Two-Way Orthogonal Partial Least Squares (O2PLS) Analysis of the Lipidome and Transcriptome in Prediabetic and Diabetic Neuropathy

Abstract

Peripheral neuropathy (PN) is a common complication of prediabetes and type 2 diabetes (T2D) for which there are limited treatment options. Using mouse models of prediabetes and T2D we have demonstrated that dietary reversal (DR) corrects the PN phenotype. To identify the mechanisms that underlie PN development and reversal, we employed an omics-based approach. This was achieved by integrating transcriptomic and lipidomic datasets derived from mouse sciatic nerve using two-way Orthogonal Partial Least Squares (O2PLS) analysis, a platform that is useful in uncovering the large-scale organization of metabolic networks. Sciatic nerve was isolated from five groups of mice that consisted of controls, high-fat (HF) fed mice (prediabetic), HF mice injected with low dose STZ (T2D), HF mice subjected to 8 week dietary reversal (DR) and T2D-DR. Tissue was processed for gene expression by mRNA-seq or for lipid species profiling by untargeted shotgun lipidomics. A total of 6,024 genes and 578 lipid species were selected and the systematic joint covariation between gene and lipid species were modeled using O2PLS. A clear separation occurred between T2D/prediabetes and healthy/DR mice. Compared to a traditional ANOVA analysis, we selected genes with the top 1% values (52 genes) as highly inter-correlated genes with lipid species; we additionally identified 28 more genes (53.8%) which also highly correlated with lipid species. The top 50 lipid species, including 29 novel species, were identified based on these analyses and primarily consisted of triacylglycerols (TG) and diacylglycerols (DG). Moreover, a comprehensive lipid-gene interaction network was constructed based on these genes and lipid species. This study demonstrates that O2PLS can provide significant and biologically relevant coordination between gene expression and lipid levels and can offer new insights into our understanding of the metabolic networks underlying PN in T2D. Disclosure K. Guo: None. E.L. Feldman: None. J. Hur: None.