Nature publishes a study of enzymes designed from scratch to cleave specific sites in proteins

Nature publishes a study of enzymes designed from scratch to cleave specific sites in proteins

On October 14, Nature published a study by David Baker’s team. The authors designed new proteins around a chemical reaction and tested their ability to cleave a specified peptide bond, examined their structures, and assessed their activity in human cells.

A protein is a chain of amino acids joined by peptide bonds. In water at neutral pH, a peptide bond can remain intact for hundreds of years, so an enzyme is needed to break it. The authors specified the enzyme’s active site, the region where the reaction takes place: cysteine attacks the bond, histidine transfers a proton, aspartate modulates histidine’s activity, and a neighboring region stabilizes the molecule as the bond breaks.

They started with a minimal arrangement of components from the natural protease Ulp1: these three amino acids, the stabilizing region, and a fragment of the intended substrate. The RFD2-MI model built new protein scaffolds around them. ProteinMPNN selected an amino acid sequence for each scaffold, and AlphaFold 3 screened for designs in which the active site and peptide-binding pocket retained the required shape both in the unbound protein and after peptide binding.

Of the 69 designs tested, 13 were active, representing three distinct protein scaffolds. Mass spectrometry confirmed cleavage at the intended bond; individual enzymes cleaved short labeled peptides, and substitutions of the catalytic cysteine, histidine, or aspartate nearly abolished activity.

After computational refinement, the Dokki_15 variant accelerated hydrolysis of its target peptide bond by approximately 30 million times relative to the reaction without an enzyme. Six X-ray structures closely matched the computational models, and searches of structural databases showed that these scaffolds differ from known natural proteases. In the final version of the article, the authors also demonstrated target cleavage in a cellular reporter assay using HeLa cells.

The authors attribute the difference in reaction rate relative to the most active natural cysteine proteases to active-site preorganization and substrate positioning; the next design cycle aims to refine these features.