 I'll follow up when I started asking David in the hallway. And that is all your localization and analysis is done with these GFP RFP fusions. And the question is, how much are they functionally compromised, and how do you evaluate whether they're functionally compromised? In budding yeast, we have, so since my lab does both, we have much more sensitivity, because the cell, usually you can make it easily the only copy of the whatever gene, and especially you know about it's what phenotypes there are from loss of function you can look for or genetic interactions. As I mentioned in the hallway, I think that you always compromise function when you put a tag on a protein. And I think people try to sweep that under the rug, but it varies. I think in most cases, you preserve the function surprisingly well. But there are things you can do like assess the pathway. If you're looking at endocytosis, you can check your line and make sure the rates of endocytosis are normal. You can look at different components of a complex and see if they have similar behavior. I think you always need to be cautious when You've generated like 120 lines. Yeah. Well, many more yeast, that's mammalian lines. Yeah. Yeah, mammalian, they're diploid, so if you tag, or if they're HeLa cells, they have six copies. How many copies you tag, it can definitely make a difference. And actually, the Allen Institute, so I'm a senior investigator or whatever, I have some role that the Allen Institute where they're doing this very systematically and very rigorously, they find that for some genes, they don't recover any diploid. Where both alleles are edited, they only get single allele edited, which is a guess is that it's because it might be lethal to edit both or at least selected against some kind of growth disadvantage. So in that case, actually, what we've done in mammalian cells and what the Allen Institute tries to do is to go to experts in the field who have like leave it to experts to study a process. If you tag a protein, because the Allen Institute doesn't have expertise in all the different cellular processes, they might go to your lab and say, what would you tag in the protease home or something like that? And what evidence do you have that that was functional? But things like what linker you use can make a huge difference night and day. So it definitely is deserving of care. What are you trying to do to just systematically tag all day? The Allen Institute, they are tagging. So you can go to their web page, the Allen Institute for Cell Science. It's too bad Leonid isn't here. I meant to mention to him they have tons of 3D imaging data for all these proteins. It's all the raw data is put on the web immediately. So people who have the computational skills can look at the images. But they originally were going to try to tag essentially everything, but they decided instead to thoroughly characterize all their cell lines and instead make representatives. So every major organelle is tagged already. Every major cytoskeletal element, cell adhesion molecules, now they're working on signal transduction proteins. And they're all going to be in the same genetic background. And it's all available to the public at their cost. So Berkeley has acquired the first set as a resource for people on campus. And it's a lot, I think like model organisms, C. elegans or flies where you have, so everybody in different labs is comparing the same type of cell, which now I know in my field, even if my lab works on a HeLa cell and the lab down the hall works on HeLa cells, they are not at all the same. HeLa cells, it's not even close to a real cell. Yeah, exactly, so that's, you know, so that's, you know, it's timed, you know, to start working on things that are more physiological. That's good. I mean that here, rapidly, we will have somehow among all the cell biology lab to use this kind of cells to be sure to publish, for instance. So is it, to be sure, I mean, somehow that everybody is talking on the same base, is it, it would be a trend or not? I think it would be a positive trend. Yeah, I think it's about that. Uniformized. I mean, I'm not in charge of all of it, you know, also my answer. One more question. The GFB, GFB is a very stable protein. So when you make a fusion, you could stabilize short-list protein, and this could have an effect which is not natural, then short-list protein could be stable. So this could be a little bit of a problem. I don't disagree with that. It's just the only, it's a way to look. No, but you can test, you can test, you can look at the half-life and test. On the system, not by over-gene expression, but stabilization. Should be used in non-American GFP, or? Yes, yes. And we try also, sorry, we try also to, like when we see something with GFP to see if it's really true or so with, even of fluorescence, where it's no, it's endogenous. This was done to overcome the over-expression, but it's not always going to do that. Yeah. Okay, somebody wanted to say something, like. No. Oh, you have the question, but yeah. Yeah, no, the point, comment in GFP is of course is good, but there is a choice of different tags as well, like halo tags or snap tags, so they have big advantages. Like you can use the floor for what you want, for whatever kind of photo oxidation or threat. It's much more efficient with these tags here, so I wonder why GFP is older? So maybe it was chosen because the project was started some time ago, but not sure that today I will choose GFP. Yeah, I mean we use all those tags, and some people like the cell atlas, the Chan Zuckerberg cell atlas, they're using a split GFP. Okay, the Chan Zuckerberg is doing sort of the opposite approach of the Allen Institute. They're this new center, Chan Zuckerberg, so all these rich people are starting institutes now in the US, and so they're using this split GFP where you put an 11 amino acid tag and then you just use a parent cell line that expresses the rest of the GFP. It makes a beta barrel, so that's a smaller tag, but it makes GFP. So there's pros and cons of different things. We had a question in the back first time. After that, Brianna, what change are you looking for? Mine's a very general one outside of the field, and I guess mainly Judith, Alberto, maybe the others. So the transport complexes have been studied mainly in the context of their original mutant phenotypes. In other words, they interrupt one step in some transport pathway. We're now starting to find them in more places. I guess from a person who's interested in more regulatory mechanism, I'm curious how general this sort of multifunctional combinatorial mechanism is going to be. And also, we've learned quite a bit about signaling out from the transport complexes, from unfolded proteins and such. How about signaling in? I mean, what do we know about signals going into the transport machineries? Which parts of the signaling machinery are regulated by what kinds of mechanisms? So the first question, I think multitasking proteins, it's an emerging team and the list of proteins that are known to have multiple tasks is growing. Also some proteins that we know very long already. For example, I have been talking about the menocytosis receptor that is involved in transport of lysoma enzymes. But it's also involved in endocytosis of IGF-2. And the fields that are looking at IGF-2 endocytosis of menocytosis receptor trafficking are kind of apart, strangely enough. But it's the same receptor. And for signaling, for example, VPS-3 was so part of the Corvette, was originally identified as TGFP-REP-1 and it's a SMED-4 binding protein and that is involved in the TGF-Beta signaling pathway. So that is what I know. But I think that we have to be very open to this, that one protein can have multiple functions. You want to add? I agree. I think it's sort of disturbing. It would be much easier to think one function, one protein, but this moonlighting proteins appear to be very many, actually. But I don't think, I don't know of any systematic database or systematic study or collection of data about this moonlighting phenomenon. You find a lot of this in the literature. So I guess there are many, but there is nothing systematic and clear done already. So for your question about the signaling, trafficking, and coupling, I think a Gould-Ford translocation is a very good example. So it's a well-known AS-1 sick take can be false-related by AKT in the insulin signaling pathway. But that's just a part of the story. So David James did a whole proteomic analysis and found that hundreds of protein could be false-related upon insulin stimulation. And it's still open. Are these false-relations important or not? It's gonna be a long way. At least there are some candidates so far. Can I just add my penis word? I think what you're asking to me, it depends on exactly what your question is. A rab is going to do the same no matter where it is. It's going to function with a different set of proteins. An escort three complex makes filaments, no matter whether it does it at the surface, whether it does it at cytokinesis or whether it does it at the surface of an endosome. It might interact with different components. So the signals don't make a circle into a square. It still remains a signal, but it can change its interactors based on where it is. A cop two coat will make a cop two vesicle because it only functions at the ER. These are the hard wired nuts and bolts. Sex 61 will only translocate cargos or newly synthesized proteins at the ER. It will not do it at endosomes, for example. So you have to ask the question based on the protein you're interested in and that can change. But certain proteins do not change their location no matter what you do to them. And they don't become something else just because they get a phosphorylation event or a glycosylation event. Maybe an easy way to solve this issue is to distinguish between the activity of the protein and its function. Precisely. An enzyme is an enzyme. One activity, eventually two, like an enzymatic activity or a cell fold activity. But this. An enzyme will remain. Any functions depending on the cell type and localization it can have many, many outputs. But when you're looking at different pathways in the cells the same function may influence very different processes. Well, if you look also, Charlie Boone doesn't seem to be in the room, but his genetic interaction network, I think, I can't remember that the essential proteins, I don't know, had nine interactions each, the non-essential seven, something like that. Which, so each protein does have many different types of interaction, functional interactions. And similarly, when you do proteomics, I forget the average protein, that pulls out seven other binding proteins. And so, yeah. What did you say? Some core, hard definition of moonlighting proteins, they have completely different functions and they use different surfaces for these functions. So those are really completely different activities. I work with a protein which is involved in membrane fission. It's also a transcription factor, exactly the same problem. And the molecular mechanism is very well known, the surfaces are different and the functions are totally different. And there are a few examples. As I said before, it's not clear how many cases of these real, multifunctional proteins we have, but from the literature I guess, it's not just the change of localization, the same protein, the same enzymatic activity does, something in a compartment and has a different effect in another compartment. That's not a multifunctional protein. But Alberto, I beg to differ, sorry. Your protein bars is recruiting an acetyl transferase in doing the fission process. And when it's inside the nucleus, it recruits an acetyl transferase to acetylate. No, it's really very well known. It's completely different. Why? I mean, does it not recruit? No. No. No? No, but I can simply, it's not the case. It's not true. It's not true. It's not true. It's scaffolded, it recruits. No, it doesn't recruit an acetyl transferase in the nucleus. What is your protein? There are many examples. There are many examples. There are very few examples of those. Very few. No, no, no, many. Like coletic enzyme put a function in the nucleus, the transcription factor, for example. Or cypherchromesine. Yeah, cypherchromesine. And many times the name of a protein, you know, the history influences how people think about it. Which function was discovered first? No, there's a canonical function. And then the moonlighting activity was discovered after all. And sometimes it's, you're not even sure if it's real or not, if it's direct or indirect. The question is how many of these moonlighting proteins really shown to directly be involved in two totally different functions? I don't know. No, I don't know. Albert was the same. As I said, I don't know how many. And also the definition of mode, the definition, the criteria should be strict. Otherwise, okay, a protein can do many things. It should be strict. Different surfaces, completely different activities. Then under those criteria, I'm not sure how many there are, but several, I think. What do you call VPS3A of moonlighting protein? I don't know enough, maybe not. But I don't know what it does, actually. No, it's not completely, we know. It's already, so I can't answer that, no. So all of you guys are working with some membranes in some way or another. And I think a few of you were asked about lipids. And none of you talked about lipids. And I'm wondering how you feel about them, both in the, maybe why it's not relevant to your research or how maybe it is and how you're interested in that or not. Repeat the first few seconds. Why did you talk about lipids in your talk? Lipids. I was talking about lipids today. No, because Patricia and I talked about lipids a lot, but it was privately. I don't know, not because. No, no, no, it's a completely different story. It's not related to this meeting. I can't answer their way that you talk about. And it is especially, it's a nature of lipids, that's important, something beyond your question is, it's a nature of the lipids, it's important for the function that you describe. I think one of the things that limit, there are many things that limit studies in the lipid field. One is that you can't monitor them in cells, where they are, their dynamics, their exact position of the different species. The same way that can be done with proteins. You can label them and monitor exactly their dynamics. That's a huge advantage. I think people got two Nobel Prizes. For lipids, it's a completely, there is no way. It's a completely unexplored territory. So imaging is impossible. It's a big limitation. And biochemistry can be done, but of course it can be done. But again, without imaging, you don't know where they are. And for cell biologists, it's a big question that can't be answered. So Tommy is not here, but yesterday he used, I think people are using to look at the lipids, the phospholipids, to use domains of proteins that recognize the lipids specifically. Yeah, but they sequester lipids. So one has to be careful. But answering his question, there are lipids that are being studied extensively, but they are usually modified lipids. Phosphorylinosatides, for example, the PIPs, the PIP2s, and the PIP3s, diacylglycerols, they've been studied extensively. But you're talking about lipids that are present in all the membranes, you know, the phosphorylalkylins and the serins and ethanolamines, et cetera, et cetera, et cetera. They are present in all the membranes. What changes is the concentration and their production at any given time. And that becomes very difficult to quantitate. You cannot easily manipulate them. And if you use these kinds of proteins, such as a pH domain of a protein that binds to a phosphorylinosatide, sure, you can monitor it, but you don't know what you do to the dynamics because you have bound something and you prevent its consumption. And there is there for that little caveat that you have to be careful. And that is probably one of the major reasons. People think that those methods are not reliable. You can monitor lipids, one with lipid binding proteins, but then, you know, you mess up completely the dynamics of that lip. Or you can replace anacyl, anacyl moiety with a fluorescent molecule like a body pile or something. But again, it's not the same molecule. So people don't, including myself, people don't trust those data, basically. Now, there is things might change now with the advent of imaging mass spec, mass spectrometry. That's going to reach a resolution it has already, but it's not commercial yet, of one square micron. And you can identify 300, 300, 400 lipid species using imaging mass spec. So then you can, you know, have big, you have big pixels, but you can't get an image there. And the result, the molecular resolution is fantastic. You can distinguish many different species of phosphoryl, call in, for instance, because they differ in length of the acyl moieties. So that's going to come. It's here already, basically. Okay, so going off of what David said, I was thinking about during David's talk about the math modeling of the endosome formation or whatever. I was curious whether you took into, I mean, did you take into account the concentration of the different kinds of lipids that could be there and how that affects the elasticity of the membrane or something? No. No, we just varied parameters that, you know, to do with the membrane tension. We treat in, for those, most of those models we used, we just modeled the lipid bilayer as an elastic sheet and then varied the properties of that sheet and didn't change anything locally, which probably happens during these processes. So that would be a refinement. But you know that in terms of budding and fusion, you're cutting membranes and you're fusing membranes and there is no space for lipids in snare mediated fusion event and there is no space for cop two and cop one mediated cutting. Why? Because it's just been very, very difficult. And in vitro, people kind of don't ever get to measure those things. And in vivo, it's very difficult because unlike proteins where you can do S-I-R-N-A and CRISPR and quantitate, you can't really do that with lipids. So it's been a technical challenge, not because of lack of interest or not wanting to study it. Well, we had exactly with Dr. Sherfield this morning. She's not here exactly this conversation. She's doing experiments. We are also doing experiments with liposomes and changing the composition of liposomes and looking at how this changes the response of certain proteins. It can be done in liposomes. You control the lipid completely. You know what's there. But when you want to translate this data into in vivo, in cell biology, you have no idea what's going on there. That's the big limitation, essentially. Now, I was going to say that I find it great to this initiative to go to the primary cell and try to redefine in terms of endocytosis and maturation of endocytic compartment, how in primary cells and not in Hila cell this is happening. Because most of what was found in Hila cell turns out to not apply to the primary cell. However, I wanted to have your opinion on how much could you estimate how much of what you're going to find in the cells you're using, which are not really primary, actually, from what I understand, is going to be general, possible to be generalized. Because my feeling is that when it comes to endocytosis, an endocytic vesicle, it's always defined by what it contains. And since all the cells have different activity in terms of receptor and internalization in tissues, finally, can we really find general rules on endocytosis and endocytic trafficking in general? That's the question I asked myself. So what do you tell to yourself? Sorry? No, I said you asked yourself, so what do you answer? I mean, we've just looked at one process, and we've found that it's remarkably different. So if you look at the literature in endocytosis, different papers using cancer cell lines report very different morphological features for endocytic sites, plaques, or round vesicles, very dynamic, very slow. And it's hard to make sense of how those differences came about. Are they physiologically relevant? Because if they are relevant, then they're likely to be important. And then there's some kind of mechanism that adapted that process for the different cell types. And I think if you extrapolate to all the different cell types and then looking in the context of a tissue where you have cell-cell contacts, apical basolateral surfaces, there's going to be differences that you just wouldn't be able to see when you look at things on glass. And then the fact that you can differentiate them into many different cell types. And I think these organoids are, they're imperfect now. Like they don't have a vascular system, but there's just so much effort into improving them that there's going to be a huge push in that direction. So I think all these things are complementary. I think that Tommy thing, looking in zebrafish, which are very translucent, and you can look in the living animal a lot of intracellular events, those are all good things to do. This is a general problem. There are 300 types of cells in mammalians. And they have some basic DNA polymerase will be probably the same in all of them. But as soon as you get a little bit away from these very, very core basic functions, there is obviously a lot of difference among cell types. So that's also a big problem when you start to look into the physiology. You know, the physiology means that cell type in that organ, in that context, we are not there. I mean, it's going to take many, many, many years to get there. It's possible to make transgenic mice with that protein. No, sure, but it's a huge amount of work, centuries of work. But listen. No, it is personal. Just, yeah, just. You, this love affair with transgenic mice. You know, if you do a, if you do a mucin PubMed, you get, I think something like 15,000 papers or 10,000 papers. And a lot of this has been done in mice. In fact, the whole cystic fibrosis model based on the mouse system turns out to be a complete flop because the human goblet cells of the airways are different from the goblet cells of the gut. We work on this blasted problem. And we follow the mouse wallace for years. And it turns out the way the airways goblet cells work in our system is completely different and has nothing to do with what happens in a mouse. But some of the issues you can, coming back to the original question raised by the fellow up there in blue shirt, that some of the questions, for example, you asked about generalization. Well, we know that basic mechanism or basic principle of endocytosis and exocytosis is conserved in yeast and in us. So some of the things are going to be conserved and there are no changes to be expected. Ah, there might be modules, you know. But the basic concept is the same. The cops do the same thing in yeast and flies and worms and in us. But there are certain things that yeast just does not do. I mean, they don't have bones, they don't spit, you know, they don't think. So of course, studying those things in yeast is liable to give you wrong. It might take you in the wrong direction. So it all boils down to what is it that you want to study, what you're capable of doing and how far do you want to stretch it? Lipids are not studied by many of us because it's blasted difficult to get at it, you know. Whereas other things are relatively easy, relatively not easy, but they're just relatively easy. And people who work on proteasomes and things of that kind, they don't have to worry about, they're no membranes. You know, they're just pure blobs of proteins. They're pure blobs of proteins, you know, so you can get away with it. So I think it's picking your problem carefully and knowing how far you can take it. It's very important. Okay, are we done? No. No. No. Is the whole the microphone? Oh, and unless somebody wants to say something about the previous question. No, no, no, that's the... For instance, in Germany, they have an enormous alliance to study liver cells, the liver and liver cells. So all the data are understandable, comparable in one system, but otherwise discussing processes like signaling, for instance, randocytosis in different animals, in different cell types, it's dangerous. One needs to know that, that's all. He was... He had a really philosophical... Please, David, he was completely... Yeah, he keeps asking, is this over? No, it's not. Okay, David, you answer one more question and then you can go. Just you. So what do you think are the big questions that you still want to see answered in this field? In life? No. In what? In your... Oh, in your life? Yeah. In your life. Oh, honey, this is... No, really, what's of the field? We know the pathways, we know the mechanisms. No, we all know the pathways. We don't? So what is the question? The folks go around telling the world that we know everything about the trafficking business, so you shouldn't be worrying about it. I mean, I'm sure David gets beaten to jelly that don't we know everything about actin? And he turns out, well, no, not really. Why, what is it that we don't know? We don't know how actin can do what it does in so many different forms. And it's the same thing. If you just study VSVG transport, or if you just look at invertase accretion, then it's kind of, yeah, maybe it's over. But if you want to look at the real stuff, then we have no understanding. And the question is, do you want to just say, well, we kind of know the basics, so it's over? Or you want to say, no, it isn't. I mean, if you were to talk to George Palladi when he was alive, he used to say, sometimes, not always, the same, don't we already know everything about... Jim Rothman said in five years we'll already figure out everything out. So we live and... We don't. So the question is, do we want to know more players? You're suggesting that we don't... Make a problem and solve it, whether it requires new players, knowing lipids, knowing stuff in organoids, doing stuff in, you know, whatever, just solve it at whatever the level. I mean, it's simple as that. I think I have, of course, an opinion here. Now, it's been around, the answer has been around for too long. And it's complexity. What we don't understand is the complexity of the biological system. That's systems biology. We have been talking about systems biology for 20 years or so, and it's been disappointing. I think it's... The thing is, that's the real problem and it has taken a lot of time and it's going to take more time, but it's the real frontier for the next, I think, 10 or 20 years complexity. Well, I think it's clear what complexity means. No, it's how the systems work, not the nuts and bolts. Well, now you mentioned Jim Rossman. Well, he's a person who tried to simplify everything. I remember he mentioned once, autophagy. He think it's just one branch or a secretary possibly. I'm not sure if Marky heard about that. But in the end, it was much more complicated than that. They earned a Nobel Prize a couple of years ago. So, my opinion, probably going to be many, many, not only players, but also posh ways. We just don't know, right? I think Jim mentioned that autophagy is just another way to degrade proteins. He didn't say that it's another form of, it's assembling compartments. The idea that you use ubiquitin to then tag proteins and then you need to clear them in lysosomes. One could argue that this was shown by Chikanova and Hershko and Varshavsky, but I agree that what Osumi did, he basically came up with a whole pathway. And it turns out that this pathway is so crucial for so many physiological processes. This goes beyond just doing this inside a cell and throwing it into a container. That is not the case. So, you know, he can say whatever. We also said openly that protein transfer, ah, we kind of know everything about it. And it's not true. But I think if you challenge him, you want him on, he would admit that we don't even know when a vesicle fuses. You, a vesicle is fusing to the target membrane and there is this pore that expands. How that pore expands is not clear. And it's a very major issue. So I think it's, you know, when you talk, sort of loose talk is very easy. And you can say, yeah, we kind of know everything. People say the same thing about transcription. And every time you find out, jeez. The question is, what do you want to know about the biological system? So for instance, Randy Shekman says, I want to understand it at the atomic level. I'm not interested in the atomic level. I think it's fantastic. But I think what I need is enough knowledge to predict the behavior of the system when we perturb it. So that problem means dynamic relationships between address and notes. And then you have to define them better. But the ability to predict responses quantitatively when we perturb a system. That's what we want to know. And I think this is the problem. It looks like a physical approach. No, physical. Just a comment on what you said, and the systems biology and complexity. I think that systems biology will continue to be disappointing for many years because we just don't know enough. And that's the whole point. So I want to agree with the panel that we are still scratching the surface of many. We're still getting a lot of surprises in terms of functions and moon lightings and all the rest of it. In terms of protein activities and functions. And so on the other hand, you have to start somewhere and you have to start trying. So if you go into systems biology, you know you're prone to fail, but there's small advances that will eventually catch up. But it's just trying to get a model of one simple cell and a mathematical model of a simple cell to predict behavior. That's the ultimate goal. But we're so far away from that because our technology is just not there. And we're just ignorant. So I think we started talking about systems and systems biology far too early. So it was bound to be a big disappointment. But I think it's the way. And drugs, it's the same principle. You just don't know enough about the system so you cannot predict. If you could predict, the pharma systems would be a lot more efficient than they are and that's the whole point. But you keep making runs, right? Well, yes, so that's, again, it's not the reason why there shouldn't be pharma companies. And we're achieving some success. It's exactly the same with systems biology. You're achieving a little bit of success but you cannot expect that you solve the whole thing. Then the mathematicians will come eventually to save us. One of the problems of systems biology is that there is a divide, a separation between modelers, mathematicians, and real biologists and real biologists. It's important really to fuse, to know exactly what a model means in terms of the molecule, the function you are familiar with. Otherwise it's going to be models for what? And the classical nuts and bolts biology. This is...