The combination of machine learning and untargeted metabolomics identifies the lipid metabolism -related gene CH25H as a potential biomarker in asthma

Abstract

Background: Lipids, significant signaling molecules, regulate a multitude of cellular responses and biological pathways in asthma which are closely associated with disease onset and progression. However, the characteristic lipid genes and metabolites in asthma remain to be explored. It is also necessary to further investigate the role of lipid molecules in asthma based on high-throughput data.

Objective: To explore the biomarkers and molecular mechanisms associated with lipid metabolism in asthma.

Methods: In this study, we selected three mouse-derived datasets and one human dataset (GSE41665, GSE41667, GSE3184 and GSE67472) from the GEO database. Five machine learning algorithms, LASSO, SVM-RFE, Boruta, XGBoost and RF, were used to identify core gene. Additionally, we used non-negative matrix breakdown (NMF) clustering to identify two lipid molecular subgroups and constructed a lipid metabolism score by principal component analysis (PCA) to differentiate the subtypes. Finally, Western blot confirmed the altered expression levels of core genes in OVA (ovalbumin) and HDM+LPS (house dust mite+lipopolysaccharide) stimulated and challenged BALB/c mice, respectively. Results of non-targeted metabolomics revealed multiple differentially expressed metabolites in the plasma of OVA-induced asthmatic mice.

Results: Cholesterol 25-hydroxylase (CH25H) was finally localized as a core lipid metabolism gene in asthma and was verified to be highly expressed in two mouse models of asthma. Five-gene lipid metabolism constructed from CYP2E1, CH25H, PTGES, ALOX15 and ME1 was able to distinguish the subtypes effectively. The results of non-targeted metabolomics showed that most of the aberrantly expressed metabolites in the plasma of asthmatic mice were lipids, such as LPC 16:0, LPC 18:1 and LPA 18:1.

Conclusion: Our findings imply that the lipid-related gene CH25H may be a useful biomarker in the diagnosis of asthma.

Keywords: Asthma; CH25H; Lipid metabolism; Lipid metabolism-related genes; Machine learning; Untargeted metabolomics.