Fig 2: Working model for ENDOA2-mediated VEGFR2 endocytosis. ENDOA2-mediated VEGFR2 endocytosis controls EC migration independently of the CLATHRIN-mediated VEGFR2 endocytosis pathway. SLIT2-ROBO1 targets VEGFR2 to the ENDOA2-dependent internalization pathway via srGAP1, leading to downstream PAK/p38 activation and cell migration

Fig 3: SLIT2 induces ENDOA2-dependent VEGFR2 internalization. a Cell surface biotinylation assay of VEGFR2 internalization in response to SLIT2 (6 nM) in Control (Ctrl), ROBO1/2 and ENDOA2 siRNA silenced HUVECs. VEGFR2, ROBO1, ENDOA2, and ACTIN expression from the total cell lysate are shown as loading controls (input). Surf: surface expression. b Quantification of internalized VEGFR2 normalized to VEGFR2 surface expression before stimulation (N = 7 independent experiments; two-way ANOVA test: *P < 0.05, **P < 0.01, ***P < 0.001, ns P > 0.05). c, d Antibody feeding assay to measure VEGFR2 internalization in response to SLIT2 (6 nM) in Ctrl, ROBO1/2, and ENDOA2 siRNA silenced HUVECs (c) and mLECs from EndoA2+/+ and EndoA2−/− mice (d). e, f Quantification of internalized VEGFR2 fluorescent intensity shown in c and d, respectively (N = 4 independent experiment, at least 103 cells analyzed per experiment; e one-way ANOVA and f Mann–Whitney U test: *P < 0.05). g SIM images of HUVEC lamellipodia stained for ENDOA2 before and after SLIT2 stimulation (6 nM for 2′30″). h SLIT2 effects on EPA formation at the cell migration front (N = 16–17 cells per group analyzed from three independent experiments; t-test: ***P < 0.001). i SLIT2 increases the pixel overlap between VEGFR2 and ENDOA2 fluorescent signals (N = 8–10 cells per group analyzed from three independent experiments; Mann–Whitney U test: ***P < 0.001). j SIM images of HUVEC lamellipodia stained for ENDOA2, VEGFR2, and CLATHRIN heavy chain after SLIT2 stimulation (6 nM for 2′30″). k–m Overlapping pixels between VEGFR2/ENDOA2 (k), VEGFR2/CLATHRIN (l), and ENDOA2/CLATHRIN (m) from the image presented in j are shown in white. Boxed areas are magnified to highlight VEGFR2/ENDOA2, VEGFR2/CLATHRIN, or ENDOA2/CLATHRIN overlap (white). n Quantification of overlap between VEGFR2/ENDOA2 and VEGFR2/CLATHRIN fluorescent signals (N = 8–10 cells per group analyzed from three independent experiments; Mann–Whitney U test: **P < 0.01) (right panel). Error bars represent mean ± s.e.m. Scale bars: c and d 20μm, g 2μm, j 1μm

Fig 4: VEGFR2 pY1173/PLCγ signaling in experimental tumors affects vascular leakage, response to therapy, and antitumor immunity.(A and B) Immunostaining (A) and mean fluorescent intensity (MFI) quantification (B) of pPLCγ Y783 in Vegfr2+/+ (WT) and Vegfr2Y1173F/+ (Y1173F/+) B16F10 melanoma tumors; n = 5 (WT) and 3 (Y1173F/+) mice, ≥ 3 fields of view/experiment. Scale bar: 100 μm. Unpaired 2-tailed Students’ t test. (C) Tumor growth of WT and Y1173F/+ B16F10 tumors; n = 5 (WT) and 3 (Y1173F/+) mice. 1-way ANOVA. (D) Quantification of vessel density (MFI) from A; n = 5 (WT) and 3 (Y1173F/+) mice, ≥ 3 fields of view/tumor. Unpaired 2-tailed Students’ t test. (E and F) Representative image (E) and quantification (F) of extravasated fixable 2,000 kDa FITC-dextran in B16F10 tumors; n = 5 (WT) and 3 (Y1173F/+) mice, ≥ 3 fields of view/experiment. Scale bar:100 μm. Unpaired 2-tailed Students’ t test. (G) Fluorescent intensity of 70 kDa TRITC-dextran extracted from B16F10 tumors; n = 5 (WT) and 3 (Y1173F/+) mice. Unpaired 2-tailed Students’ t test. (H and I) Tumor growth (H) and tumor weights (I) of WT and Y1173F/+ B16F10 tumors treated with 5 mg/kg Temozolomide (TMZ) or DMSO control; n = 9–10 mice/group. 2-way ANOVA (H), 1-way ANOVA (I). (J–M) Percent B cells (CD19) (J), helper T cells (CD3 and CD4 costaining) (K), regulatory T cells (CD3, CD4, CD25, and Foxp3 costaining) (L), cytotoxic T cells (CD3 and CD8 costaining) (M) in WT and Y1173F/+ B16F10 tumors; n = 16/genotype. Unpaired 2-tailed Students’ t test. (N) mRNA expression of TGFβ (Tgfb1-3), IL10 (Il10), IL35 (Il12a, Ebi3), IFNγ (Ifng), and TNFα (Tnf) in WT and Y1173F/+ B16F10 tumors; n = 5 mice/genotype. Unpaired 2-tailed Students’ t test. (O) Tumor weight of CD40- or IgG-treated WT and Y1173F/+ B16F10 tumors harvested at day 12; n = 9–10 mice/group. 1-way ANOVA. Data represent mean ± SD or SEM (H). *P < 0.05, **P < 0.01, ***P < 0.001.

Fig 5: VEGFR2 pY1173 signaling regulates VEGFA-induced vascular permeability in the healthy skin.(A) Representative images of immunofluorescent staining for VE-cadherin (VEC) and pPLCγ Y783 in the back skin of Vegfr2+/+ WT and Vegfr2Y1173F/+ heterozygous (Y1173F/+) mice, subsequent to intradermal injection of PBS or VEGFA. Scale bar: 20 μm. (B) Quantification of mean fluorescent intensity of vascular pPLCγ Y783 from A; n = 5 mice/genotype, ≥ 2 fields of view/mouse. 1-way ANOVA. (C) Representative images of Evans blue leakage in response to PBS, VEGFA, bradykinin, or histamine in the back skin of WT and Y1173F/+ mice. (D) Quantification (620 nm absorbance) of extravasated Evans blue from C. Values are shown as fold of PBS-treated control, normalized to tissue weight; n = 3–5 mice/group. 1-way ANOVA. (E) Representative time-lapse image of VEGFA-induced vascular permeability of 2,000 kDa FITC dextran in the ear dermis of WT and Y1173F/+ mice. Scale bar: 50 μm. (F and G) Number of leakage sites/100 μm in WT and Vegfr2Y1173F/+ mice, in veins (F) and capillaries (G); n = 3 mice/genotype, ≥ 2 fields of view/mouse. Unpaired 2-tailed Students’ t test. (H and I) Ear dermis leakage in response to VEGFA in tamoxifen-treated Vegfr2fl/fl; Cdh5-Cre– (fl/fl), Vegfr2Y1173F/fl; Cdh5-Cre– (Y1173F/fl), Vegfr2Y1173F/fl; Cdh5-Cre+ (Y1173F/–), Vegfr+/fl; Cdh5-Cre+ (+/–), Vegfr2fl/fl; Cdh5-Cre+ (–/–), quantified as leakage sites/100 μm, in veins (H) and capillaries (I); n = 3–5 mice/genotype, ≥ 2 fields of view/mouse. 1-way ANOVA. Data represent mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.