 I'm R.A. Warmflash. My background is actually in theoretical physics, but sometime during my PhD, I got excited about developmental biology and also got excited about the prospect of sort of combining theory and experiment. And so I went to do my postdoc with Ali Brievenlu and Eric Sidge at Rockefeller. Actually, when I got to Ali's lab, it was sort of in the midst of transitioning between working on Xenopus embryos and working on human embryonic stem cells. I actually came there to work on Xenopus embryos, mostly. But while I was there, I got really excited about the idea of using human embryonic stem cells to model embryonic development in vitro. And that's what I ended up doing and what my lab does now. And so during my lectures today and tomorrow, I hope to convince you that sort of modeling development in vitro with embryonic stem cells is sort of an exciting platform for studying development, both because it provides an opportunity to do things that I think really are impossible in any developmental system, but also because it provides a window into human development that we obviously don't have unless we study cells that are actually human cells. And while obviously it's ethically impossible to experiment with human embryos, to the extent that we can take human cells and use the human cells to recreate what's happening in the human embryo and experiment on those sort of embryo-like structures that we can create in vitro, we can learn something about human development. So the picture back here shows the sort of somewhat, well, anyway, shows the correspondence that I'm sort of excited about right on the left is a cartoon of a human embryo. And on the right is the kind of patterning systems we can make in vitro with human embryonic stem cells. And the cells in the cartoon are sort of roughly colored the same way as they are in the actual image of the staining here. And so you can see while it's not a perfect correspondence, we can recapitulate a lot of this sort of patterning in vitro. And I think that's providing us opportunities to study things that would, especially about mammalian development, that would really not be otherwise possible. So what I'm going to talk about today, I'm going to give a brief overview of early mammalian development. I know Cat probably covered some of this material last week, so I won't dwell on it for too long. But I'd like to sort of highlight the things that are important for the systems I'm interested in studying and also highlight some of the things that I think have proved to be really challenging to understand about mammalian development working in in vivo systems. We've learned a tremendous amount, but there are also other things that are just very difficult to do. I'd like to tell you then a little bit about embryonic stem cells, sort of different types of stem cells and the molecular properties. And then get into sort of the main part of what I'd like to talk about, which is the uses of these stem cells for understanding aspects of development that are difficult to otherwise understand. OK, so a brief overview of mammalian development. So as you're all probably aware by now, mammalian development is this process where the fertilized egg travels down the fallopian tube. It's dividing as it goes down the fallopian tube. It eventually implants in the uterus. And at some point after that, in most of the species we study, initiates the process of gastrulation. So you can see these series of cleavages. And this all happens over the process of several days. So compared to things like the worm or the fly that people have been thinking about, this process is very slow. If we were to zoom in on these sort of early cleavages that take place, these cleavages are very slow. Compared to species like the fly or the fish or the frog, Stefano talked about the mid-blastula transition. That doesn't really happen in mammals. And these mammalian systems, the zygotic genome was activated early, which means they start transcribing zygotic genes very early. And that means that the initial cells are basically all the same. They're dividing, they're transcribing, and they rely on the zygotic transcription to make all the patterns that are made. Which in a sense means that all the patterns that are made in mammalian embryos, there's no pre-pattern laid down by a mom that has to be obeyed later on in development. The initial cells are essentially identical, and they divide and divide as identical cells until they form this clump of identical cells. And then all the cell fate decisions and symmetries that have to be broken have to happen in this kind of cell-autonomous fashion after that. And so I think it's an exciting system for studying how these sort of self-organized patterns are formed during development in the sense that they're everything that happens is self-organized. And this is sort of what gives us the potential to recreate this thing from stem cells, because you can imagine if every embryo had to have the mother make a particular asymmetry, right, it'd be difficult to imagine a cell culture system where that asymmetry always exists, or the experimenter would somehow have to create this asymmetry, but to the extent that these things are self-organized, that we can imagine that they can self-organize themselves in culture just like they do in the uterus. And so that gives us the opportunity to study these processes in vitro. And so these cleavages happen, and then at some point there are a series of early cell-fade decisions, and these happen really before the processes that I'm interested in, and I'm gonna be telling you about happen, right? So the first thing that happens is you have these cells that are all, I can't do this, that are all identical. Around the 16 cell stage, they divide into these two lineages. One is the trophectoderm, which will go on to form the placenta, and the other is the inner cell mass that will go on to form the entire embryo proper. And then the tissue inside this ICM makes a second decision, where it makes the epiblast, which again goes on to form the embryo proper, and the primitive endoderm, which goes on to form other extra embryonic tissues, although people have shown that it also contributes to the embryo proper as well. And so the stem cells that we're gonna be thinking about is I'll get into either come from this ICM population or come from this epiblast population. I mean, that's important for understanding what they really should be doing sort of in vivo and what we expect them to do in vitro. So this is just a sort of summary of the hierarchy of lineage decisions, right, so you have the blast assist, which divides between the trophoblasts and the inner cell mass, as I just said. The trophoblasts has its own differentiation program, which I won't go into. And then the inner cell mass divides between the hypoblasts and the epiblast. The epiblast is essentially what gives rise to the embryo proper, right? So, and in particular interest to me is the epiblast eventually gives rise to all three embryonic germ layers, so the ectoderm, and then cells which go through the structure called the primitive streak, which emerges during gastrulation, which I'll get into in a minute, and those cells become the mesoderm and the endoderm of the embryo, right? So roughly speaking, you have this series of two cell fate decisions that gives rise to the epiblast, and the epiblast goes on to give you the entire embryo proper, and the first step in giving rise to that embryo proper is to form these three germ lineages, the ectoderm, mesoderm, and endoderm. So I like to show this movie, this is not a mammalian embryo, this is a Xenopus embryo, but it's really, I think, still not possible to make these kind of movies in mammalian embryos. And this is a movie of the process of gastrulation happening. So this dark spot that you see here is the dorsal side of the blastopore lip of the Xenopus embryo, so the blastopore lip is where the invagination's gonna happen, so you're gonna see basically all the tissue that belongs on the inside of the embryo go onto the inside of the embryo, and this movie actually continues through neuralation, so you'll see the neural tube form and close along the back of the embryo. And so here you see this gastrulation process happening, all the mesoderm and endoderm is invaginating, and the ectoderm is coming down to cover the surface of the embryo, and then this is the neural tube forming along the back of the embryo, and the head forms here, and the tail forms here, and you can start to see the eyes starting to form here. So I think like Stefano said about the fly movies, this is the kind of movie that I sort of never get tired of watching. And so the sort of basic processes which guide gastrulation and sort of the rough idea of the cell movements is similar between the Sanphibian and mammals, but the geometry that it takes place in is actually quite different in mammals. So here's a picture of the mouse embryo. The human embryo is again, somewhat different from this, but we're getting closer. And so what happens in this embryo is that you have this extra embryonic endoderm surrounding the embryo. On the inside of that, you'll have the epiblast, which at this stage forms an epithelial sheet that covers the extra embryonic endoderm. The primitive streak, which is the structure I mentioned before, is the site of gastrulation. So it's sort of equivalent to this lip where you saw these Xenopus tissues invaginating, but instead of making the circle around the embryo, it makes this streak down the side. And at this site, you get differentiation into the mesoderm and endoderm as they move through this region of the primitive streak. And then if you were to imagine peeling away this extra embryonic layer, so you could see what's happening, these will migrate around the embryo this way. And so you end up with this kind of trilaminar structure where the epiblast stays in the middle and the cells that stay in the epiblast and don't go through this process of gastrulation differentiate to the ectoderm, and then you get this mesoderm and endodermal germ layers wrapped around that. So that's sort of the process of the sort of mechanical process of gastrulation which takes place. So now I'd like to tell you sort of a story about the sort of signaling and how that instructs this gastrulation process and what we think we don't know about it. Before I do that, just to make sure we're all sort of on the same page, I wanted to sort of outline the hierarchy of regulation. So, much like the signaling pathways we heard about in the last lecture, but different from things like bicoid which just float around in this syncytium, all of these signaling pathways are thought of as morphogen signaling pathways where there's some extracellular ligand. The extracellular ligand is produced by the cells of the embryo themselves, possibly with particular regional specificities, so possibly in only particular regions or particular tissues. When that diffuses around and binds to the surfaces of cells, it engages the receptors. The receptors activate signaling effectors. The signaling effectors typically move to the nucleus and turn on transcription, right? So, things like notch delta or ERC or the pathways that I'm gonna be talking about like nodal and BMP and WENT all follow this sort of same basic structure, although the details obviously are different, right? There could be multiple signaling effectors or a cascade of signaling effectors. Sometimes this involves a positive regulation, sometimes this in the case of WENT, it involves a double negative regulation so it'll turn off an inhibitor of the pathway and then that will allow something to translocate to the nucleus, but roughly the structure of all these pathways is the same or sort of the same general idea. Any questions? I'm in your way. I don't know where I should, should I go here? Is that better? Okay, I can do that. Okay, so how does gastrorelation work? So, in the mouse embryo, the sort of pictures I showed you before, this cup shape is here and this zoomed in, this is the epiblast cells here and this is the visceral endodermia. So, this is extra embryonic tissue which is not going to give rise for the most part to the embryo proper but it's thought to be involved in patterning of the embryo and that nodal signaling through mechanisms which I think are not entirely clear, somehow specifically induces inhibitors of nodal signaling in the population of cells at the bottom of the embryo which is called the distal visceral endoderm for the reason that this axis is called the proximal distal axis and proximal means close to the placenta, right? So, the placenta is up here, this is proximal to it, this is distal. Somehow, nodal signaling specifically induces these inhibitors of nodal at the bottom and those serve to feedback and inhibit nodal signaling at the distal tip of the embryo. Then, migration event happens and this is a very important event for the symmetry breaking of the embryo where it's a prior you only have this proximal distal axis and the symmetry event, this migration event induces a symmetry breaking which actually makes the anterior posterior axis of the embryo and so these cells which express these nodal inhibitors migrate to what is now going to be the anterior side of the embryo and so the region of low nodal is defined as the anterior side of the embryo and the other side of the embryo will be posterior, right? So, the presence of the inhibitors on the anterior side of the embryo it causes restriction of nodal signaling to the posterior side, right? So, you can see that here, right? These cells, they move over, the serverus and lefty are nodal inhibitors so they're expressing it, they're inhibiting nodal which was initially expressed all throughout here, gets restricted to the posterior side and more proximal so up in this corner of the embryo, okay? And then once that occurs, people have shown that there's a positive feedback loop which involves three pathways so nodal is this pathway that we've been talking about, a second pathway which is very related to nodal is the BMP4 pathway but it actually shares a number of receptors and signal transducers with nodal although its target genes are different and some of its signaling effectors are different and BMP4 is activated actually in the extra embryonic cells up here and that turns on Wnt signaling down here and Wnt turns on nodal, right? So, what you really have to realize is that you've sort of generated this asymmetry, this asymmetry has restricted some of your signals to a particular side of the embryo and then there's a positive feedback loop between these three signaling pathways up on the posterior side of the embryo and proximally that sort of create this self-sustaining loop and initiate the side of gastrulation at the posterior side of the embryo so this is actually where the primitive streak is gonna form, so in the primitive streak I have high nodal, high Wnt, well high BMP possibly coming from this extra embryonic tissue. Okay, so that's sort of, I think a sketch of what sort of thought about how gastrulation works in mammalian embryos but there's sort of a lot of incomplete aspects of these theories, so what are the issues with these kind of models? So one is the state of the data, right? I've told you all these things like, okay, I know where nodal is, I know where the nodal signaling is but actually we don't know really well where the nodal is and where the nodal signaling is, right, so most of our evidence comes from experiments that look like this so this is already a pretty old example but they've engineered a regulatory element of nodal to drive laxie and see where it's expressed and then you can do the staining for laxie, right? And you can sort of see that the nodal is restricted proximal posteriorly but it's pretty fuzzy and it'd be pretty hard to be quantitative at all about this stain, right? So here's some idea of where nodal is but do we really have confidence in sort of being quantitative about this gradient of nodal, oh, I don't know. And then, so, and this is about the state of our data for where nodal is, we have no data about nodal protein, right, so these are basically marking the cells that are making nodal, right, because this is where the mRNA for nodal is but we don't know where nodal protein is because we can't visualize the nodal protein and then we actually don't know where these signals are received because aside from a couple of reports which are sort of hotly debated, it's proven very difficult to report on the sort of signaling effectors, right? So I told you the signaling effectors go to the nucleus, in theory I could do a stain and watch that the signaling effectors are in the nucleus but that's actually proven to be quite difficult in whole embryos and for the most part we don't have any data on that either, right? So we have this, in theory, there should be a cascade initiated by the production of mRNA to the making of protein, to the secretion of that protein, to the diffusion of that protein, to the reception of those signals by other cells and that signal transduction but we actually have no data on any of that besides for this. And then so that leads to the point that it doesn't mean we don't know anything else, it means that most of our evidence is indirect and comes from doing genetic manipulations on these systems, right? And so I can't measure, say, the levels of these signal transducers so here are two particular signal transducers for nodal, SMAD-2 and SMAD-3, they function somewhat redundantly so they can both, when nodal activates its receptors, both SMAD-2 and SMAD-3 will go to the cell nucleus. There might be some tissue-specific expression so SMAD-2 might transduce these signals in some tissues in SMAD-3 and others but for the most part they're redundant and what people have done is made a series of genetic inactivations of these things where as you can see, okay, if I have both SMAD-2 and SMAD-3 active I get a certain amount of hypothesized nodal activity so the y-axis of this is not actually anything that's measured either, right? What we really know is there's a series of cell fates here and I lose these cell fates in a progressive way as I make these mutations, right? So I make mutations, I think these mutations are inactivating in these nodal signal transducers and so I hypothesize that I can order these mutations by some aggregate level of nodal and then I can rank these tissues for how much nodal they require, right? So, and that's sort of summarized in this paper here, right? So if I have both, I have hypothetically this amount of nodal signaling and I get all of these things, if I inactivate one copy of SMAD-3, well, presumably I have less nodal signaling but I have no way to measure that but my pattern is still normal, right? And then at some point as I inactivate either both copies of, actually both copies of SMAD-3 still gives a viable mice but then if in that background I start to inactivate SMAD-2 or I inactivate SMAD-2 altogether, I actually start to lose some of these tissues which are the mesodermal and endodermal tissues, right? So what people have done is basically put together a story from this kind of expression data and this kind of sort of summary of genetic data to try to explain how these signaling pathways are producing this sort of sequence of events in the embryo. So that's sort of the state of our data. And then one can ask questions about do we really understand, does the story that we tell about nodal, does it really make sense, right? So are the expression patterns consistent with its function, right? So if you remember what I told you, the earliest event, the earliest sort of function of nodal here is to signal to these cells at the distal tip to induce this asymmetry that will make them, these cells, express nodal inhibitors and then these cells will migrate. But how is it that nodal specifically induces these cells at the tip where we don't see any great, in order to do that, the gradient should be highest at the tip and lowest away from the tip but actually all the observed gradients either go the other way or homogeneous throughout the epiblast, right? So what we see, if you look early enough is that nodal sort of homogeneous through here and then if you look later, nodal is up here. So the idea that nodal is somehow specific down here is only from this function of inducing the DVE down there but we don't really understand why nodal would specifically induce the DVE down there. And nodal is expressed in the epiblast throughout gastrulation, right? And like I said, we don't have any information on the nodal protein or the nodal activity so does where nodal RNA mean that you have nodal protein there mean that you have nodal activity there? We don't really know. And we know just from looking at the nodal gradient and looking at how much these cells move that the nodal gradient that any individual cell experiences must be very dynamic, right? It's not really possible to think of a single cell sitting there experiencing one particular level of nodal. The nodal levels are changing, the cells are moving and so we need to know something about how cells interpret this changing signal. And then finally, if we care about human development, right? Human is obviously not equal to mouse, right? So in mouse I have this cup-shaped embryo which I think is actually somewhat particular to rodents and when we think about the human embryo, the human embryo is actually much more similar in structure to the chick embryo, some of you are familiar with that. So you get this sort of flat disc here with the epiblast sitting on top of the hypoblast and instead of cells, instead of one side of this creating the primitive streak and cells moving around this cup, you get a line down the center of the disc and then cells move around the disc. So it is sort of like this kind of flattened out but we don't know whether this sort of, sort of rather dramatic difference in geometry actually makes a difference in terms of how these morphogens are passed around and how things happen during development. Okay, so that's what I wanted to say about early mammalian development, any questions on that? So I'd like to argue that there's a lot of things we can do in cell culture that are difficult to do in the embryo. So I mean, first in order to do this at all, right? We need to be able to make cell fate patterns, right? If we want to study how cell fate patterns are generated we need to be able to make them. And as I'll show you later on in the talk we can make very reproducible patterns of cell fate which resemble those in the embryo even if they're not exactly the same as those in the embryo. We can study dynamics much better. So I told you that we have almost no data on how nodal signals are received in the embryo but in cell culture we can do things like this. So these are cells that express our reporter for the nodal signaling pathway and the way this reporter works is that we've used CRISPR technology to tag one of these signal transducers with GFP. And so the signal transducer should move to the nucleus when these cells receive a signal. And so these holes that you see here are the nuclei of the cells. And if I play this movie, right? You see these nuclei fill in over time as the cells are supplied with signal and then they actually adapt to the signal and go back. And that's something I'll talk about more tomorrow. So we can actually study the dynamics and the input-output relationships of these morphogen signaling pathways in these embryonic stem cells because the imaging conditions are good and now it's quite easy to make genetic manipulations in these cells. We can play with the geometries in which we grow these cells. So I've showed you circular patterns but we can make any kinds of patterns we want, right? So we can ask questions like, okay, what happens if I make an embryo-like structure that's half the size? How does that perturb the pattern that I get? It's quite difficult to make a half-sized mouse embryo but it's quite easy to make a half-sized stem cell colony and ask things about the scaling of the pattern. And then we can modulate the dynamics very finely, right? So people have gotten quite good at genetic manipulations. I can do things like inactivate particular genes and particular tissues but if I really want to say, okay, I want to inactivate this pathway for an hour and then turn it back on, that's quite difficult. But in cell culture, right, there's all kinds of microfluidics that people have noted. So this is an example of a microfluidic device where there's a bunch of culture chambers pipetting different food dyes into different culture chambers and so you can imagine that you can very precisely modulate the dynamics with which you expose stem cells to ligands and sort of probe these input-output relationships. And so I think these sort of capabilities of working with stem cells open up the possibilities of doing lots of things with stem cells that are difficult otherwise. So I'd just like to point out that so the idea of taking embryonic cells and studying development in vitro is not a totally new idea. Of course, people have been doing this for a long time and in particular, people working with Xenopus embryos a long time ago, one of the most attractive features of the Xenopus embryo was that its tissues are particularly hardy. So you can actually take them, chop them, paste them onto other embryos or study the cells of the embryos in vitro. And so that underlies sort of the famous Spamon experiments if you're familiar with that. Someone more recently, right, this was one of the first demonstrations of morphogen effects really in any system which is the idea that the same molecule can give rise to different cell fades depending on its concentration. And so what these people did is they took cells from the animal cap of the Xenopus embryo which is sort of like the stem cells of the Xenopus embryo and that they're capable of turning into any of the cell fades of the embryo. They exposed them to different doses of activin and they saw what they got. So these are markers of different cell fades and the band means that gene is there roughly so there's not really quantitative technique but tells you about presence or absence. So if you don't have any activin, these cells turn into the epidermis of the embryo so they express epidermal keratin. And as you increase activin, what these fades are is not particularly important but what's happening is these cells are becoming mesoderm and then they're becoming more and more dorsal within the mesoderm. So bouscoid is a marker of dorsal mesoderm. And so it was difficult to prove in the embryo that activin could serve as a morphogen but by taking the cells out of the embryo and then treating them with different doses in a conserved way, right, it was able, these people were able to show this morphogen effect sort of for the first time. And you can not only do things like that but people have shown that morphogenetic movements can be recapitulated in vitro in the same kind of system. So if I take these animal cap cells from Xenopus and just cut them out and let them form into a ball, they'll form these nice round balls. If I treat them with activin that induces this mesoderm differentiation and along with that is this process of axial elongation that goes together with gastrulation. And so these will actually make an elongated axis in vitro. And so people have been doing things like this for a long time and now we're sort of starting to look at doing these things with amelian systems and doing them quantitatively. So before I tell you about that in more detail, I wanna tell you about a little bit about the background on stem cells. So I guess you sort of surmise. Stem cells come from early embryos, right? So we have this series of decisions in the early embryo where we have identical cells and then the inner cells versus the outer cells and then the inner cells make this division. And typically how you get stem cells is people will take these embryos, they'll take cells from this inner cell mass and they'll put them in culture dishes and start them growing. And with some success rate, these things will grow and they'll grow essentially forever if they're grown under the right conditions and you can maintain these cell lines. And it's sort of interesting to note and people don't really understand why, but that if I do this in a mouse and I do this in a human, I get a different outcome, right? So if I take mouse stem cells and I grow them in a dish, they basically retain the properties of this inner cell mass, right? If I take human stem cells, even if I take them from the inner cell mass and grow them in a dish, they look like epiblast, right? So human stem cells, when you see papers about mouse stem cells or human stem cells, the human stem cells typically represent a later stage of development than the mouse stem cells. Although people have now understood enough about the signaling requirements, underlying these things that they can force the stem cells back and forth, sort of it will. Okay, so we get these stem cells from early embryos. How do we know that these are stem cells? So people have devised a number of techniques to check if these things are stem cells and what do I mean by stem cell? I mean something that can grow forever in the dish. And that's easy to tell, where you just keep passing it and it keeps growing. But I also mean something that's capable of giving rise to all the tissues inside the body. So it maintains the property of pluripotency that it's able to give rise to every cell type in the body. And so people have a number of assays to check for that. So one is just you take the cells, you let them clump up. When they clump up into these things called embryoid bodies, they'll differentiate. And when they differentiate, you see all the different derivatives they make and they'll make derivatives from all the different dermal layers, right? So here you just see stains for things that are ectoderm, mesoderm, and endoderm. So inside these embryoid bodies, you have this mass of different tissues, which represent all the different tissues of the body. People will essentially do the same thing in vivo, which is a little gruesome, but if you take a mouse and you inject these stem cells in it, they basically get a tumor made of these stem cells. And these tumors will also have derivatives of all these different dermal layers inside the tumor. Sort of amazingly and most rigorously, there are mechanisms of basically taking mouse embryos and manipulating them such that they're tetraploid, so that they have four copies of the genome, so that they can't then go on to give rise to normal development. And if I take these tetraploid embryos and I was called complement them with the stem cells, so I inject these stem cells into the tetraploid embryos, then the entire embryo has to come from the stem cells that I injected. And so people have shown that if I take embryonic stem cells and I do these manipulations, I can get a whole mouse out of this thing. So that's a very stringent test of pluripotency because I've actually grown an entire organism from these stem cells. Okay, so just a little bit about what these stem cells are like. So the original, the first mouse stem cells were sort of isolated in 1981 in this pair of papers. They kind of look like this when they grow and people were, if you look around them, you can see they're growing on this other layer of cells for the sort of classic way of growing these cells is you put down this layer of feeder cells with mouse embryonic fibroblasts and then you grow these cells on top and somehow this layer of feeder cells was secreting signals to keep these cells happy and they grow in kind of these balled up dome shapes which makes sense when you think about those pictures of the ICM, the inner cell mass and the embryo, they're also not flat epithelial layers, they're sort of these balled up of shells and these colonies sort of mimic that. It took about 17 years later before people were able to do this from human cells but they've done it from human cells and this is a picture of a human stem cell colony, you can see it looks quite different, right? They grow as these flat epithelial layers which are reminiscent of the epiblast growing on top of the hyperblast in the embryo. And even later than that, right, people realize that, okay, if I'm careful about how I do it and I take the mouse stem cells somewhat later, I can actually derive these epiblast-like cells from mouse, right? So you can see if I derive them earlier, I get these sort of rounded up morphologies. If I derive them later, I get these flatter, more spread morphologies but they both express markers of stem cells. So there are markers that will differentiate between these types of stem cells but I'm not gonna get into it. And then even more recently, so only in the last couple of years, people have figured out ways to take human stem cells and revert them to this ICM-like state which looks like mouse stem cells. Well, this is the recipe for doing that so it's quite complicated and why mouse stem cells will just sort of naturally grow in this state whereas when you take human cells, you really have to hit them hard to make them do that. I think people don't really understand. Okay, I just wanted to briefly mention, maybe Kat covered this, I'm not sure, but I shouldn't talk about stem cells without mentioning that stem cells have sort of already precipitated a revolution in studying mammalian development because what people quickly realize after you can grow stem cells, people realize, well, if I can grow these stem cells in culture, then I can modify the DNA of these stem cells in culture and then this was really what allowed people to do genetics on mouse embryos, right? So within a few years after people were growing stem cells, people had shown that I can do homologous recombination in these stem cell cultures. So right, so here's just an example of correcting a mutation in some gene and they're able to select for the cells that correct this mutation with the meal bison gene. And this paper basically sort of laid out the program now that we're able to do this, we're gonna be able to make embryos from the cells and if we can make embryos from the cells, basically we can make any mutation in mouse we like, essentially, and we can study mouse development, we can study mouse sort of equivalents of human diseases and so a lot of this program has sort of been realized, maybe I'll skip this. And so that led to the sort of development of the first transgenic mice only a couple of years later which of course was awarded the Nobel Prize recently. So stem cells have sort of already allowed us to manipulate genes in vitro and then stick those cells back in vivo and grow entire mice and then that's sort of the whole basis for the idea of knockout mice and then they're underlying mammalian development. Okay, so this stem cells, when they grow, they have particular requirements to grow and it's interesting to note that sort of this fate of stemness is not sort of the natural or default fate of cells and what people realized a long time ago by studying other developmental systems is that the normal fate of cells or sort of the default fate of cells is to become neurons. So if I take cells, these are for example, this is earlier experiments done in Xenophus embryos, I take these stem-like cells from Xenophus embryos, I just take them and dissociate them and set them outside the dish, they'll just turn into neurons and this shows that there's a particular signal as BMP4 signal, so here's the neural marker as I apply more and more BMP4, I lose this neural marker and then I gain this epidermal marker. So here, the default is neural and then diverting signals can either divert myself to epidermal or divert myself to other germ layers like mesoderm. Nothing, yeah, so actually it's a, yeah, so what, yeah, so Ali is fond of saying that the culture media for Xenophus embryos is pond water. So when you normally grow embryos, you grow them in just a buffer with a little bit of salt and things like that. When you dissociate cells, so you actually will lower the salt concentration and make them fall apart and then you can put them back into higher salt or not, but they'll grow, I mean, they're not gonna grow forever, right, but they'll grow for several cell divisions, execute somewhat normal developmental programs. But you're readily taking, what people do is they'll dissect the tissue from the animal cap of the embryo, which is the future exoderm, but it's capable of giving rise to all the germ layers. They'll put them in this low salt media, which will dissociate them. They'll move them into a culture media in which they'll be happier and they'll just leave them there, right? And things happen very fast in Xenophus, like I guess they do in the fly, right? So within a few hours, you'll see neural markers in these cells if you dissociate them. But if you add these ligands, then you won't see neural markers. And actually if you take them to dissociate, you do this, but then you re-aggregate them and grow them as a ball, you also won't see these neural markers. So they'll create signals between themselves that divert them away from the default fade. But isolated cells will adopt this default fade. Okay, so stem cells basically enable us to do the exact same experiment in humans. It's done a little bit differently in the sense that we actively inhibit these signaling pathways like BMP and nodal. And if you inhibit those two pathways, what you see over time is Ox4 is a stem cell marker. Pac6 is a neuronal marker. And rather rapidly you lose the Ox4 and you gain almost pure cultures of these Pac6 expressing cells. So if I want to make neurons from stem cells, this all I have to do basically is to remove their signaling. Which tells me that I need some particular signals to maintain cells in the stem cell state and people have figured out what most of them are. So quite some time ago already, Austin Smith, whose name you'll see in a lot of publications, published this paper where they found the cytokine called Lyft is capable of growing, is capable of maintaining mouse stem cells. And so you remember before I told you that you had these cells growing on this feeder layer. So what they realized was if I wanna remove the feeder layer, what I really need to grow them on is the cytokine called Lyft. So one of the things that they were supplying was the cytokine called Lyft. But they're still grown. These cells are grown in serum. So presumably there was another signal needed. And people realize that the second signal for cells to grow can be anyway, BMP4. And so it was a little bit complicated, but what it shows is that if I add Lyft and I add BMP together, right, I maintain these oct4, which is a marker of pluripotency. If I have only BMP or only Lyft, I don't maintain this oct4 very well. And so there's sort of two signals necessary. More recently, people have claimed that you can get stem cells just by inhibiting pathways. So they call this the ground state of embryonic cell renewal. I'm not supposed to be sure if that's supposed to be some quantum mechanical thing or not. And so they're inhibiting basically roughly two things, which are the MEK arc pathway, which signals through FGF and also GSK3 beta. But whether this is really just inhibiting things is sort of up for debate because what happens when you inhibit GSK3 beta is that it elevates signaling through the wind pathway mostly. And so other people have shown that basically stem cells require wind protein. So it's a complicated graph, but the point of it is basically is whether I have this GSK3 beta inhibitor or this wind protein, those are the sort of pairs of two lines that lie on top of each other in terms of how well these stem cells grow. So I can replace this, I can replace this inhibitor of one pathway with an activator of another pathway. So it's not really clear that any stem cells can grow in a state where you're just inhibiting pathways. And this sort of brings up another point, which I think is very important to realize when you're thinking about growing stem cells in culture is that sort of this pair of studies identified wind signals as a really important pathway for growing these stem cells in vitro. But more recently, people have made genetic manipulations in the mouse embryo and shown that wind pathways are not important at all for these early stages of mammalian development. So this is a mutation in a gene that is basically essential for processing all wind ligands. So these porcin knockouts basically can no longer produce and secrete wind ligands. And the people think that basically there's no wind signaling left in this embryo. And consistent with that, these embryos have severe defects at the time of gastrolation, right? I've told you that wind signaling is very important for gastrolation. But when we look early on in development, when they looked early on in development and you look at these sort of marked the three cell fades, right? The blue is this tropectoderm and the primitive endoderm and the inner cell mass, there's essentially no difference between these embryos. They all go on to develop normally through this gastrolation stage. And so there's always some caution required when thinking about cells in culture where you can see these things like, okay, these cells really want wind signaling, they need wind signaling to grow and then you look in the embryo and that's not really true. Or it's true for the stem cells, but not true in vivo. And this shows that it's just a completely different set of pathways that maintain the human stem cells, right? So whereas the mouse stem cells are dependent on Lyft and VMP or Wnt, the human stem cells are dependent on nodal and FGF. And it's interesting that both of these pathways are important in gastrolation, right? So I have these pathways nodal, which I need to initiate gastrolation, FGF, which is essential for the cell movements during gastrolation. And if I inhibit either one of them, so these show the inhibition of them, I lose this OCT4 differentiation marker. Okay, so I'd like to say just a little bit about the sort of transcriptional networks that maintain pluripotency in these cells and how they do it and how people discovered them. And so there's a series of transcription factors that are important in stem cells, one of which is this OCT4 transcription factor. And it's interesting to know how it was discovered. So the people knew that OCT4 was expressed in the early embryo, so Austin Smith's lab made this knockout of OCT4. And what they found was when they did these crosses and they grew out the mice, they couldn't find any homozygous knockout for OCT4 among the progeny, so that indicates that they're probably embryonic lethal. And they went quite early, right? So even looking at gastrolation stages, you couldn't find any homozygous knockout so they're embryonic lethal very early. So they finally take these very early embryos and I think E3 and a half or something and start sticking them in culture and looking at the outgrowth from these embryos. What you see here is normal embryos, they have two things in the outgrowth, right? This is the inner cell mass growing out, these clumps of cells and these flat cells are the trophactoderm surrounding them. And if you look in the minus case, it's all trophactoderm, they're just missing the inner cell mass. So they discovered this super important factor that actually everyone uses in the stem cell field by looking in embryos and doing this sort of very laborious process of going further and further back into development until they could actually recover the embryos and see that these things were missing. A few years later, they cloned another gene which is essential for pluripotency and they did this in a totally different way. So instead of doing this, they took stem cells, they put them on conditions where they would differentiate and they over expressed a library of cDNAs looking for things that would cause the stem cells to remain pluripotent, right? And what they found was there are particular cDNAs that they could find that even under differentiation conditions would give you these stem cell colonies. And it turned out they encoded this gene called Nanog and then when they went back to the embryo and looked Nanog is expressed precisely in this epiblast tissue, right? So this is actually, I think, turned out to be a much easier way to discover things, right? Because you can take stem cells, you can do this unbiased gene screen for over expressing things and then ask which of these, and then once you've discovered these things that when overexpressed maintain pluripotency, ask if they're actually involved in embryogenesis. And so people have gone on to identify sort of a core circuitry in stem cells which involves these two genes in another one and these genes, they all turn on themselves and they all turn on each other and then they all turn on various targets. And so the story sort of emerging from this was that there's sort of this self-reinforcing network of pluripotency where you have a series of transcription factors and you have all these activating connections, right? Everything is activating everything else. And that's kind of, well, in a sense, it makes sense, right? Because you want stem cells to be able to grow as stem cells but it's also kind of troubling because you think, okay, how is this state ever dismantled when you need to go on and differentiate, right? Somehow I need to turn these things off and be able to turn on differentiation genes. And people have more recently realized that the situation is much more subtle than that, right? So people have realized that it's not correct to call all these genes like Ox4 and Nanog and Sox2 strictly pluripotency genes that they're involved in differentiation as well. And so this is sort of a complicated network but the point is that particular genes promote particular lineages. So if I look, right, Sox2 is promoting ectoderm and suppressing alternatives. Ox4 is promoting mesoderm and suppressing alternatives and Nanog is promoting endoderm and suppressing alternatives. And so even though these genes all reinforce each other they're actually all pushing on particular differentiation genes and suppressing other ones which leads to this sort of picture that there's a balance of sort of suppression of different lineages which is important for pluripotency. One of the most interesting results I've seen relatively recently is that where people have shown that if you want to reverse cells to the stem cell state you can not only do it with these stem cell genes but you can do it with taking genes from different lineages and overexpressing them together. If you overexpress mesoderm and exoderm genes together somehow the cell interprets that as reverting to pluripotency. It's the state where you have balance sort of repression and activation of these different states. And that's reflected people have shown in the dynamics of these genes upon exiting pluripotency rate. So when I leave the pluripotent state I have to turn off at least some of these pluripotent genes. And when people look for example this Ox4 gene if I do neuro exoderm differentiation so this is following this in single cells Ox4 goes down but if I do mesoderm differentiation Ox4 goes up. And so there's this sort of connection between pluripotency and differentiation where these genes where how I exit the pluripotent state is very important for how I end up in the for where I end up and what differentiated fates. And right side would be remiss if I didn't mention right of course that we now know that right differentiation is not a sort of one way street right that you know the Nobel Prize was awarded it's four years ago now to Shinri Yamanaka and John Gerdin and what Yamanaka discovered was through a similar sort of screen that I told you about for finding Nanog he was looking for factors that he can throw into regular cells and have them turn into stem cells. Right and he started out with a lot of these things and it sort of worked and then he narrowed it down to particular factors that you need. He ended up with four genes you know some of these critical pluripotency things I've been telling you about like Oc4 and Soc2 so I can take these and throw them into any cell and revert that cell to a stem cell and so that's shown sort of the plasticity of these cells and he was awarded the Nobel Prize for this along with John Gerdin who actually showed about 50 years earlier that if I take a nucleus from a differentiated cell and put it into an O site that O site will somehow magically reprogram that nucleus into something that can give rise to a whole organism so it's a way of cloning organisms so it took 50 years for Gerdin to get it but only sort of six years after Yamanaka did this and people have now gone on to show that actually you can do the same thing Gerdin did 50 years ago with human cells so I can take a human nucleus put it inside a human O site and it will reprogram the human O site into something pluripotent which is pretty amazing. Okay this is two. Yeah. So it actually was not in, it's not in his original screen I think that people have shown that it will enhance reprogramming if you include it but it's not essential for reasons and I think people have also shown that it's actually it's not totally essential for the pluripotent state itself and so if I knock out Nanog I can actually grow those cells as pluripotent although they'll be less robust. So but why that is in terms of the molecular circuitry I don't think it's very well understood. Other question. Okay so now I wanna sort of get into understanding development with stem cells with a couple of examples before I go on to tell you about my lab's own micro patterning systems and so one case that has been sort of studied and provides another sort of cautionary tale about doing these experiments in vitro is people wanted to understand how you get this cavity forming in the embryo where the embryo starts with the sort of more solid thing and it cavitates and actually if you take an embryo body so you take this lump of stem cells and you allow it to just grow it does essentially the same thing, right? So this is sectioned through an embryo body in time and you see this cavity forming in the middle and people said oh this must be exactly like what's happening in the embryo and what they realized was that the cells in the middle are undergoing apoptosis and only the cells sort of in contact with these outer cells are rescued by apoptosis from some signal from these outer cells and so this sort of seemed a compelling mechanism for making these cavities and then it wasn't until much later that people realized that actually this is not what's happening at all if you take much smaller aggregates of cells and grow them like a lot of epithelial cells will they'll actually normally form an epithelium with a lumen in the middle and they'll grow and that lumen will get bigger and the sort of mistake of this earlier study was to take these really large embryoid bodies that have more than a thousand cells and try to understand the development of the embryo which at this stage has a hundred or less cells so you really have to think about whether your sort of in vitro system is the equivalent of your in vivo system or else you can get mechanisms that are sort of not correct. Okay so I wanna get into now can we get stem cells to form patterns and can we use these patterns to study embryonic development in vitro and I think that this is the first paper that I know of that really looked at this question and this lab is a lab that focuses on wind signaling and what they noticed was if I just take these embryoid bodies and grow them and I look at a winch reporter so these are normal cells growing with a winch reporter and this is a laxie stain so it turns blue where there's wind at three days I have this is pretty maybe it's hard to see from where you are but I have these little spots of wind signaling typically I have one within each embryoid body and what that told them was that as I grow these embryoid bodies they're sort of self polarizing in a way that I have wind signaling on one side and not on the other side and it's a sort of reminiscent of the primitive streak in the embryo where I'll have high wind and nodal on one side and nothing on the other side right eventually this wind signaling sort of just takes over and goes to the whole thing and then if I look at cell fate patterns and things like that this is sort of fuzzy but what I'll see is right so the wind is in red and the mesoderm marker is in green I sort of see these polarized aggregates that make wind on one side and typically the mesoderm marker is correlated with the wind signal so somehow what they've shown is that these cells will self organize into these sort of patterns where wind signaling is high and mesoderm differentiation is high on one side and interestingly so remember I mentioned there's this positive feedback loop that involves three pathways wind, BMP and active in or nodal active in sort of the equivalent of nodal that many people use in experiments because it's potent and cheap and what they found was that indeed I can add any one of these three proteins and get this polarization and this is consistent with this idea of a feedback loop if I just have a feedback loop that I need to kick on then I should be able to take any of the proteins in that feedback loop and kick it on and get these polarizations where my reporter is high on one side of the aggregate but not on the other side and this is always completely suppressed by this DKK which is a wind inhibitor so I need to have this feedback loop intact and then I can kick it on with any of these proteins so people have followed up on these kinds of studies and actually shown that sort of analogous to these Xenopus experiments where you can get this axial elongation in aggregates of cells you can actually do this in mouse embryonic stem cells so if you have the right size aggregate and you treat it with the right factors they form these elongated things which people believe correspond to an interior posterior type axis and then if I look at these elongated type things I have this wind signaling and this brachiary on one side of the embryo so these kind of studies of sort of getting this polarization in a self-organized way sort of formed the backdrop for what we wanted to do in my lab but we wanted to sort of improve on these initial studies so one thing we were interested in is well can we make this patterning reproducible one of the great things about setting embryogenesis is that every embryo basically looks the same and that gives you a way to compare between things and you know this when we started every stem cell culture basically looks different where you get some fates out but the precise pattern of fates always looks like well a mess basically we want something where we can actually quantify and put numbers on the resulting patterns and say things like are all these treatments where you treat with these different ligands and kick on this positive feedback loop are they really the same or are their signatures or something different and we want to know can we really use these systems to dissect patterning in vivo so we started sort of doing similar things to what they were doing which is taking cells we're growing them in 2D culture not in 3D treating them with growth factors and asking what we get and what we found is that basically you know you get differentiation you get some locally correlated differentiation which is similar to what they see right I look at this cells where I have a whole field of cells here particular cells become mesoderm and if a cell is mesoderm its neighbor is clearly much more likely to be mesoderm but there's still no rhyme or reason to why I have this swath of mesoderm sort of curving arbitrarily through this field of cells and no mesoderm here or here so we started to think about well what's the source of this variability and when we think about how you grow stem cells so this is sort of a zoom out of a stem cell culture so this is like a millimeter so these big colonies have thousands of cells down to these little colonies which have a few cells when you think about this you have all kinds of sizes and shapes these cells are gonna grow they're gonna merge into each other especially when they differentiate right funny things are gonna happen on the boundary and none of this stuff happens in an embryo right an embryo it always has the same geometry it doesn't have these sort of boundaries you know you don't have two embryos fusing under any kind of normal circumstance and so could this be the source of the variability that we see we notice that when we look in these colonies which we first tried to sort of take the colonies and say okay what if we just separate them and we keep them well separated so that you have distinct colonies and ask whether they pattern found that colonies the size of embryos don't really pattern when colonies get somewhat bigger you start to see these kind of sketchy sort of fuzzy patterns emerge but these are actually several times the size and contain several times the amount themselves in the embryo so can we improve upon this and part of what we're thinking was that okay we still have these irregular geometries and also these cells as they make these sort of large objects are just spreading and spreading and spreading whereas the embryo doesn't do that obviously the embryo is confined to sit in a particular geometry and so what if we use micro patterning technology to grow ourselves like this instead right so here we've taken a culture service we've made complimentary patterns of things the cells like to grow on and things the cells won't like to grow on and then when you see the cells they form these perfectly circular colonies or you can make colonies of any shape you like and then the other thing that turns out to be very important is that the cells are then confined to these regions and so that tends to increase the density of the cells and increase their interactions with each other another nice thing about the system is right so I have many copies of these identical colonies right so I can zoom in and think about any one colony the imaging is good enough that I can identify most of the cells and then if I think about zooming out so here each color is a different image and each dot is a cell I can zoom out and zoom out zoom out and zoom out until I have this whole cover slip composed of thousands of images and I've quantified what every cell is doing in you know millions of cells and thousands of colonies and so this is a way to ask you know how reproducible my patterning is in a colony of a given size and also think about when it deviates from that and why okay so the first question is well can we grow stem cells like this and the answer is yes we can grow stem cells like this so all the cells will still grown on these patterns for a couple of days will still express all these markers of pluripotency that I told you about interestingly they have sort of pre-patterns within the colony which you can discern when you grow them this way so all these markers are somewhat higher on the border of the colony than the center which we hadn't realized before and if you look at the sort of spatial scale of that elevation it's actually quite consistent across colony sizes so that's another way to say that the sort of hundred to hundred and fifty microns at the edge of the colony are always different and then after that it doesn't care about the edge of the colony so they're not wells it's basically what you do is you take a piece of glass you coat that piece of glass with something the cells don't like to stick to which in this case is PLL peg then you take a mask and a UV lamp and you burn that PLL peg off in a particular pattern and then when you come in and coat with things the cells do like to stick to like laminin or like matrigel it won't stick it's all right it's also protein phobic so the protein won't stick in those areas and so you've created this complementary pattern of we're now using a specific form of recombinant laminin but you can also use matrigel so you've created this complementary pattern of PLL peg and these islands of laminin in the middle of it other hand class yes so we don't do that yet so I mean basically the stiffness right now is the stiffness of the glass so it's sort of infinitely stiff we're actually we have a collaboration with a company who is now making these things with they're not really as soft as we'd like them to be but they can do it in the range of thousands of kilo Pascal's and so we're gonna look at that it affects the patterns but I don't have data on that okay so the important thing of course is whether this actually works you actually get patterns and we were quite happy to find that it does so if you just look at the patterns of sort of where cells are you get colonies that look like this where you get sort of sparser cells in the middle and then this pile up of cells in a ring and then sparser cells around the outside when you look at cell fades these middle cells stain for markers of ectoderm and then you get markers of mesoderm in a ring around that and markers of extra embryonic tissue around that I'm gonna get myself in trouble with cat if I call these trophoblasts but I think they're trophoblasts but maybe that's yeah we can talk well I think we've talked about that a lot okay yeah so they could be extra embryonic mesoderm as well this is a millimeter across and there's about there's about two to three thousand cells in here okay so I think there's two okay I'll show you a little data on that I don't think the cells on the border need something mechanical so that well I mean one part is just that you know you need reproducible input to give you reproducible output otherwise you get a mess I think another part is that you the density of cells is very important and what we've seen is the lower the density of cells the more cells will respond to the extrinsic signals in the media versus talking to each other and in particular so the the fates in these colonies that rely on particularly high densities of cells are the exoderm in the middle and the reason is because they have to be inhibited but right so the signals that we're supplying will divert them away from that exodermal fate and so you need inhibitors to produce that exodermal fate and if the cells don't achieve a high enough density they can't sort of locally produce high enough concentrations of those inhibitors to do it and then the second fate so I was about to show you where it's so I couldn't stain for everything at the same time but you also get endoderm in this ring between where you would get mesoderm and the extra embryonic tissue around it and this endoderm is also the most sensitive to these density effects right and I think the reason for that is as I'll show you a little bit later these mesoderm and endoderm are formed by secondary signaling through the nodal pathway so we add BMP4 that indirectly turns on nodal and that nodal is crucial for forming these cell fates and in theory the highest levels of nodal are what give you endoderm although we haven't formally proven that in this system yet and so what I think happens is unless you get these sort of high cell densities in this ring you can't accumulate enough nodal to make these patterns right so I think in the embryo right you have this confinement you have these high cell densities and it forces these interactions between the cells in flat culture the cells just spread and spread and spread and you talk more to the culture media and the interactions that require the sort of highest cell densities just don't happen Yes, all right I think that it looks like you think that what gives you the right or mimics the embryo is just because you have enough with that geometry in the monthly combination of concentration and geometry and sort of mimics the human embryo or you can actually say that you make something that is the size of the geometry of the human embryo well so the size matters I'll show you the data in a minute the size matters because you need enough room to make all the cell phase and I think the edge matters in that it biases particular phase to form a particular place so I think you have yeah I don't think we have enough data to really prove this yet what I think is happening is that you have some self-autonomous patterning system that you can kick off at high enough densities and certain regions of it are more sensitive to those densities than others the cells are more sensitive to these signals at the edge and so that will right if you have some instability if you have this sort of pre-pattern right that will bias how your instability makes the pattern in the tissue but if you don't have that pre-pattern the thing is still unstable and you'll make that same pattern but it will be sort of more random the way it forms so I'll show you a little bit of that in a minute other questions of this that I don't know you're saying the cells are more dense or the marker is brighter yeah so I'm not sure that's actually so one thing that can be misleading when you look at these pictures is that so if you look at the cells in here they're sort of more flat and spread and the cells over here are pushed up against the boundary and they tend to get smaller and their nuclei are more condensed so they can sometimes just appear brighter like if you look at the DAPI right these are dimmer and these are brighter so this okay so I'll show you in a minute this is epithelial and it's pushing up against the border of the colony if that's what you mean but that like I'm not confident these are real differences in expression of the marker what I think is happening is you have similar expression of the marker but then when the cell gets compressed and the nucleus gets smaller you have the same amount of the marker in a smaller nucleus and it just appears brighter and so when we quantify these things we're always normalizing we're barely normalizing these markers to this DAPI right because otherwise right there's a clear pattern in the brightness of this DAPI which mostly relates to the cell density and size and which has to be normalized out to quantify these things properly okay so you can you can put numbers on all these things and it works pretty well right so you get ectoderm, mesoderm, endoderm and this extra embryonic at the border you can put error bars on them these are standard deviations across colonies so I think this is actually quite good so the the colonies really look the same almost all of them okay so this goes to your question right so what happens if I take the colony and I shrink it down right and so what we see in these circles is that when I shrink the colony size down I make essentially the same pattern from the edge inward but then I run out of room to make that pattern and so I I sort of stop at the mesoderm here right so here I have the screen territory I have this red territory here I have this green territory if we look quantitatively it's the same size this red territory is in the same place but here I've run out of room and I can no longer make this territory in the middle right so if I look at that quantity yeah both I think so some cells are missing here because so this is a there's four layers here really right so there's ectoderm, mesoderm and extra embryonic and precisely in this gap is where you would see the endoderm right so where you see these kinds of markers right so these cells the cells that you see light up here are going to just be dark here because I don't have these marker we just can't stay in for them all at the same time but but I also there are higher and lower density regions of this culture that result from patterns of the migration of the cells so as I'll get to in the final time but this region well so this brachiori is also it's a marker of the primitive streak where this mesoderm differentiates and what happens in the primitive streak right is that the cells do this gastrulation movement and move into this primitive streak so this higher density in this region we think results from these gastrulation like movements into this region and then actually the the cells can undergo a really natural gastrulation like movement because they're on this hard culture surface so the direction of this sort of invagination that would happen at gastrulation is into the tissue culture surface and then they'll sort of move from there but they're very restricted in how they can do it so you end up getting a pile up right there right so they so this density different so yeah so I mean the answer to your question is basically both are happening some are you're missing some cells that are actually here because they're just not labeled by these markers but you actually really do have the densest area here which reflects these cell movements resulting from this sort of primitive streak formation yeah saying would we do better in 3D than 2D is that the question I think we might and we're trying to do that but so the the hard thing to do in 3D is to have geometry this reproducible right so I think the best that people have done in 3D is sort of the experiments I showed you before I showed you these experiments which is you can make 3D aggregates they'll self pattern but they'll you'll get these sort of random patches of self patterning and nobody's really been able to do it in such a way that I can you know actually quantify at how big is the territory of say mesoderm and put an error bar that it goes from here to there there's just no standardization in that sense the other thing I'll say is that so what I think about these 2D systems is they're actually quite a good model for the starting point right because the starting point of this differentiation process is you have this flat epithelial sheet of epiblast on top of a hypoblast which is mimicked by our culture surface and then where when you start to get these cell movements right that's where it starts to break down so what you'd really want is to start in the same place right to start with this flat 2D sheet but then to allow the cells to sort of move and invade more naturally when they leave that 2D sheet so and I think that's a I don't know that it's impossible but it's a challenging bioengineering problem to figure out what the substrate is that will sort of allow you to grow the cells on top of it and then invaginate into it as gas relation happens yeah so we've tried growing cells in 3D aggregates and we've actually we've put a fair amount of effort into like you know using wells and controlling the size and shape of those aggregates and we still don't get nearly as reproducible results as this yeah that's something we're still looking at there is so yeah I mean I think it's a good question because like sort of one of the main opportunities of this system is actually to watch live what's happening and there is clearly some cell movements that happen here and they're sort of well they sort of mimic these gas relation movements but maybe there's some cell sorting but the the tricky part of doing that is that we need to be able to follow how the cells moved and then also follow their final fate and so we're just now generating the reporters where we're going to be able to you know see cells light up for SOC17 or Brackuary and follow their trajectory they actually sell tracking in these colonies it's I'm sure it's easier than in an embryo but it's still quite challenging so still work in progress yeah yeah so these are radial averages right this gives you some idea this is these are single colonies where each color each dot is an individual cell and it's color coded by how much the cell expresses that marker yes you can plot that so the density plot looks not too dissimilar from this Brackuary it maybe it's a little the peak is a little broader right so it peaks in this sort of mesodermal region which is consistent with the idea that the density is highest in the region that mimics the primitive streak yeah so and so the way these are made is we segment the individual cells like you see here each cell is assigned in expression level and then for each sort of bin of you average over all the cells in that bin so this shouldn't this should not depend on density because you're right if you have more cells you're just taking the average over more cells but it doesn't give you a higher number right but yeah you could make a density plot does that answer your question yes sir everything here is is nuclear all transcription factors so they're all in the nucleus okay so actually this gets to a little bit of what we're talking about what about cell movements and then sort of gastrulation events inside the the micro pattern culture so do they mimic these things that happen in the embryo or so this is a nice electron micrograph of an embryo where you see this is the epiblastic sheet and these are the cells leaving it and sort of migrating this is cut off but it's so it's known that there's a signal in cascade that works through this FGF pathway that up regulates now that down regulates deket hearing and so we asked whether this sort of pathway is intact in these cells that would lead to these cell movements and indeed right so if I look at where the ERC pathway is activated it's specifically activated right in this ring where you see the highest cell density where you'd imagine this gastrulation is taking place if we look at snail which is this marker of epithelial mesenchymal transition you see that in a similar region and overlapping with the back here and then if we look at deket hearing it's actually a little bit more complicated so but what we see here is that the deket hearing in a lot of the culture is primarily on the cell surface and then in this region where the cells are migrating it's getting pulled off the cell surfaces and becoming cytoplasmic although we don't see large changes in its intensity and so then this gives you some idea of the 3D structure of the colony at the end of these experiments right so you indeed have this sort of piled up region here where the primitive streak is in the center you have this flat monolayer and on the edge you have this flat monolayer and if you're interested in the epithelial properties of this right so what you see is that if you take confocal sections right these sort of monolayered areas are epithelial and in the region where you have this pile up the top layer is epithelial and the bottom layer is not right so this is consistent with this you have this sheet this epithelial sheet where gastrulation happens cells move into this epiblastic region and then they invaginate underneath the region and at that point they leave the epithelium and stop expressing these epithelial markers so we see these kind of hallmarks of the beginning of gastrulation and then cells are sort of getting stuck piled up under the culture we do see a little bit of what we think is cell migration because if you look at SNAL which is a marker of this EMT having taken place and so you see the main sort of streak of SNAL here if I look at the culture a little bit away from there I look at the top layer it's entirely SOX2 positive which is characteristic of this epithelium which is going to become the ectoderm but if I look underneath it you see these sort of scattered individual SNAL positive cells which are presumably migrating away from this primitive streak so they're sort of recapitulating this gastrulation process okay so now I'd like to talk a little bit about the question of how do these patterns form at the level of signaling right we really want to understand you know how is it that we take these cultures we treat them with a single ligand and then that single ligand somehow induces the cells to make these patterns along the radial axis of the colonies so to tell you about that I need to tell you about sort of what's known about the signaling pathways we're interested in and fates and very potent cells and I've told you a little bit of this already right so if I inhibit all pathways I get neural differentiation or more generally I start going towards ectoderm I need a little bit of this nodal signaling to give me pluripotent cells but if I up-regulate nodal perhaps together with BMP I'll get mesendoderm and if I up-regulate only BMP well I'll call this extra embryonic virtual ectoderm so the sort of more sharply focused question is we're taking what we're doing the proximal manipulation we're doing is just to push hard along this axis right so we take these cells we dump in a lot of BMP4 so why is it that I don't get trovectodermal differentiation only I get all these cell fates in some kind of ordered pattern along the embryo so to start answering that question we actually started looking at the patterns of signal transduction that we get and what we see is that so this is an immediate read out of the BMP signal that we add so you take BMP you add it in SMAD1 is the signal transducer for the BMP pathway and when it's phosphorylated it's activated and what you see here is that it's particularly elevated at the border of the colony so this signal is primarily received only at the border of the colony and that's something that evolves in time right so initially it's received in sort of a broader gradient across the colony which is shown in red here and then it becomes more and more restricted such that you only have sustained signaling of the colony border and so that makes sense with this diagram I just showed you right because BMP signaling is supposed to push me towards these extra embryonic I get BMP signaling in a sustained way only in one particular region and I get extra embryonic differentiation in that region but what's going on in the rest of the culture well if I'm getting mesendoderm differentiation I should be getting nodal signaling and so if I look at a readout for nodal signaling indeed I see elevated nuclear SMAD2 in a much broader ring than the elevated SMAD1 and this is consistent with the territory in which I'm getting mesendoderm and this actually is a sort of well it's always elevated here compared to the rest of the colony but I get really high levels of it in the sort of transient first after I add the BMP and then it adapts back down and so our hypothesis was okay only the out somehow only the outer cells respond to this BMP signal the response to this BMP signal leads to secondary nodal signaling the region of the nodal signaling is broader and that creates these mesendodermal fades from the rest of this colony so you have this sort of two-step cascade where the primary signal delineates inner versus outer and then you position these additional layers in between so that hypothesis is pretty easy to check in cell culture because I can just add an inhibitor to the secondary active and nodal signal and ask what happens so if I have the BMP alone I form these multi-layered patterns and indeed if I have the active and nodal inhibitor right I still respond and make this outer territory and the entire rest becomes this inner territory so I haven't made the secondary nodal signal and then I don't make these cells along the colony axis we have so you can go back a step further and ask why is it that the BMP signal is restricted to this region that becomes green and why is it that the nodal signal then ends up restricted to the these regions where you get the mesendoderm and we've looked into that using some knockdown experiments so here's a control where I get these nice rings if I knock down BMP inhibitors I see this pattern totally encroach upon the center right so I need to express these extracellular inhibitors of BMP they prevent the signal from being transduced at the center of the colony and then if I don't have them there right the signal gets transduced all over and I get these things and interestingly if I knock down inhibitors of nodal I actually get a spread of these fades all over the colony as well and so these phenotypes they look similar but they're actually subtly different and the main difference is that here the red mesodermal differentiation only spreads inwards whereas here it spreads in both directions so what we think happens is that you have cordon and noggin which are these BMP inhibitors they're concentrated at the center of the colony they that they prevent the BMP from being transduced there so it's only transduced at the outside at the outside you respond to this BMP signal so you make this green territory you also make nodal but you make a system of nodal and its inhibitor lefty which is thought to be this sort of self-patterning system and that positions this stripe of mesoderm differentiation here if you make only nodal but not lefty then the signal just takes over the entire colony and you get mesoderm basically everywhere so what we do is we we will take cells we'll do the s-i-r-n-a typically we'll do the s-i-r-n-a we'll wait one day then we'll pass the cells onto the micro pattern they'll grow overnight on the micro pattern so at the second day after the s-i-r-n-a transfection we're hitting them with the BMP ligand and so the experiment with the cells will typically take place on the micro pattern will take place two to four days after the s-i-r-n-a which is basically the usually if you're doing s-i-r-n-a and you're looking for knockdown that's sort of the window where it's most effective yeah we haven't done knockdown directly on the micro pattern and then you take you always take the feature after the load yeah okay so yeah so I sorry I should have been clear of no no so these patterns take two days to form right so you treat them with a different so right they'll at time zero they'll be these pluripotent cells they'll look pretty homogeneous right so I showed some there's a slight pre-pattern of these pluripotency markers towards the edge of the colony but then you treat with these ligands and you get these markers appearing over the course of two days and there's there's actually some sort of spatial dynamics and how these markers form these final patterns which I if you're interested I can tell you about but I don't have here okay I just wanted to mention so that this was actually not done in my lab but this was done in Ali and Eric's lab after I left so they've actually done some nice cell biology to show that basically these inhibitors are they are part of the reason that you get this pattern but they're not the only reason that you get this pattern that there's actually effects of receptor accessibility so here you're looking at stains for the receptor and white here and this 01 is a stain for tight junctions between the cells and so what you see is that the colony edge you can see these receptors are on the apical surface of the cells and so they're available for ligands to bind to and that's part of why these cells particularly sense these external BMP ligands if you look at the colony center you no longer have these receptors on the cell surface but they're only between the cells and then you've got this tight junction marker on top so basically the receptors are all in such a place that they're completely sealed off from receiving the ligand which explains in part why these cells in the center don't respond to this ligand right that's sort of shown in the cartoon here and they've actually looked at a bunch of different TGF beta receptors and found the same pattern in all of them and so they went on to show that actually this does make a big difference you can grow cells on a porous membrane and then you can look at the difference between stimulating from the top and the bottom where if you're stimulating apically or stimulating basally what they found is that if you stimulate at low density it doesn't really matter right I stimulate at the top I get response which is shown at red here I stimulate the bottom I get response but if I'm at high enough density on these porous membranes I only get response from the bottom right so in these culture substrates where the bottom is rigid right I'm not gonna get I basically can't access the cells anymore and that's part of what underlies the pattern essentially either either either mechanism is sort of sufficient if you push it hard enough right so they've been able to take cells which are knockouts for these BMP inhibitors and then if they're dense enough this receptor sequestration will take over and then if they're but if they're at low density I can still make some semblance of a pattern by sort of overexpressing the inhibitor so there's this combination of inhibitors and receptor sequestration that gives these patterns okay so I think I'm this is not a bad stopping point I'm basically out of time so I wanted to say more about these patterns but luckily I have another lecture tomorrow so maybe I should stop here and take any last questions