All right. Welcome. We've got Taylor Schreiber from Shattuck Labs joining us this morning. Thank you so much for being here, Taylor.
It's my pleasure. Thank you.
Awesome. Let's jump right into it. Shattuck is working on a novel mechanism in a space that is pretty familiar to some folks. It's on the TL1A axis. Can you walk us through what your target is, why it's different from other TL1As, and maybe some of the biological rationale for making that difference?
Sure, so TL1A is an inflammatory ligand, and it signals through a receptor, which is called DR3, and like many receptor-ligand interactions, there can be one reason, if you look at many pairs in the TNF superfamily, there might be a good reason in some cases to go after the ligand and not the receptor if you have a ligand that binds multiple receptors or a receptor that binds multiple ligands. That is not the case in the TL1A pathway. TL1A only signals through a single receptor, and folks focused on the ligand, I think, for two primary reasons. One was that Stephan Targan developed the original TL1A blocking antibodies that became Tulisokibart. And once something works, there's, of course, always a mentality to stick with what works. But it's also true that the engineering challenges associated with building a ligand-blocking antibody are simpler than those associated with building a receptor-blocking antibody.
And maybe we'll get into that.
Certainly. Well, let's get right into that right now. I mean, people sometimes get very nervous about going after receptor-blocking antibodies. We're hearing about paradoxical activation, safety liabilities. So can you talk a little bit about what gave you confidence to go after the receptor and maybe what data you've seen so far that give you confidence you're not going to run into those problems in the clinic?
Yeah. So correct. The concern is that you'll build an antibody that you believe is an antagonist that ends up having residual agonist properties. And the reason for that concern is that all TNF receptors, including the TL1A receptor, have to assemble into trimers in order to signal. And when you have a monoclonal antibody that has two arms, each arm of that antibody is able to cross-link two different subunits of the receptor. And so when you are developing and screening antibody clones, you can look for a few things to give yourself confidence that there is no agonist risk. The first thing you look for when you're screening clones is to find antibodies that don't cause your receptor to be internalized. That ends up being an early indicator that in subsequent cell-based assays, you'll see agonist signaling.
Unless you're making an ADC.
Yes, correct. The second thing you do is, so you weed out a whole bunch of clones by doing that. And then later on in vitro assays, you use primary cell-based assays to see whether your antibody titrated up to very high doses in any way replicates the co-stimulation that TL1A delivers. And again, you will weed out many of your previously promising candidates by doing those sorts of screens. Ultimately, if you had an antibody that activated DR3, the TL1A receptor, what you would see is proliferation of different lymphocyte subsets about a week after you give the antibody. And when we took our lead antibody into monkeys, we only took it into monkeys because it looked great in all of those preclinical assays. And in the non-human primate studies, there was no evidence of receptor agonism in any animal at any time, at any dose.
And so we were very confident about the profile going into humans. And now we are about halfway through our phase 1 study, and so far, so good.
So those are some of the liabilities, maybe. But let's talk about some of the potential benefits to targeting the receptor. If everybody's going after the ligand and we know it works, what's the potential differentiation going after the other side of the equation?
Yeah. It relates to two potential areas of superiority. One is easier to understand, and it relates to a better immunogenicity profile. The second relates to a superior efficacy profile independent of whether the immunogenicity profile is superior. And so I'll start with the immunogenicity side. What every TL1A blocking antibody measures as their pharmacodynamic biomarker to pick a dose to move into phase two is the extent to which your TL1A blocking antibody leads to an increase in the total concentration of TL1A in the blood. At the same time that is being measured, every antibody shows you that the concentration of "free" TL1A goes to zero. And so a two-log increase in total TL1A in the setting of eliminating free TL1A proves that what you are measuring is an immune complex between the therapeutic antibody and soluble TL1A.
You mean that you're stabilizing the soluble TL1A by binding it with the antibody.
Correct.
That's what's driving the concentration increase.
Correct. And so that is a shared source of immunogenicity for every TL1A.
Taylor, let me just make sure I get it right. You're saying free TL1A knocked down, and you're talking about total TL1A, which includes the soluble TL1A, going up twofold?
Yeah. So basically what these folks are measuring is they're saying in an ELISA assay that can detect TL1A at an epitope not bound by the antibody, concentration goes up two logs. In an assay that detects the epitope where their antibody binds, concentration goes to zero. So it's telling you that everything you're measuring is antibody-bound TL1A.
But every single TL1A program is showing significant increases in soluble TL1A. So my sense just was that's like the bound.
Exactly. Yeah, that's exactly right.
Right. But do you think that's an immunogenicity biomarker as well?
Immune?
That may not mean it's an antidrug antibody necessarily.
It is a source of antidrug antibody creation.
Oh, I see. Okay.
Immune complexes are known to be internalized through, interestingly, neonatal Fc receptor binding.
Right.
That leads to ADA formation.
Okay.
And so that is a shared source of immunogenicity. And that's the reason why every single TL1A antibody has seen greater than 65% of patients develop ADA.
Got it. Could you remind us, Taylor, just for the purposes of this conversation, so you're binding the receptor, not the ligand. We know the TL1A ligand goes through this receptor. But in our sort of simplified cartoon images, we don't know what other ligands go through that receptor as well. Do we have a sense for that?
There are no other TL1A ligands.
There are no other ligands.
Correct.
Okay. And then I guess the other point is we know the ligand is implicated in a bunch of various settings, which obviously includes both the various cytokines that are downstream affected, some fibroblasts as well. Do you think by hitting the receptor, all those mechanisms are technically at play?
Yes. Yeah. I mean, it's a monogamous receptor-ligand relationship. So anything that any inflammation, fibrosis that TL1A mediates, it mediates through this receptor.
Teva was really leaning into the DR3, and I don't want to get into the DR3, but DR3 versus DcR3 and how they thought that could actually selectivity would help them sort of avoid some of the bads. Is that do you think they were selective enough to demonstrate that? Because you were obviously very specifically selective for DR3.
Yeah. I agree with Teva that if you have an antibody that spares that allows decoy receptor 3 to continue binding and degrading TL1A, that is better than if you have a TL1A blocking antibody that interferes with that interaction.
Got it. But my understanding was all these TL1A MABs, ligand-targeted, were at least threefold selective. And I think in Teva's case, they were maybe hundredfold selective, something like that. So I guess, is there incremental gains from having that bias at a receptor when we do it at a receptor level instead of doing it at a ligand level?
I think the gains are independent from whether a TL1A blocking antibody is decoy receptor 3 sparing or not.
Got it.
Because Tulisokibart has the same immunogenicity liabilities as every other TL1A blocking antibody.
Right.
Just to sort of finish up on that point, the receptor is not shed. When you have a receptor-blocking antibody, there is no risk of immune complex formation.
Right.
And so, for example, the Tulisokibart GLP-tox studies. You can find this in the Prometheus phase one. They had lethal immunogenicity. In our GLP-tox studies, there was zero toxicity.
Right.
This is already an indicator that there might be a better safety profile here because of reduced ADA.
So in your ongoing phase one, what can we learn, which will be an early indicator that this could theoretically be a step function better in terms of efficacy theoretically? Because I realize, I think maybe just remind us, what are the type of patients in the trial? Are you measuring some downstream cytokines, etc.? Because we're not getting modified MALE scores right now, obviously.
Yeah. These are healthy volunteers.
Correct.
So what we will learn from phase one is this the least immunogenic way to block this axis? That will be a definitive takeaway.
Where are you with enrollment?
We're about halfway through the study right now.
Would you have known on immunogenicity by now?
We have indicators.
Got it, and you continue to stay super confident on a very differentiated immunogenicity profile.
Yes.
Which should open up the efficacy window.
Yes.
The other point that you were going to make is on potential for efficacy differentiation. What's the mechanism there? And is there something we can learn from healthy volunteers that would give us confidence in efficacy moving into patients?
So the efficacy hypothesis has to do with the fact that the receptor is expressed by lymphocytes. And those lymphocytes are found both in the peripheral blood and in tissues of patients, for example, with IBD. And when a lymphocyte has decided to turn on the receptor, it never turns it off again. So it represents a stable target. Pardon me.
You mean it's expressed consistently.
It's expressed consistently. Yeah. The ligand, on the other hand, is not constitutively expressed. It's induced in a pulsatile manner primarily by tissue-resident antigen-presenting cells, and so when an antigen-presenting cell in the gut, for example, encounters a bacterial cell wall component because you have inflammation, you have barrier disruption, TL1A gets turned on and reaches maximal expression within about eight hours, but then the next day, that cell is no longer expressing TL1A, so you get the short pulse of protein, all of which needs to be blocked in order to prevent it from binding locally abundant DR3, so when you're trying to block TL1A, the point here is that you're chasing a moving target, especially in inflammatory diseases that can migrate from previously non-inflamed to inflamed status, and when that happens, when you get new TL1A expression, that sets off a race.
Does this cell that wasn't previously expressing TL1A and now is, is that TL1A able to bind locally abundant DR3 before that TL1A gets 100% bound by a TL1A blocking antibody?
You need to have systemic TL1A antibody levels high enough to cover every inch of the gut and all possible cells that are going to start pumping out TL1A at some point.
That's correct. But even when you do have the right concentration gradient from the blood into the tissue, there is still a kinetic battle because in a tissue where TL1A has not been expressed, there is no means of retention for TL1A blocking antibodies at those sites.
I see. So the hope is by binding the receptor, you can have stable complexes block the receptor, whether there's local TL1A or not.
Correct.
The battle is already won before it gets started.
Correct.
Okay.
And this is just to point out a couple of proxies. I mean, right now we're seeing interesting data in the BAFF receptor domain. That's another TNF receptor ligand pair. Benlysta is an anti-BAFF antibody that's been on the market for a decade. You now have Novartis and Jade developing anti-BAFF receptor antibodies. Those antibodies have to tackle the same engineering challenges that we had to. And at least in the Novartis case, seems to be performing in Sjogren's better than Benlysta ever did.
What could we see in healthy volunteers? Obviously, they're not experiencing these pulses of TL1A. So are we going to be able to see data in healthy volunteers, whether that's through receptor occupancy or some other measure that could give us confidence on that efficacy hypothesis? And maybe cytokines?
Yeah. So what you hope in a healthy volunteer study, because there is no ongoing TL1A-driven inflammation to interrupt, is that all you see is receptor occupancy. And so from an efficacy perspective, what I would assert is that Pfizer published a paper in Lancet Gastroenterology this year where they reported efficacy and exposure by ADA quartile in their Crohn's disease patient studies. This was in the data supplement for that paper. And what they showed is that efficacy and exposure went down based on your ADA quartile. And what that highlights that I think will become more important over time is that even if many of the ADA that these TL1A antibodies generate are not neutralizing, they still accelerate clearance of the therapeutic antibody.
The more ADAs you have, the less effective you are. If you have no ADAs, you ought to be more effective. And there's almost separately this persistent receptor blockade thesis on efficacy.
Correct.
So, between lack of ADAs and high receptor occupancy in healthy volunteers, we ought to have reasonable confidence moving into patients.
That's right. We won't answer the definitive efficacy sort of independent of superior immunogenicity hypothesis until we get those phase two data, but phase one, we'll know if it's the best in class on the immunogenicity. We'll know RO. We'll know if it's equivalent safety to the TL1As, and for anybody who still has lingering concerns about residual agonism, that is definitive coming out of phase one.
Now, the other thing that we're seeing in the TL1A space is very active development towards half-life extension, active development in bispecifics. I know you're working on both of those things behind the lead asset. Can you tell us a little bit about where those programs stand and how rapidly they could catch up?
Yep. So 425 is our YTE-containing variant of 325. That's the lead DR3 blocking antibody. And those two antibodies are more than halfway through a chronic tox study right now where they are being tested head-to-head against one another. And so both of those antibodies will be through chronic tox and able to go into clinical studies immediately after that. What our plan is, is to look at our phase one healthy volunteer PK and receptor occupancy, look at the 325 versus 425 non-human primate PK and receptor occupancy, and ask ourselves, is there a path to get to sub-Q maintenance dosing with a non-half-life extended version, no more often than Q8 weekly or not? And there's reason to believe that we will be able to get there because this antibody doesn't cause internalization of the receptor, so you expect highly durable occupancy. We'll see how durable.
Monkeys, it was at least a month. But if we can't get there, then what we will do is we'll run a healthy volunteer phase one study with the half-life extended version while we are running the phase two study in IBD with the non-half-life extended version and bridge it into phase three.
The moral of the story is you have a half-life extended version and could potentially lose very little time doing that swap as you get into a phase three study.
That's right.
Taylor, where's your primary focus? Because there's a lot of data coming next year for TL1As, not just in UC registrational, but also a lot of new indications. Merck will have SSC-ILD. There's some RA trials ongoing at Spyre. There's possibly even an additional one or two indications. We might have perhaps up to four indications' worth of data next year. How do you intend to prioritize it? And are you waiting for the backup molecule or maybe not?
Yeah. We're not waiting for the backup molecule. And in our recent pipe, we included in our use of proceeds a non-IBD phase two study. And all of the indications where these folks are pursuing make sense from a translational biology, a preclinical data standpoint. I believe some are higher risk than others. I think, for example, there's a risk that the atopic derm study may look more like the asthma study for Teva than the IBD study for Teva.
Right.
If Merck's SSC study hits, in my opinion, that is a huge upside surprise.
Right.
I'm not banking on it.
And then what about combos? Because, I mean, I don't think Spyre's a different story, obviously, but there are biosimilar IL-23s that you could theoretically contemplate as well. I guess the question I'm really, the direction I'm going is there's been very positive market receptivity to the phase two program Spyre's put up. They have six readouts coming up this year.
Yeah.
I guess one of the questions for you is, could you be in a position within the next couple of quarters to put out a large phase two sort of platform trial where you have multiple indications and/or possibly even some biosimilar combinations all embedded within the same trial? Your 2027 kind of looks like the type of attention the Street's paying to Spyre into 2026.
Yeah, that's possible. And the other thing is I agree with Spyre that TL1A is not an approachable target in a bispecific format because the ADA issue is compounded.
Right.
So you either have to co-formulate multiple antibodies or you have to build DR3-directed bispecific.
Or you could just run the trial for now on two and then you could keep co-formulating later.
Correct.
And so is that a consideration?
That is.
IL-23 is unencumbered theoretically.
Yeah.
I don't think TNF combo would make any sense. I think there's mechanistic overlaps.
I agree. That's a consideration, but we're also excited about a DR3 bispecific.
Okay, got it.
Oh.
Oh, DR3 bispecific with IL-23. Oh, wow. Which, of course, we get to just as we run out of time. Wait, I just want to be clear. That is not on your slides.
It's a combo that makes sense.
All right. Sounds good.
Awesome. Well, thank you so much for joining us, Taylor. Lots to think about as we head into 2026.
Yep. Appreciate it.
Thank you so much.