 In the last lecture Dr. Josh Rebair provided his expert opinion and comments not only the researcher but also as a clinician who works in the areas of breast cancer. In today's lecture, Dr. Josh will continue his lecture about studying auto antibody signatures using NAPA technologies which could be used for early direction of breast cancer. So, let me welcome Josh again for his lecture on auto antibody detection for breast cancer using NAPA arrays. Okay, so I talked to you a little bit before about good study design and so when we went to do our breast cancer study that is what we did. I think I may have mentioned this already but we had around 5000 antigens that we studied. We studied 50 cases and 50 controls. These are women with breast cancer and these were women who were going to a clinic for routine mammography who did not have breast cancer. Okay, so these were healthy women in the population. From there we identified around 760 antigens possibly different between the cases and controls. So, at this stage of the game from here to here we did not want to be overly selective. Any possibility alive, we just wanted to exclude everything that was not even possible. So, even though we knew that there would never be anywhere near 760, we had already eliminated well more than 4000 and that was already good for us. Then for this set we printed a new array, duplicates on the array and we compared 50 cases to 39 controls in this case where the controls had benign breast disease. So they had cysts and other things in their breasts but no cancer and so that allows us to distinguish between cancer and controls. I will tell you that it's a little odd when you look at it that we had 50 and 39. You would have thought that we would have 50 and 50 right? It turns out that we did originally have 50 but about 11 of them turned out to have cancer after all and so we had to take them out of the study. So that kind of thing can happen, you have to be prepared for that. And then we did a third, said these women were different from these women and these women were different from both of these women. And this was the final validation, so three phases of validation. In this case we did it blinded so we didn't know who was what. And then we identified 28 antigens that even blinded were predictive of breast cancer. And I think in the end what worked for us was the fact that we were so careful and going through all of these phases that what we got in the end really did make a difference and that's why it was licensed by this company to make this VEDESSA which is a blood test that's based on our biomarkers and a couple of other markers that they added to detect breast cancer. And this has now been through a couple of clinical trials. I think I mentioned that earlier. You can see that in this clinical trial and this clinical trial they've reduced false positives dramatically while at the same time detecting all of the cancers. So the combination of the marker plus mammography found every single cancer and at the same time reduced a lot of biopties in women who didn't need it. Okay, so that I think I told you a little bit of that story before but any questions on this part so far before I go on? No, we haven't commercialized the rapidalizer. We have a core facility that will help you do it if you want. Yeah, it's not hard to do. An academic lab can do it and we've written up the protocols. You know, I think they're kind of anywhere between certainly anywhere up to I would say 100 nanomolar maybe if you use SPR you can get up to micromolar affinities but anything below that any tighter affinity than micromolar you can detect on the arrays. You can, yes, yeah. So micromolar is a little iffy. Certainly anything sub-micromolar you can detect, yeah. All right, so we moved on and I mentioned to you before that we know that breast cancer is a heterogeneous disease that means that there's multiple subtypes but there's the basal-like subtype, the HER2 new subtype, luminal A, luminal B, clotting-like, low, so these are all various subtypes of breast cancer and we were especially interested in this one here, the basal-like breast cancer because it's estrogen receptor negative, progesterone receptor negative and HER2 negative. It also tends to have either EGF4 or cytokeratin positive disease and this particular type of breast cancer is very lethal. It tends to occur in younger women, tends to occur in women of color and it is often not detected well by mammography and so all of those features make it a disease that would benefit from early detection because it doesn't have estrogen receptor and it doesn't have HER2 receptor. A lot of the drugs that we have today like Herceptin and the different inhibitors will not like Tamoxifen and that sort of thing will not work on that disease because the companion diagnostic is negative, right? So we thought wouldn't it be good if we could get some markers for that disease? So to do that we collaborated with Jonine Figueroa, she was at NIH at that time and she was running this large Polish health study. This study had roughly 2,400 breast cancer cases in Poland along with around 2,800 age and location matched controls. So we took her study and we sorted it for women who are estrogen receptor negative, progesterone receptor negative, HER2 negative and who had either CK positive or EGFR positive and so from all of these we got down to around 148 cases of true basal like breast cancer and then she was able to get us 150 age matched controls. Okay so I won't go through the details of the study design here but I will point out that we did very much the same that we did last time. So we had this sort of multi-phase study where we basically did two stages of discovery and then verification and then did a third level of validation where we tested a set of 26 proteins on a sample set that had never been seen before, completely independent sample set and we did that blinded and it turned out that we did not get a lot of markers that held up all the way through but we did get a couple. This is CTAG1A and you can see these are different types of breast cancer, luminal B, luminal A, HER2, these are control women who are healthy. This is the cut-off value for the test and you can see that at least for the basal like there are definitely a number of responders here. Now not everybody is a responder but all of these were responders, right? Here's another protein you may have heard of P53 before, it turns out that it's a good auto antibody protein and again it was particularly positive in these basal like breast cancer cases. Okay so we actually participated in a national study to see how these markers held up against other markers for triple negative breast cancer. This is through the early detection research network at the National Cancer Institute in the US there were I think 14 different laboratories that participated in the study. It was a fully blinded study so that all of the participants were given samples that were unmarked and asked to test them and predict cancer for them. There were over 80 markers tested and after all of that and all the studies that were done the only markers that held up were our markers, the antibody markers that I just showed you. And those markers, well there they are. So P53, CTag A and B, here's the AUC curve. We didn't really talk about AUC curves, I think you're going to get that soon when you get your statistic lecture but this gives you some flavor of that. And then we added that together with CA125 to get a slightly better performance here in red. And so this right now I think is the best blood test for basal cancer. Okay, any questions on that? So I'm going to enter the last section of the talk. I'm going to talk a little bit about where we're headed with breast cancer studies and we think that the future of biomarkers is looking at post-translational modification. So here's one form of post-translational modification which is going to come back later. This is called citrullination. It's when you take an arginine and you treat it with this enzyme peptidyl arginine deaminase and it converts the arginine into this thing called citrulline. And this is the citrulline molecule here. So you get this ketone here. Okay, so what we mostly have been interested in is glycosylation. So you guys are all familiar with glycosylation, right? So what fraction of proteins in the human do you think are glycosylated? I heard someone say 70. Certainly more than 50 anyway. I don't know if it's as high as 70, it could be. But certainly more than 50 to 60% of proteins have sugars on them. So it's very common. And you know that glycosylation occurs N-linked and O-linked. And today I'm only going to be talking about O-linked glycosylation. Mostly because most of the studies done in cancer have been done on O-linked glycosylation. So typically what happens with glycosylation is you have these sugars attached to specific amino acids and they form these very complicated branch structures, right? So they have this, you know, many many many sugars stacked upon one another. That's pretty typical in normal cells. In cancer two things happen. The first thing happens is that you get truncated sugar structures. They get truncated because there's an enzyme called cosmic that's missing and that is the one that only adds the sort of fourth level sugar. So you get three levels and then it sort of stops. So it's like getting a crew cut. It gets very short hair, right? The other thing that happens in cancer is it gets more promiscuous and so instead of only a couple of places on the protein you start adding sugar at multiple places on the protein. So to someone like me that looks like an opportunity, right? Because there are two things now that are different about these modifications. They create two new shapes that could induce an immune response. The first is that you get these short sugars and so the immune system might see that as different from normal and the second is that you now have amino acids that are modified that didn't used to be modified. So both of these are potentially inducers of an immune response and that would mean that for me they're an opportunity to look for new auto antibodies. So to get our platform to work well with looking at post-translational modification we had to clean up the platform a little bit. And so consequently we developed this method here which is called contra-captor. What we do is instead of printing our DNA on a glass slide like the way you're doing it, right? We print our DNA in these soft wells of a PDX membrane and then we put the DNA in these wells and we add the lysate to make the protein and then we cover the well with a glass slide that has the capture agent on it. So we separate where the capture agent is from where the DNA is. Normally at Napa you print the capture agent and the DNA at the same place. Here they're on opposite ends of the sandwich. The other thing that we're doing here is we're not using anti-GST anymore. We're using the halo tag. Are you guys familiar with halo tag? How many people know what halo tag is? Not so many. So halo tag is an enzymatic tag. It's an enzyme that normally is supposed to bind to haloalkanes. That is, you know, alkanes are structures that have haloalkanes have a chloride at the end or a bromide at the end and the enzyme in its normal function binds to that and removes the bromine or the chlorine. In the case of the halo tag, the enzyme has been mutated so that it binds to the chloride but it can't remove it. And because it binds to the chloride in a covalent attachment it now becomes stuck, permanently stuck to the chloride. So it's essentially a way of causing a protein to form a covalent linkage with a specific tag. The reaction is very specific and so you can selectively pull one protein out of a mixture of millions by having a haloalkane on the end of it. So we put the halo ligand on this glass slide and the proteins that we're producing have the halo tag and so when they see the glass slide they form a covalent attachment to the glass slide. What we like about a covalent attachment, of course, is that it's much more permanent. You can watch it really hard and it'll stay stuck. Alright, so then we express the protein. The protein goes up, it links up into the halo tag up there. It forms a covalent linkage. We lift the cover slip off the well. And now we can throw this stuff away and we're left with a glass slide that has these proteins covalently attached to them. And they're absolutely clean. There's nothing on this slide but the tag and the protein. So now we can come in with an enzyme and we can add a phosphate group, a sugar group, whatever we want to add and we can modify the proteins and then study them. And I'll just show you a couple of examples. Here's some NAPPA arrays. Here, remember I mentioned the citrullination. Here we've either citrullinated the array or we did not citrullinate the array. And here's a patient blood. This is a patient who happens to have rheumatoid arthritis. In the case of rheumatoid arthritis patients make antibodies against citrulline. And you can see that this patient has all these antibodies to citrulline proteins but does not recognize the proteins when they're not citrullinated. You'll notice that we have this ebna protein. That's our positive control. So that's a different antibody and not citrullinated, right related. And yet it shows us that the assay was working but not detecting all these other proteins. Here's another patient, same idea. And again, only signal are present in the citrullinated array. Identifying new interactors based on the citrullination. Here you can see some responses here. They were not previously known. And again, they're citrullination positive or specific. This is the same general approach but now we're doing with glycosylation. So here we've added some sugars to the protein. Again, these are different patients, Sira. And if you look carefully, you see a strong response when it's glycosylated but not when it's not. Here's a couple of responses that you see on glycosylation that are not present when they're not. And so, and here's another one down here. So the hope is that these responses will give us a new opportunity to find biomarkers or specific disease because of the profound differences in glycosylation that occur in cancer. Okay, so these are some of the diseases that we've studied using the NAPA. I've talked a lot about breast. We've also done ovarian and lung and also head and neck. All of these that we've identified, some markers that have been published over the years. Type 1 diabetes is an autoimmune disease. And so we've looked at that. We've looked at inflammatory bowel disease. We recently published a paper there. We're looking at autism that's early stage and also interstitial lung disease. And then we've looked at a series of different pathogens over the years. This one we just published earlier this year. And actually now we're increasing the list of viruses dramatically. So with that, I'm going to stop for today. I can take questions. Right, right. So how do we avoid diffusion? My last lecture, I'll probably talk about that a little bit more. You do get some diffusion. At the spacing that we do on the arrays and the size of the features that we create, the amount of diffusion is pretty limited. It's about, we actually measured it. It's about a couple percent compared to the spot itself. So the immediate neighbors will have around two percent of what you would have. So we felt that that was low enough that we could get away with it. Where it becomes an issue is if you try to make the arrays at a much higher density and move the spots closer together, then you start to see more diffusion to neighboring spots. And so when we get to that stage, we've induced a different technology that involves creating nanowells. That's a good question. Well, first of all, I don't know the numbers here. Certainly there could be a number of factors. Certainly there could be risk factors that are different in that population than this population. It's also possible that it's detected more often in the U.S., possibly because of more aggressive screening programs, which may mean that, like, I don't know what the overall per capita mortality is from breast cancer in India versus the U.S. Does anybody know that? Well, it's different from mortality or just the disease. Because one of the issues that people raise a lot with breast cancer, not unlike prostate cancer, is are we detecting disease that doesn't need treatment? So as you know, in prostate cancer, that's a big issue. Because of the use of the prostate-specific antigen, we're detecting a lot of men with prostate cancer who will die with prostate cancer, but not because of prostate cancer. And so they end up getting treated when they probably didn't need to be treated. And so that may be true. People have argued that's true with breast cancer as well, that we may be over-diagnosing it in the U.S. And we don't really know yet. So, yeah, I'd have to look at the statistics, but I don't know offhand that any major differences. But I could imagine a number of them. I don't think the markers would help us in that particularly. Mostly in part because these markers were developed in the U.S., right? So they were based on a population in the U.S. I don't know if that would necessarily tell us about breast cancer here. It might, but we just don't know. Yeah. With what? Yeah? Oh, interesting question. Well, yeah, so a couple of points there. First of all, I should just make the point that we don't use wheat germ, we use human, cell-free. We do that because we're looking to make human proteins and we think better chance of getting good folding. That said, all of the cell-free lysates tend to be a little finicky. They definitely are. If you think about what they're doing, they're very complex. Within that tube, you have everything to do, both transcription and translation. So you're asking a lot. First you're making, you have a promoter binding transcription factor to produce RNA. Then you have to have ribosomes bind to the RNA, tRNA is recruited, amino acids added, and then you have to have an energy generating system because as you know, translation requires ATP usage. And then you have to have chaperone proteins present to fold the proteins in the natural folding. So you're asking a lot. And to get all those components in a single concentrate that works, it doesn't entirely surprise me that it is temperature sensitive and that it's fragile. That said, believe it or not, you can lyophilize the cell-free lysate from bacteria. So you can make it a powder. You can add water and make protein from it. So that's pretty stable. Efficiency is low and it's not good for large proteins. So wouldn't be my favorite choice by far. But I think it doesn't surprise me that such a lysate would be a little bit sensitive. It would be cool to get an extremophile to do that. I don't know if anyone's tried. Okay. Yeah, I don't know. For the CML. Yeah, that's interesting to me. I mean, usually when you get cancers in young people, that's a sign of either genetic translocation, yeah, chromosomal abnormalities, because that's a pretty young age to have just sporadic mutation. Okay, I think we're done. I'm sure you have enjoyed both these lectures delivered by Dr. Josh Lebert, talking to you about utility of a technology, especially NAPA technology platforms, the insights which are required while performing these assays, your details for testing the reproducibility, thinking about experimental design, and then finally, the outcome which one could obtain from these experiments can be so remarkable which could be utilized for the patient care. You must have understood the clinical significance of early direction of biomarkers. You also studied about the biology of cancer in some detail and the tests that are now being used in the clinical trials. You are introduced to the concepts of contra-capture protein arrays which could be utilized for studying post-translational modifications. And you also got a glimpse of how various diseases could be studied using NAPA technologies. In the next lecture, we'll continue our discussion about use of novel technology platforms for various biological applications. And you will have series of interesting examples and illustrations to convey the utility of these technologies at the same time what entails to obtain the success from these experiments which is your careful experiment, your quality control checks, your data analysis and your insight and understanding about how to make a meaningful biological experiment. We'll continue these in the next lecture. Till then, thank you.