Notch signaling regulates UNC5B to suppress endothelial proliferation, migration, junction activity, and retinal plexus branching

Abstract

Notch signaling guides vascular development and function by regulating diverse endothelial cell behaviors, including migration, proliferation, vascular density, endothelial junctions, and polarization in response to flow. Notch proteins form transcriptional activation complexes that regulate endothelial gene expression, but few of the downstream effectors that enable these phenotypic changes have been characterized in endothelial cells, limiting our understanding of vascular Notch activities. Using an unbiased screen of translated mRNA rapidly regulated by Notch signaling, we identified novel in vivo targets of Notch signaling in neonatal mouse brain endothelium, including UNC5B, a member of the netrin family of angiogenic-regulatory receptors. Endothelial Notch signaling rapidly upregulates UNC5B in multiple endothelial cell types. Loss or gain of UNC5B recapitulated specific Notch-regulated phenotypes. UNC5B expression inhibited endothelial migration and proliferation and was required for stabilization of endothelial junctions in response to shear stress. Loss of UNC5B partially or wholly blocked the ability of Notch activation to regulate these endothelial cell behaviors. In the developing mouse retina, endothelial-specific loss of UNC5B led to excessive vascularization, including increased vascular outgrowth, density, and branchpoint count. These data indicate that Notch signaling upregulates UNC5B as an effector protein to control specific endothelial cell behaviors and inhibit angiogenic growth.

Keywords: Cell–cell adhesion; Endothelial migration; Endothelial proliferation; Notch effectors; Retinal angiogenesis; UNC5B.

Figure 1

Characterization of rapidly regulated endothelial Notch targets in the developing mouse brain. (A) Diagram of the tamoxifen administration timeline and RiboTag^EC mouse harvest at postnatal day (P) 8. (B) Validation of RiboTag recombination efficiency and specificity in brain endothelium. P8 RiboTag^EC forebrain sections stained for vasculature with Isolectin B4 (IB4, green) and tagged ribosomes (anti-HA, red). Scale bar, 115 μm. (C) Time course of suppression of previously characterized canonical Notch target genes Hey1, Hey2, Hes1, and Nrarp after GSI treatment. (D) Diagram of the experimental workflow of RiboTag-based isolation of immunoprecipitated (IP) endothelial ribosomes and bulk brain homogenate for RNA extraction and analysis. (E,F) Heatmaps of RNAseq analysis of IP compared to forebrain homogenate. The IP fraction is highly enriched in endothelial markers (E) and depleted of neural markers (F). (G) Heatmap of the 20 most significantly downregulated genes in the GSI-treated brain endothelium. Known Notch targets Hey1 and Dll4 (blue arrows) and novel target Unc5B (red arrow) were significantly suppressed in GSI-treated animals. (H) GSEA analysis of pathways enriched in the IP endothelial fraction. All heatmaps indicate a z-score.

Figure 2

UNC5B is a novel target of Notch in endothelial cells. (A,B) Induction of Notch signaling in Human Umbilical Vein Endothelial Cells (HUVEC) using DLL4-Fc coated TLA plates significantly upregulate expression of Notch target genes HES1, HEY1, NRARP, and DLL4 (blue bars). This effect is blocked by GSI treatment with Compound E (CpE, pink bars). UNC5B expression is significantly upregulated by Notch signaling and this upregulation is significantly blocked by CpE (right). Expression levels were evaluated by qPCR. (C–E) EGTA induction (black lines) and CPE inhibition of EGTA (red lines) of canonical Notch target HEY1 (C), UNC5B (D), and GAPDH (E) in HUVECs. (F,G) HUVECs were lentivirally transduced with control RFP, ICN1, or ICN4 expression constructs. 24 h after lentivirus infection, cells were harvested and analyzed by qPCR for NOTCH1, NOTCH4, and UNC5B expression. Two-way ANOVA (A,B), multiple unpaired t-tests (C–E) and one-way ANOVA (F,G), presented as mean ± s.e.m. from at least 3 different biological replicates per experiment.

Figure 3

UNC5B regulates endothelial cell proliferation and migration downstream of Notch signaling. HUVECs were lentivirally transduced with scramble control (shCNT) or shRNA targeting UNC5B (shUNC5B). (A) qPCR for UNC5B expression 24 h after lentivirus infection. (B) Representative images of shCNT and shUNC5B HUVEC cell morphology. Yellow boxed inserts show zoomed images for cellular morphology. (C) shCNT and shUNC5B HUVEC proliferation measured by MTT assay. (D) Percent closure of scratch wounds in shCNT and shUNC5B HUVEC monolayers. Asterisks indicate the time points for significant differences in migration between shCNT and shUNC5B. (E) Proliferation of PMVECs transduced with ICN1 or ICN4 expression vectors and shCNT or shUNC5B, relative to shCNT-RFP control. (F) Cell migration values of PMVECs transduced with ICN1 or ICN4 expression vectors and shCNT or shUNC5B, relative to shCNT-RFP control. Unpaired t-tests (A,C) and multiple comparison unpaired t-tests (D–F), presented as mean ± s.e.m. Each dot represents an independent experiment with 4 replicates per experiment. Scale bars, 150 μm and 35 μm (zoomed).

Figure 4

UNC5B regulates endothelial cell morphology, alignment, and junctional VE-cadherin under static and flow conditions. (A) Representative immunofluorescence images of shCNT and shUNC5B-treated HUVECs stained with anti-VE-cadherin (VE-cad, green) and DAPI (white). (B) Quantification of VE-cadherin area and fluorescent intensity at endothelial cell–cell junctions and endothelial cytoplasm under static conditions. (C) Representative images of shCNT and shUNC5B-treated HUVECs under high laminar flow (20 dyn/cm²). (D) Quantifying changes in VE-cadherin area at endothelial cell–cell junctions and endothelial cytoplasm under flow conditions. At least 60 representative fields of view were evaluated from three biological replicates for the graphs in B and D. (E) A representative schematic of the scoring for cell orientation angle. Direction of flow (red arrow) and cell axis (green arrow) are indicated. (F) Quantification of cell orientation angle in shCNT and shUNC5B cells under laminar flow. A total of 120 cells were evaluated per condition across three biological replicates. Unpaired t-tests, presented as mean ± s.e.m. Scale bars, 50 μm (A,C) and 33 μm (E).

Figure 5

Endothelial cell-specific Notch activation increases Unc5B expression. (A) Representative postnatal day (P) 5 retina distinguishing the immature vascular plexus of the angiogenic front region from the central region, which has increasingly mature arteries, capillaries, and veins. (B) P5 retina stained with anti-Unc5B (red) and Isolectin B4 (IB4, green) to label the blood vessels at the angiogenic front. (C) P5 retina stained with anti-Unc5B and VE-cadherin (VE-cad, green) to label the maturing vessels. A = artery and V = vein in all panels. (D) Diagram of tamoxifen administration to ICN1^IOE-EC mice and harvest at P5 for analysis. (E) Detection of ICN1 in endothelial nuclei of ICN1^IOE-EC retina with anti-GFP antibody. (F) ICN1^IOE-EC mutant and control mouse retinas stained for Unc5B (red) and vasculature (VE-cad, green; IB4, blue). (F’) Isolated Unc5B channel depicted in grayscale. (G) Quantification of the percentage of Unc5B fluorescence intensity in arteries, capillaries and veins. At least 17 representative fields of view were examined and averaged from five distinct control mice and six distinct ICN1^OE-EC mice. Multiple comparisons unpaired t-test, presented as mean ± s.e.m. Scale bars, 230 μm (A), 100 μm (B–E), and 100 μm (F).

Figure 6

Endothelial Unc5B suppresses angiogenesis in the retina. (A) Diagram of tamoxifen administration to Cdh5-CreER^T2; Unc5B^flox mice and harvest at postnatal day (P) 5 for analysis. (B) Whole-mount P5 retinas from Unc5B^+/+, Unc5B^fl/+, and Unc5B^fl/fl mice stained for the vasculature using Isolectin B4 (IB4, gray). Green arrows illustrate the degree of vascular outgrowth in Unc5B^fl/+, and Unc5B^fl/fl retinas, while blue arrows represent the average vascular outgrowth of control retinas for comparison. Red boxes indicate magnified regions of the vasculature in B’. (C–G) Quantification of vascular outgrowth (C), vascular density (D), branch count per area (E), branchpoint density (F), and tip cell density (G) normalized to Unc5B^flox/+. Each data point represents the average of 4–8 measurements on a single animal. One-way ANOVA, presented as mean ± s.e.m. from at minimum three distinct animals per genotype. Scale bars, 300 μm (A) and 100 μm (A’).