Fig 2: VEGF-induced shape change is accompanied by dephosphorylation of myosin II light chain at the junction. a LSM of VE-cadherin and VEGFR2 in HUVECs treated with 50 ng ml−1 VEGF or PBS. VEGF-induced invaginations (arrows) were mostly found in cells with high VEGFR2 intensity (stars) but not in less VEGFR2-expressing cells (crosses). b SIM of VE-cadherin- and Phalloidin-TRITC-labelled HUVECs after VEGF demonstrates that VE-cadherin invaginations largely co-localise with actin filaments (arrows). VE-cadherin and actin at cell junctions (arrowheads). c LSM of phospho (ser19)-myosin II light chain (P-ser-MII-LC) and VE-cadherin in P6 mice retina. P-ser-MII-LC was largely seen in the junction of centre vein but not in front ECs. d HUVECs treated with 50 ng ml−1 VEGF and 10 µM Y27632, respectively, were immune labelled with P-ser-MII-LC and Phalloidin-TRITC. VEGF induced a partial loss of P-ser-MII-LC at junctions accompanied by a partial loss of JAAF (arrowhead) and wave-like junction remodelling. Stress fibres (open arrowheads) were P-ser-MII-LC positive. Arrows indicates intact JAAF. Y27632 treatment caused a total loss of P-ser-MII-LC and also induced wave-like junction remodelling. Quantification of e number (n = 411 VE-cadherin invagination for 305 cells for PBS, 815 VE-cadherin invaginations for 338 cells for VEGF) and f length of VE-cadherin invaginations in VEGF- (n = 98 VE-cadherin invaginations) and PBS-treated (n = 92 VE-cadherin invagination) cells. Quantification of (g) VE-cadherin-positive invaginations (n = 422 and 280 VE-cadherin invaginations for 143 Y27632 and DMSO treated cells respectively), (h) invagination length (n = 227 and 232 VE-cadherin invaginations for Y27632 and DMSO separately), (i) cell perimeter, and j Rel-VEcad-C in Y27632-treated cells for 1 h or DMSO based on VE-cadherin immune labelling (n = 143 cells); (compare Supplementary Fig. 5b). Unpaired student’s t test was used for (e–j). k, l HUVECs pre-treated with 10 µM Y27632 for 1 h were applied with VEGF for another 18 h. Quantification of (k) migration velocity (n = 100 cells) and (i) elongation factor (87, 97, 87 and 87 cells were used for Y27632+VEGF, Y27632, VEGF, and control, respectively) using Fiji (compare Supplementary Fig. 5c); one-way ANOVA. Y = Y27632. Error bars indicate ± SEM. Scale bars represent 30 µm in a, c, d and 5 µm in b

Fig 3: VEGFR2 and VE-cadherin expression control VEGF-induced EC elongation and migration. Confluent HUVECs were treated with 50 ng ml−1 VEGF or PBS. a Time-lapse images show cell elongation (arrows) in a subpopulation of HUVECs (Scale bar: 30 µm). b PhaCoB Cell Tracker analysis of cell elongation 5 h after VEGF (n = 3070 cells, one-way ANOVA). c VEGF-dependent cell velocity in the first 6 h (n = 71 cells) and after 24 h (n = 77 cells). d Elongated cells displayed increased 'Euclidean distance' (ED) and 'Accumulated distance' (AD) comparison to polygonal cells in the same VEGF-treated culture (n = 30 cells from t = 24–30 h, two-way ANOVA). e Anti-VE-cadherin-labelled HUVEC cultures. Rel-VEcad-C was decreased in a subpopulation of elongated cells (arrow) but not in polygonal cells (arrowhead) (scale bar: 20 µm). Quantification of (f) cell perimeter and (g) Rel-VEcad-C based on immunofluorescent labelling using the CBT. h HUVECs were stained with VE-cadherin (green) and VEGFR2 (red) (scale bar: 20 µm). i Quantitative distribution of the relative VEGFR2 intensity under PBS (n = 59 cells). j Correlation of relative VEGFR2 intensity with elongation factor in VEGF-treated HUVEC; n = 146 cells; R = correlation coefficient. k Western blot demonstrates downregulation of VEGFR2 and Nrp1 by the respective siRNAs but not non-targeting siRNA (siNt). l VEGFR2 downregulation totally blocked VEGF-induced cell elongation as determined by PhaCoB Cell Tracker, whereas Nrp1 downregulation had only a minor effect (n ≈ 3500/time point), compare Supplementary Movie 6. m A small but statistically significant decrease in the perimeter in Nrp1-downregulated cells quantified based on VE-cadherin-labelling using CBT (n = 100 cells). n (left) VE-cadherin-EGFP overexpression blocked the VEGF-induced elongation as analysed by PhaCoB Cell Tracker (n ≈ 1100 cells/time point). (right) Determination migration velocity of VE-cadherin-EGFP (n = 100 cells) and EGFP (n = 101 cells) overexpressing HUVEC in the presence of VEGF for 24 h. Compare supplementary movie 7. o (left) Western blot of VEGFR2 expression after VE-cadherin-EGFP overexpression. (right) Quantification of VEGFR2 expression in VE-cadherin overexpressing cells (n = 3). α-tubulin served as an internal loading control. Unpaired student’s t test was used to analyse the significance difference for experiment c, f, g, m, n, o. Error bars represent ± SEM. Representative results from three independent experiments are shown

Fig 4: Distinct VEGFR2 pools are present in endothelial cells.(A) Endothelial cells stained for alpha tubulin (red) and total VEGFR2 (green). (B) Inset from panel A shows cell in more detail. Two insets (B1) and (B2) from panel B separate channels in red and green to more clearly show the distinct pools of VEGFR2 present in endothelial cells. This includes a pool of VEGFR2 on the cell surface (arrowhead), a granular cytoplasmic VEGFR2 pool (black arrow) and a pool of VEGFR2 arranged in fibrous arrays (white arrow). (C) Endothelial cells stained for microfilament beta actin (red) and total VEGFR2 (green). (D) Inset from panel C shows cell in more detail. The insets (D1) and (D2) from panel D separate channels into red and green to demonstrate more clearly distinct pools of VEGFR2 present in endothelial cells. This includes a pool of VEGFR2 on the cell surface (arrowhead, green), a granular cytoplasmic VEGFR2 pool (black arrow) and a pool of VEGFR2 arranged in fibrous arrays (white arrow). (E) Cell stained for intermediate filament Vimentin (red) and total VEGFR2 (green). (F) Inset from panel E shows cell in more detail. Insets (F1) and (F2) taken from panel F separate channels in green and red to show more clearly the distinct pools of VEGFR2 present in endothelial cells. This includes a granular cytoplasmic VEGFR2 pool (black arrow) and a pool of VEGFR2 arranged in fibrous arrays (white arrow). All displayed images are single focal planes.

Fig 5: Changes in VEGFR2 subcellular localization as a result of cytoskeletal disruption.(A-F’) Endothelial cells treated for 3 hours with microtubule inhibitor Nocodazole stained with (A) alpha tubulin, (B) total VEGFR2 and (C) merged image. (C’) High power inset shows that microtubule disruption results in loss of fibrous cytoskeletal pattern of staining. VEGFR2 stain instead appears in large inclusions (arrows). (D-F’) Endothelial cells stained with (D) beta actin, (E) total VEGFR2 and (F) merged image. (F’) High power inset shows that VEGFR2 stain does not colocalize with intact actin fibers. (G-L’) Endothelial cells treated for 3 hours with actin polymerization inhibitor Cytochalasin D (CCD) and stained with (G) alpha tubulin, (H) total VEGFR2 and (I) merged image. (I’) High power image demonstrates that total VEGFR2 remains arranged in a fibrous pattern in the absence of intact actin filaments (arrow). (J) Endothelial cells stained with beta actin, (K) total VEGFR2 and (L) merged image. (L’) High power image shows that CCD treatment results in accumulation of beta actin at the periphery of processes and that total VEGFR2 remains organized in a fibrous cytoskeletal pattern despite the disruption of actin filaments.