Logotyp: till Uppsala universitets webbplats

Mass spectrometry imaging (MSI) is an emerging technique for spatially resolving the molecular composition of biological samples. MSI frequently relies on matrix-assisted laser desorption/ionization (MALDI), in which a pulsed laser beam and chemical matrices are used to facilitate desorption/ionization of molecular species from the sample surface. MALDI matrices can be divided into two broad groups: conventional matrices that promote ionization by protonation/deprotonation or cationization, and derivatizing matrices that target specific chemical functionalities. Derivatizing matrices such as FMP-10 are charged molecules that react with specific chemical structures on target analytes to form covalent matrix-analyte conjugates, enhancing ionization and detectability but limiting chemical coverage. Derivatizing matrices may also create multiple derivatization products through serial reactions with single analytes, complicating annotation. This prompted development of Met-ID, a software tool for automatic annotation of MSI data with an emphasis on derivatization-based workflows. Met-ID incorporates matrix-specific chemistry to enumerate plausible derivative products and filter chemically implausible annotations. It includes a database of in-house acquired tandem mass spectrometry (MS2) spectra of FMP-10-derivatized chemical standards to support MS2 spectral matching. The use of ion mobility (IM) spectrometry in MSI enables collision cross section (CCS) values to be used for annotation. This motivated the development of CCSSim, an in-silico CCS prediction method implemented in Met-ID together with a mixture-model framework to increase annotation confidence by integrating m/z and CCS data. To improve spatial correlations between mass spectrometric and transcriptomic data, a method was developed to enable sequential MSI and spatially resolved transcriptomics (SRT) analysis of one tissue section rather than using consecutive sections. This spatial multimodal analysis can be performed on non-conductive Visium slides without appreciable degradation of MSI metabolite signal or SRT RNA signal. Finally, MALDI-MSI was evaluated as a sample-efficient approach for distinguishing de novo Parkinson’s disease patients from controls using limited patient material and minimal sample preparation, reducing analytical time compared to more sample-intensive workflows. In conclusion, this thesis introduces new high-throughput computational methods for automated metabolite annotation in tissue sections, demonstrates the compatibility of MALDI-MSI with SRT, and highlights the versatility of MSI for analyzing sample-limited clinical biofluids.