TRANSCRIPT
Announcement:
Welcome to The Chain. In today's episode, we bring you a panel discussion from the most recent PEGS Summit, exploring the future of biologic therapeutics.
Daniel Chen:
Alright, so let's get started. It's great to be with you here today. I'd like to start with Paul and Janine. Paul and Janine, you both individually have made huge progress for the field of antibody engineering. How is it that we didn't solve GLP-1 therapeutics with engineered antibodies?
Paul Carter:
I would just point out a few things. I think that's you know that they are. You know, GPCR targets which are harder with antibodies but not impossible. You know, I could sort of congratulate their the your you know the colleagues who are developing peptide drugs. I think that you know what they've done is is inspired. And I think it is you know a natural place for peptides because of you know just being able to take a natural like end and engineer it. I think it's a terrific you know starting point. Could you do it with an antibody? Quite possibly, but but I you know, as I kind of commented in in my opening remarks, you know, just our try trying to you know mimic a natural like end is not easy with it with it with it with an antibody because of you know the differences in valency, you know, and geometry. So I think you could do it with an antibody, but I think our you know this is a spectacular you know success for for peptides, and I share your enthusiasm and and celebrate that. So Janine.
Janine Schuurman:
Yeah, I think we need to apologize for the field, maybe, Paul. But I think I think you touched an important topic. Some targets might be fit better, and I think that's maybe a key take-home message already, might fit better to a particular modality than the other modality, and maybe the antibody fits to this approach might not be the most optimal one. Already thinking about the natural peptide underlying this mechanism. Yeah, I think you could say the provocative title uh uh already shows that it's it's far more uh nuanced than we generally think. But this is an example where maybe also we as antibody engineers might never have put that much effort in, Paul. What do you think?
Paul Carter:
I think that I think that's possible. I also kind of wanted to come a comment on their you know the choice of mode modality. And it's kind of well, what clinical problem are you are you trying to solve? And you know, what tools do you have in your you know, toolbox? I mean, if you if you're an organization and you have the luxury of expertise in multiple areas, that then maybe you try multiple approaches and you know take an you know an agnostic approach as as to you know what works best. I mean, but of course, you know, in the real world, many of us have more expertise in one area than other, so probably you're going to try that.
Daniel Chen:
But Paul, you you mentioned that you thought that agonism was one of the things that maybe peptides could do really well. But we have fantastic antibody agonists, particularly in the immune field. Janine, you've you've made some really powerful agonists to targets like DR5 and Oxford as well. So why not GLP-1?
Paul Carter:
I well I would say you absolutely you know could. And I kind of think of, you know, obviously there's a lot of great work that's been done with agonists, including from Chris Grassier. So sure it's possible. I think you know it's possible now.
Janine Schuurman:
Maybe a question, Jonah, to you. If you think back about how these GLP-1 peptides how what was the initial hypothesis to work on that? Why did it evolve? How was this modality picked? Do you have any insights in that?
G. Jonah Rainey:
Yeah, I mean, you know, the history of of GLP-1 goes back to really trying to get good glucose control for diabetics. It's an important uh metabolite that then comes back and and signals you know in order to help have the blood glucose level of diabetics controlled. But it turns out I guess that it has a lot of receptors and a lot of different tissues, including in the brain. And so some of the signaling, some of the satiety may come from brain signaling, some of the effects that you'll see reported, you know, around addiction and compulsive gambling and things like that. I think most people are assuming until we see otherwise that those are probably brain-mediated effects. But there are other you know, tissues and other receptors and and pancreas and stuff where you know it's pretty straightforward blood glucose control. But it's clear now that it does a whole lot more than that. It really has these signals where there's this you know communication between the gut, the pancreas, the brain, where you know the are you still hungry sort of sort of signals as well. And so, you know, you can have a tremendous effect with weight loss, but then that also you know feeds back into uh diabetes control as well.
Janine Schuurman:
And and the and the journey to find the the GLP-1 peptides as a drug, what kind of modalities have been tested before? Was it immediately going into this peptide approach? Do you know?
G. Jonah Rainey:
I mean, I assume so I don't really know. I'm gonna speculate. I mean, I think pip probably people tried uh tried whatever you know, whatever was at their disposal. You know, you you probably know that there's a small molecule or for glipron that we have anyway, late in, you know, late in clinical trials. And so you know, there is another GLP-1 non-peptide agonist. Um so I suppose you know small molecules and peptides can do it. So that's gonna just give Dan more fuel for his fire over there, I think. Why not antibodies? But yeah, I kind of liked Paul's hypothesis. It's probably hard to find something on an antibody that fits perfectly into that pocket. And I think particularly then if you want to make something that's multi-specific, there are additional tools that are available for engineering peptides that are, I won't say they're not available for antibodies, they're just much less available for antibodies. And these involve a lot of modifications, non-natural amino acids, things like that. Of course, there are ways to encode non-natural amino acids in eukaryotic expression systems. But the number of non-naturals you can load up on these peptides by using a mixture of like synthetic approaches as well as expression and biological systems, and then ligating the different pieces together, which is how most of these things are made. You know, the just the tools and the amount of space that you have to explore on a per amino acid basis is just so vast.
Daniel Chen:
But so how do we get here? I mean, GLP-1 is an old target, right? This is not like, oh my good, we stumbled upon this great target. It's been around for a long time. And yet it's maybe been the last 10 years where the field really took off. So what happened? How did how did people get there?
G. Jonah Rainey:
I mean, GLP-1 has been a great drug ever since it was first released, especially for diabetics. And there's always been a measurable weight loss associated with it. I mean, it's obviously, you know, weight loss is a really big indication with with associated medical benefits. So I think that it's, you know, when we've kind of come through and there seems to be something that I don't think that we fully understand about these , you know, fatty acid acylation, particularly the diacid versions, that are that are either giving them the appropriate half-life or getting them to the appropriate place or allowing them to go to the appropriate places in order to do some of some additional signaling to get these deeper these deeper effects.
Daniel Chen:
But it's funny because it's not like we haven't tried different peptide engineer peptides for other indications as a field, right? We've we've seen all sorts of different structures in for diseases like oncology go into clinical testing. But they seem to always have struggled with their half-life and exposure. So is that really the big breakthrough here? Is it the ability to get one or two week half-life, or do you think it's something else? Paul, Janine, do you want to jump in here? Yeah, go ahead.
Janine Schuurman:
I think it it all comes back to the kind of targets you like to to to to engage. What's the best fit? Paul already mentioned it. Does it fit? Should it be an antibody? Would it be an antibody target? What kind of size do you need? Stoichiometry, geometry, affinity. I think the antibody in itself is a fantastic molecule. We are we can engage also beyond the half-life of the FC, which was already mentioned, and also part of some of the peptides. We can also use the FC for additional engineering interaction, the natural IDG interacting with FC gamma receptors, complement system, or we can half-life extend it. We could do a lot with this FC in the natural context and the valency of the molecule or but in particular also the size and the type of antigens we can target. Peptides might fit to particular targets, antibodies might fit to other targets. So I think it's generally said very it all depends on what you aim to achieve about the purpose.
Paul Carter:
Yeah, yeah, I think that's kind of nicely said. I mean it's not sort of one size fits all. I mean, I think it's great as a community we have, you know, more tools in in the toolbox. And I think it's you know, maybe not uh you know simple to figure out, okay, which is going to be the preferred tool for this case, but I think it's you know, the better definition of what of what it is that you're trying to accomplish, you know, clinically including, you know, desired you know PK and is I think the better shot that you have at starting from a place where you're picking the modality or a few modalities which have have the clearest trap record in you know in that kind of, you know, problem. It's not to say that, well, you couldn't do it with an antibody or couldn't do it with a peptide, but the there are probably use cases which are are kind of more more natural choices for for one that than than the , you know, other. I mean I think with the you know antibodies and the engineering capabilities that Janine mentioned and some that I mentioned in my sort of intro remarks. I think being able to tailor you know the half-life and it you know if you want it to be able to have half-life of our you know sort of months can be a real plus in terms of being able to get to to very infrequent dosing. So it really gets down to the the details of the specific problem that you're trying to solve and what are plausibly their the the best tools that you have in your toolbox.
Daniel Chen:
But if they can push peptide half-life to 21 days using more advanced lipidation approaches, using better protection against degradation, are you ready to go and start creating novel drugs with peptides rather than antibodies?
Paul Carter:
I would say it depends what you're trying to do. I mean, you know, as Janine pointed out, you know, there there are lots of things which, you know, FC-mediated functions that are I think would be hard to recapitulate in a peptide in terms of our you know killing by secondary, you know, our immune cells or the whole kind of T cell engager field. So I think it gives us more you know versatility, it gives us you know additional, you know, options. So I don't actually I'd also want to uh at the risk of mudding the water here, they're saying that well, they're not mutually exclusive. I mean, we've kind of talked about you know peptide FC fusions or or just taking peptides and grafting them into CDR. So I think those but you know they're both have distinct flavors which can be exploited in their own right, but I think there are there are hybrid things that can be done that could also be pretty useful and combine different desirable attributes of antibodies and peptides.
Daniel Chen:
So we'll get there. Jonah. Peptides. You mentioned a little bit about brain penetration and activity. Your thoughts on that? Is that a special piece for peptides as a strength or or not really?
G. Jonah Rainey:
Yeah, so I think the jury is still out. You know, peptide penetration, because of their smaller size, they they can get into the brain better. But the, you know, the peptides that we that you know the the that are being developed spend most of their time bound to HSA. And so HSA does not go across the blood-brain barrier very well. So you know, it's tempting to to speculate, but you know, real science will have to be done. That as agonists, maybe it's you know, achieving some C max is not really what we want to do. Maybe a little trickling in is actually exactly the exposure profile that we need in order to get the effects that we're seeing. I mean, I think these are these are questions that the field's gonna have to address.
Daniel Chen:
And what what about the non-natural amino acid part? That seems like a pretty cool technology piece that can be difficult sometimes with antibodies. How powerful is that? How would you think about applying the non-natural amino acids? Janine, any thoughts on it? Not really.
Janine Schuurman:
Not really never thought about it. Paul, did you?
Paul Carter:
I mean it's not my expertise, but I think that's sort of one dimension of peptides, which which is kind of a core strength and being elf to really have a toolbox of you know dozens, if not larger numbers of non-natural amino acids to you know remove you know liabilities for you know protease cleavage or specific handles for for you know so lipidation. So I think that that's you know from looking from outside of of their peptide therapeutic field, sort of into that field, that kind of strikes me as being one one of their uh sort of notable you know strengths.
Daniel Chen:
Does it also help with immunogenicity? Like if we engineered using more non-natural amino acids, is that helpful or not sure?
Paul Carter:
I don't know. I mean, I would say it maybe. I mean, maybe you know, if you made your peptide with D amino acids, you know, maybe it would circumvent the our sort of presentation. I'll pick up a little bit on the non-natural amino acids and antibodies. When I was at AstraZeneca, we acquired a non-natural amino acid, you know, a company that that had a non-natural amino acid technology for you know incorporating in Chow cells. And, you know, certainly you can use them to put different shapes into binding sites, but you hit a limitation. You have to use one of the stop codons to encode the non-natural amino acid. And so you don't have so many options there. And then, you know, also you have potential for premature terminations and things like that that can happen. So you can't get you can't get too crazy with the number of non-naturals that you include, but you could include a few, particularly if you're using the same non-natural in multiple locations within the antibody. What we found it was particularly useful for was creating conjugation handles, for instance, for making antibody drug conjugates. Others have also, I think Sereno Therapeutics a while ago, you know, had an idea where they put conjugation handles on two different sides and used non-natural amino acids to encode ways to make bispecific antibodies by bringing those together. So, you know, they can they can expand the engineering toolbox uh for sure uh within within an antibody space and a eukaryotic expression system. But that you know, you just can't load them up the same way that you can when you have a semi-synthetic approach to making your drug.
Announcement:
Are you enjoying the conversation? We'd love to hear from you. Please subscribe to the podcast and give us a rating. It helps other people find and join the conversation. If you've got speaker or topic ideas, we'd love to hear those too. You can send them in a podcast review.
Daniel Chen:
So, Janine, you've obviously spent your career engineering around antibodies. Where do we think that antibodies are really gonna continue to dominate therapeutics because of their features?
Janine Schuurman:
Yeah, we already touched upon that. I think it's it's a fantastic natural molecule with respect if you talk about I2G1, for instance, just keep it simple here, with respect to the FC mediated effective functions, interactions with FC gamma receptors, complements. I think the molecule also is very suitable, and that's we've heard many stories already at the conference about T cell engagers where the spacing might be important, which I think might be maybe doable with a peptide. I'm not saying it's undoable because the synapse could be maybe have a benefit there. So that might be a field to explore. But if you start to think about more logical end cating or cating, we have uh even heard a story about not cating yesterday. I see that antibodies can do that probably easier than peptides, at least at this moment in time. I think also the size of the molecule in relation to particularly cellularly expressed receptors might be often a better a better fit, but not always. And we have also not explored it that broadly yet, as as what we did with antibodies. So I think it's more nuanced than just saying it can be done. But we haven't found maybe the tools I'm using your words now, Paul, the the tools in our toolbox yet to too broadly, yeah, evaluate it and test it.
Paul Carter:
Maybe I could comment a bit on you kind of look back at where we've been are particularly successful thus far with with our antibody therapeutics as a field. I think you know oncology is particularly striking, and also, you know, and I few showed a few slides from a review that we wrote this year, and to our sort of illustrate the progress in the in the field. We are commented on our antibodies targeting HER2 and B cells. And I think what particularly struck me is their our you know, the history of antibodies targeting HER2 and our B cells, it almost kind of you can see the progression in the field in terms of the improvements in the technology for, you know just chimerics, humanized antibodies, effector enhanced, you know, uh ADCs, you know, T cell uh engages, our the application of our antibodies to target cell therapeutic, so notably in CAR T. So I think I think part of the the power in the in the antibodies is is the sheer momentum that we have as a field, and it's not like well we just made 200 IGGs and we just got pretty good at it. It's just kind of like the the field has exploded in terms of you know multiple alternative ways of sort of using antibodies. Of course, we you know we've learned things along the way, things which are you know harder. We talked about sort of you know agonists. I think we've starting at a better understanding of how hard it is to get antibodies into the brain, and as a field getting at least a little bit better at do it doing that. I mean, there there are still sort of you know other areas like you know oral delivery looks which looks really hard and perhaps you know at least a little easier with peptides. So I think personally think there's a lot of mileage in our you know, we with our antibodies are just because of just this incredible knowledge that we have with as a field and that just the that the that the uh you know that they're that the the diversity of different antibody formats. I'm kind of, you know, it's always kind of stunning to come to a meeting like this and hear about kind of new applications and new kind of clever engineering strategies. So I think there's, you know, the there's our you know there's our I think the glass is more than half full in the sense that there's plenty of opportunity still left with antibodies, but I think there are are are going to be therapeutic challenges which are just maybe better better suited to peptides or other modalities come to that. I mean I think it's it seems like they're trying trying to target intracellular targets with antibodies is not something we know how to do and and I think our you know our small molecules or maybe if you can figure out more routinely how to get peptides inside cells, maybe that's another opportunity for peptides.
Daniel Chen:
So you guys all hinted at something around the physical space that we have to cover in more complex biology and maybe something about the physical characteristics of of IgGs as our base format. Can we not make those with peptides?
Paul Carter:
Can we not just start to engineer larger peptides to to bridge those spaces well I mean we you certainly could and that kind of a comment I sort of made earlier well you can you could obviously make FC fusion some people haven't been successful doing that or you can you know take an antibody as a scaffold and graft in you know different peptides into different CDRs. So so I think it's you know it's it's entirely possible. It's you know it becomes down to you know what's the rationale what you you could do it but why do you want to do it? And is it something that you could do with you know a very long-lived peptide where where that that seems to offer a unique advantage over you know uh you know antibody or something else it's I think it's almost an embarrassment of riches problem in the sense of there are just so many different ways that you can engineer antibodies peptides are other other modalities that I think are kind of a grounding in what what what it what is the what is clinical application what problem are you trying to to solve and then focus your kind of ingenuity on that rather than build it and they'll come what what do you think Jonah would you engineer more with peptide hybrids with antibodies?
G. Jonah Rainey:
Yeah for sure um yeah I kind of want to just build on what what Paul was saying which is that uh a lot of modalities can be made to work but that doesn't mean they're the best modality. So I think you know we really win when we have multiple modalities at our disposal and we figure out the best way to put them together to make the drug that's really differentiated. And maybe it's you know maybe you could make a complex multi-domain peptide or stitch multiple peptides together with linkers that bound to FC gamma Rs, bound to you know you could modify it to make it bind to HSA you could have something binding CD3 and a tumor target. You could do all that, but at the end of the day you you know you would want to look and you would want to say well is this better? Is this better in some way? Because otherwise why throw out a perfectly good modality that's been working for decades. I think we will in the future not just see peptides put on top of an FC but peptides added to antibody activity. So that you know when when the binding interaction is best with a peptide you've got the peptide and then you can get an additional activity out of an antibody fab arm or some other binding moiety that's also associated with the that you've stuck on the antibody scaffold.
Daniel Chen:
Well so with the explosion in these more complicated engineered antibodies that interact with all sorts of targets, right? We talk about bispecifics and multispecifics all the time here. Why don't we see more of these mixed hybrids with peptide ligands or interaction domains as part of our therapeutics? I don't know. Janine, your thoughts on it?
Janine Schuurman:
Maybe it's an invitation to the people here in the audience to start to explore and and think better about the design parameters and the tools you have available and very pragmatically that might also be the restriction you have depending on what kind of company you're working in or academic situation you're working in. But I think it's also an invitation to the field here to think a little bit open the brains, open the mind and think on the structural constraints on the design parameters of your best drugs and think a little bit wider than the might have been done previously.
Daniel Chen:
So you think it's just that we haven't really done it we've been sticking with what's more comfortable.
Janine Schuurman:
Yeah I think that's what we do as humans, don't we? We are comfortable but I think for innovation and really bringing the field forward it might be super helpful when we broaden our thinking a bit. Of course we are generally restricted in our labs or where we are working to the toolbox we have but if you have the possibility think it through and think what makes most sense to solve the problem you aim to achieve with your drug so really take your design criteria and the problem more serious than yeah the tools you have and maybe find a collaboration or get it done.
Daniel Chen:
Alright so let's talk about that objectives what are we trying to solve here as a field? What do you think the big problems we need to solve whether it's peptides or antibodies or hybrids what are we trying to solve?
Janine Schuurman:
Yeah maybe I can jump in. I think what the whole field is solve... having having trouble in regardless of the disease area is the therapeutic index I think we want more efficacious drugs less problems in toxicity safety so I think if we could do something there uh because most of the diseases we're working on are diseases of self so a lot of the binders we find so we're all that's also why we talk about masking why we talk about end cating or cating not cating to make our drugs more selective while forcing potency. So I think if you could come up with new ideas to work on our therapeutic index in general that could be fantastic. But there are probably also other problems. Jonah.
G. Jonah Rainey:
I'll go even broader than that I mean the top line umbrella goal really is to make patients live longer healthier more comfortable lives and so you know we've made a lot of inroads into that through curing diseases you know we're continuing to cure is a you know is a tough word you know often the drugs that we make you know it would be nice if their effects were longer and there weren't remissions and stuff from the diseases one area that I think we really as a field need to do more in is pain. So that's feeds into the more comfortable lives.
Paul Carter:
Pain I think is is something for for our generation to solve you know without the use of of drugs that have you know addictive and and you know issues around making it so you can't think clearly.
Daniel Chen:
So lot lasting effects therapeutic index. Paul.
Paul Carter:
So yes so let's try and think of something which hasn't been said yet say you know in oncology and you know beyond improved TI especially these you know ferociously potent formats you know clearly our our greater response rates better duration of response if look in say A and I you know look at the you know antibodies for inflammatory biodisease it would be great if we could do it seems like we've kind of hit a plateau in terms of response rate and it would be it would be great if we could figure out you know well maybe it's you're not going to get there with any combination of just one or two targets. Maybe you have to target more pathways. You know, looking at other our clinical settings we we've barely you know scratch the surface in terms of being able to our our address a devastating impact of neurodegeneration it seems like that that is that that is a field which is both both uh incredibly important but obviously very difficult in terms of you know clinical trials and typically very very long our duration clinical trials with many patients and so that's our an area that I hope that in sort of coming decades that we make more progress in.
Daniel Chen:
And I I guess I'll just weigh in as well. I think that we've lived our careers mostly in the era of single targeted therapeutics, right? And all almost all The blockbuster drugs go after just a single target, mostly with IgG antibodies. But I think for the future we need to be able to leverage the remaining 99% of biology to make much more powerful drugs. So that's a four big areas to try to solve I think we'd like to invite questions from the audience there are microphones if you don't mind. I don't know if we're able to pass out any microphones but obviously the we've covered a lot of ground and would love to field additional questions.
Paul Carter:
So there are kind of microphones or four they're kind of like a midway through there.
Daniel Chen:
And while we're waiting for the first one there was a lot of talk about AI here. Is AI going to help us solve the future of engineered therapeutics starting with peptides do you think?
G. Jonah Rainey:
AI is going to have its role in everything that we do right and I think time will tell how big that role is but it's already having an impact it's already having an impact in understanding the design space and coming up with potential sequence solutions but it kind of comes to you know your provocative title of this do peptides replace antibodies. I mean I certainly don't think that they will but it's a tool I love to have in my box. Is AI going to replace discovery and engineering paradigms that we use already? Not anytime super soon but it's gonna help us along quite a bit.
Audience Member #1:
Hi everyone so great discussion I'm Stalaya Patel from Constructive Bio and I want to take you back to the conversation about the non incorporating non-natural and non-canonical amino acids into peptides. I'm gonna skip to the punchline first. We're a company that can put multiple non-canonical amino acids into peptides recombinantly and also into large proteins. So my question to you and the panel is if you are if you can put multiple non-canonical amino acids and bring novel chemistries into your modalities um what what would be the most valuable and most exciting functionalities you want to bring in to large proteins and into peptides?
Paul Carter:
Well I think one obvious application which I think is already being done is the area of of ADCs and being able using your non-naturals for spike site specific conjugation or may maybe for for dual payloading.
G. Jonah Rainey:
I think as we begin to bring together some drugs that are peptides now and we want to put them together with drugs that are antibodies you know CMC becomes an issue right so if you have to synthesize some of your peptide grow some of it up in E. coli or whatever uh stitch that together then take that and conjugate it to an antibody that you've made through sort of a normal antibody platform process that becomes a really really complex set of processes. So if we're able to recapitulate some of the semi-synthetic stuff that's done with peptides in a in a fully you know genetically encoded system. I think that's going to really lower the barrier to being able to make some of these fusion drugs so I'll be extra provocative I've often joked about you know the future being we we can make silicon based nanobots with our engineering approaches what would happen if we made our entire therapeutic out of non-natural amino acids would you ever see that as a future it just sounds like a challenge but I'm I'm gonna come back to my previous comment of should we? Only if it's better.
Daniel Chen:
Only if it's better.
Audience Member #1:
Maybe if you can answer that.
G. Jonah Rainey:
Yeah do you want to answer the question Janine suggests we'd like to invite the fifth panelists to the to the discussion.
Audience Member #1:
No I think it's entirely possible right with with the technologies that we've got in expanding genetic code and also being able to produce these recompondently now in both bacterial and mammalian systems and thinking about how we can make polypeptides and I think the applications that we're thinking about in constructive bio are in therapeutics for now but could also move into other other sectors in terms of you know um agrochemicals and also into novel biomaterials as well where we are thinking about designing polypeptide chains with multiple non-canonical amino acids and you know long long proteins as well for like thick applications like more sustainable and biodegradable bioplastics for instance but yeah but the current you know the the application within therapeutics is also such a big space and there's so much we can still do with within within that.
Daniel Chen:
Thank you.
G. Jonah Rainey:
Speaking of big space if I can just add one thing in there one thing that's occurring to me throughout this part of our conversation is you know we we have a hard enough time getting through the complexity of the space with natural amino acids. If non-natural amino acids enter in and can be put in in unlimited numbers and then you can dream up any structure you want um it really is going to create a complexity uh that is going to require a computational based design in order to be able to fill it very much.
Daniel Chen:
Next question please .
Audience Member #2:
Yeah, Anatiso Beyer I mean you've mentioned all these wonderful degrees of freedoms with peptides right so that you could could do everything to a certain extent antibodies have shown a certain level of productivity obviously lots of surprises always but there is it's been treated as a platform so how do you see the how do you see the path to optimization and and going through all these degrees of freedoms do do we have everything we need how is it gonna look like?
G. Jonah Rainey:
Everyone's looking at me I mean I think I pre-answered your question actually sorry that I didn't wait for you to ask it that would have been the perfect setup I think to you know what tools are we gonna need we'll use some of the current tools that we have right so we'll use you know display and screening and things like that but I think a lot of it is gonna have to be computational in order to get into that space
Paul Carter:
Maybe I can also come comment you know that's I it feels like that's one of the challenges in this you know era of you know antibody engineering you get into you know bi-specifics trispecifics you change you can vary the the valency you can vary their you know the format you know sort of lankers and I think we've already heard some some uh really nice examples at this meeting and in trying to figure out how to how to use AI to you know simplify that design problem because you could there are so many permutations of different parameters that it it's it's sort of way beyond what we can currently do in terms of our experimental work so I think there's I think that will be it already obviously is but I think will become maybe a more important in the future with with the applications of of AI to help us figure out how to sort of judiciously sample uh you know a a vast uh design space well that sounds really complicated how far off are we I'm not sure I mean there have been some you know examples at this meeting in terms of you know designing T-cell engages you know using AI and I don't understand the AI what goes on behind the curtain enough to to know you know quite how well that works so I think you know let's see I mean it you know I think it's I you know I would like to in the in the in the AI space it's always pleasing when you kind of see comparisons of what the AI did with your best human expert and so that it feels like it's a contest rather than showing me okay well I was successful with AI but well if I did five rounds of iteration maybe I'd I would get something as good or bet better with my human expert.
Daniel Chen:
All right come on put your nickel down is it one to three years three to five years or five to ten years to what? To really being able to use it to design complex molecules with lots of variables and interactions.
Paul Carter:
Well I've got a struck by that Neo Bull quote that it's it's hard to predict especially about the future. You're not gonna phone a friend are you no but I mean there are people kind of trying to do this now. I'm not trying to weasel out and not give you an answer but well I guess I am a little bit but it also depends on you know what you consider as as their you know their how you define success in terms of well you know just uh uh our uh our what what it is that you're actually trying to to to accomplish you know uh and whether or not you're actually doing this better than sort of conventional methods. I mean I suspect there were probably no more than a a few years of being able to do that or being able to to our whether we can do it routinely within a few years I'm not sure but it feels like we're on the cusp of being able to do that.
Daniel Chen:
So Paul Carter said three to five years? Please
Audience Member #3:
Nice discussion so far. I'm Vinod Kurala from Takeda. I was just trying to touch base on the panel discussed a little bit on the quality of patient life extending the therapeutic index in that space I was just curious you know antibodies try to treat a disease which is already there how how would we try to prevent the disease from the happening from the first place? And we have some examples in the clinic uh for cervical cancer we have a vaccine for prostate cancer we have a vaccine but not others so what does the panel think about expanding that where we can prevent the disease then treat a disease?
Daniel Chen:
Well I'll take a stab first I think this this panel discussion actually focuses on the probably one of the most powerful preventative uh medicines we've ever seen and that's GLP1 agonism. Um when we talk about the effects of GLP1 on Alzheimer's or autoimmunity or on or oncology, I don't think we're really talking about curing things like metastatic disease, right? We're probably reducing inflammation broadly and it's it's a nice example of how inflammation may be so fundamental um across a core range of human diseases. Janine?
Janine Schuurman:
Yeah, I agree with you it's so this drug in particularly is definitely a game changer with respect to prevention but I think the question you ask can't we do more? Are there more peptides or more drugs we should develop which are more in the preventive space than we are currently doing? Is that is that actually your request to us?
Audience Member #3:
Oncology specifically specifically
G. Jonah Rainey:
I had an answer all you know loaded up and ready that isn't oncology but I think for a lot of things that that we're doing, you know, we're already doing this right we're already taking things the easiest thing to do to get a drug approved in the first instance is to take it into a disease modifying population because in order to run a clinical trial where you're showing that you've got prevention, you have to treat X patients for every event that you're gonna, right? It's this number needed to treat principle, right? For every event that you're trying to prevent from happening. So I think you're gonna see that as a follow-on with a lot of drugs. Certainly like in the amyloid beta depletion space, people are running clinical trials um in a preventive mode. So you make a really good point this is something that where there's a real opportunity to really have a great impact on health by you know stopping diseases from happening in the first place. And I think some of the things that are developed as a therapeutic intervention are going to work even better as a as a prevention. However, the one thing we didn't none of us came up with I guess with the you know what do we really need to do, in order to move the field forward, when you need to treat that many people, you've got to get the costs down. So in order to do that, to treat people broadly, we have to have cheaper drugs. Are you saying that would be peptides? I'm not not saying that would be peptides.
Daniel Chen:
We got two more two more questions here. Please.
Audience Member #4:
Yeah, thanks for the discussion. Yasmin Zuda Anderson, Roche. So I have a question to the peptide field. What is the experience regarding immunogenicity? So if you can't use any naturally occurring peptides, and say you would have to use de novo-generated peptides, what's the experience here? And how risky do you see this immunogenicity-wise?
G. Jonah Rainey:
Yeah, I mean that's it's always gonna be a question. I mean, anything where you can make a T cell epitope and a B cell epitope could be immunogenic. It's just something that'll have to be evaluated case by case. Um I think certainly there's a lot of experience putting peptides that have a lot of non-natural things into the clinic and not seeing ADA. So I think, you know, can I conclude from that that they are less immunogenic than than a larger molecule? I'm I guess I'm not willing to go that far, but certainly there seems to be a lot of space to use them.
Daniel Chen:
But it is true. Like you look at that trifunctional GLP1 agonist, there are a lot of modifications there. And I guess it's not immunogenic, otherwise we wouldn't be talking about it. Was that predictive model based, or is there really something special about I don't know, the length or structure of it now? No question.
Audience Member #5:
Thanks. So Daniel opened by asking why did antibodies miss the GLP agonism space? I would turn the question around actually and say, why did peptides take so long? Thank you, John. But I think maybe for discussion, you know, what are the the barriers now, apart from half-life to an expansion of peptides as drugs?
G. Jonah Rainey:
I mean I would say peptides have been doing it for hundreds of millions of years, right? So but what is not as drugs. Not as drugs. So what are the barriers? Is that the question, John?
Audience Member #5:
Yeah. Well, what yeah, what can we do to accelerate it, actually? I'm trying to get out of antibodies.
G. Jonah Rainey:
Do you have any ideas in mind while I stall and try to think about it?
Audience Member #5:
I can tell you about our little company, Maxwell Therapeutics. No, I won't I won't do it in the size, but constrained peptides, I think, a nice idea. Cyclic peptides. Cyclic constrained peptides, you know, robust, potentially cheap, potentially made in plants, etc. You know. Have you thought of an answer yet?
G. Jonah Rainey:
Sure. I mean it's gonna be it's gonna be more facile CMC things. I mean I think that semi semi-synthetic approaches and stitching things together, yeah, we can do it. Can we deliver it to you know billions of people across the world? You know, that gets to be a little bit more complex. So we have to go to simpler simpler production systems.
Audience Member #5:
So are the GLP agonists flexible peptides or are they anchored? And you know what does that how does that relate to affinity for target?
G. Jonah Rainey:
Yeah, I mean are they relatively flexible. They, yeah, I don't know. I'm not gonna speak outside of my knowledge. One last question behind you here.
Audience Member #6:
Hi, I'm Harry from Bicycle Therapeutics. Exactly addressing the question which has just come. I was curious, you mentioned about D amino acids and where you formats mirror image phage display or mirror image ribosome display or any other mirror image type display technologies might influence fully synthetic D-peptides in the future clinic setting.
Paul Carter:
I think it's an interesting idea. I think I've kind of sort of wondered, you know, with their sort of de peptides whether that that really you know gives you a lot, say, in terms of you know greater stability for you know oral delivery and a way of being able to reduce or or or avoid immunity. So I think it's worth exploring. I, you know, it's not really my eras, I don't know how much has been done there, but those are the things which kind of immediately sprung to mind.