Comparative Analysis of In vivo Endothelial Cell Translatomes Across Central Nervous System Vascular Beds (Cdh5)

Endothelial cells (ECs) display organ- and tissue-specific heterogeneity. In the eye, the retinal and choroidal vascular beds are distinct networks with different morphological and functional features with the former representing a tight barrier and the latter, a permeable fenestrated vasculature. Given that retinal health critically relies on the function of these vascular beds and that their dysfunction is implicated in a variety of retinal diseases, a molecular understanding of both physiological and pathophysiological characteristics of these distinct vasculatures is critical. Given their interspersed anatomic distribution among parenchymal cells, the study of EC gene expression, in vivo, has been hampered by the challenge of isolating pure populations of ocular ECs in sufficient quantities for large-scale transcriptomics. To address this challenge, we present a methodological and analytical workflow to facilitate inter-tissue comparisons of the in vivo EC translatome isolated from choroid, retina, and brain using the Cre-inducible NuTRAP flox construct and two widely-used endothelial cre mouse lines, constitutive Tie2-Cre and Tamoxifen-inducible Cdh5-CreERT2, which express EGFP and mCherry tags for cell-specific isolation of nuclei and ribosomes. Within each model, inter-tissue comparison of TRAP-RNAseq enrichment (TRAP translatome vs input transcriptome) showed tissue-specific gene enrichments with differential pathway representation. For each mouse model, inter-tissue comparison of the EC translatome (choroid vs brain, choroid vs retina, and brain vs retina) showed over 50% overlap of differentially expressed genes (DEGs) between the three paired comparisons, with differential pathway representation for each tissue. Pathway analysis of DEGs in the Cdh5-NT vs Tie2-NT comparison for retina, choroid, and brain predicted inhibition of processes related to myeloid cell function and activation, consistent with more specific targeting of ECs in the Cdh5-NT than in the Tie2-NT model. While TRAP enriches for EC transcripts in both models, it also captures myeloid transcripts in the Tie2-NT model. We confirmed these observations with a cell sorting approach. We suggest experimental/analytical considerations should be taken when selecting cre-lines to target ECs.