CAR-T Meeting 2026 | AI use for CAR design: engineering de novo binder-based CAR-T therapies to overcome antigen escape

Yeah, my lab in the recent years has been very excited and interested in leveraging these new AI-based tool that allows us to, in a targeted fashion, design protein binders against a target of interest. And we think that this is a very powerful technique because it helps to very much scale up this process of generating new binding moieties that we can then also employ for CAR engineering because traditional methods to do this would have been to, for example, immunize a mouse and then kind of check what immune response that mouse has and what antibody it develops against it, and so from there to derive a new binding moiety, and so these new tools that we have dramatically speed up the process at which we get new candidate binders, of course, we then have to vet them and thoroughly test them, but just, yeah, the scale and speed that we’re able to generate these new binding domains for us has been just very exciting and something that we have been fully leveraging in the recent years. 

So the way that we have been employing these AI binders is that we use them as the binding moiety of the CAR, of the chimeric antigen receptor, which is the part that really mediates what the T-cell then recognizes and attacks. Traditionally, people have used antibody-derived domains for this purpose. What we’re doing is that we’re basically replacing this with these AI de novo-designed binders. Where we think that this is particularly powerful is when we’re thinking about cancers under therapeutic pressures changing and modifying their antigens that we’re targeting, so we are very excited to specifically use these AI-based binders to design them in a way that they are able to rebind escape variants of these target antigens and or deliberately just broaden the coverage of these antigens that we’re going after. Since this is also clinically a major problem that we’re dealing with, with resistances occurring to CAR T-cell therapy because the tumor evolves and tries to escape them. So this is where I think this technology is particularly powerful and helpful. 

So the work that we are doing is right now at a preclinical level, so meaning that we’ve designed these domains and we’ve submitted them to thorough testing. The testing procedure that we do is actually quite comparable to a standard new CAR T-cell product, so meaning that we’re doing a couple of in vitro assays to test the properties, the potency of this new CAR T-cell product that we generated. And we also use them in preclinical mouse models to check their efficiency. And the next step now for us will be to move this into translation and to bring these also to the clinic.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.