Computational Characterization of Epitopes for the Development of Optical Biosensors for the Diagnosis of Pathogenic Escherichia coli

Escherichia coli accounted for 25.6% of all foodborne disease outbreaks in Brazil, according to epidemiological studies conducted between 2007 and 2021. Its diagnosis takes approximately 48 hours, highlighting the need for a rapid, low-cost, sensitive, and specific diagnostic method. Biosensors emerge as a promising alternative for pathogen detection, using antibodies as bioreceptors capable of identifying targets with high specificity. Protein that is possibly recognized as antigenic by the organism was found, and to test this hypothesis, we modeled an antibody for docking and molecular dynamics simulations to evaluate the binding affinity of the antigen-antibody complex. The present study aims to identify antigenic epitopes of the intimin protein through computational predictions for use in an optical biosensor designed to detect pathogenic strains of Escherichia coli, known as enteropathogenic Escherichia coli and enterohemorrhagic Escherichia coli. The three-dimensional structure of the proposed antibody for the biosensor was generated using various algorithms, including Alphafold, Boltz, Swiss-Model, and Abodybuilder. These models were evaluated using Molprobity, the Ramachandran plot, and ProSA. The antigen-antibody complex was generated using the software packages Haddock, Cluspro, and RosettaDock. These complexes were filtered using the Protein Binding Energy Estimator (PBEE), and the model with the highest binding affinity was selected. The physiological environment was simulated in heated dynamics using Amber22 at temperatures of 310K, 330K, 360K, and 390K to perturb the structure and verify whether the interaction interface remains stable over time, in 500 ns simulations, in triplicate. Alphafold3 modeling provided a model that was well-evaluated by the applied metrics, achieving 98% in the favorable regions and 2% in those permitted by the Ramachandran plot. The model was also ranked in the 99th and 100th percentiles for the Molprobity Score and Clash Score, respectively. Heat-treated molecular dynamics complexes initially proved stable, with a root mean square deviation (RMSD) at the interface below the 5 Å threshold found in the literature for bond interfaces.