High-Resolution Magic Angle Spinning NMR-Based Metabolomics Revealing Metabolic Changes in Lung of Mice Infected with P. aeruginosa Consistent with the Degree of Disease Severity

Abstract

Pseudomonas aeruginosa is a critical pathogen for human health, due to increased resistances to antibiotics and to nosocomial infections. There is an urgent need for tools allowing for better understanding mechanisms underlying the disease processes and for evaluating new therapeutic strategies with animal models. Here, we used a novel approach, applying high-resolution magic angle spinning nuclear magnetic resonance spectroscopy (HRMAS NMR) directly to lung biopsies of mice to better understand the impact of infection on the tissue at a molecular level. Mice were infected with two P. aeruginosa strains of different virulence levels. Statistical analysis applied to HRMAS NMR data allowed us to build a multivariate discriminant model to distinguish the lungs' metabolic profiles of mice, infected or not. Moreover, a second model was built to appreciate the degree of severity of infection, demonstrating sufficient sensitivity of HRMAS NMR-based metabolomics to investigate this type of infection. The metabolic features that discriminate infection statuses are dominated by some key differentially expressed metabolites that are related, respectively, to bacterial carbon metabolism (glycerophosphocholine) and to septic hypoxic stress response of host (succinate). Finally, to get closer to clinical and diagnosis issues, we proposed to build simple logistic regression models to predict the infection status on the basis of only one metabolite intensity. Thus, we have demonstrated that succinate intensity could discriminate the infected/noninfected status infection with a sensibility of 89% and a specificity of 95%, and leucine/isoleucine intensity could predict the severe/not severe status of infection with a sensibility of 100% and a specificity of 95%. We also looked for the interest of this model in order to predict the efficacy of anti- P. aeruginosa treatment. By HRMAS metabolomics analysis of lungs infected with P. aeruginosa after vaccination, we demonstrated that this model could be a useful tool to predict the efficacy of new anti- P. aeruginosa drugs. This metabolomics approach could therefore be useful both for the definition of biomarkers of severity of infection and for an earlier characterization of therapeutic efficacy.