 So, thank you for giving me a chance to present a different approach to all the opportunities that COMP has offered the community. These two talks that we just heard are just spectacular to see how we learn basic mechanisms from studying these knockouts. But my approach is somewhat different than how can we understand this, how it affects diseases affecting the skeleton. This is going to be much more from a clinical point of view than from a basic science point of view. So, what I'm going to be talking about primarily is we had the opportunity to learn how to interact with COMP lines when we were looking at homozygous knockouts. And I want to relate that experience and then tell you how it's going to apply to the heterozygous animals. The rationale for doing this is that skeletal disease, although you don't die from it, you cost the country an awful lot of money. Somewhere between 5% of our total GNP is directed toward diseases affecting the skeleton. And it affects people. It is the highest disease category of all diseases in terms of those that are affected. You know, it hits you in the midlife, it never goes away. Orthopedic hospitals are proliferating everywhere and just driving the cost of medicine out the wall. So, these degenerative diseases of the skeleton that we're primarily interested in, and it turns out that approximately 60 or more percent of the effect is genetic. There are a number of clinical studies that show that these diseases affecting the knee, that your joints, your bones, just your spine, I have a genetic cause for this. And being an ex-pediatrician, I think of this as a pediatric disease. These are genes that these children have inherited that we need to recognize to modify behavior and potentially prevent progression of disease so that they don't go on and have degeneration and have all these huge costs. So, we wanted to see whether or not we could use COMP to try to identify these, some diseases that otherwise are not being picked up by the current stream. So, and the reason why, and so in this screen that we did over the last four years, we looked at 220 lines agnostic-ly and turned out that about 10 to 15 percent of these animals, just homozygous knockout animals, viable, have significant abnormalities in their skeletal phenotyping. And so, and that plus GWAS data would predict that there's somewhere like 3,000 plus genes that can impact your skeletal health, either making more resistant to having these things happen or to making more susceptible to that. So, why is it so complex? Why could it be so many genes? This is a huge issue that the skeletal biology field is going to have to confront. So, when we look at the histology of bone and how it is organized, we have a bone that is resting or is being remodeled that is already there. These are basically trap cells, osteoclastic cells that are remodeling bone. They have their own set of issues. Then we have the osteogenic side of the equation in which you have cells that are AP-positive, they're laying down a matrix, and they also are mineralizing this matrix, and that can be high or low. And then in between is you have the actual action where it makes the decision whether you're going to have cells that are going to be going down more the osteoblastic route to promote formation or the remodeling moment. So, this is balanced between formation, it's all called remodeling. There are a huge number of genes involved in all this process. Now, the osteocyte, it turns out, the osteocyte, which is the cell that is buried within the bone matrix, appears to be the brains of this whole outfit. It has these dendritic processes that communicate with progenitor cells on the surface, and not particularly perivascular cells, which are progenitors, as well as the osteogenic cells that are on the endocortical surface of your bones. And this cell not only senses mechanical loading and therefore influences what the cells will do in response to mechanical loading or fracture, but it also senses your environment. It senses hormones that will regulate calcium and phosphate. It secretes hormones that will affect your metabolism, your overall metabolic activity. So, this is very much a partner in the whole scheme of homeostasis. And so, then on the other side, the other important part of this is the coupling, that is the talk that goes on between the osteoblast and the osteoclast. How do they stay coordinated? And there's a huge set of pathways now that we now recognize and diseases of these pathways, all of which could be targets of genes that we would need to understand. So, we, initially, we established our program to start the screen and look for these possibilities by having the production facility at JAX give us breeders and the animals were bred at JAX and then the bones were sent to us and we did the analysis. So, all the mouse stuff was done up at JAX with our collaborators who were there at the time. And what we learned was, number one, if we did micro CT, so we really felt we had to do micro CT in bone density because of the lack of specificity which we'll get to in a minute. We were shocked to see that the variance in just bone volume, BVTV, a measurement of how dense the trabeculia are, how high the variance was in male mice relative to female mice and that the saw that there is seasonal variation in this, we could not validate that. That did not seem to be the case at all. So, because of this, we said we had to do at least eight males and eight females in order to get statistical power to make any statements. So, these were control animals, eight males, eight females taken every month over the course of the study just so we get a very solid background of what the variance would be. And one of the things that we learned early on is that looking at the data at the end of the time is that if we plot changes just in body weight, so this is fractional increase or decrease in body weight and this is a fractional increase or decrease in BVTV, trabecular density, that there really was very poor correlation. There really was no relationship at all between bone, trabecular bone and body weight. But if you looked at the cortical size, how thick the cortices are and more importantly, even how large the bones are, there's a very strong relationship between body size and bone size. So, this really explains why the DEXA really is telling you more about how much bone you have and not how, what the quality of the bone is. It really is not telling what we need to know. So, that was important. So, we screened these 220 lines and we started to look at them as groups. So, this is the group of animals that had a low BVTV. And I go in, so these are the top 10 hits in which the ratio of their BVTV relative to control is highlighted here. So, this is the femur and this is the vertebra. These are some bone size and body size measurements. So, the one that is most striking, it came up pretty early on, was IRF-8. So, I wanted to go use that as an example of how we drill down to learn about that mouse more. So, this just shows the, how severe the BVTV value is in the trabecular bone of the femur in the female and in the male and in the vertebra of the two. So, both bones are greatly reduced in BVTV. Unfortunately, if you're looking at the IMPC site, there's bone was looked at, nothing was identified. This is the BMC content, which is probably the best measurement to look at for BVTV. We didn't, what didn't show as being abnormal. And this is the growth curve, we agree the animals were normal size. So, we did our micro CT, oh, my left, this, hmm. So, this is the data now expressed in the way that bone heads like to see this where we just do T tests between tests and control. And you can see that now that the BVTV is low in the femur and in the female in the male. And it's primarily due because they're fewer trabeculae rather than smaller trabeculae. This is the same thing in the vertebra, low BVTV, low size, a number of trabeculae essentially normal in the size. Now, we did our histomorphometry. So, in these animals, you get two doses of a mineralization dye so we can look at the line, the mineralization lines histologically and we can measure how the mineral is incorporated. And the measurement shows that the bone forming activity actually is pretty normal, not a big difference in the femur or it increased in the female. And this is another example of the dimorphism that we see in sex dimorphism. So, there is something there but not very impressive. We can't blame it on less bone being made. But if we look at the histology of this from the cellular point of view, things become very clear. So, the measurement of trap activity, there's much more trap on the surface of bone and particularly trap on the surface of bone that's actually also being labeled being a remodeling site. So, there's lots of trap activity on the degradation side. There was surprisingly little increase in the bone forming activity you'd expect that if they have all this removal of bone that you'd have a lot of formation. And we were surprised we didn't see that in this one. And then here, these are the actual measuring the cells that are actually initiating the whole process. So, there are a lot of sites that are trying to initiate these processes but the bone is not going forward to actually fill it in. And this same pattern we saw in the vertebra also, high trap activity here, relatively unimpressive bone forming activity but a lot of attempts to try to make it. So, that's all we, that's as far as we can go but the literature helped us a lot on this one. This is a pretty well studied animal. So, we know IRRF-8 is an inhibitor, is an inhibitory loop of NF Kappa B. So, without this being active, you have a lot of overexpression of NF Kappa B that drives osteoclasts into becoming osteoclastic cells. So, there's a lot of active, that would explain why all this osteoclastic activity. Clinically, this presents primarily as an immune deficiency disorder, has no bone phenotype has been reported in Omen even. Although there was a very interesting study showing that in mice, in the herozygous IRB mice, they have lost of their tooth roots. So, they're going to lose their teeth. It's always that the dental people will be very interested in this because they're going to have periodontal disease. So, how do we explain this problem of why the bone, the osteogenic side was not better? And so, it turned out that have been some targeted studies of this where if you knock out IRRF-8 in a, just in the myeloid lineage, then you don't get low bone mass. You have a lot of osteoclastic activity, but you don't get low bone mass suggesting that in this case, the bone cells are responding. And if you overexpress NF Kappa B in osteoblasts, which might be the case in this, you have a diminution of osteogenic differentiation. And with the diminution of osteogenic differentiation, that is now has a negative impact on the hematopoietic system, which particularly B cells, because of the role that bone plays in supporting hematopoiesis. So, this is an example of how we identified a kaolin that would really be one that you could drill down on much more and understand both the coupling mechanism between the two and how the osteoblast plays a role in immunocompetence, animals that could be studied in much greater detail. Now, on the other side of high bone mass, we identified the top one of this was called RIN-3, totally, has not been, it's totally unsuspected at all, no one has studied this one before. This one, however, has a various phenotype in that the trabecular bone mass is very high in the femur, but it's perfectly normal in the vertebrae. So, another example of dimorphism, okay, this is site dimorphism because your spine and your limbs come from different lineages. So, that may be the explanation. The IMPC had, didn't find any skeletal findings at all. They are reporting a low growth rate, a weight gain. We didn't see that, but that's, our mice didn't show that, so I don't quite know the explanation for that. So, if we look at the CT data on this one, we can see that in the femur, BBTVE is high, as we said before, and it's also, they're more trabecially, but the thickness is normal. If you look now at the bone-forming activity, there is, so this is the bone-forming activity in the femur, which is minimally increased at all, a little bit more in the female vertebrae, which didn't have a phenotype at the CT level. So, relatively less bone-forming activity than you might expect with all those big bones that they have. So, the histology really helped on this one. So, what we found in this one is that there is very low trap activity, very low-ass osteoclastic activity in these animals. Bone-forming activity is moderately increased in terms of the number of osteoblasts and active osteoblasts that are on the surface, but the remodeling activity is low. So, what this means is that bone is not being resorbed, it is being made at a continuous rate. This is what we call bone-modeling. This is what happens in a growing child. You make bone at more expense than remodeling the bone. So, the balance is much more towards formation. So, in this model here, this we have a high trabecular bone mass, primarily in the femur, this dimorphism we set about. It's due primarily to more trabeculi, and the increase is based primarily on low osteoclastic activity, but continued formating activity. So, this is extended modeling, if you will. So, what is known about this gene? It's a strong GWAS candidate actually in patches disease. And the GWAS data is a number of studies now showing that it actually delays the onset of symptoms of patches disease, which is a disease in which you have osteoclasts that erode the bone and then there's a very strong osteoblastic response to it. So, in this case here, it prevents that from happening. Another GWAS that just came out showed that it's associated with increased bone mass in children. That's interesting. And finally, there's another strong association with Alzheimer's and other forms of degenerative disease. Why is that important? Well, this would be an ideal candidate gene to study later on, because number one, we have now something, a coupling gene that somehow prolongs formation. So, this would be the ideal drug for osteoporosis in which you could promote formation over osteoclastic activity. And then there's a whole new emerging literature of how the CNS influences the bone axis that could be studied to. So, these are things that we've learned from this, and it really influenced us how we would go forward with doing hetcom. So, hetcom for us is that these are stronger effect genes. We're more likely to see adult phenotypes because of that. And we've enjoyed going to the embryonic call meetings. We'll try to be more vocal. We are often embarrassed to talk. So, Steve, wherever you are, we'll try to show you that we're there and we're going to ask more questions. So, we'll let you know that we appreciate what you're doing for us. And the thing that really has impressed us so far is how frequently there are craniofacial and limb developmental abnormalities. And that sort of hit us. Well, of course, because those are two tissues that are essential for making it through embryogenesis. You don't need germinolized skeleton to get through embryogenesis. You only need that when you hit the ground and you have to deal with gravity. So, we may be really having to focus more on adult phenotypes affecting cartilage and the craniofacial than on bone per se. We hadn't thought that through. We're having to do our own breeding. And, wow. So, we're both production and phenotyping and this is, wow. So, hopefully soon we'll get our first animals that will start to do this. But, because of the phenotypes, we think we're going to have to modify our phenotyping to make it more sensitive to bones as it relates to cartilage and shape. So, the CT, not only is it good for looking at the internal structure of cortices and bone, but it also is good for looking at shape. So, we can look at the shape of the bone and see the angles and shapes of the various parts that bear weight. Why is this important? Well, the adults have shown us that you can use a DEXA scan and use that as a phenotyping tool for the onset of degenerative joint disease. So, it just changes in shape alone, which probably reflected either abnormalities in how the cartilage laid out with limbs initially or how weight is being distributed is one thing that we need to do. So, we can look at these surface rendlings called SDL files and do image analysis on them to measure in 3D the various structures that are here, that are bearing weight, how are they changing in their bearing of weight. Similarly, we can look at the bone from top to bottom through all the image stacks and look at the relationships of various structures throughout the whole structure of the bone. So, we're fortunate that we have a very strong image analysis colleague that's going to help us try to deal with those issues. But the next thing is that we're going to have to deal with some histology. So, I just want to show you so we've been trying to know how to use our histology to try to pull more information out of these joints. So, the field usually looks at degenerative joint disease after the disease has happened where it's already been destroyed. We need to find a way of looking at stress that says you're going to be developing this kind of thing. And we hope that our histology is going to help. So, I want to give you a couple of examples of this. So, this is our frozen sections in which it's held onto the slide with a piece of tape and allows us to do repetitive imaging of the same section, not just with antibodies, but with other staining mechanisms. So, this is the Toludine Blue Stain. This is a Saffronone Stain that's used primarily in cartilage biology. You can look at the Saffronone Stain under fluorescence. It has a very nice fluorescent image on which you now we can map various things to see what's going on. So, this is the mineral. And this is the mineral overlaid onto that fluorescent background. This is the staining for the two mineralization dyes, green and red, that shows the Terecule labeling. The growth plate in this 12-month-week-old animal is still showing some labeling activity, but very, very little labeling activity on the cartilage or on the emphesis. I'm sure I should just give you a little more anatomy here. So, we're very interested in the articular cartilage here of the knee. So, this is the condyle of your knee, and this is the condyle of the tibial plate that it rides on. This is a ligament that's attaching. This is the emphesis. That's another cartilage structure there. All of those can be affected. So, we can look at those and we can say they are not mineralizing their quiescent. This is a normal animal. And this is trap. There's still a little trap activity, the osteoclastic growth plate, but other than that, it's a very quiescent normal animal. And this is alkaline phosphatase, which lines up all the osteoblastic cells on the bone surface, as well as the hypertrophic cells that are in the cartilage that have the capacity to mineralize if there's disease. So, now to just show you quickly that this is an animal, a 16-week-old animal, otherwise normal animal, that we are still allowed to do this. We let him hang him by his tail for three weeks, so he's running around on his front legs. His left leg, his hind legs are unweighted, so the loading of it has been totally changed. And when you do that, this is the mineral. Now you start to see that they're getting mineralization lines here. And this is with the background shot of mineralization lines here around the enthesis. We're getting mineralization lines around on his articular cartilage. What this is saying is that this cartilage is now responding, is starting to remineralize its bone. That will pre-sage the onset of cartilage disease. So we're hoping that we can use this to look for early evidence of stress, both by shape and this. And then finally, we can do this, look at a growth plate. So this is the Saffronova 3-week-old animal. This is the mineral. This is now the labeling cell. Strong labeling is in rapidly growing mice. We can use that to look at the mineralization of the growth plate. But the nice thing, we can also do EDU staining, so we can look at the proliferation index of this. So these are all things that we would like to try to implement. In terms of going with this forward. So finally, I'd just like to try to get a conversation going about how interacting with COMP, we could benefit so much if we coordinated things in a different way. So here is how we did it initially. The mice were bred at jacks, the parts were sent to us, and we did the analysis. The new version, we get the breeders, we grow them at Yukon, and then we characterize them. The advantage of that is that if we hit on an animal we like, we'll keep it alive. And we can drill down on it and get more information on it. That's the big plus. I would like to propose a different way. Because this breeding is so expensive. I would like to propose some kind of a model where our grant would pay for somebody at the production site who would harvest tissues for us as they're coming through with the production site. Let us screen them, send the bones, the ability about the bones, you can just send them to us. And we'll screen them and then we'll say, oh, this looks interesting. Do another round of breeding specifically for what we need. Mineralization labels, EDU, even harvesting the bones and sending us where we can do the marrow cultures. Because if you send us the bones on ice, we can do the marrow cultures. We can do it all that way. So it would be a far more efficient way of doing it. And multiple sites could do it. But the other ability is that if we're going to knock off 3,000 genes, we aren't going to do all that. We're going to have to have multiple sites of doing this. And given the fact that the biome is going to be different in every place, having a common production site where we send them out would be very helpful. But then have a common deposit site for the data that's bone-centric, that would be ideal. So I wish we could talk more about that kind of design. And likewise, when we discussed about the aging project, all these incredibly important bones there. We can figure out a way that we can do it to make what you have more valuable and then it would be useful information for us also. So finally, in comp, that we will do the homozygous animal. This should be heterozygous. I don't know why I said homozygous. So we're going to be using microCT and also body composition as our primary screen. Do we go further? We'll do our fluorescent imaging of the skeleton to look at it at the histological level. We plan to do primary cell cultures of osteoplasts and osteoclasts to try to get at this issue. Is it autonomous or not? We will be having monthly meetings with a panel of bone experts. So we will present our information to them to help us interpret this. But primarily to say, who do you know that would like to study this animal further? We want to get the bone community invested in wanting to take these initial findings. I was talking to my colleague, how many people, at least in our world, knock out a gene, don't see what they wanted to see and spend their life hoping to find something that they could do something with. Isn't it so much better to give them something that we know has a phenotype right off and go from there? I mean, it just makes so much more sense. So these monthly meetings are going to be very helpful. And as I mentioned in the discussion, we've made arrangements with the journal Bone to have a new electronic site where we will give an overview of how we do comp. And then case reports will be made about interesting animals that we've taken so far, so that someone could pick them up and take them further. And we could track how that happened. And the last thing that Peter May, who's not here with me today, came up with it is turning out to be very interesting. As we started a course, an honors course over at the Storrs campus, which is about 40 miles from us. Primarily for pre-med, pre-den, and pre-graduate students on heritable diseases of the skeleton. And we're using comp as the tool to teach them all this. And the goal is for them to write, help us write one of these short reports that would eventually become a short report that would go into this journal. So we know get them young, get them for life. That's what we need to try to do. Thank you. David, beautiful talk. We just finished a study that I think would segue perfectly with what you're doing with comp. So we did micro-CT scanning of about 500, 600 animals, 60 strains of B by D, where B is the same B that comp uses. Well, not quite the same. And in that study, we also computed what we called a bone ignorome, a set of 2,000 genes that have absolutely no literature associated with bone, which have highly specific expression in bone, and are great candidates. We came up with about 16 quantitative trait loci. The typical quandary of the kind of forward genetics we do, rather than the reverse genetics, is there's really no highly efficient way to bring the worlds together. But your methods bridged with the data sets we have. And I definitely will take you up on your suggestion. So if you need a breeding center to send you, we have about 150 strains of mice. They're all from B by D. They're all replicable. And they're all at Jackson Laboratory also. So it'd be fun to bring these worlds together, instead of having them do parallel play for the next 20 years. I think that the CT scanning is not as overwhelming as it would appear to be. We actually convinced the institution to buy us a second one. And if we do it at a relatively low resolution, we can get the information we need very fast. So it's a grease pipeline. It goes fast. And there's just so much information in the CT scans. Certainly, looking at X-rays and trying to pull information out of it is a laudable task. But the problem is the information is so limited that to me, we should invest in, these things should be done by CT and then really have great information to go with for. Can I ask a quick question? Yes, sure, please. I'm sorry. Are there any bones in particular that you're interested in, either E-15.5 or E-18.5? Or are all bones equal at that stage? We haven't encountered that yet to make that decision. Peter May, my colleague on this, is a developmental biologist, and that is going to be his role. So because we will be getting some sublethal animals, I'm sure, in some of the lines coming through, I think we will look at them. But we really haven't thought hard yet about other than just describing them with some size measurements. And we could, and definitely we would do the histology on them, too, to try to figure that out, too. But we haven't, I mean, the great talk that you gave Jason on, an organized way to structure your thinking about why it's failing at a certain place. We want to be there someday, but we're one place from that. So David, great talk. I mean, personally, I haven't obviously spoken to anybody yet, but I don't see any red flags in your proposal that you want to go. So I think we should talk about it. But just to clarify, so the 200 knockout lines that you were initially started telling us about, you didn't choose those based on data from IMP. You just took them out. It was on the shelf coming off from the JAX site. Got it. Okay. So when I'm thinking about what you're doing and what the embryo folks are doing, they're using IMPC data to select which ones they want to proceed in their lab. You have a more agnostic approach. You just select without any information, which is fine. I guess what I'm curious is, is there any data that's being produced by IMPC that is of value in your project in order to think about how to better select because I can see what you're proposing. And I certainly welcome your suggestion. You pay somebody or that. But I could see the cost going to be quite a barrier for this. Well, it certainly is not as expensive as rebreathing all these guys, burying them in, going through our animal care to get them into the house and so forth is very expensive. I think that we are, now that we're doing these hats, we are doing some selection because of the expense of bringing them in. We want to have the majority of them to be a positive hit. If we were taking them from production, I mean, well, I heard the knife through the heart, you cut off the legs and the spine and you throw it in the trash. My gosh, that's my life right there. So we could screen those. But for the ones that we're breeding and bringing in, we are trying to select, make choices there. It's not agnostic, but the homozygous one was agnostic. And I would prefer to do all of them because we're totally surprised in all of these. I mean, there's just so many we never expected. So there's a couple of things. So for the sub-viable and lethal lines, if you're interested in the hats from the sub-viable and lethal lines, they end up being on the shelf for quite a while longer, and we end up with a lot of extra heterozygous. So I don't see that it would really cost much to take out a bone or two and send them off to you if you wanted them fixed in some way or whatever. No, you just... Oh, okay. Yes, right. We'll do this action. You just chop-chop like you're doing a chicken. Well, no, I'm just... I'm trying to figure out if it's possible, right? Yes, it is. That's what I'm trying to figure out. Anything that we could do to promote that. We do have lots of extra hats. Right. I mean, the biggest issue is going to be this problem we have with variants that picking two, we could get totally fooled. So that's... Well, yeah, I mean, we have to understand from you what number you would need because if nothing else at the end of the pipeline, the adult pipeline, we don't use the legs for most of these things. We don't have the data yet to be able to pre-select them. Well, that's what I'm asking if he doesn't want to pre-select them. I would say we would just screen him. Yes, I don't want to pre-select him. Right. Because this is so... Yeah. I mean, it's certainly more than doing a dexa, but it is... The information is so good. Yeah. Okay. That's for the bone heads. So maybe our takeaway is the com group will meet, we'll cost it out and we'll let you know what it would be. Yeah. And then you can make the decision as whether... Yeah, okay. It's worth it. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. I agree. Good. Thank you so much. This has been very helpful. Laura, will... Will you be the contact on that? Or... Yeah. You know, who... Sorry. Who wants to... I do coordinate. I mean, that's what I mean. I facilitate the follow-up discussions. And then what I would like is... How long is it? No. That's right. Well, there may be a couple of phone calls involved, but you know, you... Okay. And yeah, just send me an FYI when that settles down. And I will map that to the action items being taken post-meeting. Right. And I'm going to watch your contact information, because I don't remember name as well, so... Yeah. Yeah. So, okay. Great. Thanks.