Mutations in Penicillin G Acylase: A 4D QSAR-based Approach for Enhancing Efficacy of β-lactam Antibiotics

Penicillin G Acylase (PGA) plays a central role in the synthesis of β- lactam antibiotics. While certain variants have been extensively studied, their catalytic efficiency remains suboptimal for industrial application, necessitating further enzyme engineering to enhance substrate binding and reaction kinetics. This study aims to rationally design and engineer PGA variants with improved catalytic efficiency and stability toward β-lactam antibiotics, using an integrated approach of 4D QSAR modeling and neural network-guided mutation prediction.

A dataset of 30 enzyme-substrate complexes involving three PGA variants and diverse β-lactam substrates was compiled. Ten complexes were randomly selected for external validation. The binding conformation of Cefotaxime to a Bacillus thermotolerans PGA variant was used as a reference for molecular docking and structural alignment. Binding site analyses identified optimal substrate orientations, followed by 4D grid-based energy profiling, which revealed 15 high-energy hotspot residues per variant. These positions were systematically mutated in silico, generating 1130 variants through a neural network-based residue substitution algorithm.

Subsequent docking studies with Cefotaxime showed a strong positive correlation between predicted docking energies and Ki values derived from the 4D QSAR model, validating the model's predictive capability. Molecular dynamics simulations (2 × 100 ns) for selected variants, particularly Sequence Id_0, Id_2, Id_5, and Id_7, demonstrated stable binding interactions and favourable atomic distances, indicative of improved substrate affinity.

In Sequence Id_11, the hotspot is Phe148. Chain A showed the best results with Val and Leu as single mutants, followed by Met56 in Chain B with Leu, and Ser144 in Chain A with Glu, Ala, Ile, and Arg. In the case of Sequence Id_03, the hotspot is Phe147. Chain A showed good results with Ala, Lys, Thr, and Ser, whereas Tyr71 in Chain B showed good results with Glu, Lys, and Thr, and Arg266 in Chain B showed good results with Ala, Thr, and Val. Those that showed the highest sum of docking scores and Ki were chosen for further studies.

The study highlights the critical role of residue Phe148 in mediating stable interactions with Cefotaxime and other β-lactam substrates. The integrated computational strategy establishes a robust framework for engineering catalytically superior PGA variants, offering a valuable basis for further experimental validation and application in antibiotic biosynthesis.