Exploration of a mutant enzyme protein with active site fluctuations at 330 K via machine learning and molecular dynamics simulations

Abstract

Control of active site fluctuations in enzyme proteins would significantly expand their application fields. Because papain typically has higher activity at 330 K, shifting the temperature to 330 K may help to expand its range of use in medical treatment, food processing, and bioelectronic devices. Mutating residues R111 and Q112 in the β-strand hinge structure that links the two papain domains may significantly influence fluctuations in its active site. To identify papain mutants with active site fluctuations that match the target temperature (330 K), R111 and Q112 were repeatedly mutated complementarily using deep neural network (DNN) and molecular dynamics (MD) simulations. Overall, 24 mutation patterns were found to bring the active site fluctuations of papain at 300 K closer to those of papain at 330 K. The decision tree identified factors that specifically influenced the active site fluctuations. These factors should be considered when designing studies that combine DNN and MD simulations.