Regional heterogeneity of the blood-brain barrier

Abstract

The blood-brain barrier (BBB), formed by specialized endothelial cells (ECs), regulates the extracellular composition of the central nervous system (CNS). Little is known about whether there are regional specializations of the BBB that may control the function of specific neural circuits. We use single cell RNA-seq to characterize ECs from nine CNS regions in male mice: cortex, hippocampus, cerebellum, spinal cord, striatum, thalamus, hypothalamus, midbrain, and medulla/pons. Although there is a core BBB transcriptional profile, there are significant regional specializations. Stra6, a retinoid transporter, is highly enriched in the BBB of the nucleus accumbens shell (ShNAc) and ventral cochlear nucleus, and is controlled by dietary vitamin A, through endothelial RARƔ. EC Stra6 regulates the deposition of retinoids specifically in the ShNAc and cochlear nucleus, and is required for the function of the ShNAc, in a retinoid-dependent manner. Thus regional specializations of the BBB can regulate the function of local brain regions.

Fig. 1. scRNA-seq of endothelial cells from nine brain regions.

A UMAP plot of all endothelial cells. Plotted points (cells) are colored by brain region from which they were isolated. The number of endothelial cells analyzed from each brain region is shown at right. Each region includes three biological replicates (independently isolated and sequenced cells pooled from 3–6 mice). B UMAP plot with points colored by clusters representing endothelial arteriovenous subtypes. Clustering strategy is shown in Supplementary Fig. 2. C UMAP plots with points colored based on expression of indicated genes. Cldn5 (pan-BBB endothelial), Gkn3 (arterial-enriched), and Icam1 (venous-enriched) are shown. Color bars indicate expression (log-normalized counts). D Fraction of capillary, arterial, venous, and tip-like endothelial cells across brain regions. Bars indicate the mean of three biological replicates and error bars indicate standard deviation. Fraction arterial endothelial cells across brain regions: F(8,18) = 1.4, P = 0.25, one-way ANOVA. Fraction venous endothelial cells across brain regions: F(8,18) = 1.2, P = 0.34, one-way ANOVA. Fraction tip-like endothelial cells across brain regions: F(8,18) = 1.4, P = 0.28, one-way ANOVA. E Differential expression analysis comparing capillary endothelial cells across brain regions, with region abbreviations as defined in (A). Volcano plots show genes with region-enriched and region-depleted expression in the region of interest compared to all other regions. Points are shown for all genes with average expression >1000 pseudobulk counts; colored points indicate genes with statistically significant enrichment or depletion (average expression >1000 psuedobulk counts, P < 0.05 DESeq2 Wald test (two-sided) with Benjamini-Hochberg correction). F UMAP plot of all endothelial cells, with points colored based on expression of Stra6. Color bar indicates expression (log-normalized counts). G Expression (log-normalized counts) of Stra6 in striatal capillary, arterial, venous, and tip-like endothelial cells. H Expression (log-normalized counts) of Stra6 and hippocampal-enriched receptor mediated transcytosis targets (Lrp8, Igf1r) in capillary endothelial cells across brain regions. Points represent average expression in each biological replicate and error bars represent SD. Source data are provided as a Source Data file. Gene expression data are provided in Supplementary Data 2.

Fig. 2. Stra6 expression within the CNS.

A–D Sagittal (A, C) and coronal (B, D) sections of adult mouse brains were stained with an antibody directed against Stra6 (red) and DAPI to stain nuclei (blue). E, F Tissue sections of the rmShNAc (E) and PVCN (F) were stained with an antibody against Stra6 (red), with BSLI-Fluorescein (green) to label endothelial cells, and DAPI to label nuclei. Stra6 displayed vascular staining in the rmShNAc (A, B) and in the PVCN (C, D). All scale bars: 200 µm. Similar results were obtained in five independent experiments.

Fig. 3. Stra6 expression at the blood-brain barrier is dynamically regulated by dietary amount of Vitamin A through RARƔ.

A Tissue sections of the rmShNAc and VCN from 6 week old C57BL/6 mice raised on either VitA-deficient diet, VitA-sufficient diet, VitA-excess diet or raised on VitA-deficient diet then transferred to VitA-excess diet for 1 week prior to analysis, were stained with antibodies against Stra6 (red) and CD31 (green). Yellow arrows indicate vessels with high signal, blue arrows indicate vessels with low signal. Scale bar: 200 µm. B Tissue sections of the different brain regions from C57BL/6 mice raised on VitA-excess diet stained against Stra6 (red) and CD31 (green). Scale bar: 200 µm. C Percentage of vascular length positive for Stra6 in different brain regions according to different diet paradigm: VitA-deficient diet, VitA-sufficient diet, VitA-excess diet and VitA-deficient diet transferred to VitA-excess diet for 1 week. (n = 3 mice per region and diet). The vascular length with both high signal and low signal was quantified for each brain region. Statistics and p-values: two-way ANOVA (region P < 0.0001, diet P < 0.0001, interaction P < 0.0001) followed by a Tukey’s multiple comparisons test: rmShNAc VitA-deficient versus VitA-sufficient, VitA-deficient versus VitA-excess, and VitA-deficient versus VitA-deficient transferred to VitA-excess, total signal ****P < 0.0001; PVCN VitA-deficient versus VitA-sufficient, VitA-deficient versus VitA-excess, and VitA-deficient versus VitA-deficient transferred to VitA-excess, total signal ****P < 0.0001. D Tissue sections of the rmShNAc and PVCN from 6 week old RARg fl/fl; Cdh5(PAC)-CreERT2 and RARg fl/fl littermate controls that were raised on VitA-deficient diet, injected with tamoxifen at 4 weeks of age, and switched to a VitA-excess diet at 5 weeks, were stained with antibodies against Stra6 (red) and CD31 (green). Scale bar: 200 µm. Yellow arrows indicate vessels with high signal, blue arrows indicate vessels with low signal. E Percentage of vascular length positive for Stra6 in rmShNAc and PVCN in RARg ECK (RARg fl/fl; Cdh5(PAC)-CreERT2, n = 6) mice and in their littermate controls (RARg fl/fl, n = 7). The vascular length with both high signal and low signal was quantified for each brain region. Statistics and p-values: two-way ANOVA (region P = 0.77, genotype P < 0.0001, interaction P = 0.88) followed by Tukey’s multiple comparisons test: rmShNAc RARg fl/fl versus RARg fl/fl; Cdh5(PAC)-CreERT2, total signal ****P < 0.0001, high signal ****P < 0.0001, low signal P = 0.097; PVCN RARg fl/fl versus RARg fl/fl; Cdh5(PAC)-CreERT2, total signal ****P < 0.0001, high signal ****P < 0.0001, low signal P = 0.17. Error bars represent SEM. Source data are provided as a Source Data file.

Fig. 4. Endothelial specific Stra6 mutant mice display spatial memory defect due to a reduction of retinol uptake.

A–L Retinols and RBP4 measurements from dissected brain regions or serum from endothelial-specific Stra6 knockout mice (Stra6 f/f; Cdh5(PAC)-ERT2 and littermate controls (Stra6 f/f) raised on a VitA-sufficient diet (blue) or VitA-excess diet (green). Biological replicates are shown as individual data points. A–C, G–J n = 4 per genotype, D Stra6 f/f n = 7, Stra6 f/f; Cdh5(PAC)-ERT2 n = 6, E Stra6 f/f n = 6, Stra6 f/f; Cdh5(PAC)-ERT2 n = 4, F Stra6 f/f n = 5, Stra6 f/f; Cdh5(PAC)-ERT2 n = 5, K Stra6 f/f n = 8, Stra6 f/f; Cdh5(PAC)-ERT2 n = 8, L Stra6 f/f n = 5, Stra6 f/f; Cdh5(PAC)-ERT2 n = 4. Statistics and p-values (A–L): unpaired two-tailed t-tests: A P = 0.17, B ***P = 0.0010, C *P = 0.012, D P = 0.55, E P = 0.38, F P = 0.72, G *P = 0.020, H P = 0.84, I P = 0.90, J P = 0.15, K P = 0.61, L P = 0.47. M Discrimination index for the novel location recognition task in endothelial-specific Stra6 knockout mice (Stra6 f/f; Cdh5(PAC)-ERT2, n = 23) and littermate controls (Stra6 f/f, n = 20) raised on a VitA-sufficient diet (blue). N Latency to fall in endothelial-specific Stra6 knockout (n = 16) and littermate control mice (n = 10) raised on VitA-sufficient diet (blue). O Discrimination index for the novel object recognition task in in endothelial-specific Stra6 knockout mice (n = 9) and littermate controls (n = 9) fed on VitA-sufficient diet. P Discrimination index for the novel location recognition task in endothelial-specific Stra6 knockout mice (n = 21) and littermate controls (n = 24) raised on a VitA-excess diet (green). Q Latency to fall in endothelial-specific Stra6 knockout (n = 7) and littermate control mice (n = 11) raised on VitA-excess diet (green). Statistics and p-values (M–Q): unpaired two-tailed t-tests: M ***P = 0.0008, N Day 1 P = 0.28, Day 2 P = 0.27, Day 3 P = 0.54, Day 4 P = 0.42, Day 5 P = 0.58, O P = 0.96, P P = 0.15, Q Day 1 P = 0.84, Day 2 P = 0.99, Day 3 P = 0.86, Day 4 P = 0.64, Day 5 P = 0.94. A–L, N, Q Error bars represent SEM. M, O, P Dashed lines indicate median and dotted lines indicate first and third quartiles. Source data are provided as a Source Data file.