A study published in the journal Microbiome introduces ARG-PASS, a novel tool designed to identify previously unknown antibiotic resistance genes (ARGs) within the human microbiome based on conserved protein sequence and structural features.
The issue of antimicrobial resistance (AMR) continues to pose a significant threat to global public health. The human microbiome acts as a vast reservoir for these genes; however, many remain challenging to detect using conventional methods due to their low homology with already known ARGs. To overcome this, the authors focused on functionally critical protein regions where key structural characteristics are preserved.
The ARG-PASS tool employs a one-class support vector machine method, trained on paired distributions of primary and tertiary structures within conserved regions of ARG-encoding proteins. Researchers applied it to six reference strains from the Human Microbiome Project, selecting nine candidates for experimental validation. All nine genes, when expressed in E. coli, confirmed functional activity and were categorized as APH, dfr, Class B and C β-lactamases, and penicillin-binding proteins.
Furthermore, the method was directly applied to the AlphaFold structure database. This application led to the discovery of the phnP gene, identified as a metallo-β-lactamase type.
Although it exhibits low homology with known β-lactamases, phnP shows environment-dependent resistance activity to ampicillin. Of all genes tested, 80% conferred resistance at CLSI threshold levels, while the remainder were classified as “preresistant” – possessing activity but not yet reaching clinically significant minimum inhibitory concentrations.
Researchers highlight that these preresistant genes are more likely to evolve into full-fledged, clinically significant resistance determinants in the future. ARG-PASS, therefore, enables the identification of unique ARGs that escape detection by methods relying solely on sequence homology.
According to the authors, this approach enhances the accuracy of resistance monitoring and could lead to more informed antibiotic usage. By integrating sequence and structural information, the method unlocks access to previously hidden elements within the microbiome.
