Spatial Transcriptomics of Schizophrenia Insular Cortex Reveals Blood-Brain Barrier Hyperglycolysis and Increased Parenchymal Mitochondrial Respiration

Abstract

IntroductionThe blood-brain barrier (BBB) acts as the metabolic and immunological gatekeeper of the brain. Since alterations in neurometabolism and neuroimmunity are found in schizophrenia-spectrum disorders (SSD) which are hypothesised to be important disease mechanisms, we aimed to investigate whether changes in BBB function could underly these findings using a novel spatially resolved transcriptomics technique. MethodsFormalin-fixed paraffin-embedded insular cortex tissue from 8 brain donors with SSD and 8 matched controls derived from the Netherlands Brain Bank-Psychiatry were selected for whole transcriptome analysis (GeoMx Human Whole Transcriptome Atlas) on the GeoMx Digital Spatial Profiler platform. Combining nuclear staining with an endothelial cell marker (CD31) allowed for the separation of BBB and parenchyma areas of interest (AOIs) for downstream sequencing on the Illumina NextSeq 2000. For each sample, biological triplicates were sequenced. Comparing SSD to control for both the BBB and parenchyma AOIs, differentially expressed genes (DEGs) were identified using a Linear Mixed Model, a heatmap was created displaying all genes with a false-discovery rate <0.01, and Fast Gene Set Enrichment Analysis was used for pathway analysis. ResultsA total of 96 whole transcriptome profiles were generated (24 BBB and 24 parenchyma for both SSD and controls). Expression of endothelial genes (PECAM1/CD31, CLDN5, VWF, CD34, ENG) was significantly increased in BBB, confirming enrichment of endothelial cells (ECs) in this AOI. Cluster analysis showed perfect clustering of BBB versus parenchyma, and good clustering of SSD samples within the BBB cluster. At a |Log2FC| [&ge;] 0.25, we identify 265 significantly DEGs in the BBB AOI and 6 in the parenchyma AOI comparing SSD to control. Pathway analysis revealed a distinct metabolic transcriptional profile in SSD, characterized by hyperglycolysis in the BBB and increased mitochondrial energy metabolism in parenchyma. ConclusionOur findings implicate the BBB in the metabolic pathophysiology of SSD. Furthermore, our findings add nuance to the existing understanding of brain bioenergetic alterations in SSD, suggesting that metabolic changes may be region-specific rather than generalized. This highlights the need for a brain mapping approach examining multiple brain regions from the same donor. Finally, the distinct metabolic profiles of the BBB and brain parenchyma emphasize the importance of spatial multi-omics in post-mortem psychiatric research and the potential for therapies targeting BBB function in SSD.