Vascular FLRT2 regulates venous-mediated angiogenic expansion and CNS barriergenesis

Abstract

Veins have emerged as the origin of all other endothelial cell subtypes needed to expand vascular networks during developmental and pathological neoangiogenesis. Here, we uncover the role of the angioneurin Fibronectin Leucine Rich Transmembrane protein (FLRT) 2 in central nervous system (CNS) vascular development in the mouse. Early postnatal FLRT2 deletion reveals specific defects in retinal veins, impacting endothelial cell proliferation, sprouting and polarity that result in reduced tip cells at the vascular front. FLRT2 interacts with VE-cadherin and together with the endocytic adaptor protein Numb contribute to the modulation of adherens junction morphology in both retina and cerebral cortex in vivo. Utilizing expansion microscopy, we visualize the altered dynamic distribution of VE-cadherin in tissue of FLRT2 endothelial mutants. Additionally, FLRT2 in cortical vessels regulates the crosstalk between adherens and tight junctions, influencing blood-brain barrier development. Our findings position FLRT2 as a vein-specific regulator of CNS vascular development.

Fig. 1. CNS vascularization is compromised after FLRT2 deletion in ECs.

a Flrt2 mRNA detection by fluorescence in situ hybridization (FISH) in P8 wild-type mouse retina in the optic nerve head (ONH) area, where the central artery and vein enter the retina, and in the superficial vascular plexus (SVP) (upper panel); and in P8 wild-type cerebral cortex (upper cortical layers and pial vasculature) (lower panel). Blood vessels were detected by immunostaining with podocalyxin (Podxl) (general vessel marker). b Coronal section of the cerebral cortex from P6 wild-type mouse stained for FLRT2 and NeuN as neuronal marker (neuronal layers I-VI annotated). Blood vessels were visualized with isolectin-B4 (IB4) staining. Arrows show FLRT2 positive blood vessels (right). c Flat-mounted P7-P8 retinas from control and Flrt2^iΔEC littermate mice injected with 4-hydroxytamoxifen (Tmx) from P1 to P3 and stained with IB4. Quantification of radial vascular length ratio (d), total retinal vessel length (e), and total number of branch points (f) per retina. n = 12 control and 7 mutant (d), 7 control and 6 mutant (e, f) animals. Two-tailed unpaired t-test, p = 0.0004 (d), 0.013 (e), 0.005 (f). g Representative images of P7-P8 control and Flrt2^iΔEC flat-mounted retinas stained for IB4. Veins (V) and arteries (A) are indicated. h Quantification of capillary network density between veins and arteries. n = 8 animals per genotype. Two-tailed unpaired t-test, p = 0.0005. i Glut1 staining of the vasculature in control and Flrt2^iΔEC brain cortices from P7-P8 mice after Tmx administration from P1 to P3. Quantification of vessel density (j), vessel length (k) and number of branch points (l). n = 7 animals per genotype. Two-tailed unpaired t-test, p = 0.028 (j), 0.001 (k), 0.001 (l). Scale bars: 20 μm (a), 50 μm (b), 500 μm (c), 200 μm (g), 100 μm (i). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2. Endothelial FLRT2 regulates specifically venous sprouting and CNS angiogenic expansion.

a Representative images of main arteries and veins (identified by their morphology) from control and Flrt2^iΔEC P7-P8 retinas stained with IB4. Red dots indicate branch points from the corresponding mother vessel. b Quantification of the number of branch points in arteries (left) and veins (right) per vessel length. n = 19 control and 16 mutant animals (arteries); 20 control and 17 mutant animals (veins). Two-tailed Mann-Whitney test, p = 0.333 (arteries); Two-tailed unpaired t-test, p = 0.0001 (veins). c Expanded P7 retina stained for FLRT2 and collagen IV (Col IV) showing FLRT2 expression in retinal vessels. Note the expression of FLRT2 at the EC membrane in the vein but its absence in the artery. d Representative images of retinal vascular front from control and Flrt2^iΔEC mice stained with IB4. Red dots indicate cellular protrusions identified as angiogenic sprouts. e Quantification of number of sprouts per 100 μm of retinal vascular front. n = 22 control and 18 mutant mice. Two-tailed unpaired t-test, p = 0.0002. f Vascular fronts from control and Flrt2^iΔEC P7-P8 retinas showing blood vessels labelled with IB4 and EC nuclei stained for ERG. g Quantification of the number of tip cells per stalk cells at the vascular front. n = 10 control and 8 mutant mice. Two-tailed unpaired t-test, p = 0.002. h Glut1 staining visualizing vessel sprouts in P7-P8 control and Flrt2^iΔEC cerebral cortices. i Quantification of the number of sprouts per vessel density in the cerebral cortex. n = 5 control and 6 mutant mice. Two-tailed unpaired t-test, p < 0.0001. Scale bars: 50 μm (a, f), 90 μm (c), 40 μm (d), 100 μm (h). Data are shown as mean ± SEM. **P < 0.01, ***P < 0.001, ns = not significant.

Fig. 3. FLRT2 controls VE-cadherin recycling at EC junctions.

a Proximity ligation assay (PLA) in HUVEC. Cells were stained with phalloidin (actin filaments) and DAPI (nuclei). b PLA signal quantification as puncta per cell number (DAPI) in the field. n = 20 and 23 pictures per condition from 3 experiments. Two-tailed Mann-Whitney test, p < 0.0001. c Immunoprecipitation (IP) of VE-cadherin and immunodetection of FLRT2 and VE-cadherin from mouse brain lysates. TL, total lysate. d FLRT2 immunoblot from HUVECs treated with Flrt2 and control siRNA. Loading control: β-actin. e FLRT2 protein quantification in control and Flrt2-siRNA treated HUVECs. n = 6 independent experiments. Two-tailed unpaired t-test, p = 0.001. f Antibody feeding assay in HUVEC transfected with control and Flrt2-siRNA treated with chloroquine. Cells immunostained for internalized VE-cadherin, total VE-cadherin and DAPI. Intensity of internalized VE-cadherin shown in arbitrary units (AU, upper panels). g Fluorescence intensity quantification of internalized VE-cadherin per cell. n = 84 control, 115 Flrt2-siRNA cells from 3 experiments. Two-tailed Mann-Whitney test, p < 0.0001. h Quantification of VE-cadherin intensity per cell-junction length. n = 22 control, 18 Flrt2 siRNA-treated pictures from 3 experiments. Two-tailed unpaired t-test, p = 0.029. i VE-cadherin and α-tubulin (loading control) immunoblots in HUVEC transfected with control- and Flrt2-siRNAs. j Quantification of VE-cadherin/loading control and cleaved VE-cadherin/loading control ratios. n = 11 (total), 8 (cleaved) experiments. Two-tailed unpaired t-test, p = 0.074 (total), p = 0.0001 (cleaved). k Primary mouse brain ECs (pmBEC) from control and Flrt2^iΔEC littermates stained for Calpain-2 and DAPI. l Calpain-2 fluorescence intensity quantification per cell. n = 28 and 51 cells per condition, from 1 control and 1 Flrt2^iΔEC littermate. Two-tailed Mann-Whitney test, p = 0.011. m Vessels stained for Calpain-2 and IB4 from control and Flrt2^iΔEC littermates. n Calpain-2 fluorescence intensity quantification in vessels. n = 5 controls, 4 mutants. Two-tailed unpaired t-test, p = 0.002. Scale bars: 40 μm (a, upper panels), 15 μm (a, lower panels), 10 μm (f, k), 5 μm (m). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant.

Fig. 4. FLRT2 interacts with Numb and modulates EC polarization.

a Proximity ligation assay (PLA) in EC cultures. Cells were stained with phalloidin and DAPI to visualize actin filaments and nuclei, respectively. b Expansion microscopy on HUVEC cultures stained for FLRT2 and Numb. c mRNA expression of Flrt2 and Numb in ECs transfected with control and Flrt2-specific siRNAs. n = 4 independent experiments. Two-tailed unpaired t-test, p < 0.0001 (Flrt2), p = 0.031 (Numb). d Immunoblot showing FLRT2 and Numb protein levels in ECs transfected with control and Flrt2-specific siRNAs. e FLRT2 and Numb protein levels quantification in ECs transfected with control and Flrt2-specific siRNAs. n = 7 (control) and 8 (Flrt2-siRNA) independent experiments. Two-tailed unpaired t-test, p < 0.0001 (Flrt2), p = 0.012 (Numb). f mRNA expression quantification of Numb from primary mouse brain ECs (pmBECs) from control and Flrt2^iΔEC littermates. n = 6 control and 10 mutant animals. Two-tailed unpaired t-test, p = 0.0499. g Scheme showing the quantification of Golgi orientation in migrating ECs. Cells were classified as polarized if the angle formed between the scratch or vascular front and Golgi located within 120°. h Scratch assay on HUVECs stained for VE-cadherin, Golgi apparatus (GM130) and cell nuclei (DAPI). Stars indicate cells polarized towards the wound. The 3 first cell rows were considered for quantification. i Quantification of the percentage of cells per image polarized towards the wound. n = 28 control and 27 Flrt2 siRNA images from 3 independent experiments. Two-tailed Mann-Whitney test, p = 0.0001. j Representative images of retinal vascular front from control and Flrt2^iΔEC littermates stained for blood vessels (IB4), EC nuclei (ERG) and Golgi apparatus (GM130). Arrows indicate cellular orientation identified with GM130 position relative to ERG staining. k Quantification of the percentage of cells polarized towards the vascular front. The 3 first cell rows were considered for quantification. n = 4 animals per genotype. Two-tailed Mann-Whitney test, p = 0.029. Scale bars: 20 μm (a, b, h, j). Data are shown as mean ± SEM. *P > 0.05, ***P < 0.001.

Fig. 5. Deletion of FLRT2 affects the morphology of adherens junctions in vivo.

Vascular front of P5 retinas (a) and remodeling capillaries in cerebral cortex (c) from control and Flrt2^iΔEC littermates stained for VE-cadherin antibody. Blood vessels visualized with IB4. Quantification of VE-cadherin activity in retina (b) and cerebral cortex (d) blood vessels. 15 × 15 μm regions of interest (ROIs) were blindly classified to a VE-cadherin activity category: low (smooth pattern), medium (irregular pattern), high (rough pattern). Example images of the three VE-cadherin activity categories in retina (b) and cortex (d) vessels are shown in correlation with the graph. n = 66 control and 56 mutant images from three litters (b), 78 control and 80 Flrt2 mutant images from two litters (d). Two-way ANOVA, p = 0.012 (low), 0.693 (medium), 0.999 (high) for b; 0.023 (low), 0.009 (medium) and 0.854 (high) for d. Scale bars: 20 μm (a), 15 μm (c). Data are shown as median (thick dashed line) and quartiles (thin dashed lines). *P > 0.05, **P < 0.01, ns = not significant.

Fig. 6. High resolution imaging of VE-cadherin junctions in tissue.

a Expansion microscopy 3D-visualization of a large vessel stained for VE-cadherin, FLRT2 and Numb in the cerebral cortex. Higher magnification (lower panels) showing the colocalization of the three proteins at EC junction. Expansion microscopy 3D-visualization of a large vessel stained for VE-cadherin and FLRT2 (b) and VE-cadherin and Numb (c) in the retina. d Expansion microscopy 3D-visualization of cerebral cortex capillaries stained for VE-cadherin and Glut1 in control (left) and Flrt2^iΔEC (right) mice. Higher magnifications (1-5) showing x-y planes (first row) and y-z planes (second row) exposing VE-cadherin pattern in a control capillary (1-2), a control tip cell (3), and a Flrt2^iΔEC capillary (4-5). Scale bars: 4 μm (an upper panel), 1 μm (a lower pannels), 15 μm (b, c), 30 μm (d).

Fig. 7. FLRT2 deficiency disrupts tight junctions and blood-brain barrier.

a Immunoblot of cytosolic and nuclear fractions obtained from bEnd.3 cells treated with control and Flrt2-siRNA showing protein levels of β-catenin and FoxO1, and α-tubulin (cytosolic) and Lamin A/C (nuclear) as controls. β-catenin (b) and FoxO1 (c) protein levels relative to loading controls. n = 3 independent experiments. Two-tailed unpaired t-tests, p = 0.342 (cytoplasm), p = 0.016 (nucleus) (b); p = 0.126 (cytoplasm), p = 0.020 (nucleus) (c). d Immunoblot showing FLRT2 and Claudin-5, and β-actin (loading control) in control and Flrt2 siRNA-treated bEnd.3 cells. e Quantification of FLRT2 and Claudin-5 protein levels. n = 5 (FLRT2) and 6 (Claudin-5) independent experiments. Two-tailed unpaired t-test, p < 0.0001 and p = 0.026. f Immunoblot showing Claudin-5, and α-tubulin as loading control in total brain lysates from control and Flrt2^iΔEC littermates. g Quantification of Claudin-5 protein levels. n = 8 animals per genotype. Two-tailed Mann-Whitney test, p value = 0.034. h Neocortical blood vessels stained for Claudin-5 and VE-cadherin. i Quantification of the ratio of split Claudin-5 junction length to the total junctional length. n = 5 control and 7 mutant mice. Two-tailed unpaired t-test, p = 0.006. j Representative fluorescent whole-brain images of control and Flrt2^iΔEC littermates injected with AlexaFluor^TM 555-conjugated cadaverine at P7-P8. k Cadaverine whole-brain intensity quantification in control and Flrt2^iΔEC littermates. n = 16 control, 18 mutant animals. Two-tailed unpaired t-test, p = 0.018. l Brain cortices from cadaverine-injected control and Flrt2^iΔEC littermates immunostained for Collagen IV (Col IV). m TEM representative images of EC junctions in brain capillaries in control and Flrt2^iΔEC mice. Note the abnormal junctions often associated with the presence of vacuoles (red arrows) in Flrt2^iΔEC vessels. n Incidence of abnormal junctions (left) and junctions with vacuoles (right) in control and Flrt2^iΔEC brain capillaries. n = 3 animals per genotype. Two-tailed unpaired t-test, p = 0.001 (junctions), p = 0.021 (vacuoles). Scale bars: 10 μm (h), 1 mm (j), 100 μm (l), 200 nm (m). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P > 0.001.

Fig. 8. FLRT2 regulates vascular sprouting and CNS barriergenesis by interacting with VE-cadherin.

a Schematic timeline representation of the vascular events derived from FLRT2 deletion in ECs. b Schematic representation of the molecular mechanisms regulated by FLRT2 in postnatal CNS vasculature. In control conditions, FLRT2 forms a complex with VE-cadherin and the endocytic adaptor Numb which allows the dynamic turnover of VE-cadherin necessary for angiogenic sprouting. By contrast, in FLRT2-deficient cells, VE-cadherin cytoplasmic tail is cleaved by Calpains and subsequently fated to lysosomal degradation, while a compensatory biosynthesis accumulates VE-cadherin at the cell membrane impairing vascular sprouts. In addition, FLRT2 deletion facilitates the nuclear translocation of β-catenin, repressing Claudin-5 expression and triggering increased size-selective BBB permeability.