Generative AI designs functional thiolation domains for reprogramming non-ribosomal peptide synthetases.
- Open access
Generative AI-designed thiolation domains enable up to threefold increased yields in reprogrammed non-ribosomal peptide synthetases.
- Why it matters: Engineering dynamic biosynthetic systems like NRPSs is challenging due to their transient, context-specific domain interfaces, limiting the development of therapeutics.
- What they did: Researchers integrated pretrained generative models (ESM3, ProteinMPNN, EvoDiff) with iterative design-build-test-learn cycles to create 76 novel thiolation domains and tested 578 NRPS variants in vivo.
- The result: The AI-designed domains supported product formation, improved hybrid enzyme activity, increased yields, and enhanced stability, demonstrating generative design as a powerful tool for biosynthetic reprogramming.