Distinct cohort of EC genes regulates the formation vs maintenance of BBB.

A) Workflow for transcriptomic analysis. Primary ECs were isolated from E13.5 and adult (2-3 months old) cortex followed by RNA isolation and mRNA sequencing and transcriptomic analysis. B) Comparative transcriptome analysis of ECs from E13.5 and adult cortex. Volcano plot depicting downregulated and upregulated genes in adult primary cortical ECs compared to E-13.5. Genes marked in red are significant (P<0.05) N=3. C) Diagram depicting the percentage of genes downregulated, and upregulated, and shows no differential expression in adult cortical primary ECs compared to E-13.5. D) Downregulated and upregulated genes were categorized with 5 important EC functions. E) Enrichment analysis revealed important BBB-related genes that were differentially regulated during deveolpment. F) Relative mRNA expression of CLDN1, MFSD2A, ZIC3, SOX17, and CLDN5. in primary cortical ECs isolated from E13.5, E-17.5, P0, P7 and adult. Significant differences are observed between E-13.5 and all consecutive stages for CLDN1(*** p<0.001, N=3/group), MFSD2A (*** p<0.001, N=3-4/group) and ZIC3(*** p<0.001N=3-4/group). CLDN1 showed significant differences between E-17.5 vs P7 and adult and P0 vs P7 and adult (*** p<0.001). ZIC3 showed a significant difference between E-17.5 vs P7 and adult es (** p<0.01). Significant differences are observed between E-13.5 vs P7 and adults for CLDN5 and SOX17 (* p<0.001 N=3-4/group).

Epigenetic regulators HDAC2 and PRC2 regulate the transcription of BBB genes.

A) qPCR of adult primary cortical ECs treated with TSA (200nm) and MS-275(10um) 48hrs showed significantly increased mRNA expression of CLDN1 compared to DMSO treated control while CLDN5 was significantly decreased with MS-275 treatment when compared to control (* p<0.001 vs Control #p<0.001 vs TSA treatment N=3/group). B) mRNA expression level of class-I HDAC family members in adult primary cortical ECs. Expression was normalized to housekeeping genes GAPDH and HDAC1. Significant mRNA expression of HDAC2 was observed in adult primary cortical ECs compared to other Class-I HDACs (* p<0.05 vs HDAC1,3 and 8 N=3/group). HDAC1 showed significantly higher expression compared to HDAC3 and HDAC8 (* p<0.05) and HDAC8 showed significantly higher expression compared to HDAC3 (* p<0.05). C) Effect of HDAC2 siRNA on BBB gene expression in adult cortical ECs. Using lipofectamine adult cortical ECs were transfected with HDAC2 siRNA (500µg). qPCR analysis revealed that compared to control siRNA treated group HDAC2 siRNA treated ECs showed significantly increased expression of CLDN1(* p<0.001), MFSD2A (* p<0.05) and ZIC3 (* p<0.05) while CLDN5 (* p<0.05) showed significantly decreased expression. N=3/group D) HDAC2 occupancy of the indicated chromatin regions in primary cortical ECs from E-13.5 and adult. Occupancy was measured by ChIP followed by quantitative PCR (ChIP-qPCR). Chromatin regions are named by the adjacent gene and the distance to the TSS. (* p<0.05 vs IgG N=3-4/group) E) qPCR analysis of CLDN1, CLDN5, MFSD2A, and ZIC3 in EZH2 and control siRNA-treated adult primary cortical ECs. Compared to the control siRNA-treated group EZH2 siRNA-treated ECs showed significantly increased expression of CLDN1 (* p<0.001), MFSD2A (* p<0.001) and ZIC3 (* p<0.05) while CLDN5 (* p<0.05) showed significantly decreased expression. N=3/group. F) ChIP-qPCR analysis of PRC2 subunit EED on indicated chromatin regions and genes in E13.5 and adult primary cortical ECs. * p<0.05 vs IgG, N=3-4/ group. G) Schematic representation of HDAC2 and PRC2 binding on indicated genes in cortical ECs at E-13.5 and adult.

BBB genes exhibit diverse post-translational histone modifications.

A) ChIP-qPCR analysis of histone marks H3K27me3, H3K4me3, and H3K9ac on 1KB region CLDN1 gene in E13.5 and adult cortical ECs. CLDN1 gene was downregulated in ECs during development. The ChIP signals were normalized to IgG. B) ChIP-qPCR showing the H3K27me3, H3K4me3, and H3K9ac density at the CLDN5 gene in E13.5 and adult cortical ECs. CLDN5 gene was upregulated during development. Data are shown as mean ± SD. ***p < 0.001, **p < 0.01, *p < 0.05 vs IgG & ###p < 0.001, ##p < 0.01, #p < 0.05 vs respective E-13.5 histone mark. N = 3-4/group. C) Schematic representation of H3K27me3, H3K4me3, and H3K9ac binding density on CLDN1, CLDN5, MFSD2A, and ZIC3 in E13.5 and adult primary cortical ECs. Shape size indicates the binding density in the indicated chromatin regions.

HDAC2 activity is required for the formation of a functional BBB while PRC2 is dispensable.

A) Effect of deletion of HDAC2 from the ECs during embryonic development. Schematic representation of breeding scheme and generation of EC-specific KO of HDAC2. Tamoxifen was delivered to the pregnant mother at E-12.5 and brains were harvested at E-17.5. B) Representative phase microscopy images of the dorsal surface and ventral surface of the brain at E-17.5. HDAC2 ECKO shows a significant increase in pia vessels and as pointed out by black arrows dorsal surface. Black arrows in the ventral surface showed dilated vessels. C) BBB permeability assay using 70KD tracer and isolectin B4 (IB4) staining to image the vessels. In HDAC2 ECKO Green fluorescent tracer was leaked out of the vessels as indicated by white arrows. 10x images are acquired and merged using tile scanning. Scale bar, 500µm D) Vessel percentage area of HDAC2 ECKO was significantly higher than WT (* p<0.001 vs control N=10). A significant increase in fluorescent intensity was quantified in HDAC2 ECKO compared to WT (* p<0.0001 vs control N=3 for W.T and N =5 for HDAC2 ECKO). E) Schematic representation of breeding scheme and generation of EZH2 ECKO. Tamoxifen was injected as same as for HDAC2 ECKO. F) Representative phase microscopic image of WT and EZH2 ECKO. As shown by black arrows EZH2 ECKO brain shows dilated vessels with no visible increase in pial angiogenesis. G) BBB permeability assay using 70KD-FITC Dextran shows subtle leakage of tracer out of the vessels in EZH2 ECKO, represented using white arrows. IB4 staining reveals the vessels in the brain. 10x images are acquired and merged using tile scanning. Scale bar, 500 µm H) Quantification of vessel percentage area (W.T N=9 EZH2 ECKO N=10) and fluorescent intensity didn’t show any significant difference between WT and EZH2 ECKO (W.T N=5 EZH2 ECKO N=5).

Wnt pathway is active in adult CNS ECs but Wnt target genes are epigenetically repressed.

A) Diagram depicting the percentage of Wnt-related genes downregulated, and upregulated adult cortical primary ECs compared to E-13.5. Selected important Wnt-related genes are shown. B) Ligand-independent transcriptional repression of Wnt target genes. In primary cortical ECs from E-13.5, activating the Wnt pathway with Wnt3a (200ng/mL) or CHIR-99021 (5uM) for 48 hrs. caused increased mRNA expression of Wnt target genes AXIN2 and LEF1 (measured via qRT-PCR). However, activation of the Wnt pathway in primary adult mouse brain ECs does not increase Wnt target gene expressions. *p<0.001 vs E-13.5 control, ns-no significant difference n=3/group. C) mRNA sequencing was performed in control and Wnt3a (200ng/mL) treated adult primary cortical ECs. Differentially expressed genes were categorized into six selected categories that are important to CNS endothelial cells. D) Immunofluorescence staining of β-catenin (green) in control and Wnt agonist Chir-99021 treated endothelial cells. White arrows indicate the nuclear localization of β-catenin to the nucleus. 20x images are acquired and cropped and enlarged. Scale bar, 1 µm. E) Adult primary cortical ECs transfected with control, HDAC2 & EZH2 siRNA showed significant upregulation of Wnt target genes AXIN2. ***p<0.001 vs control siRNA & #p<0.05 vs control siRNA. N=3/group F) HDAC2, histone marks H3K27me3, H3K4me3, and H3K9ac occupancy on the AXIN2 TSS regions in primary cortical ECs from E-13.5 and adult. Occupancy was measured by ChIP-qPCR. N=3/group ***p < 0.001, **p < 0.01, *p < 0.05 vs IgG & #p < 0.05 vs H3K4me3 E-13.5.

Low Wnt signaling epigenetically modifies the BBB genes to achieve BBB maturation.

A) Effect of Wnt pathway inhibition on BBB genes in E13.5 primary cortical ECs. E-13.5 ECs were treated with LF3(50um) for 48hrs to inhibit the Wnt pathway. Significantly decreased mRNA expression of AXIN2 confirmed the reduced Wnt pathway. mRNA expression of CLDN1, MFSD2A and ZIC3 was significantly decreased CLDN5 and SOX17 expression was significantly increased after LF3 treatment. Data are shown as mean ± SD. ***p < 0.001, **p < 0.01, *p < 0.05 vs E13.5 control N=3/group. B) Wnt pathway inhibition via LF3 induce epigenetic modifications in target gene AXIN2 and BBB genes CLDN1, MFSD2A, and CLDN5. First row-AXIN2 showed significant enrichment of HDAC2 and EED in LF3 treated E13.5 ECs when compared to control (*p < 0.05 vs IgG). Histone mark H3K27me3 showed significant enrichment in both conditions compared to IgG however LF3 treated ECs showed significantly increased enrichment compared to the control. **p < 0.01 vs IgG, #p < 0.05 vs control N=3-4/group. Second row-LF3 treated E13.5 ECs showed significant enrichment of HDAC2 in CLDN1 TSS (*p < 0.05 vs IgG N=3-4/group). MFSD2A showed significant enrichment in both conditions while LF3 treatment showed a significantly increased enrichment compared to the control (*p < 0.05 vs IgG & ##p < 0.01 vs control N=3/group). CLDN5 didn’t show any significant difference in HDAC2 binding (not shown) while active histone marks H3K9ac showed significant enrichment in both conditions with an increased enrichment with LF3 treatment (***p < 0.001, *p < 0.05 vs IgG & #p < 0.05 vs control N=3/group). Third row-H3K4me3 ChIP-qPCR on the TSS region of CDLN1, MFSD2A, and CLDN5 showed significant enrichment in both conditions with a decreased enrichment with LF3 treatment on CDLN1 and MFSD2A and an increased enrichment with LF3 treatment on CLDN5. ***p < 0.001, *p < 0.05 vs IgG & ##p < 0.01, #p < 0.05 vs control N=3-4/group. C) Schematic representation of breeding scheme and generation of EC-specific gain of function (GOF) of β-catenin. Tamoxifen was delivered to the pregnant mother at E-12.5 and brains were harvested at E-17.5. D) Representative Phase microscopy image of the dorsal brain from W.T and β-catenin-GOF. Images in the square box were enlarged to show the increase in pial vessel angiogenesis. E) BBB permeability assay using the 70KD FITC-Dextran tracer. Cortical vessels are stained using IB4.10X tile scanning images were acquired and merged. FITC dextran was leaked out of the vessels in the brain of β-catenin GOF compared to W.T. 10x images are acquired and merged using tile scanning. Scale bar, 500 µm F) Quantification of brain vessels percentage area (***p < 0.001 vs WT N=4-5/group) and green fluorescent intensity showed a significant increase in β-catenin GOF compared to W.T. ***p < 0.001, *p < 0.05 vs WT N=5-6/group.

In adult ECs, treatment with MS-275 induces partial reactivation of BBB and angiogenesis-supporting gene cohorts.

A) Diagram depicting the percentage of genes downregulated, upregulated, and shows no differential expression in MS-275 treated adult cortical ECs compared to Control. Downregulated and upregulated genes were categorized with 6 important EC functions. B) Heat map of expression values (Z score) for differentially expressed genes (*p adj < 0.05) in E13.5, adult control, and adult treated with MS-275. Five different gene categories which show a significant difference between E13.5 vs adult control and Adult Control vs Adult MS-275 treatment are presented. N=3 for E13.5 and adult control, N =2 for MS-275. C) Adult primary cortical ECs treated with MS-275 for 48 hrs. showed significant activation of CLDN1 and downregulation of Cldn5. mRNA analysis on ECs after 7 days of withdrawing MS-275 showed a significant reversal of expression back to normal. *p<0.001 compared to control 48 hrs. # p<0.001 compared to MS-275 48 hrs. D) Representative phase contrast image of human temporal lobe vessels in culture collected from epilepsy patients who underwent surgery. MS-275 treated human vessels showed significantly increased mRNA expression of CLDN1, and AXIN2 when compared to control. **p<0.01 and *p<0.05 N=5/group. No significant difference was observed in CLDN5 expression. E) Schematic diagram illustrating the mechanisms underlying BBB formation and maintenance led by epigenetic regulators HDAC2 and PRC2. EC gene cohorts that support BBB formation are epigenetically repressed by HDAC2 and PRC2 during development. Active Wnt signaling supports the expression of gene cohorts required for the BBB formation, while a reduction in Wnt signaling recruits HDAC2 to these gene cohorts to support the formation of a functional or intact BBB. Inhibiting HDAC2 in adult ECs induce the reacquisition of gene cohorts that support BBB formation and thus represents a potential therapeutic opportunity to repair damaged BBB.