 Okay, it is a great pleasure to introduce today's seminar speaker, Dr. Joel Rothman, who got his start in his professional career in Madison, Wisconsin. He actually started a little bit before that. But he was a faculty member here from 91 to 96, and he was a faculty member at just the right time that we were designing the floors in the 2012 building, and he got an image into the floor, but he wasn't here long enough to actually enjoy it, because he left before the building opened. But he actually has been for the last 22 years in sunny Southern California, and he actually said he's enjoying the rain, which is, you know, so welcome to Madison. We've got a lot of it, and I want to now give it to you, because I want to hear you. I don't want to hear me. It is a joy to have some rain. I'll tell you. I can't tell you how wonderful it is. Because I got to put up with this all the time. So I really can't overstate what an incredible joy it is to be back here. I'm just overwhelmed with nostalgia and just appreciation for seeing all my close friends and colleagues, and I have to say that there's not hardly a day go by that I don't think about this place and sort of feel this longing and pain still for the times I had here. And I've missed it for 22 years. It's really a joy to be back, and I'm really so grateful for the department and the university for getting my career started. You know, this place was unbelievably supportive. It really got my career off to a wonderful start. I have to single out Judith for being an unbelievably supportive senior colleague. I mean, Judith did so much for me that I can't state all the things she did for me. She doesn't even know all of them. Getting me connected with wonderful people in my lab and so many other things. But including, she actually gave me a couple of duers, liquid nitrogen duers, filled with about 1,000 C. elegans strains, and said, oh, you can have these dual. I mean, what a way to start my lab, and we still have them. They're in our liquid nitrogen, and we still use them. Thank you for everything, and thanks everybody for everything you did while I was here. And I went back just for nostalgia reasons, went back to my old lab area, walked by where my old office is. I see now that it's being put to much better use than when I was there. Unfortunately, I couldn't go in there, because it's now a women's room. But it's doing more important things. OK, so he set me up. Susman set me up. I was thinking about it anyway. All right, so this is not a great introductory slide, except that this problem sort of sparked my interest about 35 years ago when I was working on cell biology. And I thought, how could something like that happen? It just kind of blows my mind. So we've been studying developmental problems since I went to the laboratory of molecular biology in Cambridge using C. elegans, not humans. Still managed to get funded, even though I'm still working on C. elegans. But it's getting tougher all the time. But I'm going to tell you, if I have time, three different aspects of development. I'll try to cover three, if I can do it. One is on developmental plasticity, how cells commit and how we can change their commitment. Another is actually evolutionary plasticity. How do gene regulatory networks that underlie development change evolutionarily? And then if I have time, I want to talk a little bit about developmental fidelity. What ensures that this process occurs so reproducibly over and over and over, 50% of human conceptuses end up as humans. So I argue that there have to be systems in place working all the way through development to ensure that fidelity is maintained. And I'll give you a little bit of information on that. OK, so let's start with developmental plasticity and our work on trans-differentiation. Of course, in development, Robin is this, I guess it's, can you see that OK? I guess it's OK. Development is naturally forward driven. Cells start out as uncommitted cells of some sort, stem cells or progenitor cells. They then become specified at some point, but not necessarily committed. And then finally undergo a program of differentiation where they do appear to be committed naturally anyway. Very rarely, naturally, do cells ever change their differentiation to another differentiated cell. Hello, Peggy, to another differentiated cell type. And of course, naturally they don't back up, except very rarely. Of course, we can do that in the lab, but naturally that's not true. So we've been fascinated with this problem for a while. So I want to just introduce the system that we've used, which really started in my lab here. And that is the endoderm of C. elegans. So this is the very early development of C. elegans. There are a series of divisions that create what are called founder cells. And I'm going to say more about these at the end if I have time. The founder cells each give rise to different parts of the animal. And one of those founder cells produces a clone of cells that are the entire endoderm, the innermost of the three germ layers of a triploblastic animal, like us, for example. And so we've focused quite a bit of our attention on endoderm and mesoderm development, starting here when Peggy was in my lab. And so this just shows you that it's a very simple pattern of development. The E cell is specified through mechanisms I'll touch on. And then there's a series of events going from 1, 2, 4, 8, and so forth. There's a series of transcription factors that are unfurling, ultimately, to create the 20 cells of the intestine. And that's the whole endoderm in this animal, very simple endoderm, unlike ours where we produce liver and pancreas and lots of other derivatives. So I was attracted to it because it's a very simple developmental switch. All we have to do is figure out why that cell is different from its sister, which makes mesoderm. And we can understand an important developmental switch. So we studied that for a long time. And this actually represents a gene regulatory network. I hate these charts. I'm sorry, but I just throw it up there for the heck of it. These just show you basically interconnections between genes as this process of development unfurls to create the endoderm, this mesodermal cell, and so forth. And I don't want you to get bogged down on that. I'm going to talk later, though, if I have time, about maternal inputs, the inputs that get the whole thing started, that impinge on this complex network of genes that make these cell types distinct from one another. But let me just summarize the endoderm gene regulatory network as my lab and Jim McGee's lab and Jim Priest's lab, Bruce Bowerman contributing, worked out over the years. And that is just shown here as a series of transcription factors starting up here with a maternal factor called skin 1 that's loaded in the embryo and acts in combination with some other inputs to turn on a series of transcription factors that are of the GATA type transcription factor class. These are zinc-finger proteins that bind in their target sequence, GATA in the core sequence. And the use of GATA factors to create endoderm is pretty much conserved across all animals that have been looked at, including even very basal animals like niderians, which are diploblasts. They have only two germ layers instead of three. The invention of the endoderm is probably the first major cell type that was invented in multicellular animals, certainly one of the earliest anyway. Because you need a gut before you can do anything else. And GATA factors seem to have pervaded the entirety of animal phylogeny. So these transcription factors act in what you can see as sort of a series of feed-forward steps, where one transcription factor turns on the next and also primes, if you want to think of it that way, the next one down the line. So there's these series of sort of bypass loops. In the end, the goal is to get these two GATA factors turned on, and they maintain their own expression and each other's expression. So that's a lockdown circuitry. That's committing the cells to make intestine for the rest of their lives. And they hit a whole bunch of genes, thousands of genes, that define the organ, the intestine. OK, so one of the things that we found when we studied this, we've actually found this feed-forward circuitry, is that the circuitry creates a kind of redundancy that is designed to torture developmental biologists or geneticists, I should say, and almost prevent them from getting tenure at the University of Wisconsin. Because were it not for a lucky break that two of these were next to each other, we could delete both of them in the genome and see the loss of them. We would never have been able to figure this out. So these guys are right next to each other, thank goodness. And they give a very strong phenotype. Strong phenotype, if we delete these two right here, is that the endoderm precursor, that E-cell, switches over to an entirely different cell type, a different founder cell type, that gives rise to skin and muscle instead of intestine. But the characteristic of this is that it has what we call sequential redundancy. That is, any two transcription factors in a row have to be knocked out to get a phenotype, which drives a geneticist crazy. We do single knockouts of any of these guys all the way from here down, all the way down, we get very weak or no phenotype. If we do double knockouts in sequential guys, if this pathway is correct, sequentially, then we get strong phenotypes. Those two, or those two, or those two, all the way down, we get strong phenotypes. If this pathway is right in these bypass loops in this sort of feed-forward circuitry, and we believe that feed-forward circuitry is designed to get the whole thing deployed rapidly, if this is correct, then if we do double mutants in which we jump a step, then we might not expect to get a strong phenotype. If we knock out both of these, but leave that intact, we still wouldn't get much of a phenotype because of the bypass, for example, these meds that we identified can turn that guy on, even if that's not there, and so forth down the line. And that holds up all the way down the pathway. If we do double mutants and skip a step, we generally get weak phenotypes. So it looks like the thing is designed, we think, through a series of feed-forward steps, to basically create a strong on switch that's essentially locked down and deployed quickly. What we found about, actually, shortly after I moved to Santa Barbara from here, is that any one of these transcription factors has the capacity to turn, I'm not a teenager anymore, why did my voice do that? What the hell? Okay, to turn, I've had that happen in lectures too, and the young students think it's funny because it was only a few years ago they were doing that. To turn on endoderm in the entire embryo and turn off all other pathways of development. So we can turn the whole embryo into, normally there's a gut, only where the gut is, we can turn the whole embryo into nothing but gut. It was subsequently found that you can turn the whole embryo into nothing but muscle, and you can turn the whole embryo into nothing but skin each with appropriate transcription factors if you turn them on throughout the whole embryo. And no other pathway of development is activated when you do this, and I forgot to mention, but we can get this to happen after the founder cells are born. So those founder cells we know are specified, if you take them out of the embryo, they will do their normal thing for the most part, that's generally true. But even though they're specified, they're not committed because we can do this experiment well after the founder cells are born and after several divisions. So the founder cells are specified, they're not committed. Now having said that, you can't do this forever. There is a window of time in which this, you can convert an embryo all over to one cell type. That is the early embryo and the founder cells are pluripotent, even though they become specified, their pluripotent can make, contribute to all three germ layers. But that period of multi-potency or pluripotency goes away at what we call the multi-potency to commitment transition. There's a transition point in which embryos are multi-potent. Here, here they're committed, we cannot tell these cells to become anything except what they would normally become by exposing them to these transcription factors. So we've been very interested in that transition point. And I won't be able to say a lot about it, but one of the things we did was to ask, well, is there something that turns on commitment or that inhibits the plasticity of cells later on in development? And the strategy we took was just to take advantage of the powerful tools of C. elegans, in this case just RNAI. And the idea was if we challenge older embryos after we've treated with various candidate RNAs to expression of this gut transcription factor, can we find knockdowns that allow older embryos to be multi-potential? So can we extend the, can we delay the onset of the MCT? And we actually got a bunch of interesting leads, but one that jumped out was an old friend, an old friend from an old friend. And that is the notch pathway. And so, or the notch glip one path, glip one notch pathway. I gotta call it. Okay, so what we found, and my lab had worked on that, of course, Judith is the one who discovered glip one. She showed what its role is in lots of developmental contexts. It's one of the two notch receptors in C. elegans. And what we found is that if one, using a temperature sensitive mutant from Judith's lab, if we eliminate a glip one function, the notch function at this time in development between the four and about 28 cell stage. So we shift it up at that point and inactivate it. And then it doesn't matter whether we shift it back down, but if it's inactive in that window of time between four and 28 cells, embryos seem to remember that. They hold on to that memory and much later in development, if they didn't get notch glip one activity in that period, then if they did not get it, then the MCT occurs later. That is they remain competent to be reprogrammed later. If they did get that, they undergo this switch, this MCT, they commit earlier on. So we pursued that. Now we had already had a lot of information on this from a bunch of labs, including my lab, who did this work here in Wisconsin, in which we knew, starting with very important observations, Judith's lab had made, that there are a series of interactions that occur that specify the identity of the anterior blastomere at the two-cell stage, that's called AB. The AB undergoes lots of divisions and various cells along the way are told what to become through a notch signaling through the glip one notch receptor and later glip one and the other receptor, Lynn 12. And the signals for that, in general, come from P1 derivatives. So we talk about P1 to AB signaling. Doesn't mean these two cells are communicating. It means the descendants of these cells, the descendants of these guys send a signal, a delta-like signal, a ligand type signal, to these cells and change their identity. And we thought, okay, what we've observed is just the same thing. The specification of the AB lineage is basically causing it to lock down into fates. But it turns out that's actually not the whole way this works, because what we've found is, this is Narg Jabrayan now, who is a graduate student of my lab back then, if we grow AB by itself, so we isolate AB in the early embryo, if the signals are coming from P1, we shouldn't see that notch-dependent commitment event. But what we found, in fact, that we did see glip one dependent commitment in isolated embryos that are derived from AB. So they can't be any P1 signals, and the effect was quite strong. And so we thought, well, maybe there's another ligand involved in this process. And so we queried the known notch-type ligands, these guys here, and found two of them that when knocked down, DSL 1 and 3, gave the same phenotype, this DSL dependent commitment in isolated AB embryos, just like this. And so it looks like, okay, the signal's coming from somewhere else. And curiously, these two ligands are bizarre, because unlike nearly all respectable notch ligands, all of them really, these guys are not membrane bound. They're actually, they appear to be secreted forms. We haven't confirmed that, but they don't have a transmembrane to me. So it looks like that there's something else going on here that the glip one receptor in AB may, it seems to be responding to these secreted ligands, that's how we interpret our data, and that those ligands are somehow causing cells to commit earlier in development than they would otherwise do. And if you get rid of either the notch receptor or the ligands, you extend that period of plasticity. Okay, so that's one thing that came out of that approach. Now having said that, we got kind of a result that sort of made us think, well, okay, the MCT is a great boundary, but it's not an irrevocable boundary, because we found, so let me summarize what I've said so far. We know, if we look at all of development of the animal, from fertilization to an adult, and we, of course, we know all the cell divisions, beautiful work by John Sulston, Bob Horvitz, and Judith Kimball, that worked out the whole lineage of these animals. This period of multi-potentiality occurs actually for relatively short time. It's a long time in the embryo, in a sense, but it's sort of halfway through embryogenesis when this switch occurs. But over the lifetime of the animal, of course, cells are largely committed the rest of the way, they're exceptions, but for the most part, most cells are committed. We can't turn them into gut, for example. Well, we thought. And we can extend that period of multi-potentiality about that far, a little bit. It's a long time in the embryo, it's a short time over the life of the animal. Susan Mango's lab had shown that components of the coli-comb chromatin remodeling complex are required for that commitment, and if they eliminate those, they can extend that period a little bit. As I've shown you, if we eliminate this, not signaling, we can extend that period a little bit of plasticity, but still, commitment kicks in later. There's something else causing that commitment. But the surprising result we got after we had done this, we had the result for a while and just let it sit. Wow, why do you do that? Why don't you jump on it? But anyway, we found that in particular, this got a transcription factor could totally override everything I said, just blow away everything I've just told you, override the MCT and do a startling thing. And that is if we turn that transcription factor on at any time in development, all the way through adulthood, instead of seeing gut cells expressing a marker for gut differentiation, initially, we see nearly every cell in the animal trying to turn on gut development, at least this marker for gut differentiation comes on. That's transient, the cells fade back, and we see two organs in which this locks in, two organs which are converted to gut-like cells. The uterus, which I'll come back to in a second, and the pharynx, which is the anterior part of the digestive tract, it's a muscular organ that pumps bacteria into the intestine and it's mesodermally derived. And those are stable changes. So there's the pharynx, stably expressing this gut marker. And one thing that we see is very rapidly on, after we expose all the cells to this transcription factor, we lose pharynx gene expression quite rapidly. So obviously the pharynx has to stay as a pharynx and it maintains stable expression of this transcription factor, which is by the way also used in heart development and there's some similarities between cardiac muscle cells and the muscle cells of the pharynx. So there may be some interesting evolutionary connections there. So this transcription factor, which is made for pharynx muscle, drops off quickly. The marker for one of the muscle myosins that's specific for pharynx myosin fades away quickly. We haven't intentionally turned off those pathways, but we are doing that inadvertently as we turn on a gut differentiation. And of course the thing that's most convincing is the cells get completely remodeled. So this is across, this is not a very good view, this is a cross section through the gut lumen where these microvilli come off inside the gut lumen. There's this brush border and other structures that immediately jump out at you. And this is a cross section through the pharynx, it's got this Y-shaped lumen, it's got muscle cells that looks quite different from the gut. This is a cell in which, that was a pharynx cell, in which it's gone undergone this transdifferentiation and we get beautiful microvilli, all these little dots are cross sections through these microvilli fingers that are sticking off. Here's a close-up. And they have many of the characteristics of a perfectly well-defined gut cell as, and this was in collaboration with our colleague in C. elegans, David Hall, who did the EM. And so it looks like we're getting a transformation with this transcription factor. It's actually reminiscent of a disease in humans known as Barrett's metaplasia, in a metaplasia differentiated cell type inappropriate for the tissue it's in appears. And in this case, Barrett's metaplasia, the normally squamous cells of the esophagus are converted to these columnar-like cells of the small intestine. And that is a marker for a particularly dangerous kind of pre-cancerous state. So it's a conversion of foregut into hindgut. We don't know if this is really absolutely, it's the same kind of event that's happening, but it's reminiscent of that, foregut being converted into hindgut. And I'm gonna come back to that in a second, but let me also show you the other event that we've seen. And that is that the uterus of the animal can be also converted over into looking like a very nice gut. So this is a cross-section through the animal, again with David Hall. That is the cross-section through the uterus, where our eggs are gonna pass. And that right there is a uterus that has been converted to really a gut. So there's a little chunk of gut right there with this transcription factor. That looks essentially indistinguishable from the normal gut. So it really looks like we've gotten what we call, again trouble for this, transorganogenesis. We're converting one organ of the uterus into another organ of the gut. And we can get that to happen after cells have divided in the uterine lineages. And the cells are remodeled to look just like gut cells. So that's a normal gut and on the other side of that, that's the gut that's been created from the uterus. And the structures look very similar to gut cells. Okay, so what we've seen here is a continuum of switches. So the thing that's distinctive about, there's lots of cases of transdifferentiation. One of the distinctive things about this is that the transition from pharynx to gut seems like it follows a continuum. So we can do this without any cell division. The pharynx cells, for example, have been post mitotic forever in a worm's lifetime, a few days, three days even. They are fully differentiated, so we don't need to undergo rounds of division, like you do, for example, when you make IPS cells. There's no evidence that we have to go through a de-differentiated intermediate. There's an overlap between one cell type and the other that's transient, it's unstable, it's gonna slide one way or the other, but we do see that overlap. And we don't have to get rid of, unlike other cases we've seen, we don't have to get rid of any chromatin remodeling factors or anything like that. This just happens in one step, right in the intact animal. Okay, so we obviously wanna understand what are the mechanisms that allow this to happen, and I can just give you a little bit of information on that. First of all, one obvious place to look is to look at the factors that are involved in specifying the foregut and the hindgut, because the foregut is turning, mid-gut, I mean, the foregut is turning into mid-gut. I hope I said that right before. I hope I didn't say hindgut. This is the mid, the endoderm is the mid-gut. There actually seems to be a similarity in all of these creatures, worms, flies, and humans, in that Fox A-type transcription factors are involved in controlling the differentiation of the foregut, the esophagus in the case of humans, and the hindgut in all of these creatures, and gut factors are involved in mid-gut development. And the important regulator of foregut that is to say pharynx development that's really famous now was identified in Judith Kimball's lab. I should have you give this, Senator Judith. It was identified in Judith Kimball's lab by Eric Lambie and Susan Mango, and that is a gene called FOP4. FOP4 shows a beautiful phenotype in FOP4 mutants. Instead of seeing a nice pharynx, you get a very nice looking worm and the pharynx has just been yanked out of there. It's delete, it's gone. There are cells that are sort of partially differentiated, not fully differentiated. We don't really know what their identity is. And we were interested in knowing, are these cells, for example, more susceptible because they're less far along in their differentiation pathway to being converted into gut than these cells? And it turns out the answer is just the opposite. That the sort of semi differentiated state of these cells is insufficient to allow us to turn on gut differentiation in the wrong place in this animal, at least in the pharynx. And that's shown, if we look at FOP4 mutants, we basically don't see this effect at all. Now that is actually consistent with the fact that FOP4 Fox A factors have been shown to be what are called pioneer transcription factors. They can get in there and sort of open up chromatin in a way that primes that chromatin to be acted on by other factors, including gut factors in liver development. So in retrospect, it's not that surprising that FOP4 is necessary for this event. Okay, but we really wanted to do a comprehensive analysis of genes that are involved in this trans-differentiation event. So the strategy we've taken, and I'd love to tell you more about it, but I don't have the results yet. The strategy we've taken is to take L7 and hook it up to a switch that we can turn on and off. A switch called a degron. A degron causes the whole protein to be degraded, in this case, in response to oxen. So we steal this from the plant system. If we take the degron-fused L7 and express it in worms without oxen, without the hormone, then we can see the pharynx turning in the gut. This is, there's the normal gut, there's the pharynx turning in the gut. But what this provides for us is a very nice, small chemical switch for trans-differentiation. If we add oxen, that causes the degron to be brought into the proteasome, essentially degrade the protein. And so we get a very nice dose-dependent response, a dose-dependent switch for trans-differentiation. At very high levels of, relatively high levels of oxen, we get perfectly normal-looking worms that are perfectly viable. We've shut off this trans-differentiation event. In a dose-dependent way, we drop the oxen down to low levels, then we can turn on this differentiation switch. And this is extremely tight. So the postdoc who worked on this is, kind of goes over the top, I guess, but he looked at 12,000 worms, 12 million worms, I don't know why he did 12 million. 12 million, just see how tight this thing was. Oh, sorry, I'm sorry. Actually, no, that's wrong. This is 1.2 million worms. 1.2 million worms exposed them to Grumann oxen, so nothing happens, took them off oxen and all 1.2 million arrested. The thing is totally tight, doesn't slip through. Then he took 12 million after mutagenizing them and got 650 viable lines, which is actually, as many as we thought, out of 12 million. So we're picking up lots of mutants that can escape this switch, right? Okay, so that means in the absence of oxen, this can't be things that somehow allow oxen uptake or something, because it's the absence of oxen that switches on this trans-differentiation event and causes the worms to arrest, and these guys don't arrest. Okay, so we wanna figure out what all these things are, of course, and so the strategy we're taking, and I'd love to get some feedback on this, some help from the genomics people, is we're hoping, instead of mapping all these things and doing all the hard work, we're hoping we might be able to identify not the specific mutations in any one mutant line, but just the density of mutations across the genome by doing massive sequencing of all these mutants. So we've done this and we're trying to look for regions of the genome that are over-represented in variation. It's a hard problem, but if it works, I think, and I have some hints that it's worked elsewhere, we might be able to look at the whole genome in a single snapshot without ever really isolating a single individual mutation. Of course, we have to confirm these, but just by mutational density, predicting what parts of the genome are sensitive to these effects. Okay, the other approach we've taken, it's inevitable, is to follow what happens during this trans-differentiation process by analyzing the transcriptome. So we basically start with normal pharynx and uterine and switch them over to gut and sample over the time course, about 30 hours in this case, whole worms. This is messy, but we take whole worms and do whole transcriptomes on this. And actually turned out to be kind of interesting to do whole worms. It's a lot messier than doing single cell sequencing and so forth, but the reason it was interesting is it kind of shows us what happens in this transition, particularly as cells are pushed up and then either decide to become gut cell or don't decide to become gut cells. And so one of the things we see is if we do a principal component analysis, there are two principal components that jump right out that account for much of the variants we're seeing in the gene expression changes over this 30 hours. These are the control worms that we've treated exactly the same way and they're really developmentally matched. And it looks like there's sort of a trajectory of changes of gene expression. We don't know that this principal component refers to development, but I think it's consistent with the fact that it's following basically developmental gene expression. And the other principal component in this direction, we get some clues by what happens in the experimental guys because the trajectory there, it follows a nice linear trajectory there in both cases, three hours, six hours and so forth. That trajectory seems to be reflecting the trans differentiation event, this principal component, which counts for most of the variants, maybe reflecting what's happening to cells when they're being told to do the wrong thing or change what they do. And notice that we push them way out there and then they kind of spring back. Boy, the statisticians are just gonna freak out by me saying this, but they seem to kind of come back so that the gene expression patterns are not as far away from animals that are at the same developmental stage. So they seem to be, if this really refers to development, they do seem to be developing, but they don't develop completely and we didn't know they arrest. And they're also following what we're claiming is this trans differentiation trajectory. Okay, so if we look at individual genes, then we see things that we would have predicted. One thing is gene expression, for example, for the pharynx, like this muscle myosin or this transcription factor I've talked to you about in pharynx differentiation, unlike the controls, which stays high and certainly doesn't go way down, drops off very quickly in between three and six hours. Really at three hours, blam, both of those genes go way down. These are genes that are supposed to be staying on for the life of the animal. So we very rapidly shut off the normal developmental program as we're turning on gut development. And this one was rather striking. This is a gonad gene that's expressed in the gonad. And you can see that this thing flares up in during this window of time in the controls and it just never even goes up, hardly a squeak in these meetings in which we've just provided this gut transcription factor and of course the uterus is turning into gut during this period. Okay, so that's sort of not unexpected. The other thing we saw, which I suppose is not also unexpected, but seems to be happening across the tissues that do not undergo transdifferentiation is that a bunch of genes, not all, not all of them behave this way, but if we look at different cell types that don't undergo this event, the skin, certain types of neurons, the germline, body muscle, what we see with a lot of those genes is all of these guys see them in the first three to six hours to plummet. They go way down and then they rebound, they come back up again. And this is consistent with what we see during this plummeting. We're seeing the transcription, that we're seeing markers for gut come on all over the place. The cells are going, whoa, I'm not sure what I'm supposed to be and they turn off their normal gene expression and then they bounce back again, okay? We'd love to understand what that bounce back is and we can talk about that in the discussion. That's certainly one of the things we want to understand during this transition from one cell type to another and the ones that decide not to do this. But the other thing that jumps out of this is the class of proteins that come out and one of the major classes we've seen in this period that goes up very quickly in the first three to six hours, that's these guys. These guys are genes that are part of the proteasome complex. They're involved in protein degradation. And so this is just functional clustering and you can see this whole cluster of genes of components that make up the proteasome and degradation complexes. And that may be a freakout response that cells get when they're getting this wrong transcription factor, but we think that the activation of the proteasome, the protein degradation machinery may be important in allowing cells to become remodeled from one type to another. For a cell of one type to be remodeled into a very different cell type, there's gotta be a lot of protein turnover. We've gotta completely change the way these cells look. And our preliminary data with some of these genes, we haven't gone very far with this yet, is that when we knock down a couple of proteins involved in degradation, we attenuate the response. If the degradation machinery is not there, the worms become much less sensitive to this remodeling process, to the trans-differentiation process. So we're postulating that when cells are converted from one type to another, from pharynx or gonad into intestine, of course we turn on new transcription, we turn off old transcription, because you don't wanna keep telling the cells to do this stuff, but we also activate a program of protein degradation that allows both for the existing gene regulatory networks to be eliminated, and also for components that are, structural components of these cells to go away and to build the new components, and allow for the remodeling of cells to this new type. And we'd love to really follow the cell biology of that remodeling, something we wanna do. Okay, so that's what I was gonna say about, I think I have some time still, about developmental plasticity. Now I wanna shift gears and talk a little bit. I used to be the director of the, co-director of the embryology course at Woods Hole Marine Biological Laboratory, which got me thinking a lot about evolution, because it's really an evo-devo type course. And then I did a sabbatical one year in Paris, which was sheer misery, as you can imagine, with Marianne Felix, who is sort of a warm evo-devo type person, got thinking about this, came back to the lab, and a few years later, we started working on this problem. And that is, we know, as I said, that gotta type transcription factors, turn on endoderm in basically everybody. So this is a completely conserved tool kit. There's differences, I'm sure, but it's a conserved tool kit. However, like most pathways of differentiation, like sex determination, for example, and many other pathways. Although they sort of impinge on a bottleneck, a certain set of regulators that are conserved across metazone phylogeny, the inputs tend to be variable. So I'm gonna go back to this really ugly gene regulatory circuit that I showed you, and just point out two inputs. I mentioned the skin one transcription factor. These are maternal inputs provided in the oocyte, that work on early development. Skin one, and then there's another input, wind signaling pathway, which is also involved, both of those in mobilizing the endoderm pathway and turning on endoderm development. Those guys act redundantly with one another in C. elegans. If you get rid of one or the other, you don't get a fully penetrant phenotype. You get embryos that have gut and other embryos that don't have gut. You get sort of this bistable switch. Interestingly, a former post-accommon Morris Maduro, who's now at UC Riverside, showed that in C. brixie, which is separated by some 20 or something million years from C. elegans, instead of this being like an or switch, where either that input or the other input can work, either of those, he showed it's an and switch. If you get rid of either one of these components, skin one, or the transcription factor through which went acts, either of those, they completely knocked down the system. So we knew that there was stuff going on evolutionarily in terms of the inputs of the system. And so I thought, well, let's just have a look at to see if this thing is varying within a single species. Are these inputs actually, can we see impending changes in the gene regulatory network in short-term evolution? So we took advantage of the fact that most notably Marianne Felix and lots of other people have gone around the world and collected nematodes and it's a great way to see great sites in the world. They've collected C. elegans from all over the place and each C. elegans isolate represents a unique isotype, a unique genome. They're homozygous at essentially all positions, each isotype, but they're different from one another. Each isotype about at the same level that any two humans differ, between one in a thousand and one in 10,000 base pairs, something like that. So about the same sort of level of genetic diversity as the human population. And so we thought, okay, what we're gonna do is we're gonna get rid of skin one, this transcription factor, which we know in the laboratory strain results in this bistable switch. 30% of the embryos make gut, 70% do not make gut when we get rid of that transcription factor. That's been known for years, many years, since Bruce Bowerman identified this factor as a postdoc. And what we found was kind of startling. There is the laboratory strain, there's the 30%. What we found is across these wild isolates, some of the wild isolates almost don't care that skin one is there. They can make gut more than half the time, about almost 70% of the time in the absence of this transcription factor. And some of them, this is one species and it's all one species, this is a really important transcription factor in development. Some of them absolutely require it. You never get gut when you get rid of that transcription factor. Okay, caveats, blah, blah, blah, blah. We've done this by RNA-I and then we've introgressed, which is to cross in the chromosomal mutation into key guys. This is real, this is not RNA-I response. Okay, and so here's a strain that that's the low strain on the low end and we've done quantitative genetics to identify genes involved in this. We basically get four major regions of the genome plus two more interacting factors that account for most of the variation. And one of the regions that we found, one of the QTLs we've identified right in here pointed to an interesting gene and I cannot tell you it's definitively the right thing, but it looks curious. And the cool thing about this gene is, I hope it's right, but might not be, is that we found this gene involved in endoderm development many years ago doing this thing called biochemistry. I was trained as a biochemist, so I had this very talented undergraduate student at UCSB who's now a professor at the University of Pennsylvania, Eric Witsie, and Eric isolated biochemical quantities of early factors that bind two and one, one of these important regulatory genes and found this factor by mass spec called pure alpha or PLP1. This is actually a single-stranded DNA binding protein and there's something weird about the region it binds. It looks like it may adopt some weird confirmations, but that's data we haven't followed up on. But the reason this is interesting is right in that peak region and there is a variant at one position just outside the DNA binding domain for this transcription factor. There are two forms. There's a valine form and a spartate form at that position 112. And notice that there's a strong correlation with the behavior we see. The spartate form, like C. brixie has the same form, shows very strong requirement for skin one. Remember I said that for brixie and the strain like the MY16 strain that I mentioned that never makes gut when we get rid of skin one shows that behavior also. It's not a perfect correlation, but the guys that can make gut in the absence of skin one actually show this variant or they preferentially show that variant. So the cool thing about this is if this is right, we've linked a biochemical experiment we did many years ago with now a quantitative genetic experiment and we have yet to crisper this in. We gotta do that. We've had some struggles for technical reasons, but I wish I could tell you it worked, but we're not there yet. But this also opened up another interesting observation. That is that this transcription factor PLP1, pure alpha like protein shows differential interactions with these two inputs, skin one and the wind input. If we get rid of PLP1 by itself it doesn't block endoderm formation, but it does synergize with mutants in the wind pathway or skin one in opposite directions, which is curious. So if you make a wind double mutant with PLP1 you basically essentially turn off the system. If you make a double mutant with skin one you actually suppress the phenotype. You actually get the opposite effect. So we don't fully understand this reciprocality, but it made us wonder if there is some interesting reciprocality in the requirement for factors that act with wind and skin one. Okay, so remember I said these two inputs turn on endoderm. They act essentially redundantly. They act in sort of an ore switch in C. elegans, not C. Briggsy. And what we found is, this is Yamila Torres-Clara now who did this work. By the way, a lot of this stuff was done in my other lab at the University of Auckland in New Zealand which I had for about five years and she was a fantastic graduate student who did this work on skin one and the wind pathway. So we were interested in wondering if there's some sort of reciprocality in the requirement for these two inputs. So in fact if you get rid of the wind ligand across all these wild isolates you see a lot of variation just like when you get rid of skin one. And we wondered if there was reciprocality and there on a strain by strain basis we couldn't detect a reciprocality but on a snip by snip basis if we take the highest scoring snips that are most associated with the variation that we're seeing with skin one then we see a strong negative correlation between or reciprocal correlation between the requirement for mom for the wind ligand and the requirement for skin one. This line is sloped this way. Notice these quadrants are essentially empty. These are snips that give a weaker phenotype when you get rid of the wind ligand they give a stronger phenotype when you get rid of skin one and vice versa. And that is not only true for all the top scoring snips but across the whole genome. If we look at all the 4,000 or so snips across the whole genome for all six chromosomes this line is always sloped in the same direction. That means there is this reciprocality between the wind input and the skin one input if you need skin one more you need mom two less and vice versa and that almost suggests there's actually an active process either in coordination between these two pathways or something constraining evolution in a certain way that causes this to happen. We don't understand that at all. So that sort of summarizes this idea of this kind of reciprocal inputs. Okay. This usually goes off at like 20 after right. 10, okay, you sure? All right, okay, Jim said. People can leave, please leave if you want, okay. Okay, so I'm just gonna talk about one more thing. Developmental fidelity. I told you I'd talk about fidelity. That was developmental plasticity, evolutionary plasticity and now evolutionary now developmental fidelity of I think a really understudied aspect of development. There must be systems I think to correct errors that are made all the way along in development. Errors are gonna occur inevitably and we think there are ways to repair them. And when we've looked across these wild isolates, I'm gonna just throw a whole bunch of phenotypes at you quickly, you'll see why. We see errors being made in some of these wild isolates that aren't made, for example, in the laboratory strain. Although the laboratory strain does make this error, John Sulston, all C. elegans people's hero who passed away this year and I'm just heartbroken over it. John Sulston observed when he analyzed the lineages in the male, he observed that although males normally make this very stereotyped pattern of sensory structures called rays, there's nine on each side, beautiful structure, that he said in his paper on the male, something like half, some fraction of the males are missing rays. And we sort of rediscovered that observation and found that the rays that are missing, when you see one of these structures gone in this laboratory strain, that missingness depends on the cell death pathway. So here's the strongest effect. If we get rid of EGLE-1, which sort of initiates activation of the program cell death pathway, the apatotic pathway, made very famous by Bob Horvitz for which you got a Nobel Prize, that eliminates the missing rays. That is, EGLE-1 worms, tails, male tails, male mating structures are better looking than the ones that don't have this mutation. And if we look across the wild isolates, we see some of the strains show these losses a lot across, so it's not just the laboratory strain, a lot of wild isolates do this. Some of the strains never make this mistake. They make perfect tails all the time. And so I wanna point to you these strains, AB4 and QX, blah, blah, blah. It's one, two, one, one, but I leave the numbers off. Now, that's sort of the stochastic cell deaths that's happening. So program cell death in CLE is supposed to be absolutely reliable, stereotyped, except in the germline, but in the soma, stereotyped exactly 131 cells die every time, supposedly. So this is a case where there's stochastic cell death. The other thing that's supposed to be completely stereotyped is the arrangement of the internal organs of the animal. The gut and the gonad are always arranged in a certain left-right way. For example, Bill Wood's lab showed in 30,000 worms that you never saw a reversal of the arrangement of the gut and the gonad. The gonad's on one side, the gut's on the other side, depending on where you are in the length of the body in hermaphrodites, but in males it's absolutely reliable. And we actually found in some of the, although it's not true in the laboratory strain, we found in some of these wild isolates, we're seeing reversals frequently. That is, they don't show the stereotyped left-right handedness in terms of the arrangement of the gut and the gonad. We see these guys, which we call heterotaxies. Heterotaxies means a partial left-right reversal, because if we look at markers of cell types that reflect back to the chirality of the embryo, the handedness of the embryo, those markers are particular, these are called solomocytes. They are not reversed, but the gut and gonad positions are reversed in these guys. And so, but we're interested by the fact that that same strain that showed the errors in cell death also showed, that didn't show errors in cell death, didn't show errors in this, and the strain that showed higher errors in stochastic cell death also showed these heterotaxi errors, big deal, that's two. So we started looking a little further. We looked at the placement of the boundary, made extremely famous. Again, I should have Judith give this whole seminar. By Judith, the boundary between the mitotically proliferating germ cells in the distal end of the gonad, which you've all heard about many times, into meiosis at what's called the MRTZ boundary. So these are the proliferating germ cells, and then they switch into meiosis. And we've just looked at the reliability of that boundary, how variable it is. And what we found is, I know there's caveats to this, but we found that this strain shows very low variance in that boundary. And that strain that I said, so that was a high fidelity strain in the other characteristics, this guy showed high variance in the boundary. This is not exactly an error, it's just variation in the position of the boundary, not the position of the boundary but the variance in the position of the boundary. And then we just took the most boring phenotype we could think of, because it's easily scored, and measured the length of newly hatched first stage larvae L1 worms, and we again see the same correlation. The high, what we're calling the high fidelity strain shows low variance, not changes in the length but just variance in the length, in the length, and the QX1211 strain shows high variance in that difference. The other thing is, this strain is much more robust to removing these two factors for gut development. This guy is more sensitive to it, as if it's more poised to make mistakes. Now, this is all a stretch, but what we'd like to suggest is the possibility that either there's a common genetic underpinning to this, which we're figuring out by quantitative genetics, or at least there are strains that are naturally in the wild higher error strains and strains that show lower rates of error, based on perhaps their natural history, perhaps spoiled brat C. elegans that have grown up in C. elegans in California and never had to deal with anything. You know, it's just perfect weather all the time. Maybe can get away with being more higher fidelity and less variable, and ones that get more variable environments need to have a more natural variation. That's a possibility. Okay, I'm not gonna push that any farther, but it's an interesting observation. And we're working on trying to make that discrimination, but what I want to do, because we started thinking about fidelity, is we wanted to ask if we could actually see fidelity mechanisms acting, operating in a developing embryo. Oh, this was just summarizing what I said, so this common variability. Okay, so here was the problem that's sort of eaten at me for a long time, and that is that each of these founder cells that I told you about, each of them is distinguished because they inherit a particular cell cycle clock. So this one, and all its descendants divide the fastest, and as you go from the anterior of the animal to the posterior, the cell cycles are slower and slower across these founder cells. So the reason this bugged me is that these clocks are not synchronized with one another. They're not, but they have to be harmoniously done because the geometry of the animal is very tightly preserved, at least in the early phases of development. And so how is it that this animal can develop over a 10-fold range, almost, in speed, depending on whether you change, where you put the temperature. So at about five or six degrees, they develop almost 10 times slower than they do at 27 degrees, which is right at the limit of viability. All these clocks must respond to temperature in the same way, and so either they just know how to do that, or maybe there's communication between these that coordinates it. So we decided to test this possibility. I'm gonna make it very simple here. Here's AB and P1 we've seen. AB normally divides before P1. It's a faster dividing cell. The question is what if we force these embryos to try to switch the order of division? And we can do so by putting them in a temperature gradient. So what if we warm up P1 to a perfectly comfortable temperature? Cool and warm are good temperatures, they're not hurting the thing, but we just make a gradient difference. If P1's warmer than AB, at some point it's gonna divide before AB. If there's no compensation, if the cells don't talk to each other at all. If the cells talk to each other, we might not get the out-of-sequence divisions. The embryo may resist our efforts to try to create what we call a discordant condition. Embryos never see temperature gradients like this in the wild. These are huge gradients I'll show you in a second. We think that this is a proxy, the gradient we're imposing, that the discordance is a proxy for the errors that just creep up from intrinsic molecular noise and environmental differences. That's our proposal. So we built a device in, so we have a lot of great engineers at UCSB, very strong engineering campus. This is a microfluidics device in which we can flow in embryos, trap them into these little chambers there in which they've been carefully designed to impose across the anterior, posterior axis a very steep gradient. This was done with my colleague Carl Meinhardt in mechanical engineering. Here's the device. And so we're able to heat up one cell versus the other and we can flip them around. And the first thing we thought, we saw that we thought was rather startling is up to a seven and a half degree, this says seven, but we can go as high as seven and a half degree gradient across the long axis of the animal. These should be very discordant conditions. By the way, seven and a half degrees across 50 microns in an embryo is about the same level of gradient that if you stood on the sun and had your head in empty cold of space across two meters, that's about the steepness of the gradient. It's a pretty steep gradient. It doesn't exist in nature, right? But one thing we saw is embryos can survive that. Now, they should be driven way out of whack, but they can't be out of whack or they die. We can still get a substantial fraction of the embryos living under this gradient. The other cool thing we see is that that's dependent on the polarity of the animal. If AB is on the warm side, so it's getting told to go faster, it's able to slow down and wait. I'll show you that in a second. It's able to compensate more for this discrepancy than if P1 is on the warm side. P1, you can see all these data drop down and when P1 is on the warm side, much more than these guys. P1 cannot, if it's P1's warm, they never survive. And we looked at the cell division patterns and I won't go through, I won't go through how we do this. Let me just take my word for the analysis because I have to go through the whole mathematical arrangement. But basically, anytime you go farther up on this line, that means slower division. Relative to the predicted division timing that these cells should do based on the temperature they find them at, based on standard curves we created at constant temperature in the same device. And what we find is that the first thing we see is you stick these embryos in a device where both temperatures are perfectly fine, but there's a seven degree difference and the embryos freak out. They slow way down, both cells go way down. It's like some kind of checkpoint or something. Whoa, man, something's wrong. And the other thing that we see is if AB is on the warm side, it slows down by more than P1 does. Anything on that side of this line means AB is slowing down more than P1. They both slow down, but this one does more. And P1 slows down by more than AB if it's on the warm side, but it doesn't do as good a job as AB does. Consistent with the lethality we saw. That's those data. Those data are for the embryos that lived. The ones that didn't show this compensation nearly as well. This is wild type. There's almost no deviation when you have constant temperatures. They just, they really are clocks. These guys are the ones that died. So it looks like that compensation system to this discordance that we're imposing on them is actually doing something for the animal. We think it's monitoring deviations that occurred during normal development and correcting for those deviations. So we're suggesting that instead of getting these out of sequence divisions, embryos can recognize this problem. We think that P1 and AB have to be talking to each other and ensuring a normal developmental pattern as a result. Okay, so that's all I have to say. I don't know that I need to go through all these conclusions. I won't do that because you already heard everything I said. But the students asked me to do this and I never thought I would do this in front of Judith Kimball, but I have tenure and I'm not at this university. So I'm embarrassed to do this in a way because scientifically this is stupid, but it's really fun to think about. I was a total space geek when I was a kid. Neil Armstrong and Buzz Aldrin were walking around on the moon when I was 13. The day I turned 13. Okay, so I have this crazy collaboration with a guy named Phil Lubin at UCSB in physics. And the goal is simple, just to go to the next star. So if you take the best rocket SpaceX or NASA can make, how long does it take to get to Proxima Centauri, the closest star which is four light years away? Takes about 50 to 100,000 years. So we're screwed, we're not getting to another star unless we find wormholes or something. Except that we may be able to, because Phil has come up with this roadmap for doing this, using laser assisted transport essentially. So the notion is you build an array of lasers. This was actually on the cover of Scientific American about a year and a half ago. An array of lasers that become focused onto a very reflective sail, light sail, that will accelerate anything that's attached to it in about five minutes to 20% of the speed of light. The cost is huge, the amount of energy for this is ridiculous, but it is doable. And as the cost of solid-state lasers drop, I'm hoping this could even happen in our lifetime. But we have the exciting possibility, since we can get up to 20% of the speed of light, perhaps sending probes to the next star in a human lifetime, which is a pretty exciting prospect, I think. We don't have to wait 100,000 years, I can't wait that long. Okay, so I heard about this, and I told Phil, I asked Phil, have you ever thought about asking what happens when you send life across there? When you go various points along the way, what if you could just sort of query life along the way and see how does it work? Is interstellar travel consistent with life? Is it consistent with development, behavior, et cetera? And I think you know where I'm going. So what we're doing is we're working with him now to try to build microfluidics devices, to put C elegans and tardigrades on these little one gram chips. And if that works, they will be the, I don't have that, do I? Oh, they will be the first interstellar astronauts. We're trying to make worms and tardigrades the first interstellar astronauts. There, I said it, students, now you can forgive me. Okay, thanks for your attention, I'm gonna stop there. I knew you'd do that one. We have to create the holes in the microfluidics chips, yes. Questions? Yes, we have. So the first one is that I didn't say the other ones didn't. We didn't know they did, but in fact, the guys at the end of the cascade, L2 can do it, if we push it hard enough, L7's better, L7's a lot smaller than the others. Maybe it's size, and even N3 very poorly can do this too. The N1 absolutely cannot, cannot, cannot. Nor any of the other guys we've seen. But we did try the experiment you're talking about and so far I think the problem is you gotta get FOP4 in there early enough to sort of set the chromatin state. We did the simple experiment of just turning on FOP4 everywhere and L7 at the same time, nothing, it just looks like we haven't done anything else. We just turned on L7. We're trying now to get it on early enough, but the problem is, so it has to be after the MCT, because we don't wanna turn everything into pharynx, but it has to be early enough that we can still do the pioneer factor, pioneering kind of thing. That's what we're doing right now, but we don't have the answer. It's a great question. Unfortunately, we thought it would work, didn't work. Oh, sorry, I'm sorry, you're the moderator. There's forward pressure on it. Well, yeah, I guess that's back pressure, right? Yeah, it's being, in fact, if we don't do that, the embryos die. You have to keep a slow amount of liquid flowing through there, I guess, because they're sputing out protons or using up oxygen or something. So they actually, if you just leave them, they do get stuck reasonably well without a flow, but without the flow, they die. So we have a low level of flow going through there all the time. The gradients, by the way, have been very, we've been validated them very carefully, and the model and the predicted measurements are just spot on. So we know they're experiencing that grading, and then there's a biological response, too. So, but we do keep a constant flow through there. Well, I showed data for both orientations, okay? So we do, I mean, we follow the data independently for the two orientations, is that what you meant? Oh, yeah, yeah, yeah. So we validated the device does not cause any lethality, regardless of which direction we go. Constant temperatures across the range from 16 degrees to 24 degrees, everybody's fine. We see a very low level lethality. Done all those controls. We have nice curves for that, yeah. Yeah, Phil. Well, yeah, that's what we're claiming, but I, yeah. They're not longer, they're just more variable. So it's just all those plots, I should, were variance, right? They were not absolute values, like the position of the end of the stem cell, of the germline stem cells, that's variance, not absolute position. We don't see sort of quantitative, I mean, maybe there's some variation between the strains, but it's variance, okay? Well, I can come up with a hypothesis. It's probably, I can come up with a model, one component, right? Lots of variability, I would like to see. Right, yeah, right, right, right, right. Yeah, and again, I hope I didn't overstate, we do not know there's a common genetic origin for these variations, but if there may be strains that are more predisposed. I can think of one famous one, HSP90, which is a biological buffer, it's claimed to be a biological buffering system, let's say, you mutate HSP90, you see more variable development at the top. So linguists show this like 20 years, it was like a long time ago. That's one possibility, things like that, things that have buffering capacity. MicroRNAs, well, we're doing the cool, the convenient thing is these, the hypervariable strain and the high fidelity strain, the low fidelity and high fidelity strain, you know, they show all these parameters. Yeah, we've built RILs, so we actually have done, we're doing the QTL analysis, so far it's concentrated on the cell death effect, and we've got four genes, four peaks, not in a gene. One of them looks very interesting, we think we may have the gene in one of those peaks, but I didn't want to mention it because I'm not convinced yet. But, so that's the way we're going out, but it's kind of a freebie because we will get a bunch of these phenotypes from the single analysis, from the single set of RILs. We have the right RIL set, we have a high and low fidelity strain, they're segregating all those traits. We don't, well we haven't done enough to say there's clear co-segregate, what we want is to see if any peak is in common across these phenotypes, and they may not be. Still doesn't say that there aren't hypervariable strains, so we need to analyze more phenotypes. You didn't say fit for a second there, did you? You didn't say fit? I didn't say fit. Okay, good, because my bias is that low fidelity guys may be a lot more fit under some conditions, I sort of imply that. So we were worried that this is just reflecting some stress response or something, that the low fidelity strains are more sensitive to temperature, and there are optimal growth temperatures for different strains. So we looked at that, we have not been able to correlate anything with fecundity, viability, growth rate, any of that stuff, and temperature, and the fidelity. So at least temperature is not an important variable, okay? My guess is that the high fidelity strains are probably more variable in the wild, because the wild is a messy place, that would be my guess. And the low fidelity, the high fidelity we're seeing, may be an artifact of growing in the lab, it's all constant and beautiful, right? I mean, that's a possibility. But does that answer your question? A little bit. Can I answer it any further? I don't know if I can see it. That's in the early embryo, yeah. It doesn't make a tube. It doesn't make a tube? No, you actually get brush border sort of between the cells, but you don't get a good, we haven't done EM on them, but it doesn't look like a good, it doesn't look like it's tubes. I think you probably need some asymmetries to be able to, I don't know. That would be fun to make it into a tube. It would be fantastic. I'd love to make it into a tube, yeah. I haven't seen that, so. Well, actually, when we look, this is old data, it's not a tube, it's sort of a sinewy kind of thing. We should really do the EM on it, because it may, it does, it sort of looks, we see the brush border winding through, but it's not a tube. It's like all over the place. Yeah, Jeff. So maybe it's because of my history with sea urchins before, I worked on it. Everybody knows that I'm going to go regular to development of this in old term. Do you seem to be able to cope with the aftermath of variability? I just wonder what your thoughts are on. Yeah, well, first of all, there are nematodes, freshwater nematodes that are highly regulative, for example. Let's see, is it not a curable ladies, no, it's another one, blanking out its name. You can kill A, B and get a, no, you can kill P1 and you get a worm out of it, for example, at the two cell stage, even though it's asymmetric, so nematodes can be highly regulative like that. But I think it's also the case that C. elegans is in the sense you're talking about more kind of regulative in the sense that if you grow them, Ralph Schnabel originally discovered this in the late 90s, and then John Murray, who's done very extensive automated cell lineage analysis, has looked at variation at higher temperatures. So if you grow them at 25 degrees, John Sulston's lineage and the geometry that he mapped out breaks down. If you, between something like a hundred, I don't know exactly, something like a hundred cells and maybe 400 cells, you see a lot of variability in cell positions. Positioned cells are really way off in the embryo. You wouldn't even, I don't think John would have been able to identify cells from embryo to embryo if he'd done it, fortunately England's cold, so you know, it just didn't have that problem. The room temperature in England is like 10 degrees, so I know it because I lived there. But if you push them to 25, it's still perfectly viable. You get this tremendous variation in timing of cell lineages and in positions of cell lineage, and that thing resolves back to a normal pattern. And in fact, I didn't mention it, but in these temperature gradient experiments, we can drive these out of sequence divisions, and when we do that, when we get P1 to divide first, so we totally script the geometry, most of the embryos look like crap, but we get some embryos that get to the end, they don't hatch, but they look almost like a normal worm inside there. So they seem to be able to, after this disaster, they seem to be able to compensate enough to undergo kind of a normal morphogenesis. Be really interesting in looking at dorsal intercalation in these guys, for example. They kind of seem to be repairing themselves. So I don't know that what you're saying is a clear line that these guys aren't regulative in that sense, and see your, or whatever verbed mammalian animals are. I think these guys can adjust a lot. Yep. That's across the whole length, yes. Oh, 24 was the high end. Did they melt at 24? I don't think we did 25, I think we had 24. Yeah, I don't know, yeah. So we can shift it down, actually, so you want us to go, you want us to go 15 to 22. Yeah, we can do that. That's a really good point. Okay, so the magnitude of the gradient should not have an effect if it's not doing what you're talking about. Very good point, we haven't done it. All right, thanks a lot.