Elucidating enzyme–substrate specificity through co-folding foundation model
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (Arcadia Science)
Abstract
Enzymatic catalysis relies on precise structural and chemical complementarity, yet systematically mapping enzyme-substrate interactions remains a critical bottleneck. While structure-aware methods have advanced functional annotation, their reliance on predefined binding pockets and rigid-body docking fails to capture the ligand-induced conformational changes essential for catalytic turnover. Here we introduce Boltz2ESI, an end-to-end framework that predicts enzyme–substrate interactions by leveraging structural knowledge learned by a biomolecular foundation model. Through native co-folding, the framework inherently captures active-site plasticity without requiring predefined pocket annotations. Integrating these learned biophysical priors with global evolutionary context and geometric molecular descriptors, Boltz2ESI consistently outperforms state-of-the-art sequence-based and rigid-docking approaches. Extensive validation demonstrates that the framework accurately discriminates tight sub-family specificities, enabling effective candidate prioritization for biosynthetic pathway elucidation, as demonstrated on the withanolide pathway. Ultimately, this structure-dynamic approach establishes an actionable foundation for accelerating rational biocatalyst discovery and large-scale pathway de-orphaning.
Article activity feed
-
As the taxonomy progresses from broad, coarse-grained catalytic classes (EC .-.-.-)down to highly specific, four-digit reaction profiles (EC x.x.x.x) and targeted families,the challenge of filtering out false positives among closely related sequences scales expo-nentially. Throughout this entire taxonomic gradient, Boltz2ESI consistently exhibitedsuperior discriminative resolution compared to baseline approaches. Notably, the sus-tained precision at the deep EC x.x.x.x level indicates that by capturing the adaptivebiophysical microenvironment within the active site, our framework effectively untan-gles tight sub-family specificity that remains hidden to one-dimensional sequencemetrics or rigid-scaffold modeling
Performance claims here are uninterpretable without the evaluation design. How are negatives sampled at each EC depth?
-
Fig. 1 Overview of the Boltz2ESI framework. (a) Given an enzyme sequence and a substrateSMILES string, Boltz-2 co-folding first crops the putative active-site pocket, then the active siteand substrate are re-folded without MSA, and the active-site-level single representation, pair repre-sentation, and distogram of the predicted complex structure are extracted. (b) ESM3 embeddingscapture evolutionary context for enzyme residues, while Uni-Mol2 embeddings and Morgan finger-prints encode substrate molecular features. These priors are fused with the single representation toform a unified representation. (c) The interaction prediction module enriches the pair representa-tion with the fused single representation and the distogram, processes the result through a 4-blockPairformer stack, and outputs an interaction probability via mean pooling …
Fig. 1 Overview of the Boltz2ESI framework. (a) Given an enzyme sequence and a substrateSMILES string, Boltz-2 co-folding first crops the putative active-site pocket, then the active siteand substrate are re-folded without MSA, and the active-site-level single representation, pair repre-sentation, and distogram of the predicted complex structure are extracted. (b) ESM3 embeddingscapture evolutionary context for enzyme residues, while Uni-Mol2 embeddings and Morgan finger-prints encode substrate molecular features. These priors are fused with the single representation toform a unified representation. (c) The interaction prediction module enriches the pair representa-tion with the fused single representation and the distogram, processes the result through a 4-blockPairformer stack, and outputs an interaction probability via mean pooling and an MLP head. Theresulting scores enable two complementary application modes: ranking candidate substrates for agiven enzyme, and ranking candidate enzymes for a given substrate.
Beautiful figure!
-
However, these methods typically rely on predefined or homology-mapped bindingcavities, rendering them ineffective for poorly characterized or “orphan”
This is true. Good context!
-
. From this simulated complex, the framework extracts interaction repre-sentations and a predicted structure to capture local biophysical constraints. Toreconcile this local structure with global biological context, Boltz2ESI fuses these geo-metric descriptors with residue-level evolutionary context derived from ESM3 [28],alongside geometry-aware molecular embeddings [29] and topological fingerprints
lol, four feature streams, no ablation. Do the topological fingerprints add anything over the molecular embeddings; both substrate-side, presumably overlapping Id guess? And do the geometric descriptors actually survive/lead to signal gain once ESM3 features are present? If ESM3's structure track is used, those two streams overlap by construction, since both descend from the stage-1 co-folded pose; that's also a leakage concern. …
. From this simulated complex, the framework extracts interaction repre-sentations and a predicted structure to capture local biophysical constraints. Toreconcile this local structure with global biological context, Boltz2ESI fuses these geo-metric descriptors with residue-level evolutionary context derived from ESM3 [28],alongside geometry-aware molecular embeddings [29] and topological fingerprints
lol, four feature streams, no ablation. Do the topological fingerprints add anything over the molecular embeddings; both substrate-side, presumably overlapping Id guess? And do the geometric descriptors actually survive/lead to signal gain once ESM3 features are present? If ESM3's structure track is used, those two streams overlap by construction, since both descend from the stage-1 co-folded pose; that's also a leakage concern. Separately: stage 1 is MSA-guided, so the pipeline draws on evolutionary information twice. What does the PLM contribute then?
-
first localizes the active-sitepocket using Multiple sequence alignment (MSA) guided co-folding, and subsequentlyre-folds the active site and substrate in an MSA-free regime to capture ligand-inducedside-chain and backbone adaptations
Stage 2 being MSA-free is presumably to avoid the consensus/apo bias that deep MSAs impose on side-chain placement. But does stage 2 condition on stage-1 coordinates? If so, evolutionary information persists as a geometric prior, and the ablation that matters is MSA-free stage 2 without the stage-1 pose, otherwise you can't attribute the induced-fit gains to the MSA-free regime.
-
we introduce Boltz2ESI, an end-to-end computationalframework that harnesses the evolutionary and structural priors of a biomolecu-lar foundation model Boltz-2
Interesting idea / choice of FM.
-
