Cyclic peptides offer strong affinity and selectivity, but membrane permeability limits oral bioavailability. The authors argue that existing AI tools mostly optimize affinity and neglect permeability.
HighPlay3 (Eur J Med Chem, 16 October 2026) is a structure-integrated de novo design platform that iteratively evaluates structure and predicts membrane permeability to generate peptides with balanced binding and permeability. It was applied to two targets, MDM2 and GABARAP, followed by a multistep screening cascade and experimental validation.
Reported results: 80% (MDM2) and 60% (GABARAP) of the designed peptides showed micromolar affinity. Among SPR-active GABARAP candidates, GAB-D4 had the highest apparent Caco-2 permeability, with an efflux ratio below 1, consistent with no apparent net efflux under the assay conditions.
Affinities remain micromolar and only two targets were tested: the abstract presents a “feasible” framework for early-stage design, not an optimized candidate.
Takeaway: building permeability into generation, rather than using it as a final filter, is the right direction for peptide macrocycles; whether it holds at higher affinity remains to be seen.
Reference: 10.1016/j.ejmech.2026.119443, Eur J Med Chem, 16 Oct 2026 (PubMed).