 And I'll say, you don't need an introduction because I don't have time to give you an introduction. Okay. Introduce yourself. Great. Everybody knows me, of course. Hi. I'm Jesse Major. It's fun to be here. And yeah, I guess I'm a good first talk because actually what I've prepared is sort of halfway between my own lab interests and I think the kinds of things you all think about. Just a really brief introduction on me. I've spent most of my career, I can't see that point over there, focusing on these early little things here to understand that complete mouse on the end. So these are embryos really during pre-implantation and gastrulation stages. That's what we work on quite extensively in a few different ways. What's nice about giving this talk is I certainly don't need to introduce this data to you all because it came from you. So there are two snapshots from Steve and from Henrik. Just again pointing out the high percentage of lethality you all find, about 94% roughly and of those about 57% are in this earliest category, so pre-9.5. And I was thinking about this last night and actually that means there's 13.5% of all genes are in this category and I don't know, maybe you guys know, is that the biggest category of phenotype? So good for me. Keep me in business. That's awesome. So this award really, we developed a sort of what we call a streamlined assessment of these early lethal phenotypes. And if many of you were here three years ago when I shared our strategy as well, it hasn't really changed because it's working quite well. And so we import heterozygous mice. Once the centers tell us about some early alleles, we establish small colonies and we start our analysis at day 7.5 and we simply ask, can we find the mutant? And if so, what does it look like? Sometimes we can find them and they look normal and we'll look a day or two later, eight and a half, nine and a half, do different types of analyses. Often we will find them and they're clearly abnormal. So then we do analysis at 7.5 and 6.5. And as you'll see, the other large group that we get is that we cannot find them at 7.5. So we back up and we flash Uder Eye and we ask, can we find them at day 3.5? And again, what do they look like? What kind of analysis can we do there? So here's my Hollywood Squares version of our lines, right? It's progressed quite a bit since three years ago. It's just a 7.5 dissection from each of the lines we're working with. Obviously I don't expect you to get anything from that, but to me it's fun to look at this way, growing some nice wallpaper. We're up to about 114 lines that we've analyzed in this way. It's been sort of a steady strategy for the last four and a half years. We've got another half year left to go on this award. Actually yesterday my renewal landed somewhere nearby here at NIH, so fingers crossed we'll keep doing this work. And what I thought I'd do today is sort of show you the kind of data we're producing for each line that we're actually getting back to IMCC and some other databases. And then I'll tell you a few short stories of phenotypes we're working on in detail. And then I'll end with some just thoughts about this whole collection and what we can learn looking across all these lines. So here's an example of a gas-relation stage phenotype. It's sort of a composite data set. You can see on the left a whole mount image of a littermate and the mutant down below. I'm showing you munifluorescence after section and embedding for a few different lineage markers along with H&E. And we've developed a strategy to do multiple rounds of IF and then H&E on the same slides. So we get a lot of data out of the same embryos. That's just what's shown here. We're also providing RT-PCR expression data. You can't probably read the labels there, but they're just stages during embryogenesis. And when available, about 30% of our lines have a laxie reporter. So we're also giving back this laxie expression. And so this is the sort of data set we're compiling for each line at gas-relation. If we can't find them there again in this other large set we have, we can find at three and a half. So here's a nice mutant gene POLD2. A nice mutant blastis looks perfectly normal. So when we find those, we do these outgrowth assays. So we can grow them for a few days in culture and ask, can they make a typical outgrowth here with an inner cell mass and trafectoderm growing on the plate? Clearly you can see this mutant doesn't quite do that. Basically just shrivels up and dies. Nonetheless, we do these assays and again we're providing all of this data for each of these lines. I'm happy to say, as you heard from Henry yesterday, there are now links up on IMPC. We're going to tweak that in different ways as well. I've also been in touch with MGI, Kent and the UCD crew, and GeneCards as well. We're going to provide links to our own website where we're starting to host this just to get this data out to the community. That was really part of my original proposal and the goal of looking at these alleles that are sort of getting left behind. So that's the baseline data we're providing on every line. So as I said, I'll take you through a few short stories and then think about some larger ideas. We find a few, only 4% of our lines actually make it through gastrulation. And there are organogenesis defects. We published a couple of these. I'm not going to tell you about those today. I'll tell you one story we're developing from this group. So 49% of our lines, and this is the biggest group we get, we call gastrulation failures. And I'll show you what those look like. And I'll tell you about this sort of family that we're working on. They are mitochondrial ribosomal proteins, or MRPs. And we have a disproportionately high number of these in our set. But again, we didn't select the genes, so I'm not sure how that came about. Maybe one of you has an idea. I'm going to show you some data on these five lines. The names are here, MRPL3, 22, 44, S18C, and S22. Some are laxate containing, and some are endonuclease mediated. You're all familiar with these snapshots. What's interesting about them for us also is they all have pretty much all of them have heterozygous phenotypes as well that came out of your groups. And they also pretty much all have human disease associations as well. So that adds some relevance and significance. And just as a reminder, these are all genes in the nucleus that are transcribed and translated and then imported into the mitochondria where they function as part of this mito ribosome, which is responsible for translation of the mitochondrial-encoded genome. And there's really two subunits of this mito ribosome, the large and the small, hence the MRPL and the MRPS. There's about 70 total of these genes. So what do the knockouts look like? Well, here's a littermate for each one above, the homozygous mutant below. So this dissections are at 7.5, where typical embryos have a pretty expanded primitive streak, a node, early headfold, different morphological features, which are all absent in each of these mutants. Hopefully you can see they've formed a really nice pristine egg cylinder and they're really just sitting there. And I'll show you some more analysis in a minute. And it's a pretty very consistent phenotype, certainly with all five of these. And in this large category in general, this is what we find. So we've done a bunch of expression analysis. I won't talk through all the details just to say that most of these genes are pretty ubiquitously expressed. We were kind of imagining we might see some tissue-specific or temporal-specific expression, which could explain why they each have a similar phenotype. We were imagining in this big family there'd be some redundancy, but we actually think there's not. We're starting to look more carefully at the amino acid sequences as well. And there's really not a lot of functional redundancy, we think. So widespread expression by RT and embryos and adults. This is true both either with the laxie reporters or what we do in C2s. There is some embryonic-specific expression versus absence in the ectomebreonic. You can see down here in S22. But for the most part, they're very widely expressed. So if we perform our standard immunofluorescence assays, it's a little hard to see. But again, these are the same sections with multiple rounds of IF. There's a littermate on the top and then each mutant below in a row of four images. And what I'm showing you here is first, all the way on the left of each set is T. And I think that's the hardest to see. This is brachyury, which marks cells of the primitive streak, which the littermate, I think you probably cannot see it, has nice nuclear T staining in all the cells of the streak. And all of the mutants have absolutely zero T-positive cells, indicating that gastrulation really has not started. We've also marked them here with oct4, which as you probably know, is in the epiblast at pre-streak stages. And that's really what we're seeing, is that in each of the mutants, they have really nice oct4 signal indicating that the epiblast cells are pretty normal and they're really just hanging out, not much going on. We also look at proliferation and cell death with different IF markers. PH3 on the left side in white indicates that there are still some dividing cells in these mutants, and we see actually very little cell death, either with P53 or active caspase. So just further indicating that, yeah, they're just hanging out. Again, this is two days after we think these embryos have stopped growing. So why are they just sitting there? What's going on? We want to really understand sort of, hopefully mechanistically, what's happening in these null cells. So we start to ask about the mitochondria, of course. Are they all there? Are we losing mitochondria? One way we've looked at this is by measuring the amount of mitochondrial DNA relative to the amount of genomic DNA. It's a little sort of busy graph, but each line is shown separately with littermates, both wild type N-hets on the left and then mutants on the right for each of these knockouts and then a composite of all of them. Long story short, there's really no change in the ratio of mitochondrial DNA to genomic DNA. Conclusion, the mitochondria are still there in the mutant cells and we see this by IF as well in different ways. But what do they actually look like? So here's some EM showing controls on the left and then four of the different mutants here on the right. And hopefully you can appreciate while the wild types have these pretty dense mitochondria with lots of inner Christi layers, in each of the mutants we see very abnormal mitochondrial morphology with really these big holes in mitochondria. This has been documented before, mostly in cell lines, indicating very significant mitochondrial dysfunction or not functioning at all. Nice mood lighting, thank you. So morphology are there, they don't look very well. Are they producing any ATP, which is really what they're meant to do? My student devised a way to take one of these cell-based assays and basically we drop a whole live embryo, one into each tube. The assay works just great even though there's pretty small cell numbers. And so we're able to get a ratio of ADP to ATP and that's what I'm showing you here in this similarly laid out graph. And hopefully what you'll see along with me is that in each case the mutants have a higher ratio, quite significantly higher ratio of ADP to ATP, telling us that there's really hardly any ATP in these cells. So mitochondria are there, they look terrible and they're not performing well. Maybe not so surprising. So it's giving us some hints into maybe what's going on and why these embryos are so nice and pristine. And we think it's partly because when you have dysfunctional mitochondria often what will happen is you'll get elevated cytochrome C levels which can activate different cell death pathways including caspase. But this process is ATP dependent leading to apoptosis. But I just showed you we have no ATP. So oddly these cells are really, should be dying, but they're not. And that's why they're in such good shape when we dissect them and we don't just find little knots of dead tissue. So we're trying to divide some experiments to get at that and try and maybe rescue this phenotype by adding ATP back either in culture or in vivo to see if we can rescue primitive streak formation which would be a long shot. But my guess is we're going to add ATP and we're really going to rescue cell death if that makes sense. So we're imagining we'll add ATP, then they'll start to die and then we'll know really that's why these embryos are surviving as long as they are. So that's where this story's at. We're going to develop it a bit more. Look for that in the near future hopefully to come out. As I said, that's our largest group of mutants. This is gastrulation failures. And the other large group that we find are those that cannot be found, 7.5. So we back up and we find that we can collect them a day three and a half and these are simply termed implantation failures. So here's a typical litter. It's a rather large one. Nice glasses morphology. And I've masked their identity so we can play a little guessing game. So who can spot the mutants? This is a head-by-head cross for this gene med-20. Anybody see any mutants up there? Cat left I think, right? She would see them. If you're like me, you would say, oh look, these are on the bottom here. Those are the mutants, obviously. Not so, right? So those three are a wild type, a hat and a hat. Even though they should be glasses, they're not. There's a different issue which if I have time, I'll address at the end about this background of mouse. But here's a mutant here. Here's a mutant here and here. They look perfectly normal. So I think myself or any other embryologist will be pretty hard-pressed to look at these and see anything wrong. But we know they're not implanting, because we can't find them at seven and a half. And they're not empty-decidual either. So they're really not having a decidual reaction at all. So what happens? Waikui here in my lab took these embryos, did our standard outgrowth assays. You can see a wild type, a hat and a mutant. In this case, these med-20 homozygous mutants are actually failing to hatch from the zona pollucida. So you can't quite see it probably on the screen, but it's still enclosed in the zona. Cells aren't quite dead, but it's not hatching. So obviously if you can't hatch, it's not going to implant. Well, we do some molecular analysis of linear specification. So here's the first linear decision, which really segregates inner cell mass and trafectoderm. We're marking it with oct4 and CDX2. A wild type, a hat and a mutant down below. The first decision seems totally fine. Nice oct4 localization, all CDX2 positive cells on the outside. And no cell death going on here either. If we look at this second decision, same layout of embryos, where we're now segregating primitive endoderm and epiblast, we can mark the primitive endoderm with SOC17 and epiblast with nanog in red. And hopefully you can appreciate SOC17 looks fine, but in the mutant, we have nanog widely expressed where it should not be, indicating there's really defective trafectoderm specification going on in these homozygous mutants. It's a little surprising to have this linear-specific effect because med20 is expressed everywhere, mediator complex, presumably is ubiquitous, and acts everywhere. I'm guessing you're all familiar with that. So we've taken this analysis a bit further. I won't tell you so much about it. So this is the type of analysis we're doing on many of these pre-implantation lines and finding very similar stories where the blasts look quite good, but they fail to implant, and they all do have some kind of molecular defects that we can identify. So if you've had enough coffee this morning, you realize we're not quite at 100% yet. So there's a couple other stages that we find that I haven't introduced yet, and we'll switch gears and I'll just show you in a more pie chart format, where we have really only a very few that have early cleavage stage arrests, just a couple real early ones, a handful that make Moira, but don't compact. We think this is probably low in numbers because most of these genes are maternally expressed as well, and they probably make it through this early period with the maternal loading. I did tell you about this next group that we find nice blasts, but they fail in outgrowth assays, and we've documented both defects in the inner cell mass and the trifectoderm that lead to this failure. We have equal numbers of mutants where we find nice blasts, and they make a perfectly normal outgrowth. Those are a little harder to explain why we can't find those at 7.5, but again, when we look, and we've looked at about half of them, we do indeed find molecular defects in one or both lineages, which we think explains why in vivo, they can't perform normally, but in vitro assays, they actually look okay. There's just a couple that implant, but don't form an egg cylinder. This is also another surprise. I thought this would be a large category where they didn't plant. We'd see the decidua, but you'd have basically a dying embryo, but we really only have two of those, and the biggest category is this gastrulation failure, of our homozygous mutant embryos, just stalled at this pre-gastrulation stage. This is a wide variety of gene function. It's not all mitochondrial-related genes. As I said at the beginning, there's a handful that make it through that we've also been characterizing. That's sort of the frequency of our early lethal phenotypes. We can think about this in a few different ways. That's the wrong direction. Sorry. We can start to do different things. We can line them up. Here's a handful of these gastrulation failures. You can see they're all roughly similar size. There are some differences, certainly. Again, these are all different gene functions. At first, we started doing a lot of cell counting to see if there's some threshold of cell number. We quickly realized there were issues with the validity of counting cells, but measuring the size in these embryonic stages anyway is a really good surrogate for cell number. We've done all kinds of measurements. I'll just show you one. From the distal tip to the boundary of the ectoplasental cone. If we plot out normal embryos, I'm showing you pre-streak all the way on the left, 6.0, mid-streak, 6.5, and then 7.5 node stage embryos, you can see this nice growth of the embryo. If we add to this graph each of our different mutant lines, I'm showing you them separately in vertical groups, you can appreciate that they all are roughly the same size. You can see that we're going before the point at which you normally get to when gastrulation starts. We think this is really supporting this notion from Patrick Tam and others that there is some critical cell number that you need to reach in order to start gastrulating. We're trying to devise experiments using some of these mutants to try and get at that question. Is that really true? A little tricky thing to sort out, but we have some ideas about how to get there. These ones where the mutants make blesses but fail to implant, they're all the same size as normal ones. Here's wild types of various stages of blesses on the left, and here's all the mutants in a similar type analysis. No overall size defect or cell number defect. We actually have good cell number counting here because it's easier to do. In general, these mutants don't have reduction in cell number, but as I indicated, they clearly have lineage defects in these early failures. We can look at our data in a slightly different way. Answering the question that I think was up on one of the earlier slides, will we see a widespread and even spread of phenotypes or are there specific developmental constraints? I think it's not that surprising, but our data is clearly showing that indeed there are these two major bottlenecks of phenotypes, implantation and gastrulation, which I think if I had asked you all before I showed this, we would have said the same thing. But to my knowledge, this is the first time where we have this big collection of mutants from the same genetic background made largely by the same people in the same ways where we can really confidently state that's the case. We also think our data is telling us that this early failure is due to molecular misregulation in some way and the later failure really ties back into cell cycle. We think all these mutants somehow will tie into proliferation defects that are stalling and not continuing in development. As you all know, there's a million ways to take gene sets and do different types of analysis. What's nice, I think we're reaching a threshold by which this is useful for our gene sets as well. So if I take all of our gastrulation phenotypes and I feed them into a gene regulatory network, here's the largest node I get out of there and what I've put on the right side or on the right side are all the genes in this node are not in our collection. And then I've color coded them red that have been published as early lethal also, green that are viable and black that are no data. And it's pretty good. We could almost predict in most of these genes in this node, indeed, do have an early lethal phenotype. This is even more accurate for our earlier set and I won't show you that slide, but the largest node there, we get 100% accuracy where we could also predict almost which genes will have a phenotype. So as we increase our numbers of actual data, I think this will become more powerful. So I'll just end by admitting that I really love doing this work and I've done a majority of the dissections myself which keeps me in the lab and it's really fun and I'm looking forward to the next few years and really making some really good Where's Waldo type examples here and I'll just thank the people involved. Here's my lab members, my main collaborators and of course you all for producing the mice and NIH for letting me do the work. I'm happy to take some questions. Thanks. Yeah. That's a tour de force Jesse. Are you going to put this into one publication, this whole story? So yes, we're just starting to put that together. Okay. And yeah, I got lots of ideas yesterday chatting with a few people about how to enhance this story, thinking about human stuff and a whole bunch of other ideas. That's my second question. Have you looked and I know I've tried to do this and it's difficult. Have you looked at fertility panels and of course childhood development diseases or overlap? So a few years I thought I found someone who could do that for me because you're right, it's very challenging and I don't have any good data there. Just, you know, a few here and there that there are mutants known, but not really cool. Well, keep in touch as the paper goes along because we'd love to help you promote it because that's a really nice story. Thank you. That was fantastic. Can you tell me about the fidelity of the outgrowth assay? I mean, how reliable is it with you, if you're looking at a group like you showed us, you're blinded, you look at a group of E3.5, some look normally at their mutants. And you take those to the outgrowth assay, how well can you rely on that assay to be absolutely definitive? We've done probably thousands, so before this project I did an RNA ice cream using that assay as well. So in our hands, we find it really quite reliable and there's maybe 5% of the time the wild types will fail, not very high. And so it's a little bit of a numbers game, but again, we're genotyping each one after. So even though we might have some wild type or heterozygous failures, what we ask is, we look at all the mutants and say, oh look, they've all failed. Maybe what you're also asking, we've only had maybe two cases where it's not a clear phenotype. Some of the mutants fail, some don't. So in answer, we trust it quite a bit. So beautiful story indeed. Thank you. I will have one practical question also from my experience. How do you control the influence of maternally deposited mRNA? Because for example, we experience in one project that basically we saw something like gastrulation arrest and when we did electroporation of the SIRNA into zygote, then we basically see much worse phenotype basically affecting already the cleavage stage. You're absolutely right. Short answer is we don't. So we're just looking at the homozygous null. There's a few stories we're following up where we also do microinjection, so we compare the knockdown and the knockout phenotype. Sometimes it matches really well, sometimes it doesn't. The only thing you can do is design a conditional experiment if you really want to follow up. Actually can I have one more minute? Is that okay? Are we on time? I forgot to put this slide in the right order. There's another observation that I'm pretty interested about, that I'd love your input as well. So these are all natural plugs that we're dealing with, so no hormones. And there's a few different ways we now have a lot of data to look at this and one thing that jumped out of me that's interesting is the pregnancy rate. So here's a graph showing two different time points, 3.5 and 7.5 and the pregnancy rate. This is just a small subset of our litter, 300 or so, in each. And so at day 3.5, if you get a plug, you're 87% of the time going to find embryos. At day 7.5, you're only going to find embryos 76% of the time. So that's a pretty significant difference. So 11% change. Which means female mice are ovulating, they're fertilizing, and then the whole litter is failing to implant. And so we're pretty excited that this is a possible model for human implantation failure, which we believe is sort of the largest embryo failure that happens. And we're trying to devise clever ways to try and use this as a model for that. The other observation we've made is with these extra embryos that I mentioned in that pre-implantation litter. So if we look at our own breeding results, all the way on the right of this other graph, P21, weanlings, I think we're right in line with I think what you all find, around 7 pups per litter. If we look backwards in time, the number of embryos we get, 7.5, 3.5, 0.5, so just look at the blue line. You can see there's this pretty significant attrition. So if I collect zygotes, I'll see about 10 fertilizations. If I collect 3.5, there's only about 8 blasts. And these are all now wild-type bi-hat crosses. So take out the null issue. If I look at 7.5, we've now lost another embryo on the way. And that's what the orange line is indicating, including empty deciduous. So that's a really hefty attrition. It's 30%. And so to me, that's a pretty interesting biological question also. Like why is this happening? And so I'm thinking about trying to get some data on other backgrounds to see if that's happening. I'm guessing, like CD1, it's not happening because they're not. And again, I'm just thinking about this as an interesting normal development question. And if you all have insights or ideas about how to try and get at this question, I'd love to hear it as well. Yeah, Rob. I'd like to see on that last point, it would be really interesting to read you a chunk of this study with F1s to see if it's just loss of vigor. Sure. So I had two quick questions for you. One is, have you ever tried a mitochondrial translation inhibitor to see if you can recapitulate those mitochondrial ribosome knockout phenotypes? That's a great idea. We thought about that. The hard part is how. So how do you deliver it to the embryos at the right time? You can't just throw the drug on earlier. So you can only culture embryos for so long. And depending who you ask, if they're normal during development or not, even without a drug is questionable. So yeah, we could try and culture five and a half for a day with an inhibitor. Yeah, we can do things like that. It might be tough. I was also curious about the structures that you showed, the mitochondrial structures. Have you been able to identify the molecular composition of those? Not yet. It's clearly not my area of expertise and we're hoping you want to do it. We're hoping to find someone who can do it. Yeah. Okay, great. Thank you. I think I've used up my time.