Skip to PREreview

PREreview of Computational enzyme design by catalytic motif scaffolding

by James Fraser, Rishav Mitra, and Priyanka Bajaj
Published
DOI
10.5281/zenodo.14480475
License
CC BY 4.0

Summary:

This study introduces a novel approach called Riff-Diff (rotamer inverted fragment finder – diffusion) for designing de novo enzymes based on  optimized catalytic centers. Using Riff-Diff, the authors designed 36 retro-aldolase sequences based on a highly evolved catalytic tetrad. Biochemical and structural characterization combined with molecular dynamics simulations showed that many of the designed enzymes exhibited remarkable catalytic activities, with rate accelerations exceeding 5 million-fold over the uncatalyzed reaction. Remarkably, this study demonstrates that precise positioning of the catalytic center residues and enzymatic activity are not correlated.  Their study reveals potential limitations of static structure-based analysis of de novo enzyme designs and underscores the importance of incorporating ensemble-function relationships in motif scaffolding to develop more efficient enzymatic designs. Overall,  it suggests that either artificial motifs must be designed with even sub-Angstrom accuracy to productively sample catalytically competent conformations or that directionality/modes of displacement from the mean position should be considered in an ensemble-function relationships in motif scaffolding.

Major comments:

  1. With reference to Fig 2F in particular, can mutagenesis better discover factors directly contribute to the >100 fold difference between the kcat of the evolved retro aldolase RA95.5-8f (10 s-1) and that of the most catalytically proficient variants obtained through the Riff-Diff approach (kcat values of 0.036 s-1 and 0.031 s-1)?

  2. We found it particularly intriguing that certain designed variants maintain measurable catalytic activity, albeit to different extents, even after the catalytic lysine is mutated to alanine. A discussion on the mechanistic and evolutionary implications of catalytic performance of these variants and the variability therein would be valuable.

Minor points:

  1. The authors claim in the results (second paragraph) that “Success rates for motif-scaffolding with RFdiffusion have been shown to correlate with increasing size of the selected motifs.” Is there a reference or specific data to support this statement?

  2. Under the section ‘Scaffolding of a catalytic tetrad yielded highly proficient enzymes’, the authors mention “Sequence-structure pairs designed by the Riff-Diff pipeline generally passed structural quality filters.” This statement seems a little vague. It will be good to list those filters.

Competing interests

The authors declare that they have no competing interests.

Comments

Write a comment
  1. by Gustav Oberdorfer
    Published
    License
    CC BY 4.0

    We highly appreciate this PREreview and sincerely thank the authors for their kind words about the manuscript and want to emphasize our appreciation for the thorough summary provided.

    In regard to their comments, we want to emphasize that:

    1. The revised version of the manuscript now holds detailed activity data for variants of the most active designs, addressing the question of individual contributions to catalysis from the scaffolded active site tetrad. The new data shows that the individual residues participate in catalysis, yet to different degrees when compared to the evolved variant RA95.5-8f.

    2. The background activities most likely originate from a high imidazole concentration in the screening buffer – 96 well based screens were performed with affinity pure protein in elution buffer containing 250mM imidazole.

    3. Scaffolding success with larger fragments was an empirical observation during the course of the project.

    4. We expanded the section about design selection and provide detailed data for what the structural quality filters are.

    We implemented the RiffDiff pipeline using a python package that manages protein design tools on SLURM-based computing clusters. We added installation and usage instructions to the implementation which can be accessed here: https://github.com/mabr3112/riff_diff_protflow. The initial code for RiffDiff was written in an early, (local) version of this package and can be viewed here: https://github.com/mabr3112/riff_diff_original. The corresponding design models and data files have been uploaded to Zenodo (https://zenodo.org/records/15494858) 

    Competing interests

    The author of this comment declares that they have no competing interests.