 OK, so good morning. So my name is Marianne Felix, and I come from the Ecole Normale Supérieure in Paris. And I work, as you will see, on the elegant evolution. And so I think my part of the course is to tell you about evolution of developmental systems in particular. And so I'll have three lectures. And today, we'll take a view at a relatively large time scale to look at the evolution of an intercellular signaling network. This is work which is already a few years old in my lab. And then the two next lectures, I go to a shorter evolutionary time scale and go to more genetic approaches. Oh no, there is a bad connection in my computer. It starts already. So we can make the announcement now. Yes. Sorry. Sorry. So I'll keep pushing the computer now. This afternoon from 5 p.m., we have another presentation. OK? So it's already close to 4 o'clock. And that's the lecture coming in the weekends. So also, I want to introduce you to Petra. Petra is helping us for this weekend as the secretary of the school. So if you have a need or a request for a complete lecture, so please just read the report to her. She is in the office number three, just before the board. And I think that's it for the moment. Thank you. OK, so as I don't know, so please interrupt me. Whenever you like. As I don't know how much you know about evolutionary biology. For today, the only thing I would like you to consider is the time scales. And so you can work at very small evolutionary time scales where you can do within a species. So here you can cross individuals and do genetics. And this we will see on the other days. You can look at very closely related species. Or you can look at, for example, different families or phyla of animals or plants. And in this case, what you have that you can use is a phylogenetic framework. So phylogeny is the tree of relationship of organisms. So here you see a tree of all animals. But it could be the whole tree of life. Or it could be a subpart of this tree. For example, just the nematoda, the nematodes here, to which C. elegans belong. And when you do what's called evolution of development, evo-devo, what you do is you look at the state of the system in some of this taxa. And then you try to reconstitute what was the history of change in the trade. So this is an inference type of science. That's not, I mean, you can never prove because we can't have access to the ancestor how it was. But you can find scenarios which are most likely. And what I will show you is that a relatively small evolutionary scale is within a small part of the nematoda. So very briefly in evolution of most of, so you have genetic variations. And the part of development here is to see how when you have genetic variation, it translates into phenotypic variation in the output of the developmental system. So here we're going to talk about variations in the phenotype, the final phenotype, the morphology of the animal or whatever. We will talk about a self-faith pattern in development. And this goes through modification of development. So today we won't talk about genetics, but more our modifications of development translate or don't translate into a phenotypic variation. And I'm saying this because I think to this audience is clear, but a lot of evolutionary biologists don't consider development as being important. Whereas a lot of features of developmental biology, of developmental systems are important to understand evolution. And I think for people who are more interested in cell or developmental biology, this is probably pretty obvious. So the two main points I want to make today, if you just take home messages of my talk, is that when you look at a developmental process and you, for example, make a quantitative model of this process, you can consider whether the evolution you're seeing in a group of species corresponds to evolution in parameter space of your model. It may be the case, or it may not be the case. For example, you may need to add variables and parameters. If you have an intercellular signaling network, for example, or a transcriptional network, you may need to add transcription factors and loops and things like this. Or you may just need to vary parameters. So it's to consider whether the explanation of just varying parameters is sufficient in the group you're looking at. The second point I'll make is that, because it is the case in the developmental system we are studying, C. elegans-volva-development, or nematode-volva-development, is that you can have evolution of development here without any change in the final product. So this is sometimes called a cryptic evolution because it's cryptic, it doesn't show at the level of the final output of development, but you do have evolution in the developmental system. So the outline of the talk is here. I start with a brief history of quantitative approaches in the evolution of morphology, and then introduce you to C. elegans-volva-development, which I'm going to tell you all along this week. Then see how, considering it's a conserved, self-made output, how we can unravel cryptic differences among species, how we could use a model to see whether the quantitative changes in model parameters were sufficient to explain the evolution. Then we'll see evolution in further, further larger evolutionary scales, and then I'll give a brief overview of other examples. Then see elegans-volva-development. So first, regarding the history, if you consider, so if you consider the fact that you have evolution of morphology and that you can quantify the final product now, the morphology, not the development, so these first approaches were not considering development really in the, and developmental parameters in the way they were seeing the evolution of morphology. So this is derived from a book by Darcy Thompson on growth and form, which appeared at the beginning of the 20th century and then was edited again in 1942, where he tries to explain the evolution of shape. So here you see between apes and human skull, for example, or different fish species, just by deforming Cartesian coordinates. So here there is no development at stake here. Slightly similarly, but on a system where actually, so if you take the snail shell, it actually develops from continuously during development. And so David Raub, who is a paleontologist, tried to classify snail shell shapes that you could find in fossils by two parameters, the width, the ratio, sorry, the ratio of size of the different coils and T, yeah, is the width of the coil. And so if you vary the parameters here, you can have very wide shells or very tiny pointed shells and you have snails like this. So here again, evolution is described as a parameter change in the final shape, even though in this system it may correspond to a developmental feature. And one point that he found if he was varying these parameters continuously, he could not find in the paleontological record every type of shape. And so in conclusion, that you have a non-random distribution of shape. And so this notion of a space of possible morphology was taken again by another paleontologist, Stephen J. Gould, who is a famous evolutionary biologist who wrote a lot of books and whom you may know. And so he gave the name of morpho space to this space of possible shapes that you get by getting here these three parameters or the translation distance between the axis and expansion. And again, so the space was not completely filled in real paleontological record. Another concept that Stephen J. Gould introduced again is this time a developmental concept which is not quantitative, it's just words, but he proposed that developmental changes in evolution could result from two mechanisms and these are not exclusive. One he called allometry, relative growth between different body parts. And the other one is heterochrony, where it's the timing of development of one part relative to the other that changes. And so example of this, for example, this neotenic salamander, which developed to a reproductive adult in a shape that corresponds to the larval stage of other species. So now what I would like to introduce is a more modern version where you really look at the developmental, you do a quantitative model of your developmental process and what you're really exploring is the parameter space not only of the final shape but of the developmental system. So here I just idealize in two dimensions two different developmental parameters one and two. And so when the genotype changes by mutation and recombination or if the environment changes you may be in different points in developmental parameter space. And this may give you different points in a final phenotype space or it may not give you a different, you may have two points here in developmental parameter space that gives rise to a single phenotype here. The mapping from here to here obviously can be funneled into one phenotype here. And this is the case of the system we're looking at but it's not necessarily the case, okay? So it's this main idea that evolution can be represented as a parameter variation in molecular and cellular models of development. You have questions at this point, is it clear? Yeah, so this is a good point. So no, these type of models are at a smaller evolutionary scale and can, I mean, it depends what novelty is and I really don't want to enter in this. It's really another field of research. So we're not trying to count, I mean novelty is, you can have novelty in a quantitative developmental system in some way, it depends how you define it, I think. Other questions? Okay, so we move to C. elegans. So C. elegans is a small nematode worm which is about one millimeter in length which is totally transparent. So you can see by normal ski optics all the nuclei of every cell. And it develops in three days. So you have three days from egg to egg at the next generation. And so someone called John Tholston followed development under the microscope continuously and found out that the cell lineage, so the pattern of divisions of the cells and what they become ultimately is practically invariant from individual to individual. So this is a great tool to study development. At the end, you have 959 somatic cells, so cells that form the body. In the germline, you don't have invariant divisions. So we're considering only the soma here. And so what we will look at is just a few, a little part here, which is formation of the vulva which is the organ for egg laying of the worm. Another thing I should tell you is that C. elegans is a cell thing amaphrodite. So this is actually kind of a female that also makes sperm at the beginning of her life, of an adult hood, and then makes all sides fertilizers internally and then lays the eggs. So the fact that it's a cell thing makes it, it's great for genetics, but it also makes that populations in the lab are totally isogenic if you want them to be. That is, given adults, if it's totally homozygous at all loci in the genome, will give rise to progeny which are all identical, except for new mutation. So you get populations of individuals which are genetically identical. You also have males if you want to do crosses. So what we are going to focus here is this part of development which occurs, so up there you have the development in the embryo, then the embryo hatches and you have four larval stages that are separated by moles. And here we're looking during the third and fourth larval stage. So these cells here which are the base here, there are six of them, and these are the six cells I'm going to represent from now on. Yes, sorry? This arrow here? No, no, this, oh, sorry. This is the pattern of division of a cell. So you take the zygote, the egg on top, and then you have a first division like this. And so these are trees of genealogy of the cells in development, okay? So at the end you get 959, but you exactly know how they were born through divisions, okay? And you also know what this cell is becoming, whether it's a neuron, a muscle, and so on. And if it's a neuron, you know which neuron it is, okay? No, we'll see examples tomorrow. This is not the point of today, but yeah. And for sure, different species of nematodes have different numbers slightly. So I mean, this is not totally fixed and yeah, that's why I put quasi-inviant. I'm interested in evolution, so in variation, so if it was always the same, it would be pretty boring. So if you consider these six cells, so they're actually situated in the ventral epidermis of the worm, so here you see a little larva with the head on the left, the tail on the right, the ventral side down and the dorsal on the top. So these six cells, so there are actually 12 of them, which are called P1P to P12P, so they're numbered from anterior to posterior. And there are six in the middle, which are called three to eight, which have the potential to make volval tissue, okay? So these are really epidermal cells in the ventral epidermis. Here you can see maybe the nuclei which are in the epidermis. And there above it is the gonad forming, the formation of the gonad, and especially we'll see one cell here, which is called the anchor cell. So what these six cells do is that they form a cell fate pattern. And basically the developmental event we're looking at is formation of this pattern of three fates. So one of the variation we'll talk about tomorrow is for P3P, but for today, we don't talk about the most anterior. So for the five other cells, we have three different fates, which are color coded here, which are also called primary, secondary and tertiary. So the primary and tertiary fates are those that actually form the vulva, and the tertiary cells is kind of a backup cell if something goes wrong with the other. So if you remove, for example, the three middle one, five, six and seven with a laser, the other ones are going to take their place and actually become primary and secondary and form a vulva. Okay. And the primary and secondary fates, basically the primary fates forms a few cells which are at the center of the vulva. I think I have the cell lineage later. So they will go on dividing each with its characteristic cell division pattern. Again, we have these trees of division. The yellow cells just divide once, whereas these ones divide us three times. And then they form the vulva. We hear the anchor cell just on top of it. So there has been a lot of work since the 80s using mostly laser ablation and genetics to understand how this cell fate pattern was obtained. And I'll go in more quantitative details in today and tomorrow or the third day. Basically, this anchor cell of the gonad, if you remove it, if you remove the anchor cell with a laser, that's why I'm using a laser here, all the cells adopt a tertiary fate. So they become yellow. So the yellow fate is the default fate. And the anchor cell is sending an inductive signal which is an EGF-like molecule which is received here in P6P with a cascade of signal transduction which is called the RASP-Mapcainase cascade. So this EGF can diffuse. It's not anchored to the membrane and it can touch the red cells. But the red cells also receive a second signal through which is, so the blue signal here activates. I think I have probably this on the next one. So the blue signal activates transcription of a molecule called delta, which I represented by a D here, which activates a receptor in the next cell, a receptor called notch. So that's one very conserved signaling pathways among animals. Like EGF-RAS, we have here delta notch. An activation of the notch pathway in the next cell activates the secondary fate. So it's transcription of different effectors and also inhibits the blue pathway, the RAS pathway. In addition, activation of the blue pathway represses notch in this cell. So you have both, you have two cross inhibitory interactions between the blue and the red pathway which locks the fate of each cell. And then you have positive feedback loops as well. If you have really an excess of EGF, you're going to see the lateral cells being induced. If you don't have enough, as we will see today, you get ultrashary or you get some kind of intermediate situation. And so after these cells have divided, yes, they form the vulva, but this is not the part we're looking at. We're really looking here at the formation of this special cell fate pattern along the anterior posterior axis. Get clear? Good, so I'll come back to these pictures. So where basically what we will see is that the fate pattern here doesn't change in the species of nematodes. The first ones I at least I'll tell you about, but we'll have evolution here in the development. So the fate pattern is conserved for these five cells before P2P8P in the whole family of nematodes to which synrubditis elegans belongs. So the question becomes, how does this pattern, which is conserved itself, but does it form in the same way in all these species? And these species are really far apart in terms of molecular divergence. For example, C elegans and the closest species known to C elegans are as distant as mouse to human in terms of molecular distance. So the advantage we have compared to mouse to human comparison is that here we have exactly the same cells. So we can have what's called homology in evolutionary biology. We can actually find a cell that we can code P4P without confusion in all these different nematodes and study the same cell in different species. So this is very powerful to have this cellular level of comparison. And so now that we, so we have these species which all have this final cell fate pattern that we can follow using the division and the morphogenesis of the vulva, how do we reveal a difference in the way it's made? So to do this, there are several methods, but so the method, so one thing you can do is to try to look to quantify the signaling pathways in the different species. Technically, this was difficult to do. So what we decided to do was to actually perturb the system to reveal differences. And so to do this, we use laser ablation. So this is a very easy thing to do. You mount your animals under the microscope. This is a Novosky optics, pretty trivial. So here you're seeing the anchor cell and P5P, P6P, and P7P nuclei. And all what you have to do is to have a laser that aligns in your microscope and so you shoot the cell that you want and then for example here we kill the anchor cell and we can see the effect. So this was historically the experiment that was done by Jelit Kimball in the 1980s to demonstrate that in C elegans, these three cells then adopt a non-volval fate, a tertiary yellow fate. If you do it early. So here now what I'm going to represent here is, so the final cell fate pattern is that one. If you remove the anchor cell very early in most of the species, you get this non-induced pattern. So in all species so far, we'll see other species later, but in the senior update is genus, all species required the gonad and the anchor cell to form the vulva. The question is if you do different types of ablation, do you get some kind of intermediate state in the middle? So it's not really an intermediate state because it's the final state after the system in the middle of the induction. And you can imagine different types of cell fate pattern that you obtain. So for example, that P6P is induced sufficiently but doesn't induce its neighbors because it's not induced enough. Or it's not induced enough, so it becomes secondary red here, which is equivalent as the signal at a distance here. And actually to my surprise, the most common patterns we got are these ones with three tertiary fate, or this one which I did not expect at the beginning, but you can completely understand, you will see later, where P6P is not induced but P5P and P7P are normally induced to the secondary fate. So these are results that you can characterize basically with what happened to P6P, yeah. So what I'm going to represent later is the proportion of the three fates in yellow, red, or blue, so primary, secondary, tertiary. That's adopted when you do an ablation in a given strain of a species. So for example, we have these species called synopditis japonica. If you do ablation at successive time, you get either this or this, but never anything else. So in this case, it goes directly from one to the other. We have another species, synopditis bryxae, where we have very clearly in the middle this pattern with two, two, two, okay. Which we confirmed in many different ways with GFP reporters and so on. And then we got this surprising pattern in species called synopditis remanai, where P6P in the middle did not form a volatile tissue and a volatile invagination, but the lateral one did adopt a secondary fate. So in my coding system, it would be like this. So in this circumstance, what we asked then was to see whether these arrows were real. So if P6P was sufficiently induced here to induce its neighbor to become secondary. And so to do this, what we did was to ablate at the same time the anchor cell plus P6P to see whether its neighbors were affected. And yes, this is the case. If you do that, then they all become tertiary. So this shows that you actually have P6P that despite the fact that it's not induced itself, is able to induce its neighbors. So now I'm going to represent, as I told you, by taking just the fate of P6P. And because we have three different fates, this is represented on a triangle. Where for the different species or different strains, different genotypes within a species, you can represent, if you do this ablations over time, taking the middle window, what's the proportion of fate adopted by P6P on this triangle. And so you see some of them, which are basal ones in the phylogeny, which adopt mostly a secondary fate. Then you've got some like synoptic breneri or japonica which are directly to the primary fate. And then remanai, where you mostly have in the middle this secondary tertiary, secondary fate. And then you have a reversal here in this part of the phylogeny to a secondary, secondary, secondary fate. So this is a kind of description of the variation that we can reveal by perturbing the system. So yes, exactly, yes. We make a perturbation, right? Yeah. Yes. Oh, sorry. So the worms develop in three days. All this happens in about 10 hours because it's a short window of level development. And so we have led this during the third level stage and we look within six hours. It's actually a very satisfying experiment to do because you have your answers immediately. Sure? No, they're not. I mean, no, sorry. Yes, they are. It depends what the question is. They can produce progeny because they're self-fertilizing amaphodites. And so what happens if they don't have a proper vulva, so to get egg laying, you need a primary fate at least. So if they don't have the primary fate here, they will, I will show you pictures another day, but they don't have a vulva, but the amaphodite self-fertilizes, the eggs are fertilized inside her. So eventually, they cannot get out of the mother, but they hatch inside the mother, eat her, eat her. And so yes, she can produce progeny, yes. But she cannot lay. So the question if you did not hear is, if when you laser, if I understand, when if you laser ablate a cell, are you sure that something is not leaking out of the cell? Probably, right? So my answer is no, I cannot be sure. The only thing is that if there is an effect, it must be that I remove that this part of the signal. That's the most I can say, right? And practically the cell, you can see that they undergo some kind of drastic morphological change, and then they are usually eaten by the next cell. But for sure, you may have things leaking out. Not really apoptotic, I think. It's more degeneration. I mean, a mutant of apoptosis would not result in no death of this cell. Other questions? Yes, sorry? No, no, no, no, sorry. No, no, they don't give the same output. So what gives the same output is that if you don't perturb the system, in all cases, P6P will become primary, okay? Here what I'm showing is in cases where I ablated the anchor cell, the final product here will change. It's not a natural, it's not a normal output. I'm perturbing the system. So I'm getting to cases, for example, where, so this representation may be misleading. I have to make clear. This is ablation time, and these arrows is not time of development. Is if I ablate over time at different times, I will get as final product this, or I will get as final product this, or I will get as final product this, okay? Yeah, sorry? No, no, no, this becomes irreversible because then the cells divide and it's irreversible. Yes? I mean, this is a quantitative, you may have 80% of the animals doing this, right? So you have variation among animals, subjects to which I will return tomorrow. But basically to look at this variation, you need, I mean, laser ablation is not a good tool because there is variation in laser ablation. So for this I need genetics, basically. So that's why I'm not talking about the variation today. It's mostly, we're talking about the proportion. So that's why I have a proportion. Any other questions? Fine. So the next question we asked is knowing the network and here I have simplified this entire cellular signaling network, I showed you before. Can we guess what is going on in the different species? And so I will show you the guess and then I will show you a model that actually vindicated this guess. So if in C. elegans we have this situation here in blue and now it's coded by species. In Briggs, you can imagine that the rest signaling is able at when it's at lower level to induce the secondary phase in P6P more easily and that you have maybe less inactivation of the rest pathway laterally. And this weird pattern in synoditis remani with a tertiary in the middle, you can rationalize by the fact that it's very difficult in these species to get a primary fate here but you have transcription of deltas very easily at low activity of the, of anchor cell signaling. And so you get activation of the secondary fate here in the absence of fate here induction of P6P. Yes. No, so again, all these species, the final fate is two, one, two and yes, they reproduce normally. Now, if you ablate the anchor cell, you will get these different fates and no, this doesn't develop normally if you don't have a primary fate. I mean, they don't lay eggs. So one thing we first did was to see whether on this space of P6P fate, we could move the species around just by mutation or transgenesis and so we used, so the reference strain of C elegans which is the reference species is here and we could basically move it around in this space by changing the level of activation of the RAS pathway here with this mutation, for example, or increasing it here. We could also increase, take Briggs' C Briggs' here, increase induction by the anchor cell and make it go along this axis. So this axis looked like the level of sensitivity of the RAS pathway and then to go laterally, what we could do is to make a mild gain of function in a much pathway. So you increase the activation of the lateral induction and you actually reconstitutes from C elegans the remani output. So again, these are just looking at the final fate after anchor cell ablations. So the next thing we wanted to do was to try to make a developmental model of the system and for this we collaborated with Kerry Kim and Ed Monroe at Friday Harbor and this was the work of Eric Aorios who was shared between the two labs and Josla Milos in my lab and this is a pretty old by now work about 10 years ago. So we started from this vulva signaling network architecture and we decided to include all the feedback loops and cross-inhibitor interactions but to simplify, for example, the RAS pathway is just represented by one step with an affinity component and a cooperativity one. So the fact that it's actually a cascade of kinases makes it much more sensitive and with a cooperative activation and so this is represented in the model. So our model looks like this, which is not very telling but you have in blue again the RAS pathway here, the EGF activating a map kinase to phosphorylation and then you have cross-talks between the pathways, so in green here where you have a degradation of notch inside this cell and activation of two forms of deltas. So in C. elegans you have transmembrane forms of deltas called here like two and a diffusable one, one that can be secreted without a transmembrane domain and then this will activate notch in the next cell but also in the same cell and this will inhibit here the notch pathway and we consider that here two, the outputs are here, the primary fate corresponds to this node and the secondary fate to that one. And so the level of activity of these two nodes which are called here EGLE 17 and EEP 1 just after the name of actual reporters that we use, GFP reporters, you have here depending on the level of activation of these two reporters we define that the cell at the end as a primary or a secondary fate or if both are low, a tashary fate and then you have undecided fate in the middle. So we started with a model with several cells and with a lot of parameters actually 40 parameters and what they did was to pick at random in a large range for example of two logs, each of the parameters and to see whether after running the differential equations you were getting the final fate pattern and we asked for the pattern to be stable over time in the model and so it's an approach where you basically you have a huge, huge parameter space and you look within this huge parameter space of 40 parameters whether you can pick sets of parameter at random and find a solution that actually gives the final target pattern. And just at random, even though this parameter space was huge, we found 8% of the solution of the random parameter sets that got the target pattern which is always, so 670,000 that we stored. So now we are going to later to work on this but first what they did was to see what, so what you can do with this is see whether there is a bias in the parameter that you introduced by asking for this target pattern and there isn't a lot, it's basically compared to the range we decided at the beginning we have some biases but this is really dependent on the range at the beginning and mostly it's working through the transparent range delta and with low EGF diffusion. There was a question in the back, no? Sorry. Okay, so now the main point of the model was to take these target patterns and see whether these solutions were doing it in different ways. And so now I'll come back to discussions in the silicon-volva development field on different ways to do this self-made pattern that I briefly alluded to so far but did not enter into and also using our experiments of laser ablation whether we can mimic the evolution, I mean, represent the evolution amongst another species. And so here we keep the topology of the network constant, I mean not completely because some of the parameters can go to zero so it's equivalent to removing but we don't add something. And keeping constant again the output since we only choose these solutions that give rise to the same output. So if you think about these systems there are actually two ways where the cell fate can be patterned in C elegance. One is very simple, it's the anchor cell induces the primary fate, the blue fate that activates the secondary fate. And this was suggested by experiments of some colleagues whereby they constructed animals which were genetic mosaics. So that's when you take an animal and you can make some cells mutants and not others. I will not expand on the techniques to do this. It's different with different animals, flies and worms and different things. But here what they saw is that the secondary cells, so here I represented a circle for those with the receptor of the blue signal of DGF and the square for the mutant cell. And by looking at many animals what they found was that you really needed the receptor to become a primary fated cell but you didn't need it to become the secondary cell. But the secondary cell had to be next to a primary cell. So this gave weight to these mechanisms where it's basically the EGF signal in blue is received by the primary cell which is able to activate its neighbors through the notch signaling. Another series of experiments were done about the same time in another lab where what they did was to isolate one of the cells and when I say isolate, it's not really isolated, it's killing all the other ones with a laser. So you're in a case where you don't have neighbors to induce you. And then using a transgene, they heat shocked. So you can have a transgene which is under the control of a heat shock promoter so that's inducible by heat, by raising temperature of the animals. And what they saw is that if they were not inducing the expression of the transgene, the cell became tertiary, the default fate. If they were inducing a lot it became primary as expected but in the middle at intermediate doses you could get an isolated secondary fate. And so what they suggested is that you have a kind of morphogen system here where at high doses you get a primary fate and at low doses a secondary fate. And this actually, you do require the notch pathway so this requires basically that this cell activates auto-prime signaling of delta, so the secretion of this diffusible form of delta which has the property to be able to bind to the same cells receptor. So what we did in the model is to see whether we could find solutions corresponding to both and more interestingly what were the properties of these solutions. So we played in the model with these 600,000 solutions we then removed either the lateral signal so you could only have auto-crime activation to the cell or doing the mosaic for EGF receptor. So basically reconstituting in the computer the two experiments of the previous slides and then categorized among the 667,000 solution which one were still working in which conditions. And so we defined four categories depending on whether the sequential induction was required or the morphogenetic induction was required and whether they were sufficient or not. So you have basically cases where each is sufficient so you have redundancy, you have cases where both are required, they really need to act jointly to get the pattern, if you remove one of them it doesn't work anymore but if you have the two together it works and then you have solutions where a single one is required and sufficient. So now first of all you have to look at the numbers. So by far the most common ones are those relying on sequential induction and then you have very few where the morphogen induction compared to 600,000 is required or sufficient. And if you look at the parameter distributions among these solutions you can do statistics. And what you see, so this is one example is the rate of activation here of the expression of the lateral signal by the RAS pathway here. And you're seeing that depending on the type of solution so the red and the green rely on the sequential was here the morphogenetic works by itself and so here it's in the morphogenetic case they tend to be very weak here and on the other hand no sorry this is the diffusible one so it tends to be very strong here so very low coefficients. And so you have a bias in parameter space towards a strong activation of the diffusible ligand here. So conceptually this is a 40 dimensional space but if you realize what we did so we started with nine million random parameter sets found a 660,000 that worked that produced the pattern and within this ones we have tiny much smaller than represented parts of space which rely on the different mechanisms here. So we can also look for these different mechanisms our robust they are two parameter variation and so here are color coded the four classes of requirements exclusively sequential morphogenetic both required and fully redundant and what you're seeing here is the fraction of parameter sets out in these sets which tolerate a tenfold variation in each parameter which is plotted here and so we don't need to enter in detail so basically the morphogenetic the blue or the both required in green have a much lower tolerance to parameter variation so they are much less robust especially to variation in parameters that concern EGF receptor synthesis, half live and this type of parameters which is not surprising because to get this morphogen model going you really need sharp exact concentration so that you don't get excess activation on the sides and not enough in the middle. Whereas the lateral signaling just requires a threshold of activation of the primary fate and then the second threshold of activation of the secondary fate so you have much less precise tuning for one mechanism than the other. The second thing we did was to look whether within again this 670,000 solutions we could find parameter sets that correspond to the behavior we observed in the Synapse IT species so this is a slide just summarizing this different pattern so not only we did this anchor cell ablation we also did EGF overexpression so if you overexpress EGF a lot you get all primary, not very informative but what's informative is what we get in the middle here at low dose of overexpression and so whether we get for example adjacent primary cells here or alternation of patterns and so we categorize solutions here first using the anchor cell ablation and what you're seeing here is the percentage of solution we could accumulate for corresponding to the behavior of each species and again we could then look at the distribution of the parameters to see what was biased in which species and this corresponds basically to our intuition whereby for example if you compare so in C-brainerai where you have a direct induction of the primary you have a strong induction here of the primary by the anchor cell and a strong induction of the lateral signal because you don't get just the primary by itself it's basically something we hardly ever found whereas in C-brainerai this arrow here is very difficult to activate represented by a dotted line whereas this remains very strong so at low dose here we get activation of this lateral pathway and no activation of the primary phase and finally in C-brainerai where we find two patterns what happens is that we have a strong activation of this auto-crine loop here of auto-activation of the secondary phase by the diffusible delta no so the question is whether this can be extrapolated to other cells, no not at all this is a very it's an isolated developmental system that I'm talking about the other cell there are many other type of developmental mechanisms yes so no the question I think you're asking is whether there is some kind of pleiotropy of so if the fact that this arrow is stronger for example in these species and these species I can relate to something else in the one frankly I don't see ways to do this at this evolutionary scale but this has to do with the notion of pleiotropy so the fact that if you do a genetic change it's going to reverberate in many different tissues and so one way that this does not happen is if there is a c-stregulatory element of transcription here that's specific for this cell type and not another so you can perfectly have something which is self tissue specific but the only way I see to do this is to go at much smaller evolutionary scale where you can actually probe this directly here we cannot we so far it's much too far that we can pinpoint to the genetic change that are doing this okay which is actually the good reasons to go at a smaller evolutionary scale which is what we went on doing okay so based on this actually we made some predictions one of them I mean a posteriori you can rationalize it very easily but we did not expect at the beginning is to see how easy it is to get an isolated secondary in the different species and so to do this what we do is we isolate the cell by laser ablation of all the others and we look and so we isolate the PHP the most posterior one and we ask what's the proportion of the animals where it adopts the different faiths so blue, primary and red, secondary and purple being an intermediate between the two often you have one of the daughters adopting one faith and not the other and basically in she elegance we can get an isolated cell to adopt a secondary faith but in Ramanai we cannot which looks odd because we saw that in Ramanai you have this I should have written this before maybe in Ramanai you have this so at first sight you have isolated secondary but they're not really isolated because this cell is actually signaling to them but when we actually isolate just one cell it cannot become a secondary and the same for Cibranari where you have a very strong so in Cibranari you go directly to this to the final faith pattern because you have a very strong induction here laterally we also use cell faith reporters so this is a GFP reporter for the notch pathway for the lateral pathway and you see that in C. elegance it's utterly activated in the two red cells on the side but in C. Briggs it's actually activated at a much more even level in three different cells so this fits also well with what we expected this quantification so in conclusions from this what we showed is that the quantitative variation in the same network encompasses the two mechanisms of self faith patterning in the row of six cells that different labs had been fighting about sequential induction and morphogen induction but it shows that in terms of robustness to parameter variation the first one is much stronger yet it is a fact that DGF diffuses and that's therefore morphogen induction may also participate in a kind of redundant fashion and that we also saw that we can account for the behavior of the different synorability species by just quantitative variation in network parameters so this is a sufficiency argument it doesn't show that it is the case but it's actually possible now we go back to a more evolutionary aspect to try to map this evolution on the phylogeny and actually moving away in the genus and further out of the genus do you have any question? Why not? So what you're seeing here again is only the cell is not the final self faith is always this one what I'm representing here is always this faith of cells after you remove the anchor cell at intermediate time points so what I showed you before is remind for example, brain array, Japanica, elegans is more or less like this and then Briggs and his sister species here with three red cells we went on and went out at species that branch more basally in the genus and they all like this and you will see actually outside of the genus as well so this is where you can use the phylogenetic relationship and try to map the changes that have occurred and so the inference here is that you have a direct induction here of the primary faith so you get right away a strong activation of the primary faith and lateral signaling here and you probably have a reversal in this branch that's the most likely scenario so if you go further out you will see so we did the anchor cell ablation in many other species and genus in different genus and you see that overall you always get this intermediate self faith pattern with three secondary cells I will see later with four whereby so this is equivalent to saying you have a first wave of induction of these three cells and then a second wave which is induction of the central faith so you can describe almost the mechanism as two different waves there is however an exception and this is work from a half summer when he was in Paul Standback's lab it says these species I'm going to tell you where surprisingly if you remove the anchor cell even very early in development you have absolutely nothing different happening so you get a normal self faith pattern again this self faith pattern in these species so here you have a completely different mechanisms whereby the anchor cell is not required so here you're no longer in the in the quantitative changes in parameters that I showed you was sufficient to explain evolution here you are in totally different situation and so what happens in these species is that the vulva doesn't form in the middle of the body below the anchor cell as in the other species but it forms very posteriorly so here that's the rectum of the animal and this occurs because the vulva precursor cells so P4 to P8P these five cells migrate posteriorly and yes at the end the anchor cell will connect with them but at the time where the self faith is pattern and the vulva forms it's actually very far from the anchor cell and so in these species if you remove the gonad very early even before the anchor cell is born you get a perfect normal self faith pattern so here you have to find and to be frank we still don't know exactly how this happens but from experiment from Ralph at the time what he showed was that these cells from the start are not equivalent to each other so they don't have the same potential to become primary secondary and tertiary and the most likely thing is that you have like as in any animals a patterning along the entire posterior axis especially through hoax chains through wind pathway signaling which makes these different cells different and you probably have lateral signaling between them in lateral inhibition for example you have a single primary one so here you have a very different way of making the same thing so the output of this is that there are many basically ways to go to Rome or to Trieste maybe but at the end you get the same self faith pattern but the ways to get there the developmental routes to get there can be very different so we saw the sequential mechanism and the graded mechanisms that's maybe co-occurring in C. elegans we see that in many species we seem to have more like a two step induction where you have first induction of the secondary faiths and then the second induction which induces the primary central faith and then things like which also occur probably at low level in all species but are particularly required in mesoabditis is the fact that you have some kind of self-organizations through reciprocal lateral inhibition ensuring you have a single cell that becomes primary and a pre-pattern so the fact that these cells are not equivalent at the very beginning so the conclusion of this wider evolutionary studies is that the redundancy of the different developmental mechanisms may favor rapid evolution of these mechanisms and I think it can also favor new things and so one example and I don't know here what's new and what's old because we have only two points what I showed you so far was this these are all the new methods the three of old new methods and so far we've stayed at the top here now we're going to consider the next suborder or whatever this next group in green here and there is a huge difference now in self faith pattern between this one in brown and this one is green is that in the group in brown we had this self faith pattern where P6P is primary, five and seven are secondary what we so far saw now in this other group what happens is that the central faiths the central cells ultimately in the vulva are now shared between one daughter of P6P and one of P7P and now you have it's a symmetric still self faith pattern so now you have four cells induced all together okay so here and you may have four and nine being able to replace them but not in all species actually so in this case you cannot rely on the self faith pattern patterning mechanisms where the anchor cell induces one cell that induces its neighbors and lateral inhibition whereby this one also prevents the next one to become primary because you have two adjacent cells daughter cells here which become primary so here again you have a different mechanisms to see elegance but you can imagine that this I showed you that you have so what happens in this species here if we remove the gonad we also have no induction you also have this intermediate state with all secondary and we only have later the primary faith so basically you have a one stage which where the gonad induces four cells instead of three and this we don't know exactly this may be related to positioning or to previous sensitivity of the cells most likely to receiving the signal and then we have a second signal and now the anchor cell instead of going in between the two daughters of P6Ps goes here so the fact that we have this two step mechanism that we saw is dominant basically many species in the phylogeny I showed you before have this induction it allows this displacement of the pattern in other species in another book again I don't know what's the direction of change between the four cells and the three cells induction and centering between the daughters of P6P and in between here the two daughters but it allows changes in mechanisms and in self-feed pattern in this case so to scaritize basically here many developmental routes lead to the same final pattern you have the same pattern X here and you may have changes in the rate of weights and requirements and sufficiency of different developmental routes here in these two arrows and ultimately this may allow for a change in the pattern for example if for example the red route of development doesn't allow you to reach another pattern and the green one so pattern Y and the pattern Z so the possibility of further evolution depends on the developmental routes here and I showed you this two step induction may allow for the situation in the family of Panagralamus where you have four cells and two cells actually sharing the primary fate Any questions before I get to the presumably short collection of final outcome Right, on the route, yes Mm-hmm Is it possible or is that it's going to depend on how the processes will evolve and change over time under the constant pressure? So in population, is there anything that's possible if there isn't anything there in the pathway? Yes, so So thank you that again we will come to is that can we find variation at a shorter evolutionary time scale, right? And so I'll come back to this, the third lecture I think and so one of the quick questions behind this is yes so in evolution, why are these things changing? Is that purely by drift? So because you have mutation and the processes are going to change and as long as it forms the final pattern that's under stabilizing selection, it will work or do you have something driving evolution of development? So I'll come back to this in the third lecture basically I don't have the answer but in one case I have the answer, yes that is the second, it's not neutral it's actually under selection in another tissue it's coming back to the pleiotropy problem that if you have changes in one tissue it reflects in the other questions on the vulva system. Can I give you an answer? Yes So there is a question that you have and it's indicated for another lecture. Yeah, here we'll watch the last minute so if you look at Prandler's face and you see that there's a certain strategy in more compatible with the larger fraction of Prandler's face where his male could be much more robust and so then I was wondering is it possible or common for originality a non-robust pathway or a non-robust strategy for doing something and if they do that then are there more sensitive to evolutionary change? So you're considering the aim of it could be really what the analogy is so can you hint at this one more time to have things that are... Yes, so my answer to this would be that deepened robustness to which perturbation I think so you cannot be infinitely robust to all perturbations and so yes, there are perturbations where the system will be more sensitive and evolution can drive in this direction. Now a lot of us are studying developmental systems which are extremely invariant and not sensitive to the environment but this is not just because we chose to do this because we like to have a good output but some other of my colleagues are very interested in plastic systems where the environment actually changes the output so the environment is acting on developmental processes and actually changing the output and in this case the situation is very different so in some way looking at what you may be looking at in early Josephine embryos or what we are looking at here is a special type of non-environmentally sensitive developmental systems there are also other systems which are extremely sensitive to the environment just the size of the organism for example the growth or really switches type of thing having a wing or not a wing in aphids for example and so here we are in very different situation where we expect to be in for some parameters in parts of the system which are sensitive to perturbations but for some parameters especially I don't think robustness to me you have I mean it's to which parameter you're talking about this specify is not infinitely robust there is always a compromise yeah so what basically I'm assuming this here so the question is probably what are the experimental proofs that they are I think in this group a synod that is species so this has been mostly studied in synod that is Briggs which is as far to see elegance as the other ones are and we are pretty I mean the AGF and the not signaling pathways are there and doing grossly the same thing yes so we think it's the case if we go outside of the synod that is genus that we are also looking at this at the moment we chose one other genus and did actually mutant screens for the vulva in the same way that was done for the elegance and also my colleague have summer yet another genus um what we find is that the species we looked at which is a genus not very far from synod like this we clearly have also a GF not we don't know yet in christian because the claim of how summer's lab is that the GF is not acting I'm not completely sure of of this but yeah and not we don't know either so it's it's very well possible that at a longer evolutionary scale we are talking about very different signaling pathways okay that you really have to know about pathways what does functionality mean sorry so it's based on the fact so the fact that they are part of the vulva or not it's a question of tissue definition I think it's very clear yes it's it's whether they imagine it yes so this is clear whether they become primary and secondary is really whether they form the center or they look different now there are some species where it's much more complicated to decide because each of the daughter and granddaughters there's something different so to assign a fate to the mother sale is slightly different but these are really weird exceptions so I'll continue by giving a very just small vignettes of other examples which I think are still not many where people try to take developmental systems and look at variation in evolution of these systems in the light of a parameter variation and I think so the first example I have is a very short evolutionary scale because it's a variety of chicken and this was he studied in the lab of Dennis Headon in a demo I think and so you have some chicken which don't have any feathers on the neck so they're called naked neck I think and basically what what they showed is that the system behaved as if so you have patterning of the feathers and the position of the feather is the lateral inhibition type of system well so depending on so you get this precursor of the feather it's very much like I don't know whether you heard in Drosophila of the bristle patterning it's a similar in some way and what's special in this is that it's only on the neck that the feathers disappear so there seem to be a local inhibition here and basically I think I summarized this here what they showed is that so we have summarized here the genotype in orange the developmental parameter in green and the final phenotype that's in this case changes in blue and so here the genotype is the naked neck compared to a reference the leghorn variety and the difference here is the genotype at BMP 12 so BMP is one of the another type of signaling pathway GGF beta signaling pathway and so if you look at the expression level of this signaling pathway here it's high in this genotype and low in that one and this corresponds in terms of the final phenotype in a high density here and a low here so here you have an inverse correlation and this is explained the tissue specificity is explained by the fact that locally you have another pathway yet retinoic acid and so this variation which is cryptic in another part of the body becomes non-cryptic here in the neck because in parameter space of a model so this is a kind of reaction diffusion model they did you actually go here in a region of parameter space where you completely remove first you have much less feathers and at the end you completely remove them another example that was also based on kind of reaction diffusion patterns is colors of fish and this is work from the lab of Shigeru Komdo so these are trout different types of species of trout which can have either spots which are clear white spots or they can have dark spots or they can have a kind of labyrinthine patterns and this type of patterns you can get so you can make a model again of reaction diffusion so the exact developmental process behind it is probably not a classical just reaction diffusion model because it takes I mean you have cellular components involved not just molecules but you can make a very simple model and see that if you change parameters you go from so this area of parameter space where you have white spots this one where you have black spots on the white background and the middle one where you have this labyrinthine patterns and what I find very nice in this system is that when you cross trout which have black spots and those which have white spots you get genetically in the middle by crossing them so you get here this labyrinthine patterns in the middle so I think this genetic cross is really telling you that by doing the cross you basically take a little from the two parents which gets in the middle of the parameter space here so if you take one here and one here when you cross them you are in the middle which genetically means something but I'll leave it for now and it also gives way to this model where the black spots and the white spots are obtained through this type of mechanism where you have so that the black spot and white spots are not each determined but that it's a generic pattern mechanism pattern formation mechanism and I think I have a third example which is the work of Isaac Salazar-Shudad and Yuka Yanval who look at the shape of teeth so here you can look both at the shape of teeth within one species so our different teeth may be explained by different parameters of shape formation because as you know we have different shapes of teeth along the draw but you can also look at evolution and they looked in different species especially in some marine mammals sorry the name escapes me and again they made a model of interaction between epithelium and meson chimes through a few signaling pathways and by changing parameters you can mimic the shape of different populations of so foca which is a marine mammal depending on the parameters okay and I will stop here I think if you have other questions