Substrate-dependent epistasis probes active site intramolecular wiring

Abstract

Enzyme adaptation toward novel substrates involves the rewiring of intramolecular residue networks, yet how this rewiring differs across multiple substrates, and how it underpins functional trade-offs and promiscuity, remains poorly understood. Here, we profile all 64 combinations of six key mutations in a phosphotriesterase across nine structurally diverse substrates spanning three chemical classes (organophosphates, esters, and lactones), thus generating a multi-dimensional map of epistasis and promiscuity within the phosphotriesterases active site. We developed a statistically robust reference-based analysis pipeline incorporating error propagation and significance testing to move beyond global epistatic trends and resolve idiosyncratic, substrate-dependent intramolecular wiring in specific genetic backgrounds. Simulations confirm that this pipeline reliably identifies genuine higher-order epistatic interactions while minimizing false positives. We reveal that intramolecular network wiring varies substantially between substrates - even within the same chemical class - with notable divergences between the adaptive target substrate 2-naphthyl hexanoate and its shorter-chain ester analogs. Key higher-order networks, including d233E/h254R/l271F and l271F/f306I/i313F, exhibit substrate-specific epistatic signatures that discriminate between subtle structural features such as acyl chain length, leaving group identity, and heteroatom substitution. These substrate-dependent rewiring events account for observed functional trade-offs, particularly the strong anti-correlation between the adaptive and native substrates. Collectively, these findings demonstrate that comprehensive cross-substrate epistatic profiling, paired with rigorous statistical analysis, provides a powerful framework for dissecting the molecular basis of enzyme promiscuity and the trade-offs that define adaptive evolution. Significance of the studyEnzymes are biological catalysts whose function depends on precisely coordinated networks of interacting amino acids. Understanding how these networks change as enzymes evolve to recognize new molecules is fundamental to understanding and predicting evolution. Here, we developed a statistical framework to map these networks across multiple molecules simultaneously, revealing how subtle differences in molecular shape drive functional trade-offs during evolution. These insights could inform the rational design of enzymes with tailored specificities for industrial and therapeutic applications.