Machine learning-assisted MALDI-MSI to characterize hippocampal subregion lipid and purine metabolic alterations in depression-related dry eye disease

Abstract

Dry eye disease (DED) and depression exhibit high comorbidity, yet lipid and purine diagnostic biomarkers for depression-related DED remain unidentified. In this study, matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) was employed to conduct, for the first time, in situ localization analysis of small-molecule substances, within the CA1, CA2, CA3, and DG subregions of the hippocampal tissue in rats with depression-related DED. Our findings indicate that the 9-aminoacridine (9AA) matrix enabled visualization of over 120 diverse metabolites in each hippocampal region. Notably, the 3-O-sulfogalactosylceramides, phosphatidylinositols (PIs), phosphatidylserines (PSs), phosphatidylethanolamines (PEs), adenosine monophosphate (AMP), and adenosine diphosphate (ADP) may be potential biomarkers. Additionally, the unsupervised machine learning and Pearson correlation analysis may facilitate the identification of potential biomarkers for depression-related DED. The Mantel analysis results of this study suggest that the development of depression and DED may be related to the reduction of CDP-ethanolamine and the increase in PG(22:6/20:4), PI(16:0/18:1), guanosine triphosphate (GTP), and eicosadienoic acid. In conclusion, using MALDI-MSI with machine learning, we enhance the understanding of the pathogenesis of depression-related DED. The study helps discover potential biomarkers and simultaneously elucidates the metabolic differences in each subregion of the hippocampus.