Structural molecular modeling of bacterial integral membrane protein enzymes and their AlphaFold2 predicted water-soluble QTY variants

Abstract

Context Beta-barrel enzymes are an important area of study in the field of structural biology. These proteins serve crucial roles, acting as porins, transporters, enzymes, virulence factors, and receptors. Recent research has unveiled a novel role for beta-barrel enzymes in the bacterial integral membrane as sentinels. They remain inactive when the integral membrane is intact but activate to carry out enzymatic catalysis in response to host immune responses and antibiotics that breach this barrier. Understanding their structure and function is pivotal in grasping their sentinel role in the bacterial integral membrane. Here we present our structural molecular modeling analyses on four bacterial integral membrane beta-barrel enzymes: (a) OMPLA, (b) OmpT, (c) PagP from E. coli, and (d) PagL from Pseudomonas aeruginosa. We superposed the structures of native beta-barrel integral membrane enzymes with their AlphaFold2-predicted QTY variant structures that showed remarkable similarity despite the replacement of at least 22.95% amino acids in transmembrane regions, the superposed structures displayed notable structural similarity, indicated by RMSD values ranging from 0.181 Å to 0.286 Å. We also analyze the hydrophobicity patches and the enhanced hydrophilic surfaces. Our research provide insights into the structural similarity of hydrophobic and hydrophilic beta-barrel enzymes, validating the utility of the QTY code for investigating beta-barrel membrane enzymes. Our results not only demonstrate that the QTY code serves as a straightforward tool for designing water-soluble membrane proteins across various biological contexts, but it may also stimulate experiments to validate our molecular modeling studies. Methods All the QTY variant beta-barrel enzyme structure prediction was performed using the AlphaFold2 program ( https://github.com/sokrypton/ColabFold ) following the provided instructions. Computations were carried out on 11th Gen Intel Core i5-11300H processor with 16 GB RAM and Iris Xe Graphics, 512 GB NVMe SSD. The structures are publicly available on the AlphaFold2 database ( https://alphafold.ebi.ac.uk ) at the European Bioinformatics Institute (EBI). A custom Python script was used to extract the relevant information from the UniProt database. To predict the structures of the QTY variants, AlphaFold2 was utilized. The native sequences for these enzymes were retrieved from UniProt https://www.uniprot.org , and AlphaFold2 structural predictions were performed using the open-source implementation at https://github.com/sokrypton/ColabFold . The predicted variant structures were then superposed with the native structures using PyMOL https://pymol.org/2/ for structural analysis and comparison. This work leverages public databases PDB, UniProt and open-source software AlphaFold2 and PyMOL to computationally model and analyze QTY variant integral membrane beta-barrel enzyme structures. Graphical abstract