 All right, thanks for having me here. And it's a really interesting meeting. And so I worked with Jackson Lab to establish and improve the technology of using CRISPR to make mouse model. And my main lab actually is in Institute of Zoology, Chinese Academy of Sciences in Beijing. In there, we focus on making large animal models, as well as therapeutic applications. So when we think about precision medicine or the topic of this conference, I think it's really what we want to know is by looking at the genomic sequence of the different patients, we can identify the genetic variants that have something to do with the disease development. Therefore, we can develop therapeutics for them specifically. And another layer of complexity is the epigenetic modification and the misregulation of the genes. So for those two layers, we need better technologies to model them in the organism or in different assays. So genome editing obviously is a very powerful tool. And those are the major tools. And CRISPR-Cas, obviously, is the most powerful one. So what they do is basically you can design those protein to bind to a very specific locus in the genome and make a double-strand break that can be repaired by different repair pathways that result in annulio or you can put in pretty fun mutation there. And the reason CRISPR is so powerful is different from the other two, you don't have to design a new protein for each locus. You just design and express a small RNA that, based on the base pairing of the RNA and the target DNA, you can guide the Cas9 that go to any place to make the gene modification. So that case of work really accumulated to this landmark paper that demonstrated you can basically express those two RNA that recognizing any DNA template and generate this double-strand break precisely. And followed quickly by those two papers showing optimization of the system can be useful in the mammalian cells. And what we did then is to show you can, in the mouse embryo and stem cells, you can simultaneously now call five genes with a very high efficiency. And that kind of useful for making cell models. Also, by introducing the system into Zygote, the enzymes start to add in the genome in the very early embryos. Therefore, you generated a mouse with the mutant in one generation. And you can actually do this by now called multiple genes at one step as well as putting in multiple specific point mutations to different genes in one step. I want to point you to the supplemental data actually, which is we tried to now call type three. And the gene mapping is very efficient. So this band is a mutant band. This is a wild type. So you can see all the embryos are kind of mutated. And we already see the phenotype in F0 animals. So when you inject them day one in day 21, you see those pups showing the neonatal lethality phenotype. So that means you can actually use this approach to screen a large number of genes to see their phenotype with a very small number of animals in a very short amount of time. And in the following study we showed by introducing different type of DNA template, you can repair the break by introducing epitope tags or reporter genes as well as LOXP sites. So without that setup, the system is comparing to traditional gene targeting and making mouse model. The cost is dramatically reduced. And more importantly, the time you save is most significant. And also it's very flexible. You can theoretically work on any string you want without have to do this first in the good MRSM cell string and then stream algorithm. So still the one bottleneck of this whole procedure that does not allow high throughput is really the micro injection process. You need a very expensive setup and also the most limiting factor is really to find a very good micro injectionist can do this very, very efficiently and consistently. So to solve this issue, we developed this zygote laceration of nucleus technology. So basically, we treat the embryos as any other cells we equate in the labs. We just put them into a QVAT, equate them in the commercial available equator. And that allows you to deliver the CRISPR region uniformly to hundreds of embryos in the same QVAT simultaneously. And after optimization, this is one example. Actually it's one of the rather difficult locus we are targeting. So here we are introducing two restriction sites to this locus. And so if it's targeted correctly with the precise palmitation, you would digest this PCR product with either enzymes. So that's exactly what you see. So you can actually achieve a really 80% and 90% of efficiency to guiding your funder animal by just select reading them. So with that, I think we are really now can do this with a high throughput manner. You have one technician knows how to handle embryos. They can quickly go through 20, 30, 50 projects in one day. And then you can just transfer them and do your phenotyping. So I think with those, we can quickly screen candidate genes in funder animals that either it's from multiple genes that GWAS hit or it's many new genes you found from axonome sequencing. And also you can generate a human genetic variant in the mouse orthologue by using a single oligo as a template by the equation. So in addition to genome editing making mouse models, I think the Cas9 is also extremely versatile protein. You basically they have two independent nucleus domains. So if you mutate either one of them, it become a NIC case that only nicking one strand of the DNA. And if you mutate both of them, they become a reprogrammable DNA binding protein. They can guide it by RNA and binding anywhere you want but doesn't cut. So by doing that, you can actually fuse a protein that have any function to do a very specific almost anything to anywhere in the genome. So we basically develop a system called CRISPR-ON. So that's a very simple idea. You basically make a fuse or a D-Cas9 with a very strong activator. This is a 10 copy of the VP16. And we show that you can actually bond to a specific promoter and activate the indulgence gene very efficiently. And you can activate three indulgence genes simultaneously. You can also control the reshow of their activation by controlling the amount of RNA you're expressing. And also we showed if you introduce a system into one cell zygote, you can actually turn on the reporter you co-injected in the early embryo. So you can do this potentially in vivo. But the system is still not perfect if you think about the complex disease or complex transcription network because if now I want to activate certain genes and repress certain genes, you cannot really do this in the same cells because if you co-express those effectors, they don't know which gene they're gonna repress and which gene they're gonna activate. So you have to have a mechanism to do this in a multiplex way. So the solution we thought of is this combination of this Cas9 protein and a familial protein. So this protein is really interesting. It's an RNA-binding protein. So it's like a tail if you're familiar with the tail end technology. It's like a tail. So they have this repetitive domain, each domain recognizing one nucleotide of the RNA. So now you can do is you can engineer this domain to recognize a specific A-base part of the RNA. Now you can have this RNA, A-base part binding sites linked to the end of the guide RNA. So now you have a molecule that have the... Okay, so the specific binding sites will recruit a specific flavor of this puff protein fused with a specific effector. So then with this single RNA molecule, you combine the information of their function and their target on the same molecule. So now you can do this with multiplex manipulation. So the first thing we tried is by just simply adding up to 47 copy of the binding site, A-base part binding sites, that doesn't interfere with the CRISPR system. So this is not fused and those are fused with different copy number. They perform equally well. And then the real experiment is we express the system together with different flavor, different puff protein and fused with the VP64. And we express, say the puff A with four different, with the guide RNA fused with four different binding sites. And they only activate when you express the guide RNA that have the puff A binding sites. So that means it doesn't activate in other sequences. That means the system is very specific and they can work independently with each other. So we showed, go ahead and this system can be used to activate in the audience gene. And because we are not recruiting many copy of the factor, not just one copy by direct fusion, you can activate the genes much more efficiently. So this is the, this is the cassilio activation. This is the direct fusion activation. So this is in OCT4 and SOC2. And more importantly, the experiments to show, you can simultaneously activating one gene and downregulating the other gene. So that's basically proved the concept that you can actually have this information. The puff A was recruited to the gene they want to activate because the puff A fused with activator and the puff C, I think, or W, whatever. A recruited to the gene they want to repress and with the puff B protein fused with the repressor. And more than activating the promoter, we showed you can, the system can recruiting histone modifier into the enhancer and therefore manipulating the gene expression from binding to the enhancer and modified histone acetylation. So here we showed you can use the cassilio system to binding to the enhancer of OCT4. And that actually interestingly activating the gene more efficiently than the direct fusion of the decasinan and the histone acetyl transferase fusion. And one last thing is really, I think it's very interesting to study the chromosome structure because that's another layer of epigenetic regulation. And here we show because you can recruiting many protein to one locus, you can get better signal to noise when you label a specific genomic locus. So we show that when you use more copy number of the binding sites, you get a better labeling of the telomere when you use the Jeff P to label it. And there's a quantification showing when you use more copies, you got more signal to noise and also a better number of foci being identified that's closer to you would predict. And by co-standing, we show the labeling is specific. It's co-localized with telomere associated protein. And the same thing can be shown for labeling centromere. And obviously the system can work simultaneously so you can co-label two genomic loci with two different colors simultaneously. And because this optimization feature, we can now use much small number of guide RNA to label unique sequence. I think that's a goal. And we do observe a very nice labeling of unique sequences, but they still need to be fun tuned to really work well. So with this platform, I think it's very versatile. You can use this to do multiplex function and directly multiple gene. You can recruit multiple protein to a specific locus to labeling or do other function. And the other thing we want to explore is to really use this to recruit a complex that can work in a synergistic way. So with that, I think we can potentially in the future to model epigenetic abnormality as well as gene regulation networks. So one last thing I want to raise, just it's a question really. Because we think about from the bench to the bedside. Other than diagnostics, I think a very important thing is therapeutics, especially for a lot of those medialine disease, we have strong, strong evidence of this gene. This gene is a causal gene and this mutation causing the disease. So now we have the tool to really go there and crack that endogenous gene. That would be the, I think, the fundamental solution of those disease. And should we actually think about, identify a list of actionable disease and genes that is really, it's really bad. So when you see this in a small kid, you want to act on it right away. And will there be a mechanism we can develop to a relatively cost effective model for the therapeutic development? Because each patient, most likely gonna be unique. They're gonna have different, maybe the same gene, but different mutation. How do you develop a therapeutic approach that can cure most of the patients? And whether we will have a mechanism to faster approve those therapeutics for some of really devastating disease. I think it's, we have the tools, people are dying. I think people should work on it. And also how do we evaluate the risk and benefit? Because when you do genome editing, I think nobody can assure there's no off target whatsoever. Nobody can ever be sure. You don't make any other genetic mutation. So if it's, if the modified cell is gonna unit our body for decades, then there is a chance to develop cancer from it. How do you know it's not from genome editing, but from some editing mutation? So I think it's really dependent on the disease and how devastating that is. But this is something to be considered. And even one step further is how do we define the norm of the human genome and what is normal human? What is normal genome? And where we draw the line to make that modification? If the risk allele is worth editing or not, and do we go to germline at all? I think that's an interesting question. And it become a reality. I think we people should think about it. So with that, I'd like to thank a lot of people involved in this for really, it's a great collaboration within JAX, with the GT group. There are a lot of key people contributed to developing the Zing and improving it, as well as the reproductive science group and my lab there. And also have a set up in Beijing that I'd like to thank my director there and the fundings. All right, thank you. Thank you very much. Questions? Once you've picked your job. Dan? Yeah. With this, the first technology you can overexpress, you can really drive up expression of any gene to sort of massive levels in a, so is that, is, I don't usually think about increasing gene expression by 20,000 fold, which is what you can do with this technology, is that? I think that's artifact because it really depends on the basal level of the gene, right? Because if it's not, if it's zero, then you can infinitely increase it. Right, right. So, right. My question is, have you been able to think about or see deleterious effects of that kind of overexpression in, across cell lines or across genes? We struggle with the idea of, what would happen if you increased the number of sodium channels in a heart cell, not one-fold or two-fold but 25-fold? Would that be a bad thing or not? And can a cell, would a cell ever do that? I think that's really interesting. I think there's so much unknown there and also think about the azoforms. With some very interesting thing about when you activate the gene a few hundred fold and the azoform totally gets skewed, right? And there is all unexplored and I think it's very interesting to look at. No answer. But it doesn't, like we activate all four, for hundreds of folding T cells and that does have activating other genes unrelated. I think it's through the secondary fact. So I have two, one question, one comment. So the imaging that you did of the nuclei, is that in an amplified target situation or could you use that to visualize the presence or absence of a single nucleotide change? Oh. I mean, could it be a fish for a single nucleotide change? I don't think so because right now what we showed most of the data are in telomere or centromere, there's still a lot of repeats. And the single, the unique sequence, you still need more than five guide RNA to towel it. So I think we're not there yet to show that type of resolution. And in terms of potential catastrophic disease targets, I would suggest IPEX, immunodisregulation, polyendocrinopathy, enteropathy, X-linked, which babies are born with a severe autoimmune disease. There is a treatment, which is bone marrow transplant, but they're often not well enough to get the transplant initially and have a very difficult course initially. And one could think about trying to use this to correct a very small fraction of T cells, which is all that you need actually to reverse the autoimmune disease. Yes, we're actually doing a lot of work in the T cells and C-34 cells. So I'll write down the name of this after. Any other specific questions?