architecture with these proteins by encoding them genetically, we can then program their self-assembly into very defined structures through making use of this phase separation driven assembly. And so this interesting phase separation driven assembly is particularly striking in the context of proteins because of the vastness of the protonogenic world is really dramatic how expansive of a system it is, and it is almost daunting but also exciting to be able to explore this kind of purpose in such a diverse system. And we're really thinking about it in the context of protein-based materials. So how can we then harness these kinds of phase separation driven assembly processes to control non-structure assembly with biomedical applications in mind? So here, for instance, we have nanoparticles that either assume a nanoparticle warm time morphology or nanoparticles that assume a spherical morphology. And more recently, we've been thinking about how this phase separation driven assembly might be playing a role in the context of the assembly of tissues or the purpose of tissues, and particularly we've been interested in the skin. But before I go into the details of our work dealing with nanomaterials and these processes of phase separation and living tissues, let me back up a little bit and just drive you through, you know, how we might be thinking about this process in the context of cells and the context of biology. As I mentioned, these are proteins that will go from a solute to an insoluble stage in response to the number of stimuli that can be temperature, enzymes, pH, but of course also concentration. And so if you think of the cytoplasm and you're having these proteins being soluble under certain conditions, you might think that at some point if the conditions are met, they might be able to undergo these kinds of phase separation events so that they would be mixed from the complex milieu of the cytoplasm forming very defined compartments within the cell even without the need for a membrane. And so it's become a part now over the last few years that there are many structures within the cell. And here I just highlight a few, like the nucleus in the nucleus for instance where we know these kinds of structures within the cell do not have members and yet they're very, very stable. But I think a turning point for cell biology and intersection with the biophysics of phase separation really came back to 2009 with the pioneering work of Tony Hyman and Cliff Brownwin at the Max Planck Institute in Germany where they look at germ growing in C.L. and embryos and what they observed was if they were to squeeze these embryos between two glass lights, these germ granules would start flowing in the span of seconds as if they were liquids. And so I got people thinking about these membranes organelles within cells not really as odd, but as potential really exciting structures within the cell that will have unique material properties like these liquid light behaviors. There's a good people thinking about, you know, how could that be interpreted by the cell programming this in cellular mechanisms. Excuse me, I don't know if we could do something about it. People in the chat saying that your slides are pixelated. I don't know. I must say that there is sort of difficulty to read the captions and the figures. I wonder if it has, I wonder if it has to do with the video sharing format. Let's see if, let's see if I stop sharing and do not optimize for video that improves it. See, because I haven't optimized for video before. Does it look better now. Yes, yes. Interesting. It's the optimization for video that is messing it up. So we'll see when I play a video first, I'll ask you how good the quality of the video is so that we stop it. And please, please warn me that I questioned the chat because I don't have the chat on my on my screen. And so, you know, this turning point back to 2009 really a pioneering science paper, people thinking about liquid liquid behaviors within cells. But a lot of traction and over the last five years, seven years. There's been really a lot of excitement and around interpreting these kinds of phase separation events in the cell in very intriguing mechanisms like, you know, my current name is a silencing teacher response. The assembly of heterochromatine, the sensing of DNA inside the cell, the spinning assembly, the assembly of genomic enhancers, and more recently, emerging roles in tissue biology, particularly tan junction assembly, and so I was planning my own work, skin barrier formation. And so really, this idea of how phase separation driven processes might be playing a role inside cells is really touching a lot of different aspects of biology at this point. Here's an outline of my talk, however, I'll begin with discussing some of the efforts that we made over the years to learn how to encode phase separation in in physical sort of proteins or physical IDPs. And then I'll move on to how that worked and enables our ability to develop a squeeze of ways to examine these behaviors in tissues, particularly on the context of skin. And I'll finish up with, with I think what what it's going to become a frontier for the field which is to understand not necessarily the equilibrium processes that that underlie phase separation but, you know, perhaps more excitedly now, the non equilibrium aspects of this process is to. So let me just clarify what I mean by by phase separation behavior, and there are two different flavors that you need to be aware, and they're for sure known as LCSC or UCSC phase separation behavior. LCSC stands for lower critical solution temperature and UCSC for upper critical solution temperature but you'll become clear in just a second. Think of two different polymers. Polymer A and Polymer B. And here what I'm plotting is the dependence of this interaction parameter on temperature. And for Polymer A, if you heat the interaction between these polymers is winding down, is decreasing. That's what for Polymer B, if you increase the temperature, the interaction parameters going up. So that has a consequence in terms of phase separation. Polymer A, because interaction parameter is decreasing with temperature, you end up with this kind of phase diagram in which above a new CSC or upper critical solution temperature, there's only one phase. But below that, there are strong interactions because that interaction parameter is strong, and it leads to phase separation. The opposite is true for LCSC. There is a lower critical solution temperature above which there are two phases because interaction parameter has increased, but below that is only one phase it is not phase separation happening. So we have a phase diagrams to really understand LCSC or UCC behavior is complicated. We've taken a very simple approach in which we simply look at cloud points. So imagine here a solution having a polymer given concentration, and we simply try to treat temperature. So we change temperature of the solution measured absorbance or triviality. And what we can detect in this case is that it's a short transition from our solar states and absorbance to an insoluble state that gives us this scattering of the light. And the opposite is true for LCSC and UCC. So you can see that for LCSC, these polymers will become insoluble as you increase temperature, whereas for UCSD, these polymers are going to become soluble as you increase the temperature. So we're not necessarily determining full phase I, we're looking at this cloud point temperatures. But how do you then, you know, start understanding how to encode these behaviors into proteins that as I mentioned the protein engineering world is fascinating but it's also a vast and hard to span. So what we've done is we've taken an engineering approach in which we come out with a, I think a clever approach which is to look at repetitive proteins. So these are the proteins with a simple repetitive architecture in which we take motifs that are based on an architecture composed of crawling residues and glycine residues that are spaced by a number of high residues. So for those of you who are not familiar with proteins, proline and glycine and two amino acids that are well known to be disruptive of secondary structure in proteins. So with this we were hoping that would end up with a blank slate. Polymers that were IDP in nature because of this proline and glycine being present, but with sufficient spacing between those kinds of punctuation marks that would allow us to then span this sequence space so that we could change amino acid composition. And with time change syntax. So even for a fixed amino acid composition then it starts changing the order of those amino acids. And we built many, many of these kinds of polymers over the years. Again, spanning this proline and glycine with different kinds of spacing. And with composition so that we would expand very hydrophobic sequences or very hydrophilic sequences again to sort of introduce new flavors of interactions into this blank slate of disorder IDPs. And with that, excuse me, may be precise in the previous slide so basically you fix the pro and and glee and yes and inside this X and would be various combinations of N. Yeah, so you can see for instance down here with by no spacing and equal zero and equal one and equal two three to four. And these are just amino acids that we now can add in. Between pro and glee module. Exactly. And then there are many copies and then there are many copies of each of these motifs so we then play with the length and I'll show you just in that in that a second in that. That is the that is the syntax of the individual repeat unit and then we can make however many repeats we want. But it's a very simple way to keep the structure disorder and allows us to introduce different kinds of interactions by different flavors of amino acids into that blank slate. We first show that is a very robust blank slate. So if you know this is a sample of different employees that we made it. And at room temperature, if you look at cyclodecoism, which is a very simple way to look at disorder. This is a typical sort of spectrum that you would see for for a sort of protein. And so no matter what kinds of spaces between probably analyzing we had or what kinds of things we added, we would always end up with a disorder protein. So that was good for us. But the interesting thing really came in that, depending on the class of amino acids that we introduced, we were able to find a broad range of LCST behaviors. So here you have, you know, all the way from 80 Celsius to 30 Celsius. And so that repeat length of these motifs as we make different lengths of these proteins would also tune that behavior. So really rated tunable LCST type behavior. And the same was true for UCST type, where depending on the interaction that we introduced, we were able to actually span a wide range of UCST cloud points, again, using molecular weight to tune or repeat number to tune that behavior but also composition. And although I don't have to go over the details of the heuristics or the rules that we uncover. Largely, we should say that for LCST, whenever we introduce hydrophobic interactions or pure hydrophobic interactions, these components will exhibit LCST type behavior. But whenever we introduce cation pie, charge charge, hydrophobic and precisely a combination of those, we wouldn't end up with chroma double XCAC behavior and I should clarify that all of these measurements were done in in simple phosphate buffer buffers. So they're actually sort of physiological buffers and like a lot of the, the work that have been done is the polynomial literature in which those were usually done in the conditions that were not physiological. So we're interested in trying to understand the behavior dispersing conditions that were close to physiological. And I encourage you to look at this nation materials paper which summarizes some of those characteristics. What I think for us was that at the beginning of a lot of these efforts, we knew very little about how to sort of go over the sequence space of amino acids to program these behaviors. But after a lot of this effort, we realized that there was a large sequence space that we could expand to encode these behaviors both LCST and UCST, even, even there will be regions in which those behaviors will start interacting. And that was exciting to us because it pointed to the fact that in this humongous space, they had to be native or naturally occurring IDPs that perhaps were fulfilling all those sequence requirements. So we began to write our heuristic as simple rules, a simple script that we could plug into our search of the proteomes to identify new IDPs that might have this interesting phase separation behaviors. So in our prediction, one of our first predictions was an interesting protein named filaggrin that we thought had all the requirements to exceed with UCST type phase separation behavior. And here is a snapshot of what that protein looks like is it's highly repetitive so R indicates repeat one repeat two repeat three so very repetitive architecture similar to what we had before in the poems that we're making. If you look at this plot of disorder, you'll see that the blue domain is the only the order domain, all the structure that the other domains of the protein are here at one which is that you know basically fully disorders expected to be fully disorder. So very intriguing architecture, but also low complexity so the entire composition of the protein is dictated by only a handful of amino acids. So basically we're intrigued by the presence of histidines this age. This history needs is one potential source of those cation pi and pi pi interactions that we know were important for UCST. And not only that but they were really really large protein so both in the mouse or in humans, we knew that these were among the largest proteins in those proteomes, which we thought was important to encode a potent phase separation behavior based on what we know about phase separation behavior at that point. And most intriguingly to us, these proteins do contain a lot of charged residues but they're exclusively by a source our gene and residue so there are two naturally occurring positively charged amino acids, our gene and lysine. And these proteins as they were used lysine they just prefer to use our gene, which we knew was important to encode phase separation behavior and the physical conditions and not only this was true for filagrin in different species, but these proteins that I'm listening to RPTN, horny and tracheal, tracheal, these are patologs of filagrin. So these are proteins that are in the same genome loss genome loss eye as filagrin. And they are supposed to be closely related in function to filagrin so not only are we seeing the filagrin is fulfilling conditions for being a CST type. But we think that our proteins in the skin and also build like that so it really got us thinking about, wow, this could be a really interesting protein to study to start encoding a potential work for phase separation in a tissue in a male tissue. So these brings me back to my outline. So I already told you just a very quick overview of our efforts to sort of sample the sequence space of amino acids so encode phase separation in IDPs. And now I just told you how we're not pivoting into thinking about using this knowledge of how to encode phase separation IDPs to then uncover phase separation behaviors in exciting biological systems and our focus on skin. And it was exciting to us because there was a raise some exciting work from the genetics field suggesting that mutations in this particular protein filagrin were strongly associated with skin barrier disorders, particularly one known as a topogenomatitis and these mutations are nonsense nonsense means truncating mutations. So it's actually a stop codons that appear. And these mutations are all over the place, but the most common mutations are stored as the end terminus of the protein. And these mutations again lead to where strongly associated with this disease activity my tie is the skin doesn't look normal. It's highly environmentally dependent which was intriguing to us because I mentioned that phase separation behavior isn't very similar dependent. And so people with this disorder when they go to call or dry environments their skin is not happy. The functional defect, however, wasn't clear at the time of this work. And we suspected that perhaps they could look at the separation and defects in the behavior of this protein lagging might be implicated. So, where is filagrin then you know that was kind of the next obvious questions for us and let me just work you through what skin looks like. And so here is stem cells of the skin of the epidermis and these cells like to move towards the surface of the skin. And they as they do so they differentiate and they form different layers are they're called the spinos layer the granular layer and then the corneum is the surface of the skin at the very top. What was intriguing though is that this this layer known as granular layer was known as such because of the presence of these kgs per carotid granules which I depict in red. And these cataract line granules is exactly where filagrin was known to recite. So here now we had a case where we predicted you see a C type of behavior. We look at the literature and we find that the protein is actually residing in really intrigued in membranes grants within the skin. And this is what it looks like in in TM. This is the part of the skin that determines the better function of the skin, because it kind of layer with these black deposits that account for this cataract granules, and then you have the corneum. So this static view people didn't really know what these granules did in the skin and we thought that it was a really interesting system to study and perhaps uncovered an interesting role for face separation and biology. And to give you just a review of what's coming. So we developed to then go from this static view to then be able to do dynamic studies of how this process is happening in the skin. So we developed this technology known as face separation sensors. And what I'm showing you here is a view a confocal microscopy view of the skin looking from the top down the microscope is kind of on the surface of the skin. So as I played this video, you'll see that over the span of about half a day, a lot of these compartments of granules within the cell disappeared to then lead way to the formation of the surface of the corneum and the skin surface. Roman, is the video displaying properly. Yes, yes video is running. Okay, great. So just to give you an overview of what's coming. Sorry, Felipe, could you, could you repeat the video because I'm not sure it runs. Yes. Could you point to what are the granules, this green balls. So yes, so the face separation sensor, this is an image from the top down through confocal microscopy. This sensor is localized into a couple of granules, and I'll explain that a little later in more detail about what you'll see is that over the span of about half a day. We're seeing tremendous remodeling of those structures within the cell as the grinder cells transition to the corneum and I'll go into more detail I just want to give you a sort of a little peek into what's coming. So we see how this green granules smear out. Exactly. That's where the liquid liquid separation stops, right. Exactly. So what I'm trying to suggest is that we went with, you know, at the end of this talk, we realized that we went from having this static view in which we know very little about this schedule and granules and how they were important to skin biology to then develop in a this dynamic view of what's happening through the lens of face separation, but I'll, again, I'll go into the details of how we did that and we actually uncovered. Okay. So these are some of the questions that we had. We know these granules were present in the scheme, but you know what are the material properties where the liquid like as perhaps will be suggested by the UCC, the UCC behavior of that we predicted for phyllagine, where they store in things. How are they interacting with other organelles within the cell? What is the composition? Some very fundamental biophysical questions and then of course more of the deal with the mechanisms. And we approach this work in two parts. In the first part, we took a very sort of engineer approach in which we said, let's just tag phyllagine itself or variants of that phyllagine to probe the architecture of the protein and its face separation behavior particularly with regards to the mutations that we saw in patients. And in the second part was okay, but how can we move this work from these simple systems to the skin itself? And so we had to come up with a strategy that we call face separation sensors to then be able to uncover a lot of these processes in the endogenous skin tissue without improving the endogenous process of trying to get as physiological as possible by looking directly at the skin as it's being developed. So the first part, you know, is a very simple engineer approach. We basically make genes encoding for different kinds of repetitive proteins that look like phyllagine. So these are eight is a sequence taken from human phyllagine, the eighth repeat, which looks very much like the 10 and the eight and the seven and the six because they're very, very similar in composition. So we can make many copies of those repeats and at any iteration of this assembly process we can add different non-repeat domains. So we can add fluorescent proteins to be able to follow their behavior. For instance, we can add this S100 domain that I mentioned is the only structure part of the protein of the variant end terminus which was written down to be a dimerization domain. And these are really large genes that we need to build, but we succeeded in building those about 16, 17 kilobases in size, fairly large. So here's the actual data and what we think is interesting. So here's a protein in which we are fusing a red protein, a fluorescent protein MRFB, with one of our engineering constructs in which we have end copies of these repeat eight unit from human phyllagine. And we're going to set the phase separation propensity of these constructs inside the cell from anywhere from one repeat of these R8 domain to 12 repeats, which is the full length of phyllagine. And a really interesting genetic trick as well to be able to be quantitative. And what we're doing here is we're using a signal that is coming from viruses to be able to couple the expression of the translational level of two different constructs. And so for every MRFB R8 molecule that we make, we're also going to make one copy of these chromatin protein bound to GFP so that we can visualize the nucleus well and for every molecule of this we'll have a molecule of the MRFB R8. And so here's what it looks like in the cell. For any equal one, if we express this in keratinocyte, these are the cells of the skin. In culture, you'll see that the red signal, the magenta signal is diffusing the cytoplasm, there's no phase separation. If we have even two copies of this, and this is again at the same concentration level we're using the GFP signal in the nucleus to judge concentration. We also don't see signs of phase separation, these molecules are diffusing the cytoplasm. But as long as you get to four repeats of that R8 domain, now you start seeing some signs of phase separation. Some cells would exceed the size of phase separation. And now when you go to eight repeats, you basically always see signs of phase separation, at least at this concentration. But this is all shown at a fixed concentration, so I can also do these experiments over a wide range of concentrations. And this is what it looks like. And it also surgesed taking a peek at the critical concentration for phase separation inside the cell. So here what I'm showing you is. Excuse me, can you explain what is the, so phase separation as a person, what does it measure? That measures the percentage of the signal in a given cell that is within residing within a granule or within a phase separated domain. So how much of the signal total within a cell is residing within this phase separated compartment. So here for instance, there is nothing within that distinguishes being phase separated. Yes, here only 14% of the signal is to our to what we can just be in phase separated here. Yes, yes, thank you. And when we do that over a range of concentrations, we get a sense of the, of the actual phase separation behavior. So take a look at this for any cool one so that they are in black. This is concentration on the x axis. So different cells expressing different levels of these proteins. And even if a cell expresses a lot. There is basically no cell that we can detect with signs of phase separation. And this is true for any cold to as well here showing red. But once you go to any cold for as I showed you before, then we see some cells that when they express a lot of the protein, we do see a strong size of phase separation. And this deepness of this of this transition increases as you go to any cold eight and equal 12. So for proteins that are as large as what we see in humans, which have anywhere from 10 to 12 repeats of these domains. We see that phase separation is very, very likely to occur at very low concentrations. And what I'm showing you here is is just copies of our aid, but I also mentioned this blue domain, which I said is the only structure part of the normal protein that we find in humans. And that curtain was going to be summarizing in a custom dependent manner. And so we decided to also test that and the reason why it's important to look at the domains without it is because we also knew that in the process of skin differentiation, these domains cleft. And we want to know how it behaves when it's absent. We also wanted to know what happens when it's first made and it's attached and it hasn't been clipped yet. And what we can see in those cases is that cells that don't that they still have that is on a domain that structure part of the protein. If you look at any cold to for any cold to and red. So no, no, it's not a domain. There is basically no phase separation, but if you add this kind of domain now in purple, you see a slight sign of a separation so not very much happening but if you go to an equal for so in in gray. In blue first you have no it's not a domain. There is like my face separation, but if you add this kind of main now in in gray for an equal for you see that is a much sharper face of personal behavior so that's not a main when we add it. It does increase the potential for face separation, which makes sense to us because initially when the protein is made, you know, the protein needs to undergo for separation, but most likely then it gets cleaved at a time when it's no longer needed. So we can quantify specifically to know how much the protein is needed to get quantitative. And so here in any cold 12 is what our normal human would expect I would have and we we see that as little as one micro molar 1.6 micro molar this protein is sufficient to keep the stress and to undergo phase separation. But for someone with a mutation that will cleave the protein to say have only two repeats, even if it hasn't started to remain the critical concentration now goes to almost 2 million molar. So basically you have to make so much of this protein in no longer really exhibits phase separation behavior or we predict that a person with a mutation that then leaves such as more protein wouldn't have to to to trigger phase separation within with those cells. So what about the behavior of these granules that are formed by phyllagin. So here's a sale, and we have a granule that is formed by phyllagin which is stuck with MRFP, and we want to know the dynamics of that protein within the granule. So here's a large protein, you know about 400 kW in size all the way to 500 kW in size. And we can do for a leaching experiments where when we bleach this particular granule, only 13 seconds there we begin seeing that the signal has recovered so there's a lot of mixing happening, even though these proteins are very large and moving very quickly. We quantify that. So you see here this is a leaching event. There's a loss of fluorescence, but on the span of a few seconds you see that it's perfect recovery so these dynamics are happening within the granules are really or some sort of very liquid like dynamics. Now the mutation that we see in humans, when you go from having, you know, many repeats like we see in humans having very few repeats of course as you shrink in this, they shrink the protein because there's a lot of pain mutations, the liquid dynamics that we then can measure with FRAP are affected because these are diffusion processes that are multi-size dependent. But how about, you know, what are they actually doing inside the cell? So here's an image of cells that we engineered to express some of these constructs. And what I want you to focus on is in this region. I'm going to play a video over this, you know, imaging every two minutes to give you a sense of how these granules behave inside the cell. So here now, and I want you to see what happens when these granules collapse and they're like droplets of water inside the cell that then they keep undergoing fusion events. So indicative of some of the signs of a phase of liquid behavior that have been described in literature and in line with the very rapid mixing dynamics that I just show you through FRAP. And although here we were imaging that every two minutes are not very high resolution because we're trying to avoid photo toxicity, whenever we capture these with, you know, millisecond resolution. If here at time zero, there are two granules that are about to collapse. Only six seconds later, I mean, there is really nice fusion that is happening. So these diffusion dynamics are very fast. So we can use a little better tools. And in this case, we turn to atomic for my formoscopy to probe the mechanical responsiveness and the liquid behavior of these granules. And so here what I'm showing you is, is a granule inside the cell that is labeled with a GFP here shown in white, and we're dropping an AFM probe from the top. And we're very gently tapping on these granules. So here I'm playing the video to show you how they respond mechanically. And we're doing this very gently to be able to measure the stiffness of the cell surroundings around that granule. And we can do this for many different flavors of constructs. And we can map the stiffness of the cellar plasma and the stiffness directly in the vicinity of the granule. And it was intriguing to us that for the constructs that we can say wild type, and that are really processed to lack this kind of domain, they're very, very, you know, not very stiff. But whenever they actually retain this kind of domain, again, the domain that is known to be processed, they become very stiff. So this really pointed to us that a potential explanation to why in the normal differentiation problem of the skin, you may want to get rid of this kind of domain soon after the assembly has happened because otherwise the stiffness of these granules increases rapidly. But instead of tapping gently, we can also tap strongly. So here's an example where we take the AFM probe. Here's a granule that is fairly spherical in shape. But when we tap strongly, we begin deforming the granule so that eventually in the span of only a few seconds, it flows around the nucleus as if it was a liquid. So again, pointing to these equalized behaviors that we were excited about based on the imaging studies that we did previously. So just to summarize this first part, based on these studies, we were confident that the truncary mutations that we were seeing in humans were likely speaking to drastic changes in the critical concentration for face separation. And that the architecture of these proteins, both based on a domain and a domain that I haven't talked about on the tail domain, are likely to be governing the specific material properties, I mentioned the stiffness of the granules as they are tuned to accomplish the biological function. And at least in the context of the cells in vitro proteins that we did and in culture, these granules do appear to have the liquid like behaviors that we were excited about. There is a question within the granules, do they chemically aggregate or do they just stay in close proximity? So do the granules aggregate into large droplets, so to speak. Yeah, so what I just showed you in the video earlier was a sale with multiple granules, and whenever they were running into each other, it would be a rapid fusion event. So in that case, there's coalescence. And so they're growing through that fusion or coalescence process. So they essentially grow through that. I'll show you later that what we ended up discovering the concept of the actual tissue in the skin is very, very different. And so it really points to the importance of probing phase separation in their native cellular environments. But at least in the context of these in vitro systems in cell culture, these engineer constructs that form Kettle high light granules, they do grow primarily through fusion events or coalescence. Now, how do you then go forward and start studying this process in the context of the skin? Now I just show you what I just showed you is essentially this. We can have a sale that if we have a phase transition protein like lagging and we know how to target like the idea with RFP in my previous examples, then it's easy to visualize what's going on as soon as the protein gets expressed, there will be phase separation happening. Excuse me. Yeah, there was a refinement to the question. I did not get it right. So I mean, Andran asks, do the proteins chemically aggregate inside the granules? Oh, that's interesting. So what I was showing you earlier with the fluorescent recovery of the photobleaching experiment is that within those granules, for as long as we can see those cells, if we do this photobleaching experiments, you can see these very rapid dynamics, this rapid mixing. And so based on those kinds of studies, my answer to your question would be that they're existing in a very amorphous highly dynamic state. So I wouldn't describe it as really as an aggregate within those granules rather, they're in a very sort of weakly interacting state that allows them to exhibit this liquid like dynamic. So it's a very interesting material state, but it's definitely not solely like aggregate rather it's more like a liquid. Thank you. At least in that context. Yeah. So here's the issue. If you look at a cell. So again, it's easy to probe these behaviors. If you tag the protein that you suspect exhibit phase of behavior, that's what I just did for filaggrin in the context of the previous experiments. But we know that wasn't ideal. In fact, when we ever quantify phase separation behavior for filaggrin, and depending on the kind of protein that we use for tagging, we would see different behaviors. So more or less the conclusions were strong and robust, but we knew that the kind of tagging was affecting our conclusions. So we didn't want that approach to be able to move into skin. And we thought the new tools were important to develop. So here's a tool that we propose what we thought of maybe it was possible to think of a phase separation sensor. This is a protein that we thought will be sold on the cell. Yeah, we have done zero. There's no expression of this native IDP, but our sensor is already expressed. And after 60 minutes when the native IDP becomes expressed in our sensor will be able to sort of signal that there is a change in phase separation behavior because it would then partition into those emerging grounds for compartments. And the rationale for our design was that based on our knowledge of how to encode phase separation behavior in IDPs, we should be able to encode really weak ultra weak phase separation and specific interactions that then would allow us to detect these phase separation dynamics. And I don't have a lot of time to go into the details of how we went about this, but I'll just give you a flavor for it. So here is R8, the same repeat from Filaggin that I showed you earlier, and the thought was very simple. It was let's think of ways to tune the phase separation propensity of R8. And so we came up with a different number of designs. But the important concept here is identity percentage. So we're doing this in a way that we're rapidly going away from the original repeat sequence. So that repeat identity between R8 and say 4 and 6 is only 20%. So essentially these new constructs are very different, but they have a relationship to the original sequence, and in composition, and their phase separation propensity is different. And to these domains we can add lab imaging tags, different kinds of fluorescent proteins and I won't talk about this, but we can also add more tools. And these are the two designs that we came up with, sensor, I call them sensor A and sensor B. This is the architecture, just a fluorescent protein that is green. And one of these domains, the 1, 4 and 6 here. The interesting thing to note here is that the identity between these two, so sensor A and sensor B, if I compare sensor A and sensor B at the sequence level, they're not identical to each other, they're very different in their sequence. The composition is similar, their sequence is very different. Now, why is this important? How does it work? So here's a cell that expresses philagrin that is tagged. So this is kind of just like benchmarking the system. If I were to simply add a fluorescent protein to one repeat of philagrin, I think that could be a sensor of some sort. Well, notice that this GFP signal does co-localize with the signal from philagrin. But it does so very poorly. So the partition coefficient, the measure of the abundance inside or outside of the compartment is fairly low. And we knew that this wasn't good enough to be able to move into skin. But if I show you how sensor A behaves, our optimal sensor, here is a cell with philagrin express and also our sensor. Now notice that the partition coefficient is exquisite. So about 25 times inside versus outside. So really very sensitive to be able to detect these nascent assemblies within the cell. Importantly, whenever philagrin is absent, both sensor A and sensor B are soluble. So unless there is a phase separation protein like philagrin inside the cell, these sensors, as I predicted, as we assigned them to be are soluble. And I don't have time to go into it is how we validated these sensors, but I want to show you now what we do with them. So because they're genetically encoded, these are proteins that that can be encoded into very simple genetic constructs, we can put them inside lentiviral vectors. So here's a lentiviral vector in which we have our sensor under a promoter, and we use different kinds of promoter sequences. But these lentiviral vectors also have an interesting trick, which is that they also carry silence in RNAs that we can use for knocking down specific mRNAs and eventually proteins. And we do this experiments by transusing embryos, these are mouse embryos in utero, so that once they're born, they already carry our genetic constructs. So this is an embryo right before it's born. And you can see the skin now is fluorescent, strongly in green because it carries our sensor. Now, what does it look like? Here's a live image in view of a cross section of the skin of these mouths. And again, this is what the skin looks in cross sections. And you go from the basal layer to the granular layer, you see that we expect to have these granules. And that's exactly what we see. So we do an optical sectioning of our live image data. You see that in the basal, there's really no much happening. But as you go towards the corneum, there's this strong signal that is co-localizing in spherical structures. And it's easier to see if we go planar. So if we take a planar view of the skin, particularly the different parts of the granular layer, so blue or orange and purple here, different sections. You see that early on there's very few granules, but middle and particularly towards the skin surface, these granules have grown tremendously. To us, this was really, really exciting to see. We did not expect to see such a dramatic event of a separation of the skin. Now, of course, this is live image data so we can do experiments and do this over time. So we first asked, well, you know, I think this goes back to one of the questions from the audience as to whether these granules were potentially going to fusion or not. And so here sales, I want you to focus on this particular part of the video. And as I played, you'll see that over the span of half a day. We don't really see much evidence of these granules coming together and fusing. That was really surprising to us. They're growing. And in fact, we can quantify this here on the right from zero minutes to a hundred minutes. We do see that a given cell acquires new granules and some of the existing granules get much bigger. And we cannot even quantify that if you look at the normalized volume or the span of these 12 hours. I mean, there's a lot of growth in volume, but it's not, it's not really happening to fusion. And that was really puzzling to us because all of our previous experiments suggested that these were liquids in the cells. So how come they're not fusing? Well, one of the first things that we had to do as well, let's probe the liquid like behavior a bit. So our sensors, you know, as they're localized to these compartments, but we can do for a religion experiments. So let's do a for a religion experiment and let's see what happens with the recovery. And if we quantify that both for sensor and sensor B, you know, the sensor have half life is only a few seconds, 1.5 seconds recovery. So these dynamics are very, very fast, right. So we bleach, and only two seconds later, it's almost recovered. These things are really mixing within those compartments that liquid like at least in their dynamics internally. The cells as they mature. So here is a cell that goes from the early granule layer to the late and you can see that they're morphology changes and they acquire more and more granules. So we were intrigued to know if there was any sort of maturation of the material properties of these compartments. So we start doing for a religion experiments at these different layers. And if you see at the early granule layer, the sensor reveals that they're very, very liquid like. And this is where we start. And as they move towards the middle, particularly at the late stage, the viscosity, the relative viscosity keeps increasing as revealed by this lengthening of the prep recovery of half life of the sensor. And intriguingly to us, if we were to compare our sensor behavior in the skin per se, in the mouse versus what we were doing in vitro in our cell culture experiments, look at this difference. So the same sensor, now deploying two different systems, but really displaying very different dynamics. And so somehow in the skin per se, the grinders are exhibiting a relative viscosity that's different. Now this was in the mouse. So we say, well, maybe it's because we were announced we were proud of working with with with human feel again. Well, we need these experiments with huge sales as well. So this is using a process of building skin in the lab. And you can see that if we repeat the experiments both in the mouse or in the human skin. So no longer in the cell culture by rather building trying to build human skin. We see very, very similar behaviors. So the viscosity of these kettle hand granules very much optimized systems for the biological function in the context of the tissue. So just to summarize, we now show that these kind of technologies face separation sensors can illuminate very intriguing dynamics of a separation in the skin. And remarkably, it points to the fact that even though these compartments may be thought of as being liquid like, and they suddenly span a range of discussities, they can indeed crowd this out of plasm, which we were expecting we're expecting all these fusion events and I don't have time to go into what we did from cover. Some of them, their opinions of that crowding, but I'll mention that we uncovered that catatine networks. So these are intermediate filament networks that exist in the skin are particularly good at isolating the granules so that they don't fuse. So there is another network of filaments catatines in this case that are sort of in a tug of war with the with the liquid like behavior of the granules that allows them to be crowded this out of plasm. I'm thinking that's important for the structure of this out of plasm. So he may give me let me give you a brief biophysical outlook of what's going on here so we were intrigued by this crowding of this out of plasm. I've told you, you know, maybe catatines are playing a role there. But you know you can think of the same as, you know, the catatines are being pushed out, they're certainly preventing the ground from fusing but they themselves have been pushed out so perhaps there is some effect on the catatines themselves in their assembly, but think of our organelles as well you know there's the nucleus and the nucleus is losing space. This is could evolve in effect. And certainly true for our organelles that are membrane bound like mitochondria and others. So very intriguing to think about the consequences of this crowding that we see in the scheme because of the liquid like conformism the cell. But I want to do attention to one particular really intriguing aspect of the scheme which is that the skin is unique with a few other teachers in that when it goes from the stem cell state to the very surface and the full differentiation state, the cells actually lose their nucleus so they have to lose their nucleus when they go from the granule layer to the corneum. And we thought that there was a good chance that these crowding and this particular liquid like granules in the skin perhaps were playing a role in this new creation step. Excuse me, there is a question. Can you please repeat what's the implication or use of LOPs inside tissue? Yeah. So I think that that's what I'm trying to get at now. So what I've shown you so far is that we see signs of liquid look at separation in the skin but I haven't yet told you, you know, why we think that's important and that's what I'm about to do. Yeah, yeah. So we thought that we thought that we thought that inucleation could be a potential a way in which these grounds were playing an important role and that's because we noticed that if you were to image, so this is image in skin is an image from the mouse skin as is being developed as is developing. And if we image with a red tag on a chromatin protein H2B together with our sensors, we noticed that whenever the inucleation was just beginning. And there we see signs of initial compaction of the chromatin. We then see immediate signs of loss of sensor signal from within these compartments. And so you go from having nice phase separation to having signs of a different phase separation state in which there is a loss of that phase separation dynamic. And so over the span of only just 200 minutes, you know, you see that it's a complete remodeling of these cells. So you go from the ground up state into what we know is this quaint state or these dead cells that are the surface of the skin. So here an implication for a coupling of the dynamics of the liquid, liquid phase separation dynamics in the skin with a process inucleation that we know is important for skin biology. Now, is that really happening? And why would it be important? So here what I'm showing you is this is normal skin in the mouse using a control herping RNA that is not targeting anything. You see beautiful phase separation. But whenever we use our lingual constructs to also knock down philagrin, then we will lose that phase separation behavior. So here's a case where we create a mouse skin depleted of philagrin, and we see no signs of phase separation. So what's the consequence of that we asked. And the first thing that we noticed was that if you look at H2B RFP, so the nuclear signal in our experiment, within, you know, two to four hours in the control case, there is complete loss of that signal. But whenever we depleted phase separation by depleting philagrin, we didn't see a lengthening of that process. So no longer this removal of the nucleus is very effective or very efficient. So that was the first sign that perhaps phase separation was an important role in controlling the process of inucleation. And more specifically, we measure an important outcome of the skin barrier function, which is the barrier quality as measured by water loss. So this TEWL. Philip, we have like eight minutes before you have to conclude. So we're getting there. This is the transfer of water loss. So this is a measure of how much water is lost in the skin. And so in the scramble case, you see that, you know, there's this baseline value, but if you deplete philagrin loose phase separation, suddenly the skin that you get is, is a lot of water that is being lost. So not only we're seeing a lengthening of these lots of the nucleus, but we're also losing seeing a loss of this very quality as measured by the water loss that we can measure. So this is just in our view, you know, we started thinking, you know, what are the potential mechanisms for this? How could the phase separation be important for the loss of the nucleus and the skin? And we did notice that what type type philagrin was very good at deforming the nucleus. So here on the left. And whenever we see mutations that are, you know, that we know are associated with interpretive antibodies, those resulting compartments, even if they form, they're not very good at deforming the nucleus. So there was this aspect of crowding and deformation of the nucleus that we thought was important that we also would see in the skin of the mouse. But what we would always see, however, very consistently was this thinking, as I mentioned, of the dynamics of the sensor within these granules and the nucleation process. So it would be then that the things that might be sequestered inside these compartments are released at the right time to orchestrate a process of nucleation. So perhaps this liquid dynamics is very sort of dynamic phase separation driven events being very good at actually creating a biological process like nucleation. And we thought that there was a chance that this was perhaps happening in the sinking of phase separation and nucleation might be happening because of a pH responsiveness. And I mentioned very early on that one of the signature features of a phase separation is its stimuli responsive behavior. So the skin surface is actually acidic. It's about 5.5 in humans. And there's a suspected a pH gradient that goes all the way to the base of the skin. And this is intriguing because the transition point which would be expected to happen somewhere in the granule layer will be around 6.5, which happens to be the PKA of histidine, which I mentioned is one of the most abundant and aromatic residues in philagrin. So I really got us thinking into, hey, could it be that philagrin is actually optimized to sense a pH shift in the skin to then actuate the process of nucleation. So this is very quickly just showing you cells. We went into this first in vitro where we have cells co-expressing sensors and philagrin at pH 7.4. They co-localize beautifully. But if we then change the pH and procedurally in our experiment, then we see that it is a granted sort of disassembly of the granules and we see both ejection of philagrin and of the sensors from within those granules in a pH responsive manner. And this is actually, we play with the pH range. We see that this is happening precisely at pH 6.3, which is again the PKA of histidine. So it does seem that these granules are optimized for sensing pH. Now, that's not saying that skin actually has that pH shift. So we wanted to test that. And we went to pH reporters that are genetically encoded and we had two, a red one and a green one, and we put those into genetically engineered mice. And what we see here is this is again imaging data from mouse, a skin in which we have a pH reported in red and sensor in green. And this is T equals zero, both are in the same cell. But just 20 minutes later, you see that the pH went to basically very low about, we suspect about 6.3. There's, these sensors work by a loss in fluorescence signal. And so you no longer see a signal from the, from the reporter, pH reporter. And the same can be, can be seen from the other angle. If you look at a pH reporter that is green, so there's plenty of fluorescence, the chromatin is intact. A pH shift leads to a loss of pH, of signal from the pH sensor reporter. And now you see that there is no longer a signal from the reporter, so indicative of a pH shift, but the chromatin is still not compacted. And 20 minutes later, plenty of compaction with the chromatin and these onset of the nucleation. So pointing to the fact that there's a pH shift intracellular in the, intracellular in the skin that is upstream of both inoculation and the changes in LAPS dynamics that we were predicting were important for inoculation. And we actually tested that directly by playing with the pH in the skin. And so here is a controlled skin in which we have pH 7.4 in our media. And when we shift the pH of the whole skin to 6.4, you see there is a dramatic compaction of the chromatin. But when we do these experiments now in the context of genetically modified mouse skin that lacks phyllagrine. So it lacks a lot of these compartments in the cells that are transduced with these antiviruses. We see that even a pH shift is unable to trigger a lot of compaction. So again, pointing to the fact that there is a strong argument for pH shift being upstream of the new phase separation dynamics that then lead to the inoculation process. So this is just a summary in which I just show you that we uncovered really intriguing phase separation dynamics in the skin. As cells move towards the surface, there's this beautiful process of phase separation that then eventually leads to a simple step that we think is important for the formation of the surface of the skin. We've seen that these dynamics are actuated in an environmental responsive manner so that a pH shift at the granular layer stage enables these liquid-like domains within the cell to then disassemble, likely actuate in the process of inoculation. Of course, there are many questions that remain. I mean, we're very intrigued by other kinds of stimuli. We mentioned pH, but, you know, the skin is at the surface of the phase of the environment. So temperature could be a trigger. How is it that forces might be impacting on the liquid-like behavior of the structures? And so a lot of these questions are things that we are very interested in continuing to study. What are those environmentally-realistic stimuli and how do they interact with their keratoline granules? We haven't yet uncovered what are the components between those compartments that are released at the critical stage of inoculation to then actuate the process. And of course, because these are strongly linked to disease, we're very interested in thinking about can we target these phase separation dynamics that we uncovered in the skin to then think about new therapies for skin-based disorders. So this brings me towards the end of the talk, and I won't have time to go into a lot of the non-equilibrium stuff, but I do want to very briefly touch upon that because I think I have a few minutes. And I'll just say that because phase separation in the context of the skin, in the context of tissues and cells is, of course, fueled by a lot of dynamics that consume energy, it's obvious to begin thinking that a lot of the things that we're going to see in the cell are not necessarily in the equilibrium, but perhaps in the non-equilibrium state. And so, in fact, when we began to very quickly look at these processes, if we go back to the libraries of extensive kinds of hormones that we've made over the years that are IVP in nature, and we begin to sort of look at their phase separation behavior over multiple cycles of phase separation to begin to probe their equilibrium dynamics, we can actually see already three kinds of behaviors, the polymers that exude LCSD behavior. And this is a bit busy, but it's easy to follow. If you go to the blue example, the full circles are at the heating cycle, and the cooling circles are at the open circles. And you'll see that once you heat and cool, these polymers go back right away into solution. But if you look now to the second example, the one in red, you see that when you heat the phase separation, but then when you cool, they actually take a long time to go back. And then if you go to a third example, there's even more of what we call hysteresis, this difference in the behavior between heating and cooling, so that for the green example, once you heat, there is phase separation, but once you cool, there is actually valial signs of reversibility. And we've been sort of parsing out a lot of the underpinnings of these different kinds of hysteresis behaviors in IDPs. And we think these behaviors are highly reproducible, so really tools that we could begin to use to examine phase separation in engineer systems to harness these behaviors for new nanostructures, the same way the new nanostructures, but we think that these kinds of behaviors might also be playing a role in orchestrating the complexity to the dynamics that might be happening in the context of tissues. So I don't have to go into the details of this. I'll just I'll just leave you with an example where if you think of even nanostructures, these are that local polymers that are made from either non hysteretic polymers or hysteretic polymers, you can begin to assemble really intriguing nanostructures where you know they assemble, so the data is shown here that the RH data is in orange, so this is higher DLS data, life starting data to measure the size of the particles. There's about 100 meter particle in size, they're actually not, they're monomers initially. If you follow the orange diamonds, they go high temperatures into this assembly state about 100 meter inside nanostructures. But as you cool down below the original assembly temperature, you see that the particles don't disassemble, and they remain stable for a long regime of that temperature range because of this hysteretic behavior. However, at very low temperatures, they might go back. It's interesting because, you know, we've seen that they're assembled into very defined structures. Here at road like that we can see through cryo TM, these are different that different that the morphology that we've seen in a lot of that examples, with the more non hysteretic polymers that we built in the past. So we think that by accessing these kinds of non equilibrium behaviors will begin to lock them in structures in very intriguing and potentially biometrically useful non morphologies. And we have to begin to exploit these dynamics to encode new biological behaviors. And we think that that could be potentially also happening to the cells in natural current IDPs as well. So overall, I think I hope I give you a flavor of how we now have sequence here is six to encode and tune these behaviors in IDPs. So we can use these behaviors to then tune self assembly of the concept protein polymers. I think it's really exciting to just see so many examples of how phase separation go over is interested our self assembly. And in particular the concept of the skin it seems to be at the root of all the skin bare formation and barrier disorders. And I think a skin with what we just show you is is skin as a nice and beautiful system to study phase separation question and dynamics in the context of tissues. And one of the things that I'm interested in asking question moving forward are the fact that we still need new tools to be able to probe this dynamic behavior in the concept of the tissue biology. I showed you this big difference between our experiments in cell culture versus what we actually in that I discovered in the concept of the skin tissue itself. So really we need tools to be able to prove these behaviors in tissues. And I'm moving forward I think that we'll have to do much more towards understanding this non equilibrium dynamics in the concept of IDPs and so on the concept of the cell because we know several mechanisms are driven by a lot of ATP driven in nature in many cases and, and these kinds of differences between the clearing systems and non equilibrium will have to be taken into account to really fully uncover the complexities that we see in biology. So with that I want to acknowledge you know this is work that I've done over the years in many different places and certainly at Duke. I want to acknowledge the contributions of my mentor, one of our collaborators at NC State at England. The work was done, particularly the skin work was done in the lab of Elaine Fuchs, a work of fellow university. I want to acknowledge James Fury, who is a biomedical engineer with an expertise in mechanical biology. He was instrumental to work on using AFM to probe these behaviors in skin. And John Edwards, who conducted a lot of the new work, looking at this length of our transaction of mice to create this interesting sense to look at face separation and skin. And my lab at General Tech Emery, who are following up on a lot of these intriguing findings that we have in the concept of the skin but also in the context of the opening tools particularly to pro face separation tissue biology. So with that we have a few minutes for more questions. Happy to answer those. Thank you for your attention. Thank you very much, Philippa for such an interesting talk. I will ask if anyone has any questions in the audience. All right, well, we're waiting for incoming questions. I have a few questions of my own. What is the role of viscosity so you said that there are filaments that prevent granules from fusing is one of mechanisms that prevent fusion of granules but what is the role of viscosity in this separation of granules. Yeah, that's that's an excellent question and so far. So I'll give you a bit of a long answer so far people have been thinking about this idea of liquids and under viscosity perhaps going into a less highly viscous state perhaps a solid life state. And thinking of that as some sort of like pathological transition. This is true particular in the context of new generation where people have been seeing for a long time aggregates within the cell that are solid like in our context, we're seeing something slightly different. We're seeing that this that is tuning of the viscosity as self mature and move towards the surface. And we're also seeing that in people with mutations the viscosity is highly reduced. And so in our case at least, we have evidence to suggest that the viscosity is being tuned. We think that this viscosity likely plays a role in these interactions with the other organelles like the nucleus I mentioned the deformation of the nucleus. So you can imagine that you know you need sufficient viscosity to begin to sort of have these deformation events or push in other parts of the cell away. But we certainly having figured out in detail, you know, what is the specific viscosity that is needed or why is that because it's so important. But I think that the emerging evidence that we have is that is that the question in this operation biology is not as simple as liquids versus solids versus, you know something between but rather that the system might be optimized to heat very specific to the viscosity for functions that we still are not completely clear. Something that I will add as well is that I mentioned that this assembly of the granules and the actuation of what might be released from those compartments to then trigger inoculation and our events that have to happen. Well, you can imagine that the viscosity will also control the timing of the dynamics, right, if they are, if the quality is very low, perhaps things are released too fast. Perhaps higher, perhaps the solution is slower. So we haven't measured those things, but it makes sense to start thinking about that this quality as being a tuner of the dynamics itself. Thank you. Well, if there are no other questions, I will go with another one myself. What about so the nucleus in this model that sorry for my ignorance is considered as a bubble is a basically liquid state of liquid bubble. And so, do you have an idea like what is actually happens when this granules releasing something? What is actually, is it a chemical reaction, physical chemical reaction that leads to endocleation? Yeah, that's an excellent question. So what we're focused on right now is, so beyond this biophysical aspect of how there might be the deformation of the nucleus. I do think that there needs to be a focus on what are those components from within those because we know the composition is heterotypic. I mean, I'm showing you there's a lot of philagin present inside them. That's not to say there's nothing else inside them. We suspect there's a lot more inside them. Now it is very tricky to figure out what's inside of them. And the reason for that is that for these kinds of organelles or components within the cell, because they're membrane less as they don't have a liquid membrane. It's not as simple as break the cell apart and purify these things and figure out what's inside. It's actually very hard to study. However, there is an emerging set of tools, which is generally known as proximity dependent proteomics. And what they entail is the idea that you can specifically label the proteins within those compartments before you break the cell apart. And you particularly label them with biotin. And so, you know, biotin is one of my favorite tools to use in biotechnology. So once you label specifically within the compartment, the protein or the components of the compartment with biotin, you don't worry if this assembly is going to happen. You break the cells apart, you break the tissue apart, and then you purify those biotin label proteins and then you do proteomics work to see who was within the proximity of the compartment. So those kind of tools are the ones that we're developing now to specifically there's a question of, hey, who is present in the compartment prior to this assembly. And I think that will give us the map of the biomolecular composition, which we would say we expect that would be proteases and DNA says, and players that you know are known to be present in the skin, and playing a role in bio in the biology of the skin but they are not not known to localize to a granule. And it's again very hard to study this particular part of the skin because it's specific architecture of the skin. So my chemistry is not particularly good either. But I think proximity dependent proteomics are going to work well. And I'm happy to say that the tools that we have with the face operation sensors we have adapted them to that kind of proteomics work. So we really are very close to being able to get that data, but we don't have it yet. So what are those molecules that are being released that are actually in the process. I would love to know we'll know soon. All right, let's hope we'll know soon. If there are no questions from the audience. We are actually a bit overdue. No, thank you so much for the invitation. Yeah, thank you very much you and if anyone has any questions or, you know, think of questions after the talk and certainly once you look at the record seminar I know I can go a bit faster with the different aspects of the work. Please email me. I'm happy to interact and answer questions and get inspired by thinking from a physics perspective and quantitative perspective opportunities in this field. So I hope some of you do do join. All right. Thank you very much.