DigestAI news desk

Cut through the AI noise.

Research

Generative AI designs de novo thiolation domains that boost nrps yields up to ~3‑fold

Researchers combined pretrained protein generators—ESM3, ProteinMPNN and EvoDiff—with iterative design‑build‑test‑learn cycles to create 76 new thiolation (T) domains for non‑ribosomal peptide synthetases (NRPSs). The designs were inserted into 578 recombinant NRPS variants spanning minimal, full‑length and hybrid assembly lines and evaluated in vivo.

1 source primary source

Key points

  • 76 de novo thiolation domains were generated with ESM3, ProteinMPNN and EvoDiff.
  • 578 NRPS variants were built and tested, with the best designs increasing product titers up to ~3‑fold.
  • The AI‑2 design showed 12 °C higher melting temperature and higher soluble expression than the native domain.

Across the experiments, many AI‑designed T‑domains supported peptide production, and the top performers raised product titers by roughly three times compared with the native T‑domain. A representative design, called AI‑2, also displayed improved biochemical properties: higher soluble expression, efficient refolding and a melting temperature about 12 °C above the wild‑type carrier. Molecular dynamics simulations suggested that while overall fold stability was retained, the engineered domains reshaped state‑dependent interdomain contact networks, explaining their context‑dependent activity.

The work demonstrates that generative protein models can engineer dynamic, multi‑domain enzymes whose function depends on transient interfaces, offering a new route for reprogramming biosynthetic assembly lines and expanding the toolkit for synthetic biology and drug discovery.

Read the original at Nature Machine Learning primary sourceOpen source ↗
Topics · follow one to build your own front page
ESM3ProteinMPNNEvoDiff

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Research

All →

Related stories