Development of novel mass spectrometric methods for point-of-care mucosal diagnostics

Abstract

Human mucosal surfaces act as key interfaces between microbiota and host. As such, mucosal sampling using medical swabs is performed for diagnostic purposes that most commonly rely upon subsequent microscopy, culture or molecular-based assays. These approaches are limited in providing information on host response, which is a critical facet of pathology. In this thesis, I sought to test the hypothesis that both presence of specific microbes as well as their interactions with the human host are reflected in the mucosal metabolome and that this information could be exploited for mucosal diagnostic applications. The study aimed to develop a method for rapid, direct metabolic profiling from swabs using desorption electrospray ionisation mass spectrometry (DESI-MS). Method optimisation was conducted to elucidate optimal instrumental and geometrical conditions essential for the swab analysis. The application of the method for mucosal diagnostics was then assessed by characterising the metabolic profile of multiple bodysites (oral, nasal and vaginal mucosa), vaginal mucosa during two different physiological states (non-pregnant vs pregnant) and to detect a pathological state (bacterial vaginosis). Correlation of DESI-MS vaginal metabolic profiles with matched vaginal microbiota composition (VMC) characterised by 16S rRNA-based metataxonomics during pregnancy enabled to robustly predict a Lactobacillus dominant from depleted state but also major vaginal community states types (CST). The predictive performance of DESI-MS based models was comparable to “gold standard” LC-MS based models. Additionally, bacterial metabolite markers predictive of specific microbial genera were identified through matching to a spectral database constructed using pure cultures of commensal and pathogenic microbes often observed in the vaginal microbiome. In summary, DESI-MS has the potential to revolutionise the current way of mucosal based diagnostic by reducing significantly the time-demand needed for the characterisation of VMC, drug or inflammatory response to only few minutes and therefore could enable a faster decision making on patient’s treatment.