Founder Stories: Ross Chambers, Integral Molecular & Targeting Conserved Antibody Targets

A good morning or afternoon to you. Thank you for joining us. I'm Jan Riker, Chief Operating Officer of the Antibody Society. Today's webinar is one in a series designed to inform and educate our members as well as the broader scientific community about topics relating to antibody discovery and development. Our expert speaker, Dr. Ross Chambers, We'll discuss the use of divergent species to access difficult and conserved antibody targets. Please note the webinar is being recorded. Please do add any and all questions to the Q&A box in the viewer, and those questions will be answered at the end of the webinar. Without further ado, I'll now turn the show over to our speaker.

Thank you so much, Janice, and welcome everyone to today's webinar. So, uh, yeah, I lead the antibody discovery group here at Integral Molecular. We're a Philadelphia-based biotech company, and I'm going to tell you about, uh, the use of divergent species, things other than mice, for the purposes of developing antibodies. So with that, I'll start. So first, a few words about our company in case you're not familiar with us. So we are Integral Molecular, We're the industry leader in developing antibodies against undruggable targets. So these are— what that usually means is complex membrane proteins, things like GPCRs, ion channels, transporters. Many of them are often highly conserved, and many times we're tasked with finding, you know, functional epitopes or epitopes with rare properties. So we've been around for more than 20 years working on membrane proteins. In the last few years, we have been developing a pipeline of therapeutic antibodies against these complex targets, and we have licensed out a number of assets already to a number of companies. So over those 20 years, you know, we like to focus on good science. We publish a lot. Our technologies have been published a lot. We have a number of technologies that we work on in addition to antibody discovery. And we, we publish in top-tier journals. So the story I'm going to tell you today is based on a paper we published just recently in MABS journal about this topic of using divergent species in antibody discovery. And I think there might be a link to that paper somewhere here, but it's on the MABS journal, December 2023 to follow up with any details that you may hear from today. Okay, so the story I'm going to tell you today is first about, you know, gaps in the antibody space today and the role that divergent species can fill those things. Then I'll tell you about 3 of the leading divergent platforms, so rabbits, camelids, and chickens, and then follow up with some case studies of our own experiences using chickens for developing antibodies against challenging targets. So first up, you know, the key problem that we are addressing here today is conserved targets. So, you know, why are conserved targets difficult for antibody discovery? Well, when you immunise an antigen into a mouse, say, the conventional source for making antibodies. You shoot that antigen in, lymphocytes recognise it, proliferate and respond producing antibodies. But of course, those animals are also full of other proteins that they completely ignore, and they ignore them because of mechanisms called immune tolerance. So self-proteins, self-antigens, they are— any lymphocyte that recognises those are either deleted or the cells are inactivated or somehow suppressed. There's multiple mechanisms in play, so it means that the immune system only focuses on things that, that are foreign. So when, of course, you shoot in a highly conserved antigen, some parts, small parts, that may be different that it may respond to, but largely most of it may look like self. And of course, the immune system immune tolerance will kick in and suppress the response to that. So what that results means is that if you put a highly conserved antigen in that you may either get no response or a very limited response to only a small subset of epitopes, and it may not be the epitopes that you want. So in general, these highly conserved proteins make poor immunogens in animals. And if you look at the current pipeline of antibody drugs, you can see that bias of highly conserved proteins in the pipeline. So on the left is an analysis of all antibody targets and dividing them into either conserved or highly conserved or low conserved. And conservation here is greater than 90% identical to the host animal, mouse in this case. So you can see that of all the hundreds of antibody drug targets, only 14% are highly conserved. But if you look at all drug targets, you know, things to small molecules and things like that, half of them are highly conserved. So really, antibody drugs are really underrepresenting in the highly conserved targets. So if you look to see in more detail the nature of that problem, if you take all of those drug targets, the 700 or 800 drug targets, and you compare them to mouse, the most common host animal for generating antibodies, you see that, you know, half of all the targets are very highly conserved. Another good chunk are 80 to 90% conserved. So many of proteins that are in humans are highly conserved to mice. Now the reason people use mice a lot is a number of reasons. They're small and convenient, but also really the antibody technology started with mice when, you know, almost 50 years ago mouse hybridoma was invented that enabled people to make monoclonal antibodies from mice. And that has, you know, started a huge wave of antibody development, drug development, using mice as the main platform for that. Since then, however, there's been a number of technologies that have been invented around antibody discovery that have enabled us to broaden outside of mice. So, you know, phage display, yeast display, the ability to humanise antibodies. There's a rabbit monoclonal hybridoma developed and also B-cell cloning. And what these technologies have enabled us to break outside of using mice. And you can see that in therapeutic antibody development where, for example, the first chicken antibody to enter clinical trials was back in 2017. The first approved antibody from a camelid was in 2018, from rabbits in 2019, and sharks— the first antibody from sharks to enter clinical trials was in 2020. Now, I should mention with phage display, that's also enabled the use of naive and synthetic libraries to pan and discover antibodies without using immunisation. However, for membrane proteins, that has proven very difficult. You know, antibodies in these naive synthetic libraries, the hits are usually low affinity and very rare. And when you're panning on a complex antigen like a GPCR that's usually in low concentration, it's very difficult to pull them out. So in practice, it's been very difficult to discover these antibodies using those kinds of things. So really, immunisation has proven to be the, the most efficient way to get antibodies to these difficult targets. So you can see that the growth of divergent species for MAB discovery grows significantly in the last 20 years. So back, you know, 20 years ago, there was only a very small number of antibodies from different species apart from mice, but you can see that's grown significantly over those the last 20 years, with largely camelids but also significant contributions from rabbits and chickens. And the reason that's driving people to use these other species has been avoiding immune tolerance that I mentioned, but also the longer HCDR3. Mice have very short CDR3s, unusually short amongst all animals. But the longer CDR3s are useful that I'll explain briefly, as well as getting cross-species reactive antibodies that work across different species that is very useful for preclinical development. So, uh, first up, just to delve a little more detail into avoiding immune tolerance. So, um, yeah, divergent species enable you to do this. Now Mammals in general are all closely related. So you can see here that, you know, the difference, evolutionary distance between humans and mice, are about 90 million years. But any mammal really is about the same distance, about 90 million years. Mammals as a group generally, they evolve very rapidly altogether, diverge very rapidly, and So jumping across mammals does help you a little bit, but really going to a species like chicken makes a very large difference because of the much greater evolutionary distance—310 million years. You can go even further into other animals, but there you— chickens still have a normal immune system, a very similar immune system to mammals. But if you go even further distant, you— they're not immunoglobulins anymore. They're different structures, and they're they're not as useful, or they don't translate as easily to antibodies. And for example, if you can see here, if you look at chickens, if you could do that same analysis I showed you before where you compare drug targets to mice where, you know, over half of them are 90% conserved, if you do the same analysis for chickens, that now drops down to 15%. So a large number of those highly conserved proteins are not highly conserved in chickens. Yeah, so what that means is that, you know, chickens— if you've got a highly conserved target, chickens are going to enable you to get access far more epitopes than you would find in a mouse. All right. Okay, the second major point is the longer CDR3. Now, CDR3 is one of the most important parts of an antibody. There are 6 CDRs, but the heavy chain 3 is the most important. It makes the greatest contact with the epitope. It's the most diverse in sequence as well as length. And the CDR3 also really determines the paratope shape of an antibody, whether it's concave or convex. And analysis some years ago, Ramsland looked at, you know, a large number of antibodies and determined that you basically, if you have a CDR3 that's between 1 and 9 amino acids, it's mostly non-protruding or concave. 10 to 13, it's variable but flat. And then really 14 and above, you need to have a protruding paratope that enables the antibody to bind into pockets. And if you project these bins or categories here onto lengths of existing therapeutic antibodies on the right-hand side, you can see, and most of these antibodies have been derived from mice, you can see that most of the antibodies are either non-protruding or flat structures, whereas only a very small fraction of them are protruding, enabling access to different epitopes. And you can see that in some examples here, what that really means is, for example, Here's an example of a very short paratope with a 4-amino acid CDR3. This is nivolumumab binding to PD-1, and it binds that protruding structure on PD-1. And that very short CDR3 creates a little pocket so that that little loop can bind into there. Situximab has a typical length for a mouse, 11 amino acids, and it's a very flat structure that binds this flat paratope on situximab. But erinumab, which wasn't derived from mice, that has a 21-amino acid CDR3, and that gives that protruding structure which was critical for finding this activity. This is very novel epitope that gave them an antagonist function on this GPCR binding to this pocket here. That's really the power of these longer paratopes. And you can see across divergent species how this influences your ability to find these kinds of antibodies to give these interesting properties. So mouse, for example, as I mentioned, they really are the sort of the outlier. They have really quite short CDR3s. They're an average of 9 amino acids, so they're generally going to make antibodies that bind to flat epitopes. However, if you go to animals like the rabbit or human, they have, you know, longer ones. And but really the ones that really stand out in terms of CDR length are llama and chickens, where they have a lot more of these longer CDR3s, and a significant fraction of their antibodies now have these protruding paratopes. A third point is that you're able in these divergent species to generate antibodies, um, again, across species, uh, reactivity. So, um, as you know, uh, there's a large number of different animal models being used for, uh, drug development. And here's a panel that we took from a paper it's really talking about Nobel Prize winners, but gives you a sense of the different kinds of animals and proportions that people are using in research. And you can see that, you know, obviously the large majority of this work is done in rodents. We know that rodents are very convenient. They're small, easy to house, and available, a lot of infrastructure around them. But people use a lot of other animals— dogs, rabbits, non-human primates, of course, etc., and pigs as well. So this is great, you know, you can use all these animals for, for research, but really the problem with antibodies in that they are generated in mice means that cross-reactivity to these other species is often very rare. And what this means is that it really forces people's hands that they have to do a lot of their testing in non-human primates, or they have to make a surrogate antibody, which is again another itself that can be very difficult. And non-human primates is a little challenging in the sense that, you know, they are a lot more resource intensive to use these things. And also they've become in short supply and they got quite acute during COVID days. But still, you know, they're still difficult to access relatively. And actually, the FDA doesn't actually require you to use them, but often people's hands are forced to use them because they don't have the required reactivity to do that. So getting antibodies that can react across these species would, would really greatly facilitate these studies. And you can see here in this panel, this example of a number of antibodies generated in divergent species, that very often they are able to get reactivity in non-primates. And many of these examples are in chickens because of their great evolutionary distance that's made it very easy, but you can get success also in rabbits and camelids as well, getting reactivity to these, these other species. From our own experience using chickens, here's 3 examples that that illustrate the point. Um, the, uh, these three examples, they all differ by how highly conserved it is. So obviously the more highly conserved the target, the more easy it is to find cross-reactive ones. And the question is, you know, when does it really start to become really difficult to find these cross-reactives? So in our experience, you know, we've found that, you know, down to 70% identity, um, uh, you're able to find these cross-reactive antibodies. So in Claudin 18.2, you know, that's 90% identical, we are able to get cyno and mouse cross-reactivity. In GPRC5D, which is 82% identical to mouse, we found cyno and mouse cross-reactivity. And even CCR8, which is 71%, we got human and marmoset. So This really, really using chickens and other divergent species really enables you to find these species. And in the literature, even, even as the lowest example is 60% identity, they made an antibody to PD-1, which is only 60% identical between humans and mice, but they managed to find a cross-reactive antibody. Okay, so let's switch gears and now talk about some detail about these different animal systems for discovering antibodies: rabbits, camelids, and chickens. So a brief overview of rabbit antibody discovery. You know, rabbits, you know, got a long history in using— being used as immunological hosts, and they're very popular, they're widely available. People use them widely, especially for polyclonal antibodies. One nice feature of rabbits is they have a single germline single VH and VL germline because they generate diversity using gene conversion. And so that makes all the cloning and humanisation and engineering relatively straightforward. There's one germline to mess with. A number of discovery techniques have been developed. Of course, you can use phage display, of course, but most of it has been around B-cell cloning and rabbit hybridomas. There have been quite a few rabbits in clinical development. And thanks again to the Antibody Society for helping providing us with these details. They track all of these therapeutics and they're a very helpful resource for finding what's going on in the pipeline. But you can see there's a number of rabbit antibodies that have been approved in the US. And many more in clinical development. Camelids have been very successful, very popular, and one of the standout features for camelids is the discovery some years ago that they produce these heavy chain only antibodies, and they do not have a light chain, just the heavy chain.. And you can even make— cut them down to the smallest binding unit, the VHH, this so-called nanobody. And that has made them very useful for a number of things, especially the engineering of bispecifics, which of course are very popular now. They also have longer CDR3s, which enable them to bind into pockets. And a lot of people have used them to make antibodies against GPCRs that bind into pockets. Map discovery, it's largely using phage display and yeast display, and there's no hybridoma cell line or anything like that for them. However, some of the downsides of using camelids is that, you know, they are rather— they're a large animal, so the logistics around them is a little more difficult. You know, they're— it's difficult to immunise large cohorts of them and like you would do for other animals. So you have to be more selective. You generally use more antigen, which may be a problem if it's precious. And then you generally just tap into the resource of blood. People don't usually sacrifice and take spleens and bone marrow like you would do for other animals. So that does limit the repertoire that you can, that you can tap into. There are a large number of antibodies from Camelids in clinical development and a number of antibodies that have been approved. So, yeah, they've been very popular in the clinic. Chickens are the platform that we have focused on ourselves. The standout for chickens, of course, as I pointed out earlier, is this much greater evolutionary distance. So they really make a significant difference on being able to make antibodies against highly conserved targets. Despite that great evolutionary distance, they still make antibodies very similar to humans. The antibodies themselves have been very highly conserved, so that makes humanization very easy. And like rabbits, they use gene conversion and they only have— which means they only have one germline. So there's just one germline to worry about, and it's most closely related to very well-behaved human germlines, the VH323. So they are very easy to engineer and are well-behaved. In terms of CDR length, they have very long CDR3s that I'll show you in a moment. And for MAB discovery techniques, it's mainly phage display and B-cell cloning. And the first, uh, chicken antibody into the clinic was, uh, SIM21 against PD-1. Okay, so, um, let me now shift gears and talk about some examples from our own group that we're using chickens to develop, uh, interesting antibodies to give you a sense of what these different animals can do. So briefly, just let me introduce our platform. So we, again, we focus on complex membrane proteins, things like GPCRs, ion channels, and transporters. And we, uh, the antigen that we use for our platform is different in that we rely on DNA, actually more recently RNA immunization, as well as like particles. These are virus-like particles that contain the antigen in the membrane. These are critical for getting success with membrane proteins because they keep the protein in the membrane and do not perturb the native structure. As soon as you try and extract the detergents and things, you get misfolding and all kinds of trouble ensues. So that our platform is very different in the antigen we use, RNA and like particles. But really central to our platform is the use of chickens. We have used many other animals in the past, you know, mice and rabbits and things. But really what drove us to chickens was that many of these targets are highly conserved, and we're often tasked with accessing very rare epitopes. And chickens, we started exploring chickens, and they've just delivered success, and we've just followed that success. These antibodies from chickens, as I mentioned, they're very easy to humanise. You get high affinity antibodies, and they're very developable. And we develop our company, we take antibodies through to IND. So we'll, we have all the capabilities to do preclinical development up to, up to IND. So here's some examples of what highly conserved animals can do. So here's 4 examples that are very highly conserved. You can see here in this top line of the panel is the identity to mouse. So these antibodies, these targets are like 88% identical, 96, 97, 95, very highly conserved to mouse. And, uh, but if you compare the identity to chickens, it's significantly lower. This is pretty typical for chickens. You'll often see that it's 20% lower in chickens. And although in this case with CV1 it was only a few percent lower, but when you've got very highly conserved proteins, every little bit counts to help you get success. And you see that in these four examples, even though they're very difficult membrane proteins, we were able to generate high titer responses against them using chickens. So let me walk you in detail through a couple of examples. So one, uh, one of our most advanced programs is against the target, uh, Claudin-6. This is an oncology target. Uh, it's, uh, not expressed in normal human tissues but is expressed in fetal tissues but also expressed in cancers. So it's a great, uh, target for selectively targeting cancer. But it presented a number of technical challenges. It was highly conserved, as I mentioned, so that gave us pause. Highly conserved means that you may get a weak response or very limited epitope diversity. And in this case, we needed epitope diversity because of the second technical challenge, which was there's a very closely related protein Claudin-9 that only had 3 amino acid differences in the extracellular domain. So we needed that extra, that epitope diversity to enable us to find discriminating antibodies that could bind selectively to Claudin-6 and not Claudin-9. So, um, uh, so, so we, um, we have developed therapeutic antibodies to this, um, IND was actually just filed through our— we licensed this out to a partner, Contex Therapeutics, and they have filed an IND last month, and we're expecting it to enter clinical trials the middle of this year. But to give you a little look at the discovery campaign, we use chickens to generate the antibodies. We actually use phage display, but we deselected on Claudin-9 to really scrub out all those cross-reactive antibodies we didn't want. Despite that, we still got a very large number of antibodies. We got 68 sequence families generated from these animals that bind Claudin-6. But the key thing was specificity. We had to find a molecule that could bind Claudin-9. So we did do some screening on the clones, but really the key tool we used here was the membrane proteome array. This is actually a service that we also provide as a company for specificity testing of antibodies, the MPA, or the membrane proteome array. This is a collection of essentially all human membrane proteins that are expressed on living unfixed cells, so you can really probe native epitope cross-reactivity. And we have a platform to enable us to screen antibodies across all of these proteins to find highly selective antibodies. You can use this for for lead selection, or if you are filing an IND, you need the data for off-target binding. So our membrane protein array contains all of the, not just CLaudin-9, but all 24 Claudin family members. And you can see the results here. So these are 4 of the lead molecules that we had focused in on after screening thousands of of clones. And you can see these 4 clones that are screened across the membrane proteome array. You know, many of them had off-target binding to other claudins, but we managed to find a small number of these antibodies that were highly specific, like IM171, that selectively bound claudin-6 and ignored all other claudins and all other proteins.. And in fact, we did— we published a paper on this last year or 2 years ago showing the mechanism, and it's able to discriminate a single amino acid in binding between Chlordon 6 and Chlordon 9. And I think the reason we were successful here is because we discovered this antibody from a very diverse panel of antibodies that were binding different epitopes across that molecule. Okay, for a second example, uh, well, I can't tell you too many details, it's through a partner, uh, Merck, uh, but they wanted to discover, um, an agonist antibody. Now agonist antibodies are very exceedingly difficult to develop. You've got to hit just the right spot on the, on this, on the protein to enable to, to, uh, to get the signaling activity you want. And Merck had tried all the usual tricks over a number of years all the usual things— immunizing mice, phage libraries, all kinds of things— and were unsuccessful at doing it. So they tried it through our platform, and we were very delighted that we were able to succeed here. So on the right here is actually a functional GPCR assay showing that it's an agonist. So you can see here in black is the native ligand triggering signaling in this, on this GPCR.. And in blue and red are two of the antibody agonist antibodies we're able to discover. One was exceedingly potent in picomolar potency, very similar to the native ligand. And this was generated in our chicken platform. And as well as that, of course, as I mentioned, cross-species reactivity was important for them too, especially dog. And we were able to get an antibody that could bind non-human primates, mice, rats, and dogs. Okay, the third example is the glucose transporter SLC2A4, otherwise known as GLUT4. This is the insulin-responsive transporter that's upregulated when you take glucose in. And we published this story some years ago in PNAS. And again, using chickens, we screened for a number of antibodies that bind to this GLUT4. We were successful in binding this antibody to this complex target. It's a 12-transmembrane transporter with very small loops on the outside and was very highly conserved. It was, I think, 95% identical to mice. But what was really interesting was we found a very unusual properties amongst these antibodies. This antibody here that we're highlighting here, LEM48, this antibody was able to bind in a state-specific manner. So the transporters like SLCs, they operate in a teeter-totter mode. They oscillate between an inward open state and an outward open state. So in the outward open, the glucose molecule diffuses into the binding pocket. It teeter-totters and closes into the inward open to push the glucose into the inside. And we found that LN48 could selectively bind to the outward open state. And this antibody had an exceedingly long CDR3. It had 26 amino acid CDR3. And when we epitope mapped this using shotgun mutagenesis, alanine scan across it, we found that it bound into the little pocket that's formed when that outward open state opens up. So we think that was a critical feature for enabling us to find this rare property. So as I mentioned, chickens do have longer CDRs. I just want to show you a bit more detail of what that means. Again, here's this graph. That shows in blue the current therapeutic antibodies that are either approved or in phase 2 or beyond, and showing the length of them. And as again, most of them are around 10 amino acids, which is typical for a mouse. In orange is the distribution of CDR lengths from chicken antibodies. You see it's, you know, it's right shifted, and you get far more of these protruding shapes of antibodies. And some of these examples I talked about, the SIM21, the antibody that is in the clinic now, a chicken-derived antibody that had a 17 amino acid CDR. The Claudin-6 antibody I mentioned, which is the drug called CTEN76, that had between 18 and 20 amino acid CDRs. And this GLUT4 antibody, SLC2A4, that had 26. And even for chickens, you see, that's quite rare for a chicken, but that would be almost impossible to find in a mouse because this— next to none of these links you'd find in a mouse. So as I said, we've been, last few years, been developing our own pipeline that we license out. This is our pipeline here. All of them have in common is their complex membrane proteins. Many of them are highly conserved, and we think chickens have played a key role enabling us to find the right antibodies to develop a drug. And yeah, we're very excited about this platform. I think chickens have played a key role in the success of this pipeline. So just to summarize some takeaways and learning points today, really the industry has largely focused on mice. And I understand, you know, people do what people do. It's mice are the status quo, right? Everyone's comfortable using mice. But really there are significant problems with using mice. You, um, just many proteins are highly conserved, and mice are either going to not give you a response or very selective epitope coverage. And people are recognizing that, uh, they're starting to exploit many other divergent species apart from mice that give them these capabilities to find drugs. The evolutionary divergent species are going to give them much stronger immune responses. We see very strong responses to highly conserved proteins that would be very difficult in mice. In addition, the long CDR3s really enable you to find additional epitopes that may be off-bounds in a mouse because this can't reach them. So binding into pockets and finding useful epitopes that are functionally important. And third is that these animals are going to give you species cross-reactivity, and that's going to give you a lot more flexibility in using animal models in developing your drug. And our own experience using chickens as a platform is we've had a great deal of success with it. We get very good epitope diversity. It's enabled us to find very unusual antibodies that would be very difficult in any other system. So we find all these unusual properties like I showed you, agonist antibodies and state-specific antibodies and highly selective antibodies like the Claudin-6. For those who are interested in accessing any of our technologies, you know, working with us, if you— I did mention the membrane proteome array for specificity testing. We offer that as a fee-for-service. We also sell like particles, which are membrane proteins in these virus-like particles that are very useful for immunization. And we have epitope mapping services as well. We do antibody discovery in a partnership model, and we have various therapeutic antibodies to license. Very recently, we have a spin-off company called Cell Surface Bio, and this company is going to be focused on reagents, so generating antibodies to various targets. But the difference here is that, as you know, there are a lot of rubbish antibodies out there in the catalogs, and people— there's a lot of stories published of frustration of researchers with poorly performing antibodies. The difference here is that all of these antibodies that we generate, they will all be screened on that membrane proteome array, the 6,000 human membrane proteins, to show that they only bind one target. And on top of that, they will all be validated. We'll all test them in Flow and other applications to make sure that they actually work, as well as they'll all be recombinant-based so that there's no lot-to-lot variability. And if you have a particularly difficult target that you've always wished you have an antibody for, you can scan that QR code that you can see on the screen and tell us what antibody you wish you had, and we would consider building it and adding it to the catalog. Okay, with that, I thank you for your attention. This is our company again, but I'll flip back to the QR code in case you're still thinking about what antibody you've always wished you had, and feel free to reach out to us and tell us So thank you for your attention.

Thank you very much for that presentation. We do have a few questions, starting with the first one that came in. It's a question on immunogenicity of these types of antibodies. Structure-wise, they present differences compared to human antibodies. How or what about can you comment on the challenges of humanization of these types of antibodies?

Yeah, that's a really good question. So for chicken humanization, you know, at first blush you might think, oh, it's going to be challenging because it's so different. But actually, it was actually remarkably easy. Despite the evolutionary distance, the antibodies are quite closely related to humans and to well-behaved human frameworks. And our experience has been and others has been that they are pretty simple to humanise. You do a CDI graft and you really have to do a few back mutations, but largely they work. And there's only one framework to worry about, so you don't have to worry about all the different frameworks and that. But it's been remarkably well behaved. In terms of immunogenicity, that's also a concern. But again, people's experience so far, although it's limited, hasn't seemed to be a concern. There hasn't been any overt problems with ADA as far as I'm aware.

And there's a follow-up to that. Another question regarding developability issues. How do the different characteristics of these divergent antibodies impact their developability compared— well, maybe I'm not familiar with these immediately, but compared, for example, to Jane— oh, that one I do know— Jane 2017 or Raabold 2019 PNAS papers?

Oh yeah, yeah, I love those papers, yes, that was a great resource. Yeah, no, they've actually been very well behaved. I think part of that is because, you know, the closest framework— well, I'm going to speak to chickens, I don't— I believe the other species like camelids and rabbits, I don't think there's been any major problem there, but for chickens in particular, in particular, yeah, they have performed equally well as any other antibody. There's been no problems with aggregation or things like that. The frameworks that we use, heavy chain is VH323. It's a well-behaved human framework. So, you know, a long history in developability, and there are many therapeutics FDA approved that have the VH323 framework. So yeah, there doesn't seem to have been any major problem with developability.

Next question. With the SLC2A4 antibody that is state-specific, presumably this is functional too. Did you assess as intact IgG, that is to say with no steric hindrance then due to the Fc domain?

Yeah, good question. Yeah. Yes, we did test for inhibition, and as you expect, if the antibody is binding state-specific, it would jam the mechanism of transport. And indeed, that's what we observed. We observed that it inhibited glucose transport when we added that antibody.

Yeah. The question: when considering divergent species with peculiar, particular— peculiar is what's here— antibody structures, what do you think of bovine antibodies?

And how would they compare to Camelid? Yes, uh, yeah, cows have very unusual antibodies. Um, yeah, they are really unusual. They have, I think, up to 60 amino acids, uh, for their CDR3. Really, it's almost like an antibody within an antibody. Um, and I think, I think people have even cleaved them off completely as independent binding units. Um, but, um, yeah, it sounds, um very intriguing. I've seen some interesting case studies on that, but I haven't got much personal experience with it. But now you're really talking about a completely new protein. I guess there could be some, some issues with immunogenicity when it's that large. But again, I'm not, I'm not too familiar with that. But that's— but yeah, very intriguing. It's amazing, you know, you look in different animals and you think we just have a normal antibody, and you're surprised, and it suddenly does something different. So I really think people should start looking at other species. Maybe some— who knows what other species have. They, they're full of surprises. So it's quite amazing that these cows have these unusual antibodies.

Could you describe your humanization process for the chicken antibodies? At what stage do you pursue this in your discovery process?

Oh yeah, so initially we would discover antibodies, find leads and then do CDR graphs, just like most people would do, and 1 or 2 baby-back mutations, sometimes often none. It was so easy, like many times we would lose no affinity, that we actually— led us to develop a new method, which is actually at the library stage. So when we build our phage display libraries now, we actually humanize the libraries themselves. So we actually isolate just the CDRs from the chickens, not the whole antibodies, the CDRs. We clone them into human frameworks, so all the CDRs get shuffled, but that's mimicking the natural chicken system. Like, chickens naturally shuffle their CDRs. They're built for it. And so that, in that way, that we create phage display libraries that are fully human, and then we use that to discover them. So they're pre-humanized before we do discovery., and we'll pan for that and pull out fully human antibodies. That's what we did for the, for the Claudin-6 antibodies and others.

It's worked very well. Okay, well, I think you anticipated the next question, which was, do you humanize the Fc part of the chicken antibodies? And he— this person was thinking about the Fc-mediated antibody functions.

Oh sure, yeah, yeah, sure. The Fc region is, is fully humanized, and it'll be whichever framework is appropriate for the disease indication.

Next question: Does Integral Molecular or Cell Surface Bio have experience producing antibodies specifically for non-human species? And the thought there is companion animal biotherapeutics is certainly a growing space.

Yeah, that's something that we've thought about. We haven't done anything, but I think that's a real strength of the chicken system is that we will get— I mean, we've discovered antibodies that really, uh, that bind to any mammal. So you can have one antibody that binds all mammals, uh, assuming the gene is highly conserved. Uh, so, uh, we've seen that before. Some of our antibodies will bind, uh, dogs and pigs and sheep and bats and whatever crazy mammal you can think of. Uh, we've seen that with using chicken antibodies because we're accessing these highly conserved epitopes that are just unavailable out of a mouse.

Okay, people are now being creative and challenging you. Can we use buffalo Fc parts? Have you checked these?

I guess they would fit along in the cow family. Yeah, that's an interesting question. Uh, no, don't know. Uh, yeah, I, I encourage someone to go and look at buffaloes.

Who knows what they have. And very direct question: do you license out those chicken/rabbit humanized phage yeast libraries?

No, no, we work in partnership from— so we'll start with a target and we'll take it all the way through to lead candidates.

We don't license out things, parts of the process. Okay, one question that I skipped over a little bit that I'm interested in myself.

How do you deal with low-expressing targets? Yeah, that's a very common problem, very common. So the kind of targets that we work with, you know, GPCRs and ion channels, of course one of the major problems and challenges is their low expression. And what we have developed over the last 20 years, because we've been working with these kinds of proteins for many, many years, is we've developed this large toolkit of solutions. And it's not one tool fixes everything. There's multiple mechanisms that limit their expression, and we have developed this toolkit that we use to solve those problems. And we have a very good track record doing that. So it's always a great challenge to— you see a protein that doesn't express at all, and you can get high levels if you, you know, you— we had trafficking motifs, we We cut things off, we make point mutations, we'll make chimeras, all kinds of things. But we have quite extensive engineering capabilities to solve expression problems. That's a key thing to start the project is to get the expression up, because if you haven't got the expression, you're going to have a tough time.

Very good. Well, thank you very much. I think we've gotten to the end of the questions. So I give people a second or two to see if anything else pops up. But if not— oh, wait, wait. Nope, there was a thanks. Thanks, Ross. Thanks, Beth. In concluding, I'd like to thank Dr. Chambers for providing these insights into how divergent species can be used to access Difficult and conserved antibody targets. And I'd like to thank our audience for joining the webinar today and all of these great questions. An on-demand version will be available very soon. I will send a link to— by email to everybody who registered, so do look for that. Please feel free to watch this or any of our on-demand webinars when it's convenient. Thanks again and have a great rest of your day.