Session: Session 4

Predicting protein pathways via spatial lipidomic and proteomic correlation: toward Dry Proteomics in human Glioblastoma

The direct prediction of protein pathways from lipid analyses via MALDI MSI represents an unmet challenge in current research. To tackle this, the concept of "dry proteomics" was introduced, focusing on the spatial localization of identified clusters in both lipid and protein imaging. By establishing consistency in cluster appearance across omics' images, it becomes feasible to link them to specific lipid and protein pathways initially identified through rigorous lipidomic and proteomic analyses. Consequently, the interplay between lipid and protein pathways within these clusters forms the core of dry proteomics.

 

In this study, we refined the methodology using rat brain tissue from the cerebellum as a model system. Sequential tissue sections underwent lipid, protein, and peptide MALDI-MSI analysis. Utilizing an unsupervised clustering approach implemented through a MATLAB script, we observed that identified clusters consistently occupied distinct spatial locations, irrespective of the analyzed molecules, corresponding to the granular layer, molecular layer, and white matter regions of the cerebellum. A Python script facilitated the differentiation of lipid and protein clusters based on discriminative ion patterns. Subsequent characterization of these clusters involved spatial omics analysis. This comprehensive approach enabled the development of a predictive computer model capable of identifying clusters in various tissues based on their unique lipid signatures and elucidating the associated protein pathways.



Furthermore, we conducted a similar analysis on 50 patients with glioblastoma, confirming the ability to associate specific lipids with proteins from Lipids MALDI MSI images and establishing correlations with patient prognosis. This comprehensive approach not only advances our understanding of protein pathways from lipid analyses but also holds promise for clinical applications, particularly in cancer research and prognostication.