Rapid Diagnosis of Microbial Infection via Mass Spectrometric Phenotyping

Abstract

Microbial infection is a perpetual public threat, causing more than 15 million annual deaths worldwide. Clinical microbiology laboratories currently rely on bacterial culture-based methods for microbial diagnosis to identify causative pathogens, which is time-consuming and labor-intensive. This drives the development of novel diagnostics to identify pathogens accurately and more rapidly. Mass spectrometry (MS) now plays a vital role in clinical diagnosis due to its high accuracy, high specificity, rapidity and high- throughput capability. In this thesis, we present a multi-faceted mass spectrometry approach for rapid microbial infection diagnosis. We first used a novel sample transfer technique, surface acoustic wave nebulization (SAWN) for bacterial membrane lipid analysis, specifically lipid A. Analytical performance of different SAWN chips was characterized, and the optimized SAWN chip was used for bacterial phenotyping. Results showed that lipid A mass spectra from different bacterial species can be differentiated by dot product analysis, in turn, demonstrating feasibility of using SAWN for rapid bacterial identification. We next developed a rapid sodium acetate (SA) based method for lipid extraction, which greatly improved our lipid-based library for pathogen identification by reducing the process time to less than an hour. Importantly, the novel SA method maintained the key components of the reported lipid library method for bacterial identification. Namely, these were the ability to detect 1) antibiotic resistance, 2) microbes direct from biological fluids without culture, and 3) single microbes in polymicrobial samples. This platform can be a complementary approach to the commercialized protein-based systems to improve patient outcomes. The last objective of this thesis is to understand the proteome change in response to lipopolysaccharide stimulus in the context of sepsis, which will facilitate the discovery of new biomarkers for sepsis diagnosis. Shotgun label-free quantification proteomics results showed that 27 new sepsis-related proteins were found among 182 significantly changed proteins in the septic mouse group. A longitudinal, but not pair-wise, data analysis strategy overcame inherent heterogeneity detected twice as many significant changes using each mouse's data as its own control sample. Overall, the advances made in this thesis have broad implications in MS-based rapid diagnosis of microbial infections.