 Anthony Rangvo, visiting us from Yale University. Anthony was born and trained in Belgium and trained at Brussels and started out in an laboratory of immunology lab, but studying really metabolism of pre-B cells and B cell development. And during that work he described the role of the NAD pathway in promoting early B cell and lymphocyte development. He went from there to Rich Vell's lab in around 2006. And there he started working on two projects which have really most recently really been coming to fruition. One was a project studying innate immune recognition of signals and in particular looked at the way why apoptosis is a non-inflammatory death. Why cells can die by apoptosis without inducing much immunity. And came across the very interesting finding, again in Rich Vell's lab, but with a knockout a month. And now I guess with CRISPR-Cas9, it's about a knockout or a knock in a week. But came across the identification that genes in the caspase pathway were associated with preventing mitochondrial DNA that's released during cell death in the cytoplasm from being recognized by the STING pathway and resulting in inflammation. This was a really impressive paper published recently in Cell, which I'm sure many of you have seen. And I suspect he'll give us some insight into that work. But part of what he has been doing and what Richard's lab has been doing now for well more than a decade is trying to develop humanized mice that can be used for studying innate immune recognition. And it's turned out that that's been a particular difficult problem in terms of the failure of the innate immune system to develop in most of the humanized mice. And through a sequential series of knock-ins, which started with the thrombopoietin and now is advancing, probably one of the most useful has been the knock-in or the transgene of human serpalp, which is preventing engulfment of some of the innate immune cells. He has now a mouse model with essentially five genes, five human genes added now that really allows the study of the human innate immune system, as well as the matipordic system in mice. And I'm quite sure he's going to give us a description of that. So it's a real pleasure, Anthony, to have you. Thank you very much for this introduction and for this opportunity to present my work here. So I'm going to present today two projects, one using mouse models and the other one on humanized mice. At first, I would like to start very briefly with a description of what innate immunity and inflammatory response are. So inflammation or innate immunity can be induced in response of a number of stresses, such as infection, injury, or different kinds of tissue stress, or malfunction. And the response induced by that is the goal of restoring tissue homeostasis. And when everything works fine, this is accomplished by producing antimicrobial factors, inducing tissue repair, and triggering an adaptive immunity, mediated by T cells and B cells, that's going to develop also a memory response for future infections. However, this is also a dangerous response. If it's not properly controlled, it can result in pathology, such as sepsis, autoimmunity, or fibrosis. So this shows that it's really important to understand the cellular and the molecular mechanisms that regulate this balance between the physiological and beneficial effects of inflammatory response versus the pathological consequences. So immunologists like to use animal models to address those questions. And in particular, the mouse is a very useful one, because it's a small animal model in which we can do a lot of genetic manipulation and experimentation. But the limitation is that it's not a human. It's just a model, so there can be some differences. Humans, in contrast, are much more relevant for diseases and more representative of the human genetic diversity. But we are much more limited with what we can do experimentally on human subjects, of course. So what we would need, ideally, would be a model that combines the advantages. That would be a small animal model with the characteristic of the human species. And we believe that humanized mice can provide some of those advantages. So I'm going to start today with the first project, exclusively on mouse models, on the silent thing of the immune response of apoptotic cells. So tissue amyocesis is maintained by the balance between the formation of new cells that differentiate and proliferate from stem cells and the death of an equivalent number of cells to maintain the size of the tissue. So this is something actually important, because some tissues have different rates of turnover. But the total number of cells that has to die every day is quite impressive. So the intestinal epithelium is probably the champion. There are trillions of cells that need to be replaced every five days, approximately. So it means that trillions of cells just in the intestine have to die every five days. And if this is not done properly, then it can result into different kinds of problems. If cell death is defective, it can result in cancer or autoimmune or inflammatory diseases. If there is increased cell death, depending on the tissue, it can result in diseases such as neurodegenerative diseases or immunodeficiency. Now there are many ways by which a cell can die. And I listed here a few of the types of cell death that we can find in the literature. Each of them is characterized by different stimuli that induce them and different biochemical characteristics. One of the most best described is apoptosis, which is a highly regulated process. It's highly conserved during evolution from worms to mammals. So one of the important questions that we would like to address is why is apoptosis so conserved during evolution when we have all those other mechanisms of cell death that could just kill the cells in the same way? So what's special about apoptosis? One of the hypotheses that emerged in the past few years is that apoptosis is unique in its capacity to kill cells in a manner that does not elicit an immune response. So in most types of cell death, including necrosis, when plasma membrane is disrupted, it releases a lot of intranational molecules that have some immunostimulatory or pro-inflammatory properties. So those molecules are collectively called dams or alarmines. In contrast, in apoptotic cell death, there is formation of those apoptotic bodies. The integrity of the plasma membrane is maintained, and those apoptotic bodies can be cleared by macrophages without releasing anything. So that's one of the hypotheses why apoptosis would be immunologically silent. But the detailed mechanisms by which this occurs, and in particular now, what is the role of caspases in this process, are still mostly unknown. So that's the question we wanted to address. So first, let's have a look at how apoptosis is induced. There are two main pathways. And I have to say that this is a highly simplified representation of cell death. One pathway is the extrinsic pathway induced by engagement of the so-called death receptors by a fast-slagant or TNF, signal through caspase 8, and then the downstream effect on caspases 3 and 7. I'm not going to talk about this pathway today. I'm going to focus exclusively on the intrinsic or mitochondrial pathway of apoptosis. It's controlled by the balance between the expression and the function of BCL2 family members. So there are pro-antipropotic family members of BCL2. And depending on the balance between those, they regulate formation of a pore in the mitochondrial auto membrane by the back and back channel. When this pore is formed, cytokines is released in the cytoplasm, where it contributes to the formation of this big complex called the apoptosome, in which psychochrome C interacts with epa1 and caspase 9. And caspase 9 in this complex is activated, acquires its enzymatic protease activity, and activates the effect on caspases 3 and 7. So caspase 3 and 7 are effect on proteases, and they cleave many, many substrates in the cells to just destroy the cell. Now, one of the complications to study this pathway is that genetic deficiency in any of those genes results in embryonic lethality. So the first thing we have to do was to generate mice with a conditional deletion in caspase 9, which was caspase 9 because it is a central role here in this pathway. And we crossed it to tytoocry, which efficiently deletes the gene in hematopoietic. It is not a very specific deleter, but at least we have mice that live to adulthood with a knockout immune system. And we also generated caspase 3 conditional knockout with the same pre-specific deletion, and crossed it to caspase 7 knockout to get the single and the double conditional knockout mice. So we first looked at those mice, and surprisingly, their immune system looked quite normal. All the cell types are there in relatively normal frequencies. But when we tried to stimulate this immune system, we found a lot of deficiencies. And one of the most striking phenotypes that we found is that caspase deficiency results in resistance to viral infection. So this is shown here in an experiment where we infected the mice with the encephalomyocarditis virus, or EMCV. You can see that it's a very lethal virus. All the control mice die within six days after infection, while in the mice that lack caspase 9 in their immune system, they survive longer. And some of them even survive long term. This correlates with a much lower viral load in the heart, measured here by PCR, 48 hours after infection. We observed a similar phenotype when we infected the same mice with VSV, so the vesicular stomatitis virus, that's here an intranasal infection, and we measure viral load in the blood 24 hours later. You can see that mice that lack caspase 9 are completely resistant. We cannot detect any virus in the blood of those mice. And the same is true for the caspase 3 and 7 double deficient mice. Next, we try to recapitulate this observation in vitro. For that, we derived mouse embryonic fibroblasts from caspase 9 white-tiled and knockout mice. And we infected them with a strain of VSV that also expresses GFP, which makes it easier to detect. You can see here that white-tiled cells are infected, shown by GFP expression here by fluorescence, by microscopy, and by fax. While the caspase 9 deficient cells are much more resistant. This is quantified here. We use the range of doses of the virus, and we need to use a really high dose of the virus to detect some GFP expression. And also, when we measure viral progeny in the supernatants, you can see that depending on the dose, there is a 2 to 4 log decrease in viral production. So that's a pretty strong phenotype. Next, we looked at other molecules in the pathway. So APAP-1, which is a partner of caspase 9 in the apoptosome, and we found the same phenotype when we infected those fibroblasts. They are resistant to VSV infection, shown here by GFP expression and by viral progeny prediction. And also the same for caspase 3-7 double knockouts. Single knockouts don't have any effect. But when we knock out both of them, they have a redundant role. We observe a resistant phenotype, very comparable to what we saw in caspase 9 knockouts. So now it's well known that cell death in the apoptosis is a mechanism of defense against the virus infection. A cell that's infected by a virus dies, commit suicide, so it cannot spread the virus. But if this was the mechanism of what we would expect, the opposite phenotype was an infected cell cannot die and continues to produce virus. We would expect more viral production. So obviously, there's something else going on. So we decided to look for other described antiviral responses and an important response to viral infection is the type-point-of-front response. So this slide summarizes how the type-point-of-front response works. It's initiated by the recognition of the virus, generally viral nucleic acids, by different types of receptors. It signals through RF3 and 7 molecules to generate, to induce the secretion of interferon beta, which is released, which is secreted, and binds the type-point-of-front receptor. Then there is signaling through STAT1 and other factors that lead to the expression of a number, hundreds, actually, genes, or ISGs. And those ISGs contribute to two things. First, they amplify the signaling pathway, which results in the secretion of a second wave of type-point-of-front and amplification of the response. And also, those hundreds of genes or proteins interfere with many cellular functions. And together, they contribute to establish a state of viral resistance. So we wanted to look in our caspase-deficient cells and animals if there was a defect or something abnormal with this response. So first, we looked for type-point-of-front themselves. So in steady state conditions, it's really difficult to detect them by a regular PCR. But it's described that they are expressed in their functionalized steady state. So we use the more sensitive assay. We use the nested PCR numeration. And you can see here that while it's completely undetectable here, we can detect them now. And there is a difference between the knockout and the white type. It's higher in the knockout. But this is a really mild increase compared to that. We quantified it here. It's approximately a 5 to 8-fold induction, while this would be a 1,000 to sometimes 10,000 fold induction. So that's really weak. We confirmed it with a bioassay, which measures type-point-of-front activity. And again, it's really weak. So this shows the limit of detection of the assay. White types are just below detection. Knockouts are just a little bit higher than the detection limit. But this is highly reproducible, although really weak. Next, we looked for the expression of ISG, so those interference-tunated genes. And here it was much easier to detect a difference. So I'm showing here two examples of ISGs, ISG-15 and RF-7. In fibroblasts, again, you can see that without any stimulation at all, caspase knockout cells express high levels of those two ISGs. It's between 10 and 100 fold increase. And it's actually close to the maximum level that we can reach when we stimulate the cells with type- point-of-front cells. And we found a similar induction in vivo. This is an expression of ISG-15 on whiteboard cells from the caspase-deficient animals that I described a few slides ago. So we find increase of steady-state type-point-of-front constitutive expression of ISGs in caspase-9-deficient cells, but also in caspase-3 and 7-double knockout and in epa-1 knockout. We also tried to recapitulate this observation using simply a caspase inhibitor. So there are many of those available. We treated cells with this inhibitor, QVD-OPH. That's one of the most recent and most effective ones. And you can see that 48 hours later, when we did RNA-seq of those cells, there was expression of a number of genes here. I listed 10 or 12 of them. If you are a little bit familiar with the interphone response, you will recognize them as typical ISGs. And this shows a more complete analysis. So you can see that the treatment with the caspase inhibitor induces expression of hundreds of genes. But when we treat cells that lack the type-point-of-front receptor, there is no response at all, showing that they are actually interphone induced. So now we still need to demonstrate that caspase inhibition is responsible for the resistance to viral infection. And we did that with a very simple experiment. We harvested the supernatants from caspase-9-white type or knockout cells. We transferred those supernatants on white-type cells in the presence or absence of blocking antibodies from type-point-of-fronts. We incubated for 24 hours, then we washed, and we infected. And what happened there is that the white-type cells that were incubated overnight with caspase-9-supernatants were resistant to viral infection, showing that the antiviral activity is contained in the supernatant. And it was completely inhibited by the entire interphone antibodies. So this demonstrated that the viral resistance is mediated by type-point-of-fronts. So to summarize those first observations, the well-described type-point-of-front pathway that leads to viral resistance is under negative control by the epa-1 caspase-9 and caspase-3-7 pathway known to be necessary for apoptosis induction. So that was a very surprising observation. And it raised a number of questions. The first one is, is this related to the pro-apoptotic function of this pathway, or is it a non-apoptotic role? The second one, if type-point-of-front is induced, what is the ligand that's going to engage a receptor and lead to the expression of those interphones? And third, if there is a ligand, how is it contained in white-type cells? First, this question, the apoptotic or non-apoptotic role. So for that, we went upstream in the pathway, and we looked at the antiviral response in cells that are deficient for BACs and BACs. So in terms of apoptosis, BAC-BAC deficiency, completely phenocopies, epa-1, caspase-9, or caspase-3-7 deficiency. However, when we infected those cells, the BAC-BAC deficient with VSV, you can see that the phenotype was very different from what we saw in caspase deficiency. The cells were perfectly infected by VSV GFP and actually produced more virus, which is actually what we would expect from an infected cell that cannot die. So this would suggest that those molecules act on the interphone response independently of the mitochondrial events of apoptosis. That's actually more complex than that. And this is a key experiment to understand the mechanism. So as before, we used BAC-BAC white-eye cells, treated them with a caspase inhibitor, and this resulted in expression without any other stimulation of ISG. So I'm showing here ISG-15 and RF-7. But when we did the same experiment on BAC-BAC deficient cells, they were not able to respond to the caspase inhibition. So how do we interpret this? So this is the situation in the knockout cells. There is no BACs and BAC treated with inhibitor. There are no caspases. The interphone response is low. There is no viral resistance. Or no expression of ISG, at least. When we do the same experiment in BAC-BAC white-eye, now the channel is there, cytochrome C can be released, and also the interphone response goes up. So this suggests that when we open the BAC-BAC channel, we also release something that can induce the expression of type pointer fronts. And finally, when we have a white-eye cell without the caspase inhibitor, so now cytochrome C can activate this, the pathway is under negative control by caspases. So this suggests that there is a ligand release from mitochondria through the BACs and BAC channel. So now you're probably thinking that there's something wrong with this model because we don't treat the cells with any proper to tick signal. So the BAC-BAC channel should be closed and there shouldn't be anything happening. We believe that this is happening in the three or four percent cells that always die in a cell culture or those billion cells that die every day in the human body. And if this model is correct, then if we were able to actually open the BAC-BAC channel, we should amplify the response and trigger stimulation of type pointer fronts. So that's what we did in the next experiment. So BCL2 is an endogenous inhibitor of BACs and BACs. And this drug, ABT737, is an inhibitor of BCL2. So when we treat cells with this BCL2 inhibitor, we can actually open the BACs and BAC channel. And when we treat cells with this inhibitor, you can see here, it doesn't do anything to the expression of type pointer fronts. The caspase inhibitor alone doesn't do anything either at those early time points. But when we combine both, this inhibitor that opens the BAC-BAC channel with the caspase inhibitor, now we get a really strong induction of type pointer fronts, which is actually comparable to the type, the kind of response that we would see during a viral infection. So that's what I call the BAC-BAC-dependent caspase regulated type pointer fronts. So I'll try to keep things as simple as possible in the next slides. So we will have two protocols to induce type pointer fronts. The first one is this BAC-BAC caspase regulated type pointer fronts. We treat with this inhibitor to open the channel and the caspase inhibitor to prevent the negative regulation by caspases. And we use a positive control, which is transfection of genomic DNA, which is known to induce type pointer fronts. So now, the next question is to determine what this potential ligand could be. What's the signal that comes out of mitochondria and induce type pointer fronts? So the mechanisms by which type pointer fronts are induced in a cell are quite well described and there are essentially two pathways used. The first one is cytosolic RNA recognition through the rig I and mass pathway, or the cytosolic DNA, which involves siga sting, and both of them converge to TBK1, I have 3, 7, and type pointer fronts. So first, we looked at the involvement of TBK1 and RS. So here, we use white-up cells, which with them, we double induce the back-to-back dependent caspase regulated type pointer fronts, and we compare them to our positive control transfected genomic DNA. You can see that the double inhibitor induces phosphorylation of TBK1 and RS3, which is comparable to our positive control. So it shows that TBK1 and RS3, 7 are involved. This is confirmed by this genetic experiment where we treat cells that lack, I have 3 and 7, they do not respond anymore to the double inhibitor. So this shows that these signaling molecules are involved. Next, we tested whether the cytosolic RNA pathway was involved, for that we used mass deficient cells, and when we treated with the double inhibitor, the absence of mass didn't do anything, so it means that it excludes the role of cytosolic RNA recognition. Then, with the cytosolic DNA recognition pathway, which is mediated by this very interesting molecule, described couple years ago, approximately, so when it recognizes cytosolic DNA, it acquires an enzymatic activity which results in the formation of this cyclic nucleotide C-gamp that binds, sting, and activates it. So the first thing we did is to measure the presence of C-gamp in the cells. So you can see here that in untreated cells, C-gamp is completely undetectable, and after treatment with our double inhibitor, we have a nice peak of C-gamp that comigrates perfectly with our standard in this experiment. So this suggests that this pathway is involved, and this is confirmed with its genetic experiment. We knock out C-gas, or we knock out sting, and we completely lose the response to the double inhibitor. So this means that the cytosolic DNA recognition pathway is involved. So now, we know that it's a mitochondrial ligand, and that it binds the DNA recognition pathway, so our main candidate is, of course, mitochondrial DNA. So we wanted to test that, and for that we used a protocol described by molecular biologists 40 or 50 years ago which consists in treating cells in control with a low dose of ethyl bromide. So what ethyl bromide does, it intercalates in DNA, and this prevents the replication of circular DNA. So mitochondrial DNA, which is circular, cannot replicate anymore, but it does not affect genomic DNA. So the cells are still alive and relatively healthy if we supplement media with everything they would need, but they have an approximately 10-fold reduction in their mitochondrial DNA content. Now, if we treat ethyl bromide-treated cells with a double inhibitor or positive control, you can see that after depletion of mitochondrial DNA, we completely lose the response, the phosphorylation of TBK1 and IF3, while the cells are still able to respond to transfected DNA. And when we measure typewriter funds themselves, you can see that there's a 20-fold reduction approximately in the response to the double inhibitor, so the back-to-back dependent typewriter funds, while the response to HTDNA, so transfected DNA, is not affected by mitochondrial DNA depletion, showing that the cells can still respond to DNA sensing and produce typewriter funds. So this is our current model. So when cells undergo mitochondrial membrane transformation through back-to-back, they release mitochondrial DNA that has the capacity to activate the C-gas, sting, and antiviral response mediated by typewriter funds. In parallel to mitochondrial DNA release, cytochrome C is also released and activate caspitis, which have a negative effect on this entire pathway. So there are three conclusions from this model. So as I explained in the beginning, cell death can be pro-inflammatory by releasing in the extracellular space molecules that are supposed to be intracellular. In this case, it's slightly different. It's an intracellular ligand, which is released in the cytosol and activate a cell intrinsic immune response. The second conclusion is that back-to-back is the key point of decision to undergo apoptosis, which is supposed to be a non-inflammatory type of cell death. But what we show here is that back-to-back is actually a pro-inflammatory event. And it's only because from conclusion, there is an anti-inflammatory activity of caspitis that apoptosis mediated by back-to-back can be maintained immunologically silent. Now there are two important questions that in which we are currently working. The most obvious one is how do caspitis prevent this pathway? So if we look in the literature, we can find cleavage by caspases of any molecule we want, because that's what caspases do. They destroy the cell. So we need to figure out which ones are relevant in this particular model. There are some data that show that RSR cleave, TBK1 is cleave, but we don't really know which ones are important. Another possibility is that caspases induce the activation of caspase-dependent DNAs that's well described. It contributes to this typical lathering of genomic DNA in apoptotic cells. Those same caspase-dependent DNAs could actually cleave mitochondrial DNA in the cytosol and prevent activation of the pathway. So this is something that we are working on now. And an important question, and really interesting question, is what's the physiological relevance of this? So we observe this phenomenon when we use caspase inhibitors or caspase deficiency, but when does it happen physiologically? So we can think about it in the context of the Guar theory of pathogen recognition. So this theory consists in sensing a pathogen-specific activity rather than sensing the pathogen itself. So in this case, it is known that several viruses encode caspase inhibitors and that prevents cells from dying and the virus can continue to replicate. But now in this arms race between the host and the virus, the host would be able to sense that someone is trying to interfere with the apoptotic pathway and it should alert the immune system. Another possibility is that it's more broad at what we just described. C-gas could actually be a sensor of mitochondrial permeabilization. There could be other mechanisms, independently of back and back, during which mitochondria are stressed, really release their cytosolic, their mitochondrial DNA and this could be a way to alert an immune response because if mitochondria are damaged, it means that something is going wrong, possibly an infection. So those are questions for the future. So now I would like to move to the second part of my presentation and I'm going to discuss our efforts to generate a new generation of humanized mice that can be used to study human innate immune responses in vivo. And all this work on humanized mice is done in close collaboration with Marcus Mance in Switzerland. So what we call humanized mice is a mouse with a human immune system. And this slide described the protocol we use to generate those humanized mice. So we need a source of human hematopoietic stem cells or human hematopoietic stem and progenitor cells which are contained in the C-34 positive population. They are isolated either from cold blood. Most of the experiments I'm going to show are from fetal liver also from adult donors. We transplant those cells into immunocompromised mice. So we first irradiate those mice to eliminate their own immune system or hematopoietic stem system and we inject the cells on the day of birth or a few days after birth in the liver which is still a site of hematopoiesis for the first few days of life of the mouse. And then we wait for two to three months and the human immune system develops in the mouth and hopefully we would have a fully functional and developed human immune system. Unfortunately, this human immune system has many defects and one of the main defects is here in the malarid lineage. So there are essentially no functional macrophages or monocytes that develop in the previous generation of humanized mice. This is illustrated here from a review that we wrote a couple of years ago. So in red, the red color shows malarid cells. You can see that it's actually the majority of the white blood cells in the human species. So this shows the different model developed since the late 80s until early 2000s. So the most commonly used model is this one, this is the NSG model and you can see there are pretty much no malarid cells in human malarid cells in the blood of those mice. So we wanted to understand why there is this defect in human malarid cell development and we wanted to try to improve that. So malarid development is a very complex and highly regulated process which starts with hematopoietic stem cells that differentiate into malarid progenitors, more committed progenitors and finally the terminally differentiated cells. There are many cytokines that regulate this process and here the color code shows the percentage of amino acid identity between the mouse and human. And I'm showing here only the cytokines that are secreted by the stromal cells. So when we put human cells in the mouse, those mouse origin and the receptor would be human. So we don't really worry about the green ones which are more than 80% amino acid identity, assuming that they are most probably sufficiently cross reactive. That's an arbitrary cut off, obviously. The red one, less than 60%, I think, are very unlikely to be cross reactive and the green one in between, we don't really know. They're probably somewhat cross reactive but not fully. So with all those factors that are probably not cross reactive and are involved in malarid differentiation, it seems obvious that those human cells would not develop properly in the mouse. So to try to improve that, we decided to humanize some of those genes. So the method we use, this is Velocigen technology in collaboration with Rengen. It consists in replacing the entire gene from the ATG to the stockodon to get this humanized allele. So now we have a double humanization of the mouse, a genetic humanization for the genes encoding some of the candidate cytokines and then humanization by transplantation of human cells. So we did that for a number of cytokines and we reported them a few years ago, so I'm not going to describe them in detail, but essentially we did phomopoietin that improved HSC maintenance. IL-3 and GMC-SF, so we worked mostly on the characterization of GMC-SF that has an effect on the development of lung alveolar macrophages and MCS-F that improved monocyte differentiation. While those were significant effects, it was not extremely impressive. Actually a little bit disappointing and the reason is quite easy to understand. If we improve HSCs, HSC maintenance, maybe we get more myriad progenitors but then they don't have the cytokines they need to bring them to the fully differentiated cells. And if we have the cytokines that are going to support their differentiation, if we don't provide the pathway with the progenitors, it's not going to work either. So what we decided to do is to cross all those mice together and generate one super mouse that we call Mr. G based on the initials of all those cytokines. So now that's a quite complex genotype so I'm going to spend a minute to explain. So RAC2 and IL2 are gamma deficiency results in the absence of mouse TB and NK cells so those human cells do not get registered by the immune system, by the mouse immune system when we transplant them. Thorpalpha, that's a human transgene, a back transgene actually, induces phagocytic tolerance. So this is illustrated here. The ligand from Thorpalpha is CD47 and it's not cross active between human and mouse. And it's a negative signal that prevents phagocytosis by macrophages. So now when we humanize Thorpalpha on the mouse macrophages, now CD47 can engage the receptor, provide what is called the don't eat me signal and prevent phagocytosis of the human cells when we transplant them. And finally all the cytokines that I described that support maintenance of HSCs and myeloid development. So in the next slide I'm going to show comparison between four groups of mice. So RG, R-gamma, RAC-gamma double knockout. It's the original model we started with several years ago described more than 10 years ago now by Marcus Mont, our collaborator. NSG, it's functionally equivalent to CERB RAC-gamma. We decided to use that one as a control because it's the most commonly used in the field. So it's a good comparison. And then Mr. G's and their liter made controls that lack the CERB alpha transgene. Those two are pretty much similar. There are some differences but they are not really important for the presentation today. So we can just consider that they are the same. So we engrafted those mice using the protocol that I described with the transplantation of human HSCs in the liver on the day of birth. And we analyzed the engraftment levels in the bone marrow three months later approximately. So you can see that the RAC-gamma double knockout was good 10 years ago but it's much better. NSG completely replaced that model for an obvious reason. It supports better engraftment. Mr. G and Mr. G's you can see here are even better than NSG. So obviously this is not very quantitative because we saturate the system but this suggests that Mr. G's supports even more human hematopoiesis than NSG's and this will be important later when instead of using those fetal liver derived HSCs which are very potent and engrafting because they are fetal so they are really young. They still have potential to function. When we use adult cells, they are much more difficult to engraft and you will see that it makes a big difference in that case. So that's the bone marrow. In a periphery again RAC-gamma was not really good and between NSG and Mr. G I would say that in terms of overall engraftment levels they are pretty much similar. But now when we look at the composition of those human cells in the blood of NSG versus Mr. G you can see that it's quite different. So both mice have development of human B cells, T cells but there are very few myeloid cells in NSG's while we have a very distinct population of myeloid cells here in Mr. G's. So this is quantified here. You can see that in NSG's it's generally less than 10% in most of the mice while in Mr. G's it's between 30 and 50% in most of the animals. And overall this results in a blood composition much more like what we would find in humans which is rich in myeloid cells. So we started from 10% approximately and now we have 40 to 50% on average myeloid cells in the blood of those animals. So I'm not going to show the data on NSG's but you can see that there's also an increase in the frequency of NSG's and we also did a lot of experiments to show that they are more functional. And that's an interesting effect. Actually human myeloid cells are producing IL-15 which is one of the key factors required for the development and function of NK cells. So we have an indirect effect of the human myeloid cells on NK cells. So this is in the blood. We also find those human myeloid cells in different tissues such as lung, liver and colon while they were very difficult to detect in NSG. So this is a staining for CD68. And another thing that we were interesting in looking for is the diversity of the subsets of myeloid cells that develop in those mice. So in humans, three main subsets of monocytes have been described based on the expression of CD14 and CD16. So there is those 14 positive 16 negative, the double positive and the 16 positive CD14 low. So what you can see that in addition to the much higher frequency of those myeloid cells in Mr. G's than in NSG, the diversity of the subsets that are present is also improved. So in NSG, there is mostly this subset that should not exist. I'm not sure what it is. While here in Mr. G, we have those three characteristic subsets. So now the relative frequency of those three is not really reflecting the human distribution. So we don't know yet why this is. So this is quantified here. Very few of those CD14 low 16 positive cells that are actually extremely interesting to study. So the distribution is not really the same, but we did a lot of experiments in vitro to characterize them based on the expression of a number of markers and also their functional properties when we isolate them, including phagocytosis and cytokine production. All three subsets isolated from Mr. G behave very similarly to the same, the equivalent subsets isolated from human blood. So we are confident that those cells represent really human monocytes. So we did a few in vivo stimulations. The first one, very simple experiment with induced endotoxic shock with LPS. There's a really low response in NSG, a much stronger response in Mr. G's. Yes, both. More than 10 to 100 fold in print, sorry. We also did a viral infection and measure typewriter front in the lung. So you can see that, again, in NSG, there is a response, but it's really weak. While in Mr. G mice, we have a really strong typewriter front response. So those results show that Mr. G is superior to NSG in terms of supporting human hematopoiesis and differentiation and function of innate cells, mostly monocytes, macrophages, and also NK cells. So now I would like to show you results in which we try to model human diseases using this model. And the first model is solid tumors with a focus on the role of monocytes and macrophages. What is known is that macrophages infiltrate human tumors and from clinical observations, it's known that macrophages actually support tumor growth. So this comes from observation of strong correlation between high density of macrophage infiltration in the tumor and poor patient prognosis. So that's a bit counterintuitive that our immune system would support tumor growth. But actually what's happening is that macrophages are capable of doing tissue repair, but in the context of the tumor, tissue repair becomes tumor support. So this is an example shown here of the CDC 163 positive human macrophages that infiltrate a human melanoma. And we collaborated with Karina Paluka to try to develop a model and see if we could recapitulate this role of macrophages. So we did the first engraftment of a human immune system as before and then we transplanted a human melanoma cell line in NSGs and in Mr. G's. And the question we wanted to ask was do human melanoma cells infiltrate the tumor? Do they affect tumor growth? And if they do, by which mechanisms? So this slide shows infiltration of human macrophages, CDC 163, it's in green now. Inpatient, so there is high infiltration. In NSG, we cannot detect any macrophage infiltrating the tumor. While in Mr. G, there is a high density of macrophages, of human macrophages in the tumor. And this is quantified here. It's very easy to find them when they are completely disabled in NSGs. We also did a little bit of characterization of their phenotype. Particularly, we looked for this M2-like phenotype which is generally associated with this tissue repair or tumor-supporting function. And we, I'm showing here one marker on CDC 206. You can see that in both the patient, the tumor from the patient and in the tumor from Mr. G, all most of the myeloid cells, the macrophages also express this M2-specific marker. Now, the cells infiltrate the tumor and have an M2-like phenotype. Does it matter? Does it do anything to tumor growth? So this shows tumor growth in NSGs and Mr. Gs that were not engrafted with a human immune system. The tumors look small and quite similar. But now, when we have mice that were previously engrafted with human CD-34 cells, so they have a human immune system, doesn't make any difference for NSGs, but you can see that the Mr. G tumors, or the tumors in Mr. G, are bigger and they look very different. They're probably more vascularized and they're probably also hemorrhagic, which is something which is described for tumors. So this is quantified here. So no human immune system, no difference between NSG and Mr. G. With a human immune system, no effect in NSGs, but the tumors grow more in Mr. Gs. So those are the tumors infiltrated with human macrophages. So as we saw this effect that suggested that there was a role for vascularization, we next treated the mice with an inhibitor of vascularization. We used the VGF inhibitor avastine and you can see that we completely reversed the tumor support phenotype. So this suggests that the macrophages support tumor growth in a VGF dependent manner. Now one of the weaknesses of this model is that we used fetal liver-derived hematophilic cells and the tumor cell line. So they are completely mismatched in terms of HLA. So what we are doing now, Tineal Carbosan with Caroyle, and more recently with Robert Schreiber, is trying to reconstruct patient-derived humanized mice. So we get TD-34, so the hematophilic progeny term from the patient, the tumor from the same patient, and we also use the peripheral blood cells that contain effect RT cells generated in the patient and we are trying to reconstruct those mice. So this is a work in progress. I don't have any data to show you yet, but we're hoping to be able to study the anti-tumoral immune response in those humanized mice and eventually use them to test and maybe also develop immunotherapies. So I'm going now to finish with another disease model and that's in collaboration with Stephanie Halin. She's a hematologist at the Yale Cancer Center. So she's really interested in myelodysplastic syndromes and in a random discussion she wants to tell me that Mr. G would be a great model to study MDS. So we decided to try. So what are myelodysplastic syndromes? So it's a disease of hematophilic stem cells which is characterized by the abnormal maturation of one or more myeloid cell lineages. It's a highly heterogeneous disease and one of the criteria to distinguish those diseases is to count the percentage of blasts in the bone marrow sample. So I don't really know much about all those diseases that Stephanie's field of expertise, but what we know is that it's a disease extremely difficult to study because there is no good animal models and those samples from patients with MDS cannot be engrafted into the currently available models of humanized mice. So that's shown here. Normal HSCs are quite easy to engraft. All those MDS samples are essentially impossible to engraft and only some of the most aggressive EMLs can be easily engrafted in mice. So we tried to engraft some samples. Stephanie sees a lot of patients and she always keeps some bone marrow frozen in her liquid nitrogen. So she went back to her collection and we put them into humanized mice. So this is first showing normal CD-34. So that's from the bone marrow of a patient that didn't have any disease. So you can see that it's much more difficult to engraft those adult hematophilic stem cells than the fetal liver ones. So in this case, it doesn't work very much in NSGs. It works much better in Mr. G. So you remember when I said it would be important to have Mr. G for more difficult samples. This is shown here. Now RCMD, so this is one of those diseases with low blast. They're still extremely difficult to engraft, but in some of the mice, we get decent levels of engraftment in the bone marrow of Mr. G's or NSG. REB1 is a little bit easier to engraft. If we use a cutoff of 1% of engraftment, most of the animals reach that cutoff. So it works quite well by those criteria. And finally, REB2 is much easier to engraft, and we can even do secondary transplantation of those samples. So re-isolate those human cells from the mouth, transplant them in the second mouth. It still works better than in NSG transplantation. So you can see here that all those diseases in NSG doesn't work really well. While MDS, we start a little bit higher and those diseases become more easy to engraft. Now it's important also to demonstrate that what we engraft is actually a disease and not some normal HSCs that are still in the sample. So for that, we use the same cytogenic characterization as was done for the patient. So this example, the patient was diagnosed with trisomy 15 and we found the same trisomy 15 in the cells isolated from the humanized mouth. And in this case, it's a 5Q deletion that was for in 10% of the cells in the patient. So this is a normal cell. There are two copies of 5P and 5Q while here in some of the cells, most of the cells actually in Mr. G, there is one copy of 5Q is missing. And finally, hematologists are very happy when they see that to mean doesn't mean much but when they compare, when they see a bone marrow like this one, they can describe it as MDS based on the presence of megacryl sites and fibrosis by reticuline. You can see that what is found in the bone marrow sample of the patient is very similar to what we found in the mouth. So we are relatively confident that we actually engulfed the disease and that the engulfed sample recapitulates some of the characteristics of the disease. So to conclude that second part, Mr. G Ma is highly permissive for human hematopoiesis. They support the development and function of diverse subsets of human monocytes. Those minor cells can infiltrate the tumor and support its growth. And finally, combining the two characteristics, so efficient hematopoiesis and minor differentiation, Mr. G supports the engulfment of minor dysplastic hematopoiesis and provides a new model for this disease. So to conclude, I hope that I convinced you that Ma is extremely and full to identify some fundamental mechanisms of the immune response. That humanized mice hopefully will be useful to translate those findings to the human species, but also that humanized mice are useful to develop animal models of human diseases and eventually could be useful to develop new vaccines or new immunotherapies. So with that, I would like to thank everybody who contributed to this work. So Richard Favell, of course, was a wonderful mentor and extremely supportive. People who contributed to the work on Caspaceous, particularly Rory Jackson, who helped me with a lot of the Western Blots. He likes that, of course, the Western Blots. John Alderman, our lab manager, and the three technicians who used to do DSL work. We got a lot of help from different labs, including Akiko Iwazaki, Jerry Shaddle, and James Shen. The humanized mouse work was a collaboration since the beginning with Marcus Mance. We did a lot of experiments together with Tim Willinger here with some important contribution of Teal and Sophia and the three technicians who helped us to cross those mice and to isolate the human cells for us. Stephanie for the MDS work, Karolina and Jan for the tumor model, and finally Regeneron Pharmaceutical, who generated the knock-in alleys in the beginning of this project. So thank you for your attention and I will take any questions. Yeah, that's a good question. Squewing them, skewing them, we haven't tried yet. M1 markers, I don't think we looked in the tumors. We did several M2 markers. I don't think we did an M1 marker. Jan Martinek did the staining. That's something we are doing now. Isolate the different subsets of monocytes that we can find in the tumor. There are transcriptome, actually, to get a broader picture than just one marker that we saw. Yeah. Yeah, we only did one tumor cell line with it. So in that model, yes. Now what we are doing is we are going step by step use fetal liver cells with tumors from patients and then bone marrow plus tumor from the same patient. And we already see some differences with different donors, which is not really surprising because the tumors are probably different. But mostly we see the same super powerful tumor growth. We haven't done yet the M2 characterization in those models. So I have a couple of questions. On the first part, yeah. And physiologic concentrations of CS based on your, any of the molecules in the CS in the natural substrates? So we did Western blot on C gas, didn't see any difference. We did Western blot on Sting, and then it's a little bit difficult to interpret because we think that there is constitutive activation of Sting in caspase knockouts. And this constitutive activation results in degradation. So it's difficult to compare the levels. We didn't find any cleavage products that could indicate constitutive cleavage. So we don't know yet. We don't know yet what the substrate is. And then a couple of two questions on the kind of this. So the first is have you tried to do a BLT mouse in that background? Yes. So the BLT protocol can be done with any recipient mouse, but it consists of transplanting, in addition to the bone marrow sample, to the CD34, a small piece of fetal timers in which human T cells can develop and be educated. So what we found there is that essentially, we get the advantages of both. So we get much more T cells while in both. So we did a comparison of NSG versus Mr. G. The BLT protocol is really good to support T cell development. What we found and what your experience is that after 12 or 14 weeks in NSG, we get 80% T cells. So it's an incubator for human T cells. In Mr. G, we get 70% T cells. So the piece of timers that we put in there is still stronger than all the myeloid cytokines. We don't really get a very balanced immune system when we add the piece of timers. And we did immunization of a Mr. G with just the CD34 or the BLT protocol. And it was quite disappointing. There was not really any improvement of the adaptive response antibody production with the BLT protocol. And then the second question, the last question, I had to do with the myeloid for myeloid was plastic. So one of the concerns is that you've already knocked in genes that are producing cytokines that may well be floating out of growth. So have you looked at, do these produce just physiologic levels of the immunocytokines, or are they driving the problem? Yes, so actually, that's something I forgot to mention. The interest of knocking in the gene from ATG to STOP is that we maintain the regular sequencing, including promoter, UTRs, from the mouth. The endogenous ones. And we get physiological expression of the cytokines. So yes, we provide those cytokines, but it's not as if we were overexpressing or injecting a cytokine that would be super physiological. So that's in theory that should be physiologically of the cytokines, at least physiological mouth levels of the cytokines. We know that TPO, for example, there is a 10-fold difference between human and mouth in steady state in the blood. And what we find with a humanized one is something intermediate. So it's probably a nontranscriptional effect. We did not recharacterize that. We would need to isolate them and do MLR experiments. So presumably, they would be, well, yeah, they could be more human restricted. Yeah, I don't know. I don't really want to do the experiments on this model and wait until we can do the experiment on the same mouth, which is human HLA noctine. And we have those mice ready to start working with now, at least with HLA2. So what we did is a listeria infection. And we measured CDT cell activation based on degranulation and production of interfront gamma. And we did side-by-side NSG versus Mr. G. There was essentially no response in NSG and some response in Mr. G. So some responses can be functional. But I think T cell responses are still highly defective in all those humanized mice. Yeah, we did not try to improve adaptive responses. We could have an indirect effect, whether innate response could result in better adaptive. We did not really see, except in the listeria experiment that I just described, we didn't really see any strong improvement of the adaptive response. So when we immunized with CFAKH, for example, which is a really strong stimulus, it's not much better than in an NSG mouse. You've just done a simple experiment and we'll look at the T-cells after tomorrow. And how diverse are the T-cells that you have? Yeah, no, we haven't done it in this model. But people have done it in other models. So one of the questions, of course, if it's restricted against mouse or human or a combination of both. So I think there are some data in the literature where people did some spectrotyping of the T-cell. And they say that there is some diversity, but... Yeah, yeah, yeah. So as I said, we did not do much on the T-cell part. We are working now on knocking in a number of genes that could improve, with the goal of improving actually timing lymphopoiesies. And then we can do all those experiments. We didn't really have any reason to spend a lot of effort on T-cells when our goal was to improve innate immunity. But clearly, those are important questions. One of the questions that we frequently get, the mouse is much smaller and is much less tele-emphasized than a human. So just because of the number, would you have the same diversity of the repertoire? Can you screen the entire repertoire with a mouse that has much less cells? So yeah, that's an important question that we can address with the T-cells improved model soon, hopefully. So there are many viruses that express BCL2 and arrows to prevent cell death. There are many viruses that encode caspase inhibitors. Again, to prevent cell death. Thank you.