Transfer learning predicts species-specific drug interactions in emerging pathogens.

Machine learning (ML) algorithms are necessary to efficiently identify potent drug combinations within a large candidate space to combat drug resistance. However, existing ML approaches cannot be applied to emerging and under-studied pathogens with limited training data. To address this, we developed a transfer learning and crowdsourcing framework (TACTIC) to train ML models on data from multiple bacteria. TACTIC was built using 2,965 drug interactions from 12 bacterial strains and outperformed traditional ML models in predicting drug interaction outcomes for species that lack training data. Top TACTIC model features revealed genetic and metabolic factors that influence cross-species and species-specific drug interaction outcomes. Upon analyzing ~600,000 predicted drug interactions across 9 metabolic environments and 18 bacterial strains, we identified a small set of drug interactions that are selectively synergistic against Gram-negative (e.g., A. baumannii) and non-tuberculous mycobacteria (NTM) pathogens. We experimentally validated synergistic drug combinations containing clarithromycin, ampicillin, and mecillinam against M. abscessus, an emerging pathogen with growing levels of antibiotic resistance. Lastly, we leveraged TACTIC to propose selectively synergistic drug combinations to treat bacterial eye infections (endophthalmitis).

Structural and functional characterization of fosfomycin resistance conferred by FosB from Enterococcus faecium.

The Gram-positive pathogen Enterococcus faecium is one of the leading causes of hospital-acquired vancomycin resistant enterococci (VRE) infections. E. faecium has extensive multidrug resistance and accounts for more than two million infections in the United States each year. FosB is a fosfomycin resistance enzyme found in Gram-positive pathogens like E. faecium. Typically, the FosB enzymes are Mn2+ -dependent bacillithiol (BSH) transferases that inactivate fosfomycin through nucleophilic addition of the thiol to the antibiotic. However, our kinetic analysis of FosBEf shows that the enzyme does not utilize BSH as a thiol substrate, unlike the other well characterized FosB enzymes. Here we report that FosBEf is a Mn2+ -dependent L-cys transferase. In addition, we have determined the three-dimensional X-ray crystal structure of FosBEf in complex with fosfomycin at a resolution of 2.0 Å. A sequence similarity network (SSN) was generated for the FosB family to investigate the unexpected substrate selectivity. Three non-conserved residues were identified in the SSN that may contribute to the substrate selectivity differences in the family of enzymes. Our structural and functional characterization of FosBEf establishes the enzyme as a potential target and may prove useful for future structure-based development of FosB inhibitors to increase the efficacy of fosfomycin.

Characterization of the genomically encoded fosfomycin resistance enzyme from Mycobacterium abscessus.

Mycobacterium abscessus belongs to a group of rapidly growing mycobacteria (RGM) and accounts for approximately 65-80% of lung disease caused by RGM. It is highly pathogenic and is considered the prominent Mycobacterium involved in pulmonary infection in patients with cystic fibrosis and chronic pulmonary disease (CPD). FosM is a putative 134 amino acid fosfomycin resistance enzyme from M. abscessus subsp. bolletii that shares approximately 30-55% sequence identity with other vicinal oxygen chelate (VOC) fosfomycin resistance enzymes and represents the first of its type found in any Mycobacterium species. Genes encoding VOC fosfomycin resistance enzymes have been found in both Gram-positive and Gram-negative pathogens. Given that FosA enzymes from Gram-negative bacteria have evolved optimum activity towards glutathione (GSH) and FosB enzymes from Gram-positive bacteria have evolved optimum activity towards bacillithiol (BSH), it was originally suggested that FosM might represent a fourth class of enzyme that has evolved to utilize mycothiol (MSH). However, a sequence similarity network (SSN) analysis identifies FosM as a member of the FosX subfamily, indicating that it may utilize water as a substrate. Here we have synthesized MSH and characterized FosM with respect to divalent metal ion activation and nucleophile selectivity. Our results indicate that FosM is a Mn2+-dependent FosX-type hydrase with no selectivity toward MSH or other thiols as analyzed by NMR and mass spectroscopy.