Surface accumulation of FGFR in VHL knockdown HMVECs. A, HMVECs transfected with plasmids expressing either control or VHL shRNA (shVHL) were incubated in complete serum with growth factor supplements and processed for immunostaining with anti–p-FGFR antibody. High levels of surface accumulation of p-FGFR (arrows) are observed in VHL knockdown HMVECs (shVHL) but not in control cells. B, control and VHL knockdown HMVECs (shVHL) are labeled with membrane-impermeable sulfo-NHS-LC biotin at 4°C. Samples were immunoprecipitated using NeutrAvidin biotin-binding protein and surface receptor levels detected using indicated antibodies. Total receptor levels were detected from total input lysates (1/12 of the amount used in pull-down) using indicated antibodies. C, ligand-stimulated endocytosis of activated FGFR was synchronized and followed over time. Serum-starved cells were incubated with bFGF + heparin at 4°C for 2 h and then chased with ligand-free media at 37°C over a 40-min period. Arrows denote p-FGFR visible at or near the cell surface at 5 min. Arrowheads in control inset (20 min) denote p-FGFR that is internalized into intracellular vesicles. Arrows in VHL knockdown (20 and 40 min) denote p-FGFR appearing in large aggregates at the tips of cellular projections. D, ligand-stimulated endocytosis of activated FGFR was synchronized and followed for 20 min in control HMVECs. Triplicate wells were double-stained for p-FGFR/LAMP1, p-FGFR/VHL, and LAMP1/VHL, as indicated. Top left, p-FGFR (red) and LAMP1 (green) colocalize in some vesicles (yellow block arrows), whereas others contain only p-FGFR (arrowheads). Top right, VHL (green) and p-FGFR (red) colocalize in some vesicles (yellow block arrows) but not others (arrowheads). Bottom, VHL (green) does not overlap with LAMP1 (red).

Vhl-heterozygous animals show increased neoangiogenic response to bFGF. Vhl^+/− mice or the wild-type littermates were injected s.c. with Matrigel with or without embedded bFGF. Seven days after gel implantation, FITC-dextran was infused by intracardiac injection. The gel was dissected 10 min later. Compared with wild-type, bFGF-loaded Matrigel plug in Vhl^+/− induces more abundant neovessels with tortuous and chaotic patterns. Arrows point to examples of bulges typically found in bFGF-induced Vhl^+/− vessels. Scale bars, 200 µm. Quantification of the area covered by vessels was performed by measuring the fluorescent area of each picture containing vessels using NIH ImageJ software. The angiogenic response to bFGF is more robust in Vhl^+/− mice (n = 6) than in the wild-type counterpart (n = 4). Statistical analysis was performed using Student’s t test. *, P ≤ 0.05; **, P ≤ 0.01.

ETS1 transcription factor mediates angiogenic activity of VHL knockdown HMVECs. A, control or VHL knockdown (shVHL) lysates (1 mg) stimulated with (lanes 2 and 4) or withou t (lanes 1 and 3) bFGF + heparin are incubated with ETS1 antibody and pull-down performed with protein G–sepharose beads. Pulled-down fractions are probed with phosphothreonine (p-Thr) and reprobed with phosphoserine (p-Ser) antibodies. Total lysates (100 µg) from control or VHL knockdown HMVECs stimulated with (lanes 6 and 8) or without (lanes 5 and 7) bFGF are probed with ETS1 antibody and reprobed with β-actin for loading control. B, left, treatment of VHL knockdown HMVECs (shVHL) with 10 MOI of ETS1-specific adeno shRNA (+) reduces ETS1 expression compared with treatment with control (−) virus. Right, transwell migration assay for VHL knockdown HMVECs treated with 10 MOI of control virus (shVHL + control) or 10 MOI of ETS1-specific adeno siRNA (shVHL + shETS1). Statistical analysis was performed using Student’s t test (n = 4). C, knockdown of ETS1 expression by 10 MOI ETS1-specific adeno siRNA significantly reduces cord formation by VHL knockdown HMVECs but has no effect on control cells. D, top left, HIF-1α expression in VHL wild-type (control shRNA), VHL single knockdown, and VHL-HIF-1α double knockdown cells. Bottom left, cord formation assay comparing VHL single knockdown (shVHL + control) and VHL-HIF-1α double knockdown (shVHL + shHIF-1α). Right, quantification of cord formation assay shown on the left. Statistical analysis was performed using Student’s t test (n = 4). *, P ≤ 0.05; ***, P ≤ 0.001.

Knockdown of VHL increases cell motility in HMVECs. A, left, HMVECs were transfected with vectors expressing control shRNA or VHL shRNA constructs (shVHL). Cell lysates were Western blotted using VHL antibody or β-actin antibody as a loading control. The apparent molecular weight of the detected pVHL species (~ 30 kDa) is indicated. Middle, 90% confluent cell monolayers in six-well culture dishes were scratched with a 1,000-µL pipette tip. Wound healing was followed in control and VHL knockdown HMVECs for 24 h. Right, extent of wound healing was quantified in control and VHL knockdown HMVECs by measuring percentage of wound remaining over time using NIH ImageJ software (n = 3). B, quantification of transwell assays for intrinsic (low serum) and chemotactic (+FGF and +VEGF) cell migration over 24 h for control and VHL knockdown HMVECs (n = 5). C, schematic representation of the three-dimensional coculture system. Cord formation by the HMVECs is supported by stromal cells, such as human dermal fibroblasts, and can be observed within 72 h by confocal microscopy usually at a depth of ~ 100 µm from the bottom of the well. D, left, images taken inside collagen gel of control and VHL knockdown HMVECs cocultured with high (3.75 × 10⁴ per chamber) or low (1.5 × 10⁴ per chamber) amount of fibroblasts (FB). Right, Cord formation by control and VHL knockdown HMVECs cocultured with high (first and second columns) or low (third and fourth columns) amount of fibroblasts was quantified using NIH ImageJ software to measure total cord length per field. Assay was performed in triplicate chambers for each condition. Random fields containing cord formation were chosen for quantification from each condition. Statistical analysis was performed using Student’s t test (n = 9 for high fibroblast condition, n = 3 for low fibroblast condition). *, P ≤ 0.05; **, P ≤ 0.01.

ERK1/2 pathway mediates cord formation by VHL knockdown HMVECs. A, top left, transfection of FGFR2 siRNA (siFGFR2) reduces expression of FGFR2 protein in both control and VHL knockdown HMVECs by ~ 3-fold. Bottom left and right, knockdown of FGFR2 (siFGFR2) expression significantly reduces cord formation by VHL knockdown HMVECs (shVHL) placed in three-dimensional coculture assay under high fibroblast condition, but has little effect on the control cells. B, kinetics of ERK1/2 activation in control and VHL (shVHL) knockdown cells stimulated with bFGF and heparin and chased for 5, 20, 60, or 120 min (post-FGF) as described in Fig. 2C. ERK1/2 activation was measured by the levels of dp-ERK. Total ERK1/2 and β-actin were used as loading controls. C, VHL knockdown HMVECs (shVHL) were incubated with 25 µmol/L PD98059 (+) or DMSO (−) for 4 h and then treated with bFGF and heparin followed by ligand-free chase as described in Fig. 2C. Lysates were collected at 30 min after chase. ERK1/2 activation is ameliorated by PD98059 as measured by levels of dp-ERK. D, treatment of VHL knockdown HMVECs (shVHL) with 25 µmol/L PD98059 significantly reduces cord formation in three-dimensional coculture assay in the presence of high amount of fibroblasts. Statistical analysis was performed using Student’s t test (n = 4). **, P ≤ 0.01; ***, P ≤ 0.001.

Increased level of cord formation by Vhl^+/− mouse endothelial cells. Isolated mouse kidney cortical endothelial cells from Vhl^+/− and the wild-type littermates were assayed in the three-dimensional coculture system as described in Fig. 1. A, Western analysis of the cell lysates shows a >50% decrease of pVHL level in the Vhl^+/− cells compared with the wild type. B, the endothelial cells were preincubated with live dye CM-Dil. Images were taken at ~ 100 to 150 µm from the bottom of the well (Gel) as well as at the bottom. Vhl^+/− cells show high levels of cord formation with few remaining at the bottom, whereas wild-type cells mostly remain at the bottom. Scale bars, 200 µm. The angiogenic activity was measured by unit area covered by networks of Vhl^+/− (n = 6) and wild-type (n = 6) mouse kidney endothelial cells. Statistical analysis was performed using Student’s t test. **, P ≤ 0.01.