Latent‑Y Research Tier

Designing functional inhibitors for ion channels, zero-shot

Human ion channels govern core nervous system functions, including pain signalling, which makes them compelling targets for non-addictive pain therapies. They are also among the hardest structural targets in the field: a candidate must not merely bind the channel but functionally block its electrical signal. Until now, the field's practical starting points were natural toxins.

Prof. Vladimir Yarov-Yarovoy (University of California, Davis), protein designer and ion channel structural biologist, used the platform to design inhibitors de novo, from scratch rather than from toxins, and tested whether they worked, using whole-cell patch-clamp electrophysiology to measure whether each design actually blocked the channel's electrical signal.

Against one human ion channel, his group designed nanobodies. Of five taken into the lab, four blocked the channel, and the most potent did so at very low concentrations (an IC50 of 30 nM), a potency usually reached only after rounds of laboratory refinement, but achieved here on the first attempt.

Against ion channels integral to pain signalling, like hNav1.7, the group designed macrocyclic peptides. Of thirty taken into the lab, four blocked sodium channel current at micromolar potency.

His whole group now works on the platform, and he teaches a university seminar built around it.

“In a promising advance, we were able to de novo design nanobodies targeting an ion channel with nanomolar potency and achieving a notable success rate.”
— Vladimir Yarov-Yarovoy

Designing CAR-T cells against solid tumours, now in vivo

CAR-T therapies, immune cells engineered to specifically destroy tumour cells, have transformed treatment for some blood cancers. Extending them to new targets, though, runs into a supply problem: the receptor is built from an existing antibody, and for many promising targets, no antibody exists. Raising one is a project in its own right, months of experimental work before the first receptor can even be built and tested. And a CAR binder has to do more than bind. It has to trigger the T cell to kill on contact, so binding affinity alone is no guarantee the construct will work.

Dr. Daria Briukhovetska (LMU University Hospital) took a different route. Using designs made with the Latent Labs Platform, generated against the crystal structure of her team's cancer targets, her group built the recognition component straight into the receptor. The resulting CAR-T cells killed tumour cells in vitro within hours. On the strength of those results, the group has moved to animal studies.

"Using the Latent Labs Platform we generate new binders against our cancer targets, and use those binders to construct CAR-T cells. Those CAR-T cells demonstrated unparalleled efficacy for eradicating tumour cells in vitro, and are currently being evaluated in vivo."
— Daria Briukhovetska

Designing binders the malaria parasite cannot evade through mutation

Malaria kills roughly half a million children each year, and the parasite is a moving target, varying its surface proteins fast enough to stay ahead of the immune response. But it has an Achilles heel. To cause severe disease, infected cells must grip the walls of blood vessels. Those that fail to hold on are swept to the spleen and destroyed. They grip with a protein that binds a specific human receptor, and mutations to that binding surface cost the parasite its hold. So while the rest of its surface shifts freely, this one part cannot. A binder that blocks it prevents adhesion, and no mutation escapes it.

Prof. John Altin (Translational Genomics Research Institute and University of Sydney), whose lab studies the immune response to malaria, is targeting that site. The lab is built for high-throughput screening, testing hundreds of thousands of candidates in parallel via DNA barcoding. This time it used that setup differently. Rather than searching a library, the lab used Latent-Y to work out how to bind the conserved site and generate candidates, then put them through the same experimental pipeline to confirm which ones bound.

"We've been going after an epitope on a parasite protein displayed on the surface of infected red blood cells. Using the platform, we've generated thousands of candidates, and so far we've found almost 50 that appear to bind with high affinity to that epitope."
— John Altin

Designing enzyme inhibitors, without a computational team

Jacob Shapiro, Ph.D. candidate (Broad Institute), is developing protein-based inhibitors of glycosyltransferases, enzymes implicated in cancer. No such inhibitor exists today.

This is exactly the kind of problem that would normally demand a dedicated computational team, and Shapiro's lab does not have one. Latent-Y closed that gap. From a description of the goal in plain language, it worked through the design decisions a specialist would have made, which framework to build on, which residues to engage, and returned a shortlist ready to take into the lab. Lab testing is underway.

"I think what Latent-Y enables is a much faster design process for every single lab member that wants to design a binder against their own target. It really reduces the overhead and activation energy for these binder design campaigns, and lets us focus on the interesting biological applications we want to explore, rather than the binder design process itself."
— Jacob Shapiro