Loss of the transcription factor RBPJ induces disease-promoting properties in brain pericytes

Abstract

Sufficient vascular supply is indispensable for brain development and function, whereas dysfunctional blood vessels are associated with human diseases such as vascular malformations, stroke or neurodegeneration. Pericytes are capillary-associated mesenchymal cells that limit vascular permeability and protect the brain by preserving blood-brain barrier integrity. Loss of pericytes has been linked to neurodegenerative changes in genetically modified mice. Here, we report that postnatal inactivation of the Rbpj gene, encoding the transcription factor RBPJ, leads to alteration of cell identity markers in brain pericytes, increases local TGFβ signalling, and triggers profound changes in endothelial behaviour. These changes, which are not mimicked by pericyte ablation, imperil vascular stability and induce the acquisition of pathological landmarks associated with cerebral cavernous malformations. In adult mice, loss of Rbpj results in bigger stroke lesions upon ischemic insult. We propose that brain pericytes can acquire deleterious properties that actively enhance vascular lesion formation and promote pathogenic processes.

Fig. 1

Pdgfrb-CreERT2-mediated Rbpj deletion compromises brain vessels. a Pdgfrb-CreERT2 transgene and Cre-mediated recombination of Rbpj^lox allele. b Whole brains and coronal sections from control and Rbpj^iPC mice at P7 or P10 after tamoxifen administration (P1–P3). Scale bar, 2 mm. c Haemorrhagic index. p-values, Mann–Whitney U-test, n = 8. d Red blood cell (RBC) extravasation in brain parenchyma. HPF, hippocampal formation. p-values, one-way ANOVA and Tukey’s test, n = 5. e Vascular area and density in brain cortices. p-values, Student’s t-test, n = 6–7. f Confocal images of blood vessels (CD31⁺, white) in brain cortices. Arteries (A) and veins (V) are indicated. Scale bar, 100 μm. g Quantitation of vessel diameter in cortical blood vessels. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test, n = 4. n.s., not statistically significant. h Arteriovenous malformation in brain superficial vasculature as revealed by the presence of latex in the caudal rhinal vein (arrowhead). Scale bar, 1 mm. i Confocal images of cortical blood vessels stained for GLUT1 (red, white) and PDGFRβ (white). Normal sprouts (dashed circles) replaced by blunt-ended capillaries (yellow arrowheads) in Rbpj^iPC mice. Scale bar, 50 μm. j Quantitation of sprouting in brain cortices. p-values, Welch’s t-test, n = 4. k Confocal images of blood vessels (ICAM2⁺, red) and EC nuclei (ERG⁺, white) in brain cortices. Scale bar, 100 μm. l Quantitation of EC proliferation in blood vessels of P7 mice. p-values, Welch’s t-test, n = 7–8. m Confocal images of cortex capillaries stained for ICAM2 (red), ERG (white), and GFP (green, recombined mural cells). Right panel shows single optical sections of boxed inset in Rbpj^iPC (left column). Recombined pericytes (white arrows) around tangled capillaries with superimposed EC nuclei (white arrowheads). Scale bar, 50 μm (left panels) and 10 μm (high magnifications). n RT-qPCR analysis in sorted ECs from brain cortices of P7 and P10 mice. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test for Dll4 and Myc; one-way ANOVA with Sidak’s test for Esm1, n = 4–8. All images correspond to P10 mice unless otherwise indicated. Error bars represent s.e.m. Source data are provided as a Source Data file

Fig. 2

Ultrastructural changes and blood flow assessment in Rbpj^iPC brains. a Electron micrographs of P10 brain cortex capillaries. Note EC hyperplasia and lumen deformation in mutant. EC junctions are not compromised (arrows). Bottom panel shows higher magnification of the boxed inset in Rbpj^iPC for better appreciation of electron-dense and continuous junctions. The basement membrane between Rbpj^iPC ECs (E) and pericytes (P) is notably enlarged (arrowhead). Scale bar, 2 μm. b Electron micrographs of mouse brain cortex capillaries at P10. ECs (E), pericytes (P), astrocyte endfeet (A), and basement membrane (black arrowhead) are indicated. Note vacuolization (black arrows) in Rbpj^iPC samples as well as emission of luminal projection (white arrowheads), whereas EC junctions appear intact (dashed arrow). Scale bar, 1 μm (top panels) and 2 μm (bottom panel). c Quantitation of astrocyte endfeet coverage in P10 cortical blood vessels. p-values, Student’s t-test, n = 4. n.s., not statistically significant. d Confocal images of brain cortex blood vessels (tdT⁺, blue) showing areas of red blood cells (Ter119⁺, red) extravasation which are not associated to defects in astrocyte endfeet polarization (Aqp4⁺, white) or absence of recombined mural cells (GFP⁺, green). Scale bar, 50 μm. e In vivo two-photon imaging of brain blood vessels from P10 mice visualized by intravenous injection of Texas Red-dextran (white). Arrowheads (white, red, or cyan for capillaries, arteries, or veins, respectively) indicate the direction of blood flow, velocity (mm s⁻¹) is annotated for specific vessel segments. Scale bar, 100 μm. f Quantitation of blood flow velocities of different brain vascular segments in P10 mice. p-values, one-way ANOVA with Sidak’s test, n = 3–4. n.s., not statistically significant. g Confocal images of brain cortex blood vessels stained for GLUT1 (red), PDGFRβ (white), and GFP (green) in P10 mice. Note that despite of vessel defects in the knockout animals, mural cell coverage is not affected. Scale bar, 100 μm. h Quantitation of mural cell coverage in cortical vasculature at P10. p-values, Welch’s t-test, n = 4. n.s., not statistically significant. Error bars represent s.e.m. Source data are provided as a Source Data file

Fig. 3

Transcriptome profiling of Rbpj^iPC mural cells. a MA-plots of DEGs between P7 and P10 control and Rbpj^iPC mutant mural cells. The x-axis represents the mean normalized counts and the y-axis shows the log₂ fold change >0.5. Red dots correspond to statistically significant DEGs. Rbpj is indicated. b Venn diagram showing the overlap in DEGs between control and Rbpj^iPC mutant mural cells at P7 and P10. c Top gene ontology (GO) biological process and cellular components terms related to differentially expressed genes in P7 Rbpj^iPC mural cells (FDR, corrected p-value). d Model-based hierarchical clustering heat map of transcripts from control and Rbpj^iPC mural cells at P7 and P10. Boxed region (dashed line) correspond to gene clusters which are consistently up (1 and 6) or downregulated (2) in Rbpj^iPC mice both at P7 and P10 stages. e RT-qPCR analysis of putative pericyte markers in sorted mural cells from P10 control and Rbpj^iPC brain cortices. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test, n = 4. f Heat map representation of known pericyte markers and recently proposed pericyte-enriched genes downregulated in Rbpj^iPC mural cells at P7 and P10. g Heat map representation of arterial/arteriolar vascular SMC-enriched genes upregulated in Rbpj^iPC mural cells at P7 and P10. h RT-qPCR analysis of vascular SMC markers in sorted mural cells from P7 and P10 control and Rbpj^iPC brain cortices. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test, n = 4. Error bars represent s.e.m. Source data are provided as a Source Data file

Fig. 4

Functional defects of Rbpj mutant mural cells. a Confocal images of P10 cortical arteries stained for isolectin B4 (IB4) or CDH5 (red), αSMA (green), and Desmin (Des, white). Note irregular SMC coverage of mutant arteries (asterisk) and lack of αSMA labelling in Des⁺ regions (white arrow). Scale bar, 50 μm (left panels) and 25 μm (right panel, high magnification). b SMC coverage in P10 brain cortex arteries. p-values, Student’s t-test, n = 6. c Confocal images of P10 cortical vasculature (tdT⁺, red) stained for GFP (green), SM22α (blue, white), and αSMA (white). Note the expression of αSMA and SM22α in the Rbpj^iPC microvasculature (white arrowheads). Scale bar, 100 μm. d–f Confocal images of brain cortex vasculature showing increased mural cell-specific staining of Vimentin (d, arrowheads), Desmin (Des, e), and Nestin (f) (all in white) in P10 Rbpj^iPC blood vessels (labelled with IB4, CD31, or tdT, red) relative to control. Scale bar, 50 μm (d) or 100 μm (e, f). g Gene expression fold change (RNA-seq fpkm) in Rbpj^iPC mutants relative to age-matched controls (set as 1). p-adjusted values < 0.0001 for all with exception of Des expression at P7. n = 3. n.s., not statistically significant. h Confocal images of P10 cortical blood vessels showing increased immunostaining for phospho-myosin light chain 2 (phMLC2, white) in pericytes (PDGFRβ⁺, red; GFP⁺, green). Scale bar, 50 μm. i Collagen gel contraction assay showing increased contractile ability of lentivirus-Cre transfected Rbpj^iPC relative to control primary mouse brain pericytes. Scale bar, 4 mm. j Contractility quantitation at 18 and 30 h after pericyte seeding. p-values, Kruskal–Wallis with Dunn’s test, n = 5. k RT-qPCR analysis of Acta2 and Tagln expression in cultured pericytes. p-values, one-way ANOVA with Sidak’s test, n = 3. l Confocal images of P10 brain cortex microvasculature stained for ICAM2 (red), GFP (green), and PDGFRβ (white). Cre-induced membrane-tagged GFP expression allows detailed imaging of abnormal protrusions emerging from Rbpj^iPC pericytes (yellow arrowheads). Scale bar, 25 μm. Error bars represent s.e.m. Source data are provided as a Source Data file

Fig. 5

Increased TGFβ signalling in the Rbpj^iPC cerebral vasculature. a GSEA of overexpressed genes (RNA-seq model-based clustering analysis) reveals upregulation of TGFβ targets in Rbpj^iPC. ES, enrichment score; NES, normalized enrichment score; FDR, false discovery rate. b Confocal images showing increased phosphorylation of SMAD3 (phSMAD3, white) in P10 Rbpj^iPC brain cortex vasculature (tdT⁺, red). Scale bar, 100 μm. c phSMAD3 immunosignal intensity quantitation in P10 brain vasculature. p-values, Welch’s t-test, n = 6. d Confocal images showing increased phosphorylation of SMAD1/5 (phSMAD1/5, white) in P10 Rbpj^iPC brain cortex vasculature (tdT⁺, red). Scale bar, 100 μm. e phSMAD1/5 immunosignal intensity quantitation in P10 brain vasculature. p-values, Welch’s t-test, n = 6. f, g Confocal single optical sections showing localization of phSMAD3 (f) or phSMAD1/5 (g) (white) in the nuclei (DAPI⁺, blue) of ECs (ICAM2⁺, red; yellow arrowheads) or pericytes (GFP⁺, green; white arrowheads) in P10 cortical vessels. Note phSMAD3 presence in ECs and pericytes, while phSMAD1/5 is only detectable in ECs. Scale bar, 25 μm. h Tgfb3 expression in cortical mural cells from P7 or P10 mice analyzed by RNA-seq (top) or RT-qPCR (bottom). p-adjusted values (RNA-seq) or p-values, Student’s t-test, n = 3–4. i RT-qPCR analysis in sorted ECs from P7 and P10 brain cortices. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test (Bmp2 and Bmp4), or one-way ANOVA with Sidak’s (Nrp1), n = 4–7. j Gene expression fold change based on RNA-seq counts in P7 and P10 brain cortex mural cells. p-adjusted values. n = 3. k Thrombospondin-1 (Thbs1) expression analysis in vivo (RNA-seq and RT-qPCR for FACS-sorted mural cells) and in vitro in cultured pericytes. p-adjusted values (RNA-seq), p-values, Welch’s t-test (in vitro), or Brown–Forsythe and Welch one-way ANOVA with Holm–Sidak’s test (RT-qPCR), n = 3–4. l Confocal images showing increased expression of thrombospondin-1 (Thbs1, white) in mural cells (PDGFRβ⁺, green) of P10 Rbpj^iPC cortical vasculature (ICAM2⁺, red). Scale bar, 50 μm. Error bars represent s.e.m. Source data are provided as a Source Data file

Fig. 6

Characterization of the RBPJ-bound genomic landscape in cultured pericytes. a MEME (top) and DREME (bottom) analysis of the RBPJ-bound sites in pericytes identify the RBPJ binding motif amongst the top enriched motifs. b RBPJ binding sites were used for the identification of over-represented functional terms associated with RBPJ-bound genes. Depicted are the results for the most strongly enriched terms of the PantherDB database. The bubble plot encodes the false discovery rate corrected p-value (Binom FdrQ), the number of bound genes (Obs Genes), and the fraction of the total number of genes belonging to the corresponding term (Term Cov). c RBPJ binding in pericytes occurs at both proximal (H3K4me3 positive clusters 1–2) and distal enhancer sites (H3K4me1 positive cluster 3). ChIP-seq was performed against RBPJ, H3K4me1, H3K4me3, and H3 in pericytes and RBPJ binding sites were clustered based on the differential enrichment of H3K4me1, H3K4me3, and nucleosome occupancy as revealed by using a panH3 antibody with k-means. Two replicates (Rep1 and Rep2) of each experiment are shown. d Identification of the genes bound by RBPJ and deregulated upon Rbpj-deletion in pericytes. Significantly up-regulated RBPJ-bound genes identified in P7 and P10 Rbpj^iPC versus respective controls were selected based on adjusted p-value < 0.05 and log₂-transformed fold change > 0.5. Log-2 transformed transcriptional changes are shown as heatmap after hierarchical clustering using Euclidean distance measure and complete linkage. Black arrowheads point to relevant genes involved in increased TGFβ signalling and related ECM composition (Tgfb3, Thbs1, Fbn2, Fn1), cell–ECM crosstalk (Itga8, Itga5, Itga3), inflammation (Ccl2), and fibrosis (Tgfbi, Serpine1, Runx1, Atf3). Source data are provided as a Source Data file

Fig. 7

Molecular similarities with CCMs and impact of Rbpj deletion in stroke. a Confocal images of cerebellar white matter (dashed outline) in P10 brains. Note dilated vessels (CD31⁺, red) and gliosis (GFAP⁺, green) in mutants. Scale bar, 50 μm. b RT-qPCR analysis in P10 sorted brain ECs. p-values, Brown–Forsythe and Welch one-way ANOVA with Tamhane’s T2 test (Krit1, Ccm2, and Pdcd10), and Kruskal–Wallis with Dunn’s test (Klf2 and Kl4), n  = 4. c Confocal images showing Klf4 (white) and active integrin-β1 (Itgβ1(a), green) in blood vessels (GLUT1⁺, red) of P10 Rbpj^iPC cerebellar white matter (dashed outline). Scale bar, 50 μm. d, e Klf4 (d) and active integrin-β1 (e) immunosignal fold change intensity in P10 vasculature. p-values, Student’s t-test, n = 5. f Confocal images showing perivascular PDGFRβ (white) expression after stroke. Note abundance of PDGFRβ⁺ GFP⁻ cells (arrowheads, top panel) and morphological changes in GFP⁺ pericytes (arrows, bottom panel). Scale bar, 50 μm. g Representative T2-weight magnetic resonance imaging (MRI) 1 day after stroke. Scale bar, 2 mm. h MRI-based quantitation of space-occupying effect attributable to oedema (% HSE) and the lesion volume corrected (LVc, mm³) after stroke. FDR-adjusted p-values, one-way ANOVA with two stage linear step-up procedure of Benjamini, Krieger, and Yekutieli multiple comparison correction, n = 4. i Confocal images showing increased phSMAD3 (white; left panel) and infiltration of inflammatory cells (IB4⁺, blue; Iba1⁺, white; right panel) in Rbpj^iPC brains after stroke. GFAP staining (green, right panel) indicates boundary between infarct zone and penumbra (dashed line). Vessels are labelled by tdT (red). Scale bar, 100 μm (left) and 50 μm (right). j Confocal images showing blood vessel (tdT, red; VEGFR2, white) defects (arrowheads) in Rbpj^iPC mice 7 days post-stroke. Scale bar, 100 μm. k Gene expression fold change of pro-inflammatory chemokines in P10 Rbpj^iPC mural cells relative to controls (set as 1). p-adjusted values. n = 3. l RBPJ binding and H3K4me, H3K4me3 distribution at Ccl2 locus in pericytes. Black bar below RBPJ profiles indicates the position of peak intervals. Two replicates (Rep1 and Rep2) are shown. Error bars represent s.e.m. Source data are provided as a Source Data file