Clustering-based time series analysis on insulin response in the blood-brain barrier

BackgroundCritical functions of the blood-brain barrier (BBB), including cerebral blood flow and vascular response, are regulated by insulin signaling pathways. Therefore, endothelial insulin resistance could lead to vascular dysfunction, which is associated with neurodegenerative diseases such as Alzheimer’s disease (AD).ObjectiveThe objective of the current study is to map the dynamics of insulin-responsive pathways in polarized human cerebral microvascular endothelial cells (hCMEC/D3) cell monolayers, a widely used BBB cell culture model, to identify molecular mechanisms underlying BBB dysfunction in AD.MethodsRNA-Sequencing (RNA-Seq) was performed on hCMEC/D3 cell monolayers with and without insulin treatment at various time points. The Short Time-series Expression Miner (STEM) method was used to identify clusters of genes with distinct and representative patterns. Functional annotation and pathway analysis of the genes from top clusters were conducted using the Webgestalt and Ingenuity Pathway Analysis (IPA) software, respectively.ResultsQuantitative expression differences of 19,971 genes between the insulin-treated and control monolayers at five-time points were determined. STEM software identified 11 clusters with 3061 genes across that displayed various temporal patterns. Gene ontology enrichment analysis performed using the top 5 clusters demonstrated that these genes were enriched in various biological processes associated with AD pathophysiology. The IPA analyses revealed that signaling pathways exacerbating AD pathology such as inflammation were downregulated after insulin treatment (clusters 1 to 3). In contrast, pathways attenuating AD pathology were upregulated, including synaptogenesis and BBB repairment (clusters 4 and 5).ConclusionsThese findings unravel the dynamics of insulin action on the BBB endothelium and inform about downstream signaling cascades that potentially regulate neurovascular unit (NVU) functions that are disrupted in AD.