Scientist Stories: Magdalena Zernicka-Goetz, Human Development and Synthetic Embryos

We will have our keynote presentation right now, and I want to— it's my honor to welcome Magdalena Czerny-Koguets. Magda is very well known across the world for her work in molecular and cellular mechanisms that study cell lineages and patterning.

She has been working on mouse embryos, and she uses various techniques that include imaging and modeling. Specifically, some of her work has looked at plasticity, pluripotency, and cell fate acquisition, asymmetric and symmetric divisions of cells, transitions from maternal to zygotic development, and self-organization and pattern formation. So, and she has a very interesting career. She got her PhD at University of Warsaw in Poland under the supervision of Andrzej Tarkowski. And then after that, she moved to Cambridge, I think for a postdoc. In 1995, and she worked there with Martin Evans to study the mechanisms of regulatory nature and spatial patterning in mice. Then in 1997, she got the Senior Research Fellowship from the Lister Group to start her independent group in the Wellcome Trust Cancer Research UK Gordon Institute in Cambridge. And there she's done amazing work again in mammalian development and stem cell biology. And then in 2013, she became a Fellow of the British Academy of Medical Sciences. And then more recently, in 2019, she moved to Caltech where she is now, and we're very happy to have her join us. And thanks for waking up so early, Magda.

Go ahead.

Thank you so much, Carlos, and this is really a great pleasure to be with you today. It's a pity it's virtual, but in a sense this is how our world works, like, like, yeah, it becomes like new normality. So I am really delighted that that I can do this. And I wish to try to share the screen. Please tell me, Carlos, whether you are seeing— just a second— one slide or more. I would move it here to my right. Do you see one slide or more?

We see one slide. It's perfect.

One slide. Okay, fantastic. I will diminish our images so I can see my slides here on that screen. Super. So, um, yeah, as Carlos said, I had a quite unusual career because I moved between countries. And recently we started, just few months before the pandemic hit, we moved to Caltech as a family and as a lab. And that's where we started to in new circumstances setting up our new group. So our group at Caltech is relatively young, our group at Cambridge is much older, and, and I would present today results mainly of our group in Cambridge and the work that we are continuing, some of the unpublished work as well. So I wish to say that— start with this very big question that really is in the roots of our work: how do we start to become ourselves? And in my view, it is incredibly— really remarkable that all of the creatures on the slide, as you see them, they are so complex, so individual with their own form, shape, thoughts, habits. We all start our development very much the same. And this, the beginning of the development that really unites us, is the major subject of my talk today. So, for many, many years, we were trying to understand how the beginning of development can take place when we really start with the very simple act of fertilisation of tiny little cell egg that you can see on the schematic on the left. It's 80 micron diameter across, so clearly very small cell and yet so potent. And we are trying to understand for many, many years since I started my group how it is that this one single cell can generate many cells by the 5th day of development in the case of human embryos. We study human embryos, but our major model system are mouse embryos and stem cells, as you will see from this talk, because we can study mouse embryos and stem cells so much easier than human embryos. Nevertheless, we now have also human embryo system in the lab, so we can always verify the findings we are making on the mouse in humans. So here I'm not going to talk much about this, but I will be moving to the next stage of our life. This, this not first 7 days, but the next 7 days. But the first 7 days, what we found, the key first event is to set up cellular heterogeneity. And I mean by this, the moment when cells start to be different from each other. And we found that in mouse embryos, unexpectedly, it happens much earlier than what was thought at the time when I started as a researcher. It's become established by the 4-cell stage. And actually, just a few months ago, 2 papers were published, one Nature, one Science paper, papers that showed the same is true in human embryos. Already blastomeres at the two-cell stage show different fates. So, this early cell fate specification, at least in the mouse and most likely also in human embryos, is not deterministic, it is sort of guiding cell fate but is initiated at the time. So, I'm not going to talk about mechanism of it, although we now understand a little bit more about it, how this process is initiated. The second important event is polarisation of the whole embryo, every single cell have to polarise in apical-basal fashion. And again, I am not going to talk about it but show you an image, very beautiful short movie of that process as it happens. So, at this 8-cell stage, polarisation is very important for developing two different ends on every single cell, apical-basal end, and we found that the trigger for this polarisation is expression of two transcription factors, Epi2γ and TET4 from zygotic genome. So, the timing of this event is dependent on the timing of zygotic genome activation. And together with constitutive RhoA, those active RhoA, those three factors will establish this apical domain. So, when cells become slightly different from each other and all of them become polarised, they are really well equipped now to diverge properly in subsequent couple of days to establish these three lineages. So, by the blastocyst stage, we have those red cells that we call epiblast, and these are the cells that will give rise to the future organism. On one end, we have green cells called primitive endoderm, and these are the cells that will build yolk sac in which embryo will be growing. And the third group of cells called polar trophectoderm, shown here, is going to form extraembryonic ectoderm that will make the placenta that will combine the growing embryo with the body of the mother. So essentially, this first few days means that we have to set up the three main actors that will be now taking action and interact with each other to build this organoid that you can see here. So this is the question that I would like to discuss today. So what happens when those three main players start to interact with each other? And this has been an extremely difficult question to address because while we can culture embryos, make those movies, and many more things, use CRISPR to knock out genes, use many technologies, classical and modern, to really uncover these first few days of development. At the time of implantation, embryo would be embedding itself within the body of the mother. It's very difficult to recover, particularly at the beginning because it's still very small, but now starts to grow. So at later stages of development, it's more accessible, so we can remove it and put it into the culture dish and observe what happens at these stages of development, but not at the time of implantation. So really, to be able to understand how those three tissues interact, and this interaction will be organizing embryo into much complex structure in which anterior becomes specified posterior becomes specified and the whole process of gastrulation starts, we needed to develop a system that will allow us to culture embryo from pre- to post-implantation stages. And this is what we did first, and I'm showing just an example of the movie now that we will be able now to follow much bigger embryo and make the movies like that and many more things to really understand this process of self-organisation. So, this is the SCULPTURE system. We set it up first prototyping the mouse,. And then we found that exactly the same system can be used to culture human embryo. And also our colleague Ali Brivanlou and Eric Sikya at Rockefeller have used this for human embryos in parallel to us. So, here these images of mouse embryos and snapshot of the movie that we have taken to follow anterior signaling centre by following expression of GFP from Cerberus promoter, which is very important gene that I will be using in my talk later to specify that centre. So, see here very simple blastocyst and how it embeds within the polyacrylamide gel in this case. This mural trophectoderm is what allows invasion, and then this complex of cells start to interact and grow into this excentrical structure in which anterior is one side and posterior on the other. So, this model system shows us that this early remodelling of the embryo interactions between those tissues happen independently of the body of the mother, or can happen independently of the body of the mother in a culture. It also allows us to uncover the chemical and mechanical interactions that happen between the tissues to be able to build the embryo. This is still ongoing work, although you see already 3 papers published just on that remodelling of the embryo. But of course, this is a very complex process that we are still studying, but I would like I would like to show you just the main thing that happened because we would need this knowledge to understand how we can model embryo from stem cells. So, the first thing that happens is reorganisation of this epiblast group of cells into this polarised rosette-like structure you can see here in the centre of which lumen will open. When it happens through the positional information from the epiblast, specifically FGF4, the boundary becomes set up between polar trophectoderm and neural trophectoderm. This boundary means that polar trophoectoderm starts to grow upwards and establish this tissue called extraembryonic ectoderm. This tissue would also form the cavity through invagination, and then these two cavities will merge with each other through resolution of hybrid rosettes that we discover happen, become established on the interface between those two compartments, and finally they become joined, and this is proamniotic cavity. Or amniotic cavity in the case of human embryo. When those two tissues interact, red and blue, the yellow tissue, primitive endoderm, grows and forms one tissue that we will not talk today, parietal endoderm, shown here. And the tissue that we will talk today is called visceral endoderm. And in this tissue, a very important anterior signaling center will become established. You see here above in the form of GFP expression from Cerberus promoter.. So, this centre will send signals to protect this part of the epiblast from posteriorising signals, and I will come back to it. So, we not only were able to describe this process, we were also able to identify genes that are responsible to a certain extent for this process working with embryos per se. And knowing this, what we have learned over the last 4 years, we wonder whether we can use this knowledge to reconstruct development in vitro. So, do we know enough to be able to attempt reconstructing development in vitro stem cells. So, you can say this is just an incredibly complex act, and indeed I thought that maybe it would not be possible in one go. So, we do it, we did it, and I will describe it today stepwise. So, I will be using three types of stem cells in the rest of my talk today. The first type of stem cells is called ES cells. They are very famous, of course, they are the first ones that were discovered and isolated, but the stem cells for the epiblast, I would We would also use primitive endoderm stem cells that are called Zen cells and trophoblast stem cells that are called polar trophectoderm, that are stem cells for polar trophectoderm. And these are the papers of the colleagues in my field that isolated the stem cells. So, the first thing that we have done, so yes, so I will show you now this journey that we have taken over the last 5 years to build those more and more complex stem cell-based models of embryo development, concluding with the final model of the complete embryo-like structure at post-implantation stages. So, we started this process to really understand the principles of development, the most important principles of development from the bottom up. So, this is our first model. We set it up first for the mouse and also then for the human stem cell with the human stem cells to mimic implantation stage of development, and specifically to mimic this reorganization of the trophoectoderm, of the epiblast, from the group of unpolar cells to the group of polarized cells that form the epithelium. So, in this case, we embed individual or very small groups, and this is very important because this would be very different from embryonal body approach that has been used in the past, in extracellular matrix. In this case, we use the Matrigel,. And you can see that after 24 hours they become polarised, after 48 hours they form this stunning rosette-like structure, and then 96 hours after cell division the lumen opens in the centre. So, very simple model, but it was extremely useful. And I would like to show you two examples how we have used this model so far to understand mechanism of epiblast morphogenesis and how this morphogenesis is coordinated with a change in cell identity.. So, this is the model we will be using. So, here below you see mouse embryos are transiting from this preimplantation blastocyst structure to postimplantation, periimplantation structure into extra-embryonic structure. So, by using two-well leaf conditions that either maintain those cells in this naive state of pluripotency or removing these conditions that allow those cells to mature, you will see that they can take those transitions, they really mimic natural development.. So, the first thing that we were interested is what is the mechanism of rosette formation from the molecular perspective. And we know that integrin, beta integrin, is extremely important for embryo development at that time. So, if you remove beta integrin, and this has been done in the past, and we repeated this experiment by using this line in which we can conditionally remove beta integrin from the embryo at the time of implantation. So, just one day after blastocyst stage, you see that epiblast become unfunctional, become degenerating in comparison to what happens in natural embryo. And these are just two examples of this embryo and showing you how many epiblast cells become eliminated from development. And of course, those embryos are aborted and cannot develop further. So, we checked whether the same will be true when we use integrin beta-1-minus-minus ES cells and found indeed that this is the case. So, here are the control ES cells. And here we don't have expression of integrin because there is no integrin in the cells. And you see after 48 hours, we start to see Caspase-3-positive cells indicating death of some cells. And after 72 hours, nearly all of the cells are becoming eliminated from those structures. So, we ask what happens to the stage of those cells just before they become eliminated. And at the time of the eliminations, can those cells polarize and form those rosette-like structures. And this was interesting because we found that those cells can initiate polarization. So, if we don't have integrin beta-1, as in normal circumstances in natural cells with beta integrin, those cells polarize, for example, gamma tubulin becomes localized in between the cells, also Golgi apparatus becomes localized. But this polarity, as they are non-distinguishable, this 24 hours after seeding of cells. But after that time, 48 hours, so 24 hours later, after 48 hours after seeding cells, they really aborted this polarization, they cannot maintain it. So, this really told us by using ES cells that integrin beta-1 is absolutely important for cells to maintain their polarization. But we found that it's more than that, and this is just a summary of the paper that we just published, so I'm not going to any details, but what we found that in natural circumstances, actually, β-integrin has dual function. It suppresses actomyosin activation on the basal epiblast to ensure its confinement apically, and this enables recruitment of PAR6 and podocalaxin to secrete fluid and make the lumen. So, this is what happened naturally. But we also found that β-integrin function is to promote the survival of the epiblast cells by activating two signalling pathways, FGF and IGF to really protect those cells from dying at the stage of remodelling. So, when we remove beta-integrin, we remove those two actions that are essential for survival. And majority of those embryos, of course, will become eliminated progressively at this stage of development. So, this is one example of the study that we have used by using the simple model system. But also now we are currently carrying out the studies, and this is Antonia Webberling from my lab, a very talented PhD student, in collaboration with another PhD student, computational modeller Max Winter, and Gullion Salbry from Craig Institute in Geneva, are trying to understand how rosette forms in this wild-type circumstances and in the time when beta integrin is eliminated. So, we can now make long-term movies of how cells behave at the time of rosette formation, and segmentation of those movies allow us to quantify them, which is very important. So, we can measure the volume of the cells, motions of cell centres, we can define the polarity of the cells by expressing different type of markers as those cells form rosette that we can follow, and this can allow us to model this process. And the modelling of rosette formation in silico shows us that by increasing the adhesion force in the apical domain is what we will be able to use to really form that rosette-like structure. So, this is really work in progress, and I'm not going to stop on this any longer, but I just want to show you how using stem cells in culture in very defined conditions can be extremely important to understand much more complex morphogenesis that happen within the embryo and how we can use this knowledge and then and check whether the same is really happening in the embryo. So, we can now move from one model, stem cell model, into the embryo model very swiftly and really understand what happens at the time of our implantation. So, another example of the very important study that we have carried out a few years ago was to understand how this morphogenesis of the epiblast that I described to you can happen in stem cells. When we culture them in extracellular matrix environment relates to the changes in cell identity. So, when the cells are apolar and become polarised, they change from naive state of pluripotency to their prime state of pluripotency, in which cells are ready to initiate differentiation and initiate formation of the body. So, Marta Schapassi, a very talented postdoc in the lab a few years ago, now set up her own group, came to my lab at the time and she wanted to understand which elements of this transition from naive to prime state is responsible for which changes in the tissue architecture. And what Marta found by carrying out many experiments that are again now described a few years ago in that paper that I'll show just an example of, that when she blocks exit from naive state of pluripotency by adding these two ILIF conditions, she is able to achieve polarisation of those cells in culture, initiation of polarisation, but those cells will never open up the lumen.. So, this told us that exit from naive state of pluripotency is important for not polarisation of cells, because if we block it, the cells can polarise, but for opening up of the lumen. And this allowed us to uncover the mechanism of lumenogenesis, identify how two transcription factors work together to induce expression of podocalaxin to enable vesicle fusion and lumen formation.. So, the journey through implantation can be achieved by just adding stem cells, embryonic stem cells on their own, but we wanted to go further and induce breaking symmetry in the structure as embryo does. And indeed, when we culture those embryonic stem cell rosettes further, they initiate chaotic process of symmetry breaking and expression of mesoderm markers that will mark for us formation of the primitive streak-like structure in one side of that structure, like you can see here, but very rarely it is regular. So, it's quite a chaotic process if we do not have these extraembryonic actors. So, I would like to show you how we thought about it. We wanted to induce the natural symmetry-breaking event, and in natural circumstances, this is the role of those two extraembryonic tissues that I introduced on the first slide. So, first is this green tissue, visceral endoderm it's called at that stage, derived from primitive endoderm. This anterior signalling centre on the distal tip of that embryo is particularly important. It will migrate to one side and secretes Cerberus, I already mentioned, but also Lefty-DKK that inhibit nodal and twin signalling on that side. So, this signalling becomes enriched on the opposite side of the embryo, also due to the blue tissue derived from polar trophoectoderm called extraembryonic ectoderm that secretes BMP, for example. This accumulation of that signalling leads to the expression increased expression of nodal added sites, so nodal gradient, and this is inducing primitive streak formation. So, we decided to induce symmetry breaking by adding extra embryonic stem cells. So, in the first instance, we did it step by step. So, actually, in the first instance, we will not add green cells or stem cells for green tissue, we will just add blue cells, and we will ask whether this is enough for symmetry breaking event, which would have been counterintuitive because we all know in mammalian embryo development field that this anterior signalling centre is very important. Okay, so what happens if we don't provide it? So, here is the embryo again. So, we add ES cells to ES cells, and amazingly, those cells find themselves. We are not fostering their collaboration. And when they found themselves, they formed the structure that I'm showing here. It's called ET embryo. It's composed of ES compartment, TS compartment, that mimics natural embryo, but it's missing this layer of cells outside. It has this extracellular matrix instead. Amazingly, what we learn about embryo development when we use this system, that we can have a symmetry-breaking induction event without this anterior signalling centre at the right time in the right space. So, most likely it is dependent on the number of cells and geometry of the embryo, and we are still trying to understand exactly what happens here. But I'm showing you here the embryo that we built using ES cells that express Cerberus from lacUripromotor in this case to mark symmetry-breaking event and mesoderm specification. And you see that on one side of those red cells, those green cells become specified. So, we took those GFP-expressing cells to really identify to which part of the embryo they correspond and whether truly this is posterior part of the embryo. And we found that indeed they correspond to gastrulating embryo, embryo in advancing gastrulation process, 7-day-old embryo. And very much to the posterior side of the embryo. When we took cells from the opposite end of the epiblast, far away from that centre, those cells were much less well specified. So, they had characteristics of 7-day-old embryo but not of anterior. So, this led us to ask the question, what will happen? Would we have proper anterior specification and posterior specification if we now incorporate into that system the third missing cell type, so anterior visceral endoderm forming tissue in the form of the visceral endoderm. And therefore, we had to put together Xen cells into the mixture. So for this, I'm showing you just success. For a while it didn't work, but finally we found conditions that all three types of cells could be happy in living together, and this is a kind of geometrically constraining dish which we use. So, here we put ES cells, T cells, Zen cells together, and with time those structures develop into something that looks absolutely stunning, and we call it ETX embryos from ES cells, T cells, Zen cells. So, you see that this embryo can be of very similar size and geometry to the natural embryo, have three lineages, This lineage is derived from Zen cells, so replaces visceral endoderm. ES cells are here, replace or mimic epiblast, and TS cells are here, they mimic trophectoderm, derive extraembryonic ectoderm, and inside both structures, obviously, cavity is formed by that time. So, not all of the embryos, not all of these stem cell-based embryos form such beautiful structures, and this is one of the extremely interesting questions important for us is to understand when development fails, and we know that in natural circumstances also not all of the embryos develop properly. So, we try to use it as a model for embryo failure. I wish to say that ETX embryos are very similar to natural embryos, not only morphology as I'm showing here, but also on transcriptional level. I'm not showing this data because we published it now 3 years ago. They break symmetry and specify not only posterior but also anterior. But not all of them are perfect. So, we use this model— this is another study ongoing, which is not yet published, but I would like to show you an example of that study that we think can be taken care now— we can use to take care of specific question about self-organisation by using that model. So, here we see from cell seeding to day 5 how the structure forms. So, I'm running this movie. In black and white. So, you see here how the structure forms step by step. So, here three types of cells are together. It's a long process, takes 5 days to create new life when you don't start from the egg and sperm. Interestingly, it's also not the matter of days in the case of mouse embryo development when we start from, with sperm and egg, So, you see here, extralendrical structure outgrew the dish. So, now we have to sort of define the dishes of the better architecture to take care of those structures. And here we labelled cells, ES cells, T cells, and ZEN cells with different colours, and we follow those structures, how they self-organise themselves. So, we only now— and this is Min Bao in the lab at Caltech— is trying to understand the process of sorting of cells from each other and their consolidation and how they build the structures and try to define when the structures are formed properly, when they are not, when, for example, they can have two ES compartments in the opposite side of the embryo, two TS compartments, or Xen cells are not behaving properly. So, it's giving us incredibly rich insight into development of the embryo at that time. So, I was tempted to show you this data, it's not yet published, but decided not to because I would like to take remaining time to tell you what else we have done with those structures and published some unpublished data. So, one of the things that this structure is not accomplishing is, well, on one side I can say positive news, so it's breaking symmetry, the structure goes through the process called epithelial to mesenchymal transition to form mesoderm layer as you can see here. So, here we have Brachiori expression and this mesoderm-forming cells start to migrate between ESL compartment and ZEN cell compartment to initiate the process of mesoderm formation. They change their shape, so you can see this process initiating itself, but this process is not happening in a very good way. So it is started, but then it stops, and we wanted to really have a proper EMT, proper gastrulation. We wanted those structures behave like embryo do. So, be perfect. I'm just checking my time to not run over. So, what it is that we are missing here? And this made us realise that actually there is very little known in natural development, what are the essential prerequisites for gastrulation. So, we came back to look at the embryo at the time, and this has been the work of another two people in the lab that in Cambridge Neophytos and Christos, that were studying in the lab the process of natural remodelling of the embryo. And due to their work aimed to understand remodelling of extraembryonic structures, and they were the ones that found how the cavity comes together from extraembryonic and embryonic compartment through resolution of the hybrid rosette. During this project, and I'm saying it because it's very important to see how studying one project might lead to another project, which is maybe potentially even more important. They discovered that there are holes in the basal membrane, which we did not expect. So, look at these embryos here. These are the real embryos and these are just schematic to make this interpretation of this image on a short presentation easier. They found out that these very young embryos in the laminin that marks basal membrane are having this perforations. So, these perforations were present in the part of the laminin that surround the epiblast, but not in the part of the laminin that surrounds extraembryonic ectoderm. And with time, when embryo grows and the growth is extremely intensive, so you can see here in proportions just the growth within the 1.5 days, when you see the embryo growing, you see these perforations become redistributed from relatively evenly first position to one side of the embryo. So, we were wondering whether this one side of the embryo is either anterior or posterior to which this perforation becomes condensed to. So, for this we use a marker, anterior marker Cerberus, so the marker that I introduced at the very beginning. So, when you look at Cerberus expression and perforations, we found that in time initially when Cerberus is expressed on the distal tip of the embryo in the visceral endoderm on the distal tip, perforations are all over present but not very close to expression of Cerberus. But when it moves towards one side to set aside this part of the embryo for formation of anterior, those perforations are much stronger on the opposite side of the embryo where primitive streak will form. And indeed, we collaborated with Isabel Migot to use her very special transgenic mouse line in which this anterior signalling centre cannot move towards one side, and we saw that in this case perforations were will not become distributed. So, this anterior visceral endoderm, this anterior signalling centre directs the positioning of the perforations. So, what really happens? So, we know that one thing that happens is that anterior signalling centre is inhibiting nodal. So, nodal, let's say, is downstream of this anterior signalling centre. So, we were wondering whether nodal has anything to do with this. And to study this, we took nodal minus minus embryos and we cultured them in vitro and checked whether they have perforations and we found that they do not. And also, when we inhibit Nodal signalling in vitro by culturing embryos, natural embryos in vitro, we also find out that these perforations are never present when Nodal is inhibited. So, Nodal is clearly important for doing perforations, and this might also explain why Nodal knockout embryos cannot grow, because we need those perforations for the rapid growth of the epiblast that happens at the time. But what's downstream of NOTCH? And here the insight came from cancer field. So, we know that to remodel extracellular matrix to allow metastasis of cancer, in the case of malignant cancer, we knew matrix metalloproteinases. And we were wondering therefore whether also there are some matrix metalloproteinases that are expressed at the time that are maybe downstream of Nodal. And very rich datasets that exist in the field allow us to— some of our colleagues allow us to look at which MMPs are present at the time, and we identified as those, these two MMP2 and MMP14, and indeed confirm their expression protein level and they express exactly where Nodal is expressed and exactly where the perforations in the basal lamina are present. Present. So, how does it relate to gastrulation? Well, it turned out that basal maple perforations paved the ways for the cell ingression gastrulation. So, this really reminds us very much of this running a stocking model that first was proposed to be acting during lung morphogenesis by Don Ingber. So, essentially what happens is that these perforations become localised to the part of the embryo where mesoendoderm become specified and proliferate. And since this weakens the basal membrane, those cells now can push themselves through those perforations to initiate EMT and progression of the primitive streak. And indeed, for primitive streak to progress like this running and stocking, we need those perforation distribution, distribution of this perforation to be orchestrated in that way. So, well, so if this is true that we need for ETX embryos to have a proper gastrulation, we thought that we have to provide them with the proper AVE. So for that, we changed the design. So we changed the design, and now we are using KNOTZEN cells in the mixture of stem cells to make those embryo-like structures, but we use not only Y-type ES cells, but ES cells that would induce expression of transcription factor that is extremely important for primitive endoderm formation called GATA4. So now we, in that way, we thought we might be able to form very young type of visceral endoderm, only just initiating transition to the formation of primitive endoderm. And maybe, we thought, we were hoping for, able to build in this way much more functional anterior visceral endoderm signalling centre that will really allow us to specify the future head induction process. So, here is very similar design. So, you see how those cells come together. So, we have two types of ES cells, one Y-type, one inducing GATA4, Y-type in yellow, and TS cells to build this extraembryonic ectoderm compartment. And here's one example of those structures. So, they look very similar, they form three compartments very beautifully as the previous model system, and we called them induced ETX embryos or ET-induced Xen embryos. So, what those embryos can do? Well, they first can build anterior signalling centres. So, here you see development of the centre by looking at snapshots of the movies in which we follow ETX embryo and how the centre migrates towards one side. And this is the site where the future anterior will be specified, and on the opposite side gastrulation movements happen. So, here you see an example of the EMT on the posterior side of the embryo and proper execution of not only initiation of gastrulation but also its propagation. So, the question is how far those embryos can go, and that's really the question that we are trying to address right now. So, as a last slide with the data, I would like to show you this slide in which we analyse much more carefully those embryos. So, each single model was characterised on transcriptional level by single-cell transcriptional MEK analysis, but I did not go to the details of the studies. But here we analyse all of the models, and this was done in collaboration with a very talented student, Joe, in the lab of my friend in Cambridge, Florian Halford. Together. So, we collected the synthetic embryos, induced synthetic embryos, and natural and not induced ETX embryos, and embryos at different stages of development, natural 4-, 5-, and 6-day-old embryos. So, before gastrulation or gastrulating embryos. We dissociated them into individual cells and used encapsulation and in-drop sequencing to compare them on the transcriptional level. So, I would like to show you only one piece of data that relates to the centriole signalling centre, because it was our hypothesis that the younger type of visceral endoderm progenitors will make the centre. Indeed, we were right morphologically, but is it also reflected on gene expression levels? And yes, the good news is that indeed, when we look at this upregulate, at the genes that are upregulated in this visceral endoderm-like layer, many genes important for formation of that layer, including the genes important for formation of the centre are specified, such as OTX2, GUSCOITOMS, but also those anterior signalling markers such as DKK1, for example. However, in contrast, this Xen type of stem cells that we have used before in our prototype for this embryo, despite being morphologically amazing, they really struggle to form this anterior signalling centre. So, sometimes they did, but most of the time they don't because they are much more inclined to express genes that have parietal endoderm signatures, so that kind of more mature type of primitive endoderm derivative. So, this is the summary of, really very brief summary of my talk. So, what I wanted to say is that by uncovering self-organizational principles of natural embryos developing at preimplantation stages, but most importantly for this talk today, at implantation stages, we were able to use this knowledge to build the fully inclusive of three types of cell types that have to be included at the time stem cell-based embryo models. So, I didn't talk about preimplantation embryo models that we and others also built recently, but I focus on this more complex embryos and how we created them and what they taught us. So, first big lesson is about the mechanisms of polarization and lumenogenesis and importance of beta-intergreen signalling and importance of this remodelling for the transition from naive to prime state of pluripotency and how this transition regulates this remodelling of the embryo I showed you at the very beginning. This was really achieved by using mainly ES cells to model embryonic part of the embryo. So, then I've told you that we do not need this visceral endoderm and anterior signalling centre, they're not that naturally is present for symmetry-breaking events. So, this was not expected in the right place on the posterior. So, we can initiate mesoderm formation in a right fashion without that centre. However, when we put all three elements together to build the centre, we are able now to build the really proper-looking embryos that can undertake epithelial to mesenchymal transition in a right place that can progress. And yeah, the question is how far those embryos can develop. So these are the embryos that are really the hero of my talk today, not the embryos that are created in natural fashion that, of course, we know are stunning. So my final slide is shown here. I would like to thank every single person in my group, both in Cambridge and at Caltech, our very important collaborators. So many of the people contributed to this work, and I emphasize them during the talk, but from the people who I have here on the slide, Gianluca is the one who created the induced ETX system, and now he's working on it with Charlotte. We are also doing mouse— sorry, human embryo modeling studies, and I didn't have a chance to talk about them today. So here are very important collaborators who I've mentioned during the talk and were particularly important for that specific data I presented presented here, ex-Lab members, many of them set up their own group right now, and I call here those who contributed to the work that I presented today and our funding bodies. And thank you for listening.

So I will stop sharing right now and I'm ready to take questions. Excellent, thank you so much, Magda. That was very interesting. I've told my students that your work makes me want to be a student again or a postdoc. So, you know, I really appreciate it. Thank you, Carlo. I don't know if there's any questions. If not, I will ask something, but please either ask on the question and answer session or on the chat. So, okay, I guess I'll start. I guess I have kind of a big picture question. So there's obviously, there's a very complicated process of how all these things come together to give us a human or some organism in the end. And we've talked in previous talks, they talk about multi-scale modeling and different kinds of processes processes. How do you see modeling and experiments going together as we move forward in this area specifically, but in all areas in general?

Just a very big picture question. Yeah, it's a very, very good question. So I, you know, that's why I thought that I would today in the talk, despite that this work is not completed yet, presented some examples of the modeling that we are doing to understand these early stages of rosette formation. So how we go from the group of disorganized cells relatively not polarized in relation to each other, how those cells become polarized and communicate with each other to form embryo-like structure. And I think for this, the modeling is extremely influential in the way we shape our thoughts. So this will be an example. Another example of the modeling that we do, and I think that many people do as well, is trying to understand how three- types of cells in this case, because here when we have one type of cells, you can sort of understand maybe easier how they can organise in relation to each other because they are created at the same time points and initiate their differentiation in parallel. But if you put three different types of cells coming with different history and backgrounds, how they become organised with each other, that's also something that we try to model here in Caltech with Matt Thompson, in collaboration with Matt Thompson. So, these are examples of modelling studies that we try to use to understand how life is created when it is created at the very beginning of life. And I imagine that this modeling that we do and many other people do to understand different nature of self-organizational process at different stages of organ formation in organoid field will really all come together to understand those phenomenon. I think modeling is very important to put our thoughts together. And in many cases bring something unexpected that we haven't thought is going to shape our future in the way of understanding.

So excellent, thank you, thank you very much.

Um, I guess, uh, there's a question, Carlos. Yes, go ahead, Hannah, go ahead. Hi, this is Hannah Al-Samad. Hi, Magda, how are you? From UCSF. Welcome to California. Thank you. So, um, My question is, I mean, when I see these beautiful kind of ordered processes, I keep thinking about the resilience of these processes and the error tolerance. Can you give us an idea of how often these fail? I mean, let's say, and in vivo, not in vivo necessarily, in a dish as you're forming these rosettes and so on, what are the modalities of failure? How much noise can the system tolerate before it polarizes the wrong way or you get the wrong number of cells and so on?

Do we know? How much do we know about that? Yeah, excellent question. So, you know, that's what sort of I was mentioning a little bit that when we have one type of cells, so when we work only with mouse ear cells or human ear cells, the failure of rosette formation when we provide extracellular matrix is actually very small. Incredible, very small. So, those cells organise themselves very, very well. So, they pick up the signal from extracellular matrix on a basal— so, define basal side according to extracellular matrix. So, this is really laminin-mediated beta-integrin signalling that I expanded a little bit on this very recent publication that we have and that we now do modelling of. So, this noise here incredibly, is incredibly small, which was very surprising to us. So there is very little failure at that point. However, real embryo fail quite often at the time of implantation, in the time in their actually not only process. Okay, so I have to say that I didn't stress it, but this time of remodeling of our embryo, human embryo implantation, and mouse embryo that we use often to model this process, but actually this also we study in human embryos is very limited. So we have a certain number of hours to go through that process. If this process is delayed, those embryos will be aborted. And actually, it is believed that 30% of pregnancies failed exactly at that time. So we really think that, that this, you know, extremely fragile window of life that we have to get right, when we mimic it, when we provide all right signaling, surroundings created outside in the form of extracellular matrix that are chemically defined, and we know what exactly has to be there, this process is not failing very often. But it's enough that one of the tissues, extraembryonic tissues, is not producing those factors at the right moment or in right concentration, and then epiblast that will make the body has nothing to respond to properly, and those embryos will. This is also the time when these cells have to transit from one state of their mind, we call it naive state, naive pluripotency state, to the prime state of pluripotency.

Through—

actually, it's even more complex, I simplify it— through a formative state. So, this transitions between state of cell fate is regulating how those cells respond to those factors. So, those very simple cells are exposed to this influence of extracellular matrix, they have to transit into different states, make podocalaxin, perhaps many other things that we haven't yet discovered, to really organize themselves. And they have to do it not like cells in culture at any time point, because we are patient and we can follow them for hours, but at one particular time point of implantation. So that's the challenge. But your question has stimulated something else I wanted to say, Yes, so, okay, so this, this system works extremely well. But when we combine in the final model system that I presented 3 types of cells together, only around 20% of those structures look good. 80% are failures here, right? It's much more complex story because we have 3 actors that never seen each other before, and now you put them on stage together. Together, they make something, they assemble like a— what's first time I'm making this, how should I say, this symbolic comparison to actors, right, on a stage. But essentially, this is like if you bring 3 different people with different histories together and now ask them to converse with each other and build something, not just converse randomly, right, build something incredible. And this is much harder to achieve. So that's why only 20% of aggregates form proper embryo. But we now find conditions that are absolutely essential for many more of them to build natural embryos. And for example, cadherin coat is very important, we found. And here, modelling is extremely influential for us. So, we can model what should be the level of different type of cadherins that we are observing expressed in these different three tissue tissues and model the level of catheerin expression and see how quickly those cells now sort from each other and then start to form the structure.

Sorry, long answer to the very beautiful question, but thank you.

It is a beautiful answer. I learned so much from your answer. Thank you so much. We have two more questions.

I forget, I don't know if Christoph or Victor were first, but go ahead, Christoph. Yeah, Thanks.

Very amazing work.

I have a question going a little bit in the similar direction about the timing.

So how does the timing of development of your artificial structures compare with the natural timing? Very good question, and thank you for a nice comment. Why I'm reflecting, because we haven't really directly measure the timing side by side, but in vitro those cells take longer first to self-assemble. And I'm also reflecting on it because we didn't measure— because, you know, there are so many different processes that we observe here, right? So first, this rosette formation, the timing of this event in vitro, it takes 24 hours, and in vivo it will take around 24 hours. So here the timing is very similar. But when we think about 3 cells coming together for the first time to form the structure, it's difficult exactly to relate the timing of this event to natural timing, because in real life, all 3 types of cells are created by their mother egg progressively. So, you know, initial phase of life is just this big egg dividing to smaller, smaller cells, and then they start to differ from each other progressively at the 4-cell stage,, and even more at the 16-32. So, these 3 cells all the time coexist because they come with the one single progenitor. So, and then they start to self-organize, right? So, here we start in a very different principle. So, actually it's interesting to think how many ways we have to build a life. Here we take those stem cells for these 3 tissues and bring them together. Amazingly, within 24 hours they are sorted from each together. But in real life, they would be sorting for much longer because they are, as they are sorting, they are created as well. Does this make sense? So that's why I don't have a— it's a very good question, and I have to think a little bit more how I can compare different phases of that process with the natural phases of life. I would say the time is similar, but in one case we follow very similar event, in another case we create life in a totally different way.

There's another question from the Q&A that I will ask myself. It comes from Roberto Alonso Matilla. The question, the comment says, wonderful talk. Maybe I missed it. Does the stiffness of the basal lamina play a role in embryogenesis polarization?

Regarding the role of integrin beta-1, how much is due to biochemical signaling events and how much is due to mechanics? Okay, fantastic question, and we are just working on it right now. So stiffness does matter, and we know that the stiffness will matter also for culturing embryos. You might remember I've showed you something that few years ago we were trying to develop first system that we can culture human embryos from pre- to post-implantation stages. And of course, we use mouse embryos as a a prototype first to develop the system. Then here we found out that we have to— we cannot have too hard surface, we have to have a relatively soft surface for cells to like it and to carry on development. But we also think basal lamina stiffness matters, and this is the current project of my current postdoc, Joey, in Cambridge lab to really modify the elasticity of this basal membrane and to see how it affects development.

So, excellent question, I don't have an answer it, but I know it does matter. Perfect. And then, okay, I think that Victor will be the last question.

So we have a little bit of a break.

Go ahead, Victor.

Hi, very nice, very nice talk. Thank you. My question relates a bit with the work that I read from my colleagues in Scandinavia, where, you know, as physicists, they just put together some models and they show that, you know, just forces and elasticity is enough to get the patterns. Now my question is, when you do this, you put these 3 actors on the stage. Did you notice, I mean, is it important how you put them on the stage, where you actually place them to get a nice play, or is that a relevant thing?

Okay. Well, excellent question. I mean, by the way, you know, I just wish to say you are asking me such good questions and not many people ask me those questions, just that you know. So not always I have those sort of answers. Because I think a lot of questions relate here, right, to the, you know, physical, mechanical, and modeling of those. So if you would like to study this process of development, this biology, this profound questions in biology, join my lab. So find me and join my lab, or send your students and postdocs and colleagues, because clearly we have to do more of it, more of the modeling to understand. So this is what I want to say, which very rarely I'm saying, but I, I'm saying it today because your questions are different and we started to do this, but we have to do more. So, your question is where we put cells, whether it matters. Excellent question. You know, I was so sure when I started it. I started this process, this journey 5 years ago, and at that time I wanted to build blastocyst structure, you know, much simpler, because now we are building this more complex structures, right? And I put ES cells inside, TES cells outside, and you know, this was just really by design,. And those cells really sort themselves out, right, as they want. So we actually don't have to put them together in any particular configuration. I'm not saying that this design will not work one day, but I found out that it does not matter because ES cells, T cells, and ZEN cells, in this case the three actors, they express different cadherins, they have cadherin codes. So I didn't go into it, I was tempted at one point to show you this data, we haven't published this data yet, it's still under revision, but we found out that those cells, based on their cadherins, stiffness as well. So, actomyosin is important in one of those cells for sure. When we change the stiffness of ZEN cells, they either sort themselves out or not. So, this characteristic of those actors led them really identify the right position. But I started by designing them before I knew about this adhesion code, cadherin Wow, fantastic questions, and really I'm so pleased that you invited me to give this talk. First, when I found out that I have to give it at 6:30 in the morning yesterday, I was like, wow, I'm not going to be awake, but we changed it to 7:30 LA time, which is much better. I was more awake, and I became more awake even at the question session.

So thank you so much for fantastic input. Thank you, Magda, and thank you to all the speakers that came. Imagine that we're clapping for all of you, and I believe we'll take a