Antibiotic resistance detection and concomitant species identification of ESKAPE pathogens by proteomics

Antimicrobial resistance (AMR) of bacteria represents a significant and emerging threat to public health. In 2019, 4.95 M deaths were associated and 1.27 M deaths were attributed to resistant bacteria. Appropriate antibiotic selection for therapy relies heavily on bacterial species identification and antibiotic susceptibility testing (AST) in clinical diagnostic laboratories. Currently, the taxonomic identity of clinical isolates is determined using MALDI-ToF MS-based biotyping, which is known for its speed and cost-effectiveness. Phenotypic AST methods, such as broth dilution or disk diffusion, which assess bacterial growth in the presence of various antibiotics are time-consuming. Recently, genomics has gained attraction as a complementary approach for predicting the AMR phenotype from the genotype. Integrating molecular data for microbial diagnostics offers several advantages over phenotypic methods. It allows for simultaneous species identification and AMR detection, eliminating the need for secondary cultivation, and can predict phenotypes beyond AMR, such as virulence features. The objective of this study was to establish and evaluate whole bacterial cell proteomics for the detection of antibiotic resistance and the identification of species of ESKAPE pathogens. A representative panel of clinical isolates comprised 129 bacterial strains from 16 species, representing the full spectrum of ESKAPE genera and 30 AMR determinants. In a previous small-scale proof-of-concept (POC) proteomics study we showed that simultaneous antibiotic resistance detection and species identification from DIA-MS data is possible with high specificity and sensitivity. The diagnostic workflow was based on universal bacterial sample preparation using SPEED and data-independent acquisition mass spectrometry (DIA-MS), which potentially enables high-throughput proteomics but poses major challenges for data analysis, such as the identification of bacterial species from large peptide sequence data sets and the detection of low abundant AMR determinants. The results of this study showed, that this new workflow (see Figure 1) enables the reporting of AMR phenotypes from primary cultures within two hours with high specificity (99.9%) and sensitivity (92.3%). Furthermore, it allows for species identification from very large sequence databases with high accuracy (91%). The integration of species identification and AMR detection into a single proteomics-DIA-MS-based workflow has the potential to massively improve the current diagnostic landscape, offering faster, more accurate, and comprehensive capabilities. This innovative approach has the potential to improve patient outcomes by enabling timely and appropriate AMR phenotypes and as such guide antibiotic therapy allowing to combat the growing threat of antimicrobial resistance.