Glycocalyx regulation of vascular endothelial growth factor receptor 2 activity

Abstract

We have previously shown that knockdown of endomucin (EMCN), an integral membrane glycocalyx glycoprotein, prevents VEGF-induced proliferation, migration, and tube formation in vitro and angiogenesis in vivo. In the endothelium, VEGF mediates most of its angiogenic effects through VEGF receptor 2 (VEGFR2). To understand the role of EMCN, we examined the effect of EMCN depletion on VEGFR2 endocytosis and activation. Results showed that although VEGF stimulation promoted VEGFR2 internalization in control endothelial cells (ECs), loss of EMCN prevented VEGFR2 endocytosis. Cell surface analysis revealed a decrease in VEGFR2 following VEGF stimulation in control but not siRNA directed against EMCN–transfected ECs. EMCN depletion resulted in heightened phosphorylation following VEGF stimulation with an increase in total VEGFR2 protein. These results indicate that EMCN modulates VEGFR2 endocytosis and activity and point to EMCN as a potential therapeutic target.—LeBlanc, M. E., Saez-Torres, K. L., Cano, I., Hu, Z., Saint-Geniez, M., Ng, Y.-S., D’Amore, P. A. Glycocalyx regulation of vascular endothelial growth factor receptor 2 activity.

Angiogenesis is a complex process by which vessels develop, remodel, and expand through endothelial proliferation, sprouting, migration, and lumen formation. Angiogenesis in the adult is seen under normal conditions (i.e., wound healing and reproductive processes, such as corpus luteum and placenta formation) and pathogenic circumstances (i.e., tumor vascularization and ocular diseases) (1). Although many factors, such as fibroblast growth factor, angiopoietins, TGF-α and -β, hepatocyte growth factor, connective tissue growth factor, and VEGF, participate in angiogenesis, because of its critical functions in vascular development and physiopathology, VEGF is one of the most well studied.

A variety of pathologies, including ocular diseases, involve new blood vessel growth as a complication. Moreover, many ocular diseases are VEGF dependent, including wet age-related macular degeneration, diabetic macular edema, and retinopathy of prematurity. Accordingly, VEGF has emerged as an effective therapeutic target. Anti-VEGF therapies have revolutionized the treatment of wet macular degeneration and diabetic macular edema. However, nonvascular cells, including neural and epithelial cells, express VEGF receptors and respond to VEGF (2). Therefore, identifying a therapy that would interfere with VEGF–VEGF receptor 2 (VEGFR2) signaling specifically in the endothelium would represent a significant advance by limiting off-target adverse effects.

In the vasculature, VEGF interacts with the high-affinity tyrosine kinase receptors VEGF receptor 1 (Flt-1) and VEGFR2 (Flk-1/KDR) expressed on the surface of endothelial cells (ECs). VEGF mediates its EC proliferative, mitogenic, tube formation, permeability, and angiogenic activity primarily through activation of VEGFR2, which in turn interacts with several cell surface coreceptors, including neuropilin (NRP) 1 and 2, heparan sulfate proteoglycans, and endoglin, a TGF-β coreceptor (3, 4). Following VEGF stimulation, VEGFR2 undergoes homodimerization and autophosphorylation. VEGFR2 dimers are phosphorylated on multiple tyrosine residues (Y951, Y1175, Y1059, and Y1214) promoting the activation of divergent downstream kinases and a range of biologic responses (5) including EC survival, proliferation, permeability, and tube formation (6).

Endocytosis has emerged as a key step in modulating receptor signaling duration and amplitude (3). Although membrane receptor internalization was historically considered to be a means of terminating receptor phosphorylation and dampening its signaling, the importance of endocytosis for signaling has been demonstrated through the study of a variety of cell surface receptors, including TGF-βR, epidermal growth factor receptor, fibroblast growth factor receptor, nerve growth factor receptor, and VEGFR2 (7–11). The fate of internalized receptors is regulated by a number of factors, such as ligand concentration and interactions with coreceptors as well as intracellular adaptor proteins (7, 8, 12). Although recent reports have shown that internalized VEGFR2 continues to signal (13), the effects of VEGFR2 internalization on receptor activation and downstream functional activity are more controversial. Depletion of epsin1 and 2 results in loss of clathrin-mediated VEGFR2 internalization with increased VEGFR2 phosphorylation as well as heightened in vivo angiogenesis and enhanced in vitro EC tube formation and migration upon VEGF stimulation (14). In contrast, depletion of CDC42 results in loss of VEGFR2-internalization via micropinocytosis internalization pathways, dampened activation of phosphorylated (p) ERK1 and 2 and p–protein kinase B, and decreased VEGF-induced EC sprouting and survival. Furthermore, prevention of VEGFR2 internalization blocks VEGF-induced angiogenesis in vitro (7). In addition, loss of VEGFR2 internalization through inhibition of small GTPase ADP-ribosylation factor 6 (ARF6) protects against retinal vasopermeability in a streptozotocin-induced model of diabetic retinopathy (12). These data indicate that the biologic effects resulting from loss of VEGFR2 internalization are regulated through VEGFR2 interactions with various coreceptors or adaptor proteins.

Endomucin (EMCN) is a type-I integral membrane O-sialoglycoprotein with a serine- and threonine-rich extracellular domain (15–17) and was recently identified by our lab as a novel regulator of VEGF-induced angiogenesis. EMCN knockdown in vitro reduced VEGF-induced migration, proliferation, and tube formation of human retinal microvascular ECs (HRECs), whereas EMCN overexpression enhanced these effects. In addition, EMCN knockdown using small interfering RNA (siRNA) in vivo impaired vascularization of the developing mouse retina (18). Although these data suggest a role for EMCN in modulating VEGF-induced signaling, the mechanism of its involvement was not clear. This study defines the molecular basis for EMCN’s regulation of VEGFR2 and has revealed a role for EMCN in receptor endocytosis. Moreover, our results suggest that targeting VEGFR2 endocytosis may represent a novel therapeutic target to modulate VEGF-driven pathologies.

MATERIALS AND METHODS

Reagents and antibodies

Reagents

Nontargeting control siRNA (siCtrl, D-001810-01-05) and siRNA directed against EMCN (siEMCN, L-015860-01-0005) were purchased as Smartpools (Dharmacon, Lafayette, CO, USA). Dharmafect 1 transfection reagent (T-2001-02; Dharmacon) was used for cell culture studies. VEGF165 (293-VE-010) was purchased from R&D Systems (Minneapolis, MN, USA). Primaquine bisphosphate (PQB, 160393-1G) and Tween 20 (X251-07) were purchased from MilliporeSigma (Burlington, MA, USA). Paraformaldehyde (4% PFA, AAJ61899-AP) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Sulfo-NHS-SS-Biotin (21331), 3,3′-dithiobis(sulfosuccinimidyl propionate) (21578), avidin agarose (S1258122), monomeric avidin agarose (20228), d-Biotin (29129), 70 kDa dextran conjugated to Oregon Green 488 (D7173), and transferrin from human serum conjugated to Texas Red (T2875) were purchased from Thermo Fisher Scientific. Laemmli’s SDS Sample Buffer (BP-110R) was purchased from Boston BioProducts (Ashland, MA, USA). Lysis Buffer (9803S) and protease inhibitors (5871S) were purchased from Cell Signaling Technology (Danvers, MA, USA). Phosphatase inhibitor cocktail tablet (4906845001) was purchased from MilliporeSigma.

Expression vector and adenovirus for tagged EMCN

A cDNA fragment encoding a double-tagged human EMCN protein (myc tag at the N-terminal of the full-length human EMCN after the signal peptide sequence and DDK tag at its C terminal) was synthesized by Genscript (Piscataway, NJ, USA). This cDNA was then cloned into pcDNA4 (Addgene, Watertown, MA, USA) plasmid with a pCMV (Addgene) promoter via EcoRV and NotI cloning sites as the expression vector. The adenovirus expressing tagged EMCN was generated and titered by Vector Biolabs (Malvern, PA, USA). Briefly, this cDNA was inserted into a shuttle vector with a pCMV promoter via EcoRV and NotI, then the expression cassette was transferred into an adenoviral vector (Ad5 with E1 or E3 deletion) by homologous recombination. The adenoviral vector was linearized and transfected into 293 cells to generate adenovirus.

Antibodies

Immunoprecipitation (IP) experiments were conducted using rabbit anti-myc–conjugated sepharose beads (400S; Cell Signaling Technology) or rabbit IgG control (P120-101; Bethyl Laboratories, Montgomery, TX, USA). Immunocytochemistry (ICC)-based receptor internalization assays were carried out following incubation with goat anti-VEGFR2 (AF357; R&D Systems) and rat anti-EMCN (ab45771; Abcam, Cambridge, MA, USA) to track VEGFR2 and EMCN internalization. For endosome visualization, rabbit anti-Rab5 (3547S; Cell Signaling Technology) was used. Alexa Fluor 594–labeled donkey anti-goat (A-11058; Thermo Fisher Scientific), Alexa Fluor 488–labeled donkey anti-rat (A-21208; Thermo Fisher Scientific), Alexa Fluor 488–labeled donkey anti-goat (1:500, A-11055), and Alexa Fluor 647–labeled donkey anti-rabbit (1:500, A-31573) were used for ICC-based internalization assay. ICC in nonpermeabilized HRECs was carried out using rat anti-EMCN (ab45771; Abcam), CD31 (3528S; Cell Signaling Technology), and goat anti-VEGFR2 (AF357; R&D Systems). Fluorescent secondary antibodies Alexa Fluor 594–labeled donkey anti-goat (A-11058; Thermo Fisher Scientific), Alexa Fluor 647–labeled donkey anti-goat (A-21447; Thermo Fisher Scientific), Alexa Fluor 488–labeled donkey anti-rat (A-21208; Thermo Fisher Scientific), and Alexa Fluor 488–labeled donkey anti-mouse (A-21202; Thermo Fisher Scientific) were used for ICC. Immunoblots were probed with rabbit anti-VEGFR2 (2479S; Cell Signaling Technology), rat anti-EMCN, mouse anti-CD31 (3528S; Cell Signaling Technology), rabbit p-Y1175-VEGFR2 (2478S; Cell Signaling Technology), rabbit p-Y1059-VEGFR2 (3817S; Cell Signaling Technology), mouse α-tubulin (CP06-100UG; MilliporeSigma), or rabbit anti-myc (2278S; Cell Signaling Technology).

Cell culture

HRECs were purchased from Cell Systems (Troisdorf, Germany). HRECs were cultured on 0.2% gelatin–coated dishes for 30 min at 37°C. HRECs were cultured in EC growth medium (EGM)-2 BulletKit medium (CC-3162; Lonza, Basel, Switerland) supplemented with 2% fetal bovine serum (FBS, S11195; Atlanta Biologicals, Flowery Branch, GA, USA), 2 mM l-glutamine (17-605E; Lonza), and 100 U/ml penicillin–100 μg/ml streptomycin (17-602E; Lonza). Cells were maintained at 37°C with 5% CO₂. HRECs were used within passages 6–10.

siRNA knockdown

HRECs were seeded at 70% confluence 1 d prior to siRNA transfection. siEMCN (50 nM, L-015860-01-000) or nontargeting siRNA (siNT) (50 nM, D-001810-01-05) was incubated with Dharmafect 1 transfection reagent in Opti-MEM (51985034; Thermo Fisher Scientific) at room temperature for 30 min to allow complex formation. siRNA was added in EGM-2 complete medium supplemented with 2% FBS in the absence of penicillin-streptomycin. The following day, culture medium was changed, and cells were incubated for an additional 24 h before experiments.

Adenovirus transduction

HRECs were seeded at 70% confluence 1 d prior to adenoviral infection. Cells were infected with adenovirus expressing tagged human EMCN at a multiplicity of infection of 30 and cultured for 24 h in EGM-2 medium supplemented with 2% FBS.

VEGFR2 and EMCN localization

HRECs were plated on gelatin-coated coverslips at a density of 2.0 × 10⁵ cells per coverslip. HRECs were serum starved for 2 h in endothelial basal medium (EBM)-2, then stimulated with VEGF (10 ng/ml) in EBM-2 at 4°C. Cells were washed once with PBS, then fixed in 1% PFA in PBS overnight at 4°C. The following day, cells were stained using antibodies against VEGFR2 (1:200, AF357) and EMCN (1:200, ab45771) and incubated overnight at 4°C. Cells were then washed with PBS and incubated with Alexa Fluor 647–labeled donkey anti-goat (1:300, A-21447) and Alexa Fluor 488–labeled donkey anti-rat (1:300, A-21208) for 2 h at room temperature. Cells were washed in PBS and mounted with ProLong Gold antifade reagent with DAPI (P36935; Thermo Fisher Scientific).

ICC-based internalization assay

HRECs were serum starved for 2 h in EBM-2 and incubated with goat anti-VEGFR2 (1:100, AF357; R&D Systems) or rat anti-EMCN (1:200, ab45771) at 4°C for 1 h, followed by the addition of bovine serum albumin (BSA) or VEGF (10 ng/ml) with 70 kDa dextran conjugated to Oregon Green 488 (1.5 mg/ ml, D7173) or transferrin conjugated to Texas Red (50 μg/ml, T2875) for 30 min at 37°C. Cells were fixed in 4% PFA for 5 min at room temperature and permeabilized using 0.1% Triton X-100. At this time, cells were incubated with rabbit anti-Rab5 (1:200, 3547S; Cell Signaling Technology) to visualize colocalization between VEGFR2 and early endosomes. All cells were then incubated with Alexa Fluor 594–labeled donkey anti-goat (1:500, A-11058), Alexa Fluor 488–labeled donkey anti-goat (1:500, A-11055), Alexa Fluor 647–labeled donkey anti-rabbit (1:500, A-31573), or Alexa Fluor 488–labeled donkey anti-rat (1:300, A-21208). All experiments were conducted in the presence of PQB (0.6 μM) to prevent recycling. Five images per cover slip were imaged using an Axioskop 2 Mot Plus microscope (Carl Zeiss, Oberkochen, Germany; ×40 magnification), analyzed using Photoshop CC6 (Adobe, San Jose, CA, USA), and averaged. Intracellular fluorescence intensity was quantified using Photoshop CC6 and normalized to the total number of cells per viewing field as an indication of receptor internalization. Quantification reflects 3 independent experiments.

VEGFR2 cell surface localization following VEGF stimulation

HRECs were plated on gelatin-coated coverslips at a density of 2.0 × 10⁵ cells per coverslip. HRECs were serum starved for 2 h in EBM-2, then stimulated with VEGF (10 ng/ml) in EBM-2. VEGFR2 cell surface localization was visualized following VEGF stimulation for 30 min at 37°C in the presence of PQB (0.6 μM). Cells were washed with PBS and fixed in 1% PFA in PBS overnight at 4°C. Cells were stained using antibodies against VEGFR2 (1:200, AF357) and CD31 (1:300, 3528S) followed by incubation with Alexa Fluor 594–labeled donkey anti-goat (1:300, A-11058) and Alexa Fluor 488–labeled donkey anti-mouse (1:300, A-21202). Coverslips were imaged using an Axioskop 2 Mot Plus microscope (×20 magnification).

Biotin cell surface isolation

Confluent HRECs were serum starved for 2 h in EBM-2. Cells were then stimulated with VEGF (10 ng/ml) in serum-free EBM-2 supplemented with PQB (0.6 μM). Cell surface proteins were biotinylated with NHS-SS-Biotin (21331) for 2 h at 4°C, then quenched by washes in 50 mM Tris (pH 8.0) and PBS (pH 8.0) according to the manufacturer’s protocol. Cells were scraped in Tris-buffered saline and collected by centrifugation at 1500 g for 5 min. Pellets were lysed in Cell Signaling Lysis Buffer and incubated with 100 μl of Avidin Agarose (S1258122) rotating for 1 h at room temperature. Samples were washed 4 times in wash buffer (20 mM Tris-HCl, pH 6.8; 0.5% Tween 20) with protease and phosphatase inhibitors and centrifuged at 1000 g for 1 min. Cell surface proteins were eluted with Laemmli’s SDS Sample Buffer (BP-110R) with 100 mM DTT, boiled at 95°C for 10 min, and processed for Western blot analysis. Membranes were incubated with antibodies against VEGFR2 (1:1,000, 2479S), EMCN (1:300, ab45771) or CD31 (1:1000 14-0311-81) as a loading control for cell surface fractions.

IP

HRECs were plated in 100-mm dishes to confluence in EGM-2 complete media supplemented with 2% FBS. Cells were incubated with BSA or VEGF (10 ng/ml) in serum-free EBM-2 for 30 min at 4°C. HRECs were washed with ice-cold PBS, and cell surface proteins were labeled with NHS-SS Biotin (21331) for 30 min at 4°C and quenched with 50 mM Tris (pH 8.0) followed by washes in PBS (pH 8.0). Cells were scraped in Tris-buffered saline and collected by centrifugation at 1500 g for 5 min. Pellets were resuspended in lysis buffer [50 mM Tris (pH 7.5), 100 mM NaCl, 5 mM EDTA, 0.5% Triton X-100, and 0.5% NP-40] with protease and phosphatase inhibitor cocktail tablets and subjected to IP using monomeric avidin agarose for 1 h at room temperature. Biotinylated proteins were competitively eluted with d-biotin (1 mM). Surface fractions were incubated with rabbit anti-myc conjugated to sepharose beads (1:20, 3400S) or rabbit IgG control (1:75, P120-101) overnight, rotating at 4°C. The following day, the samples were centrifuged, and beads were washed in lysis buffer followed by washes in PBST. Bound proteins were eluted by incubation with Laemmli’s SDS sample buffer (BP-110R) with 100 mM dithiothreitol, boiled at 95°C for 10 min, and processed for Western blot analysis. Membranes were incubated with antibodies against rabbit anti-human VEGFR2 (1:1000; 2479S) and rabbit anti-myc (1:1000; 2278S).

Measurement of kinase activation

Confluent HRECs were serum starved in serum-free EBM-2 supplemented with 0.5% FBS overnight followed by incubation with VEGF (10 ng/ml) in serum-free EBM-2 for 0, 5, 10, 30, and 60 min. Cells were washed in PBS, lysed in Cell Signaling Lysis Buffer supplemented with protease inhibitor and a phosphatase inhibitor cocktail tablet, and processed for Western blot.

Analysis of protein expression

Cells were lysed with buffer containing protease inhibitor (Roche, Basel, Switzerland) and a phosphatase inhibitor cocktail tablet (1:100; MilliporeSigma). Protein concentration was determined by the bicinchoninic acid assay (23227; Thermo Fisher Scientific). Proteins separated on SDS-PAGE were transferred to nitrocellulose membranes (27376-991; VWR International, West Chester, PA, USA) and then probed with appropriate antibodies. Membranes were incubated with appropriated secondary antibodies and developed by chemiluminescence with SuperSignal West Pico Chemiluminescent Substrate (34077; Thermo Fisher Scientific) or by fluorescence method using LI-COR Odyssey (LI-COR Biosciences, Lincoln, NE, USA). Equal loading and transfer were determined by reprobing the membranes for α-tubulin. Exposed films were scanned, and the densitometry analyses were performed by Fiji (https://fiji.sc/) or using Image Studio 2.0 (LI-COR) (19). Arbitrary units of fluorescence were converted to fold change.

mRNA isolation and gene expression analysis

Total RNA was isolated from tissue and cells using RNeasy Mini Kit (74104; Qiagen, Germantown, MD, USA) and was reverse transcribed into cDNA using iScript (170-8891; Bio-Rad, Hercules, CA, USA). Reactions were performed on the LightCycler 480 II (Roche) using 0.4 μM primers and Faststart Universal Sybr Green PCR Master Mix (4913914001; Roche). PCR cycling was performed at 95°C for 10 min, followed by 95°C for 15 s and 60°C for 1 min for a total of 40 cycles. The C_t values corresponding to the PCR cycle number at which fluorescence emission in real time reaches a threshold above the baseline emission were automatically calculated by the LightCycler 480 II (Roche). Amplification of hypoxanthine-guanine phosphoribosyltransferase was performed on each sample as the housekeeping gene. All data were normalized to housekeeping control. Each biologic sample was repeated in triplicate for each set of primers, and the mean C_t value of the triplicate was used to represent each biologic sample for further analysis.

Statistical analysis and colocalization

All values are expressed as means ± sem. Statistical analysis was performed using an unpaired Student t test or 1-way ANOVA (GraphPad Prism 5). A value of P < 0.05 was considered statistically significant. Each experimental condition was conducted at least in triplicate, and all experiments were independently repeated at least 3 times. Colocalization was quantified by measuring the colocalization between 2 fluorophores using Pearson’s colocalization coefficient in the Just Another Colocalization Plugin (JACoP) in ImageJ (National Institutes of Health, Bethesda, MD, USA) (20, 21).

RESULTS

EMCN colocalizes with VEGFR2

Previous work from our lab has shown that depletion of EMCN prevents VEGF-induced HREC proliferation, migration, and tube formation in vitro and inhibits retinal vascular development in vivo, pointing to a role for EMCN in modulating VEGF-mediated functions (18). To define the molecular mechanism of EMCN regulation of VEGF-induced processes, we sought to investigate EMCN’s influence on VEGFR2 signaling. Given that EMCN and VEGFR2 are both cell surface proteins, we expected a high degree of colocalization. The purpose of this experiment was to determine whether the presence or absence of VEGF impacted the degree of surface colocalization of EMCN and VEGFR2. To determine whether EMCN colocalizes with VEGFR2 and whether VEGF plays a role in their localization, HRECs were stimulated with VEGF (10 ng/ml) for 30 min at 4°C. Incubation was carried out at 4°C to prevent receptor activation and internalization. Prominent cell-cell junctional and cell surface staining was observed for EMCN with notable cell-cell junction VEGFR2 localization in both conditions. EMCN and VEGFR2 were colocalized to both the cell surface as well as the cell-cell junction surface of HREC, and there was no change in their distribution with or without VEGF stimulation (Fig. 1A). Colocalization between EMCN (green) and VEGFR2 (white) was quantified using Pearson’s colocalization coefficient. Results show no significant difference between VEGFR2 and EMCN overlap in the presence or absence of VEGF (0.65 ± 0.03 vs. 0.62 ± 0.04) (Fig. 1B).

Figure 1.

EMCN colocalizes and interacts with VEGFR2. A) HRECs were stimulated with or without VEGF (10 ng/ml) for 30 min at 4°C. Cells were incubated with antibodies against EMCN, VEGFR2, or rabbit IgG control. EMCN (green) and VEGFR2 (white) colocalized on the HREC surface independent of VEGF stimulation. B) Colocalization between VEGFR2 and EMCN was quantified using Pearson’s correlation coefficient from the Just Another Colocalization Plugin (JACoP) in ImageJ. VEGF (−) vs. VEGF (+). C) Myc-tagged Adenovirus overexpressing human EMCN (Ad.hEMCN) was overexpressed in HRECs (multiplicity of infection 30) with or without VEGF stimulation. Cell surface proteins were isolated and incubated with antibodies against IgG or myc. EMCN and VEGFR2 interaction was independent of VEGF stimulation. IB, immunoblot; n/s, nonsignificant; n = 5. Scale bar, 20 μm.

EMCN interacts with VEGFR2

Having demonstrated that EMCN and VEGFR2 colocalize, we next sought to determine whether VEGFR2 and EMCN interact. We used IP to visualize EMCN and VEGFR2 interactions in the presence or absence of VEGF. To ensure high levels of EMCN for IP, HRECs were infected with an adenovirus-overexpressing, full-length, human myc-tagged EMCN. IP was conducted 48 h after transfection. HRECs were stimulated with or without VEGF (10 ng/ml) at 4°C for 30 min to allow for ligand binding without VEGFR2 internalization. Cell surface fractions were isolated using biotin followed by incubation with monomeric avidin agarose and competitive elution with d-biotin. Surface proteins were incubated with antibodies against myc or rabbit IgG. Similar to ICC colocalization, results showed that VEGFR2 and EMCN coimmunoprecipitated on the EC surface both in the presence and absence of VEGF (Fig. 1C).

EMCN depletion prevents VEGF-induced VEGFR2 internalization

Receptor tyrosine kinase signaling continues beyond cell surface activation; several reports have shown that receptor internalization is required for controlling signal duration and functional activity (13) and that blocking VEGFR2 internalization prevents VEGF-induced angiogenesis in vitro (7). Given that VEGFR2 plays a critical role in modulating VEGF-induced angiogenesis, we examined the role of EMCN in regulating VEGFR2 receptor internalization.

Knockdown of EMCN in HRECs was achieved using siEMCN; siNT was used as a control (50 nM). Levels of EMCN protein (0.928 ± 0.07 vs. 0.252 ± 0.03; P < 0.001) and mRNA (1 ± 0.091 vs. 0.007 ± 0.0007; P < 0.001) were significantly reduced following siEMCN transfection (Fig. 2A–C). Receptor internalization was measured utilizing an ICC-based internalization assay using a nonneutralizing antibody that recognized the extracellular domain of VEGFR2 (R&D Systems), followed by detection with a fluorescent secondary antibody (Fig. 2D). Costaining with antibodies against the early endosome marker Rab5 was done to visualize VEGFR2 trafficking following internalization. Results showed that control HRECs transfected with siNT exhibited significant VEGFR2 internalization following VEGF stimulation compared with BSA nonstimulated control (152.4 ± 29.92 vs. 519.1 ± 81.96; P < 0.001). Conversely, VEGF-stimulated VEGFR2 internalization was significantly inhibited in HRECs depleted of EMCN (1.002 ± 0.108 vs. 1.128 ± 0.132). There was no impact of EMCN depletion on constitutive VEGFR2 internalization compared with siNT control in the absence of VEGF stimulation (182.9 ± 15.36 vs. 156.3 ± 13.33) (Fig. 2E, F). To confirm anti-VEGFR2 (R&D Systems) did not neutralize receptor activity, we evaluated VEGF-stimulated VEGFR2 phosphorylation and ERK activation following incubation with IgG or anti-VEGFR2. Results show that anti-VEGFR2 (R&D Systems) did not inhibit VEGF-induced VEGFR2 phosphorylation (Y1175) or activation of ERK following 10 min (10.97 ± 2.84 vs. 11.46 ± 1.61) or 30 min of VEGF stimulation (9.65 ± 1.78 ± 11.46 ± 1.61) (Fig. 2G–I).

Figure 2.

EMCN depletion prevents VEGFR2 internalization. A–C) Immunoblot (A) and quantitative analysis (B, C) of HRECs incubated with siRNA against siNT control or siEMCN (50 nM) for 48 h posttransfection. siRNA reduced human EMCN protein (73.1% reduction) (B) and mRNA (99.3% reduction) (C). D) Illustration of the ICC-based internalization assay in which HRECs were incubated with an antibody against the extracellular domain of VEGFR2, followed by incubation with VEGF or BSA (10 ng/ml) for 30 min at 37°C. Constitutive and VEGF-induced VEGFR2 internalization were evaluated in the presence of siNT or siEMCN. E) VEGFR2 internalization was visualized by intracellular fluorescence intensity (green). Early endosome marker Rab5 (white) was covisualized in each condition. In the absence of EMCN, VEGF-induced VEGFR2 internalization (green) was reduced after 30 min. F) Receptor internalization was quantified by relative fluorescence and normalized to the total cell number per viewing field (n = 12–18). siCtrl BSA vs. VEGF, siEMCN BSA vs. VEGF, siCtrl VEGF vs. siEMCN VEGF. siNT vs. siEMCN. G–I) Anti-VEGFR2 (G) is not neutralizing. VEGF-induced phosphorylation of VEGFR2 (Y1175) (H) and ERK (I) was evaluated following preincubation with anti-VEGFR2 (1:200) or IgG control (1:200). VEGF-induced VEGFR2 activation and phosphorylation of ERK (10 and 30 min) was not altered following anti-VEGFR2 or IgG pretreatment (n = 3). VEGF 0 min: IgG control vs. anti-VEGFR2, VEGF 10 min: IgG control vs. anti-VEGFR2, VEGF 30 min: IgG control vs. anti-VEGFR2. Ctrl, control; n/s, nonsignificant. Scale bars, 20 μm. *P < 0.05, **P < 0.01, ***P < 0.001.

To confirm that VEGFR2 internalization is impeded by EMCN depletion, we evaluated surface VEGFR2 levels after knocking down EMCN in HRECs. ICC was conducted using antibodies against VEGFR2 (red) and CD31 (green), a known endothelial surface marker. In control HRECs treated with siNT, VEGF stimulation led to rapid internalization of VEGFR2 and its loss from the EC surface (Fig. 3A). In contrast, EMCN depletion resulted in the retention of cell surface VEGFR2 following VEGF stimulation, demonstrated by the colocalization of CD31 and VEGFR2 on the EC surface (Fig. 3A).

Figure 3.

EMCN depletion prevents VEGFR2 internalization and leads to its retention on the cell surface. A) Immunofluorescent localization of VEGFR2 and CD31 revealed that VEGFR2 is localized on the EC surface following VEGF (10 ng/ml) stimulation for 30 min at 37°C in EMCN-depleted HRECs. All experiments were done in the presence of PQB to prevent receptor recycling back into the membrane. B, C) Immunoblot (B) and quantitative analysis (C) of biotinylated HREC surface proteins confirmed that in the absence of EMCN, VEGFR2 was retained on the EC surface following VEGF stimulation for 30 min at 37°C in the presence of PQB. CD31 was used for cell surface loading control. Data were normalized to siNT BSA; n = 8. Scale bar, 20 μm. *P < 0.05. siNT BSA vs. siNT VEGF or siEMCN BSA vs. siEMCN VEGF.

Biotin cell surface labeling was used to quantify the cell surface levels of VEGFR2 with and without EMCN depletion. siNT-transfected HRECs showed a significant reduction in cell surface VEGFR2 following VEGF stimulation (1.052 ± 0.120 vs. 0.691 ± 0.045; P < 0.05). In contrast, siEMCN-transfected HRECs displayed no significant decrease in cell surface VEGFR2 in the absence or presence of VEGF (0.857 ± 0.114 vs. 0.706 ± 0.069). In the absence of VEGF stimulation, there was no significant difference in baseline surface levels of VEGFR2 in siEMCN (0.857 ± 0.114) compared with siNT-transfected HRECs (0.706 ± 0.069) (Fig. 3B, C).

EMCN is not cointernalized with VEGFR2 following VEGF stimulation

In light of the observation that EMCN and VEGFR2 colocalize on the EC surface (Fig. 1A) and that EMCN is necessary for VEGFR2 internalization following VEGF stimulation, we investigated whether EMCN was cointernalized with VEGFR2. For this purpose, we employed an ICC-based internalization assay using antibodies against the extracellular domains of VEGFR2 and EMCN (Fig. 4A). Results showed that EMCN was not internalized with VEGFR2 but instead remained on the EC surface (Fig. 4B). Biotin cell surface labeling was used to confirm that EMCN cell surface levels were unchanged in siNT- vs. siEMCN-transfected (0.870 ± 0.168 vs. 0.957 ± 0.178) HRECs following VEGF stimulation (Fig. 4C, D).

Figure 4.

EMCN is not internalized with VEGFR2 following VEGF stimulation. A) EMCN internalization with VEGFR2 was measured using an ICC-based internalization assay. Illustration of hypothesized VEGFR2 and EMCN cointernalization. B) EMCN internalization with VEGFR2 was measured using an ICC-based internalization assay. HRECs were incubated with antibodies against the extracellular domain of VEGFR2 and EMCN followed by VEGF or BSA stimulation (10 ng/ml) for 30 min at 37°C. Remaining cell surface antibodies were removed by washing with a low-pH buffer; cells were permeabilized and incubated with corresponding fluorescent secondary antibodies. Following VEGF stimulation, VEGFR2 but not EMCN was detected intracellularly using the ICC-based internalization assay. C, D) Immunoblot (C) and quantitative analysis (D) of biotinylated HREC surface proteins confirmed EMCN cell surface localization following VEGF stimulation. CD31 was used as a cell surface loading control. All data were normalized to BSA controls; VEGF (−) vs. VEGF (+). N/s, nonsignificant; n = 4. Scale bar, 20 μm.

EMCN depletion dampened transferrin but not dextran endocytosis

Given that EMCN depletion prevented VEGFR2 endocytosis, we next asked whether this was a VEGFR2-specific phenomenon or whether loss of EMCN impaired general endocytic pathways. To evaluate this question, endocytosis of a clathrin-dependent cargo molecule transferrin and a micropinocytosis cargo protein dextran were evaluated in the presence and absence of EMCN. An ICC-based internalization assay was utilized as described in Fig. 3A with the following modifications. Cells were pretreated with anti-VEGFR2 (R&D Systems) at 4°C and incubated with VEGF or BSA in the presence of dextran conjugated to Oregon Green 488 (70 kDa, 1.5 mg/ml, D7173) or transferrin conjugated to Texas Red (50 μg/ ml, T2875) for 30 min at 37°C. EMCN depletion did not alter dextran internalization in the presence (500 ± 64.15 vs. 357.2 ± 52.13) or absence of VEGF stimulation (398.9 ± 54.87 vs. 473.4 ± 51.22) (Fig. 5A, B). Results further showed a significant decrease in transferrin internalization following BSA (1678 ± 104.7 vs. 760.7 ± 64.2; P < 0.001) and VEGF stimulation (2392 ± 105.9 vs. 1534 ± 108.1; P < 0.001) in siEMCN-transfected HRECs compared with siNT controls (Fig. 5C, D). In both control and EMCN-depleted cells, VEGFR2 and transferrin showed stronger colocalization compared with VEGFR2 and dextran independent of BSA or VEGF stimulation (Fig. 5E).

Figure 5.

Loss of EMCN dampens transferrin but not dextran endocytosis. ICC-based internalization assay was used to quantify dextran or transferrin internalization in the presence and absence of EMCN following VEGF or BSA stimulation for 30 min. A) Dextran internalization was visualized by intracellular fluorescence intensity (green). VEGFR2 internalization was visualized by intracellular fluorescence intensity (red). Early endosome marker Rab5 (white) was covisualized in each condition. EMCN depletion did not alter dextran internalization in the presence or absence of 30-min VEGF stimulation. B) Dextran internalization was quantified by relative fluorescence and normalized to the total cell number per viewing field. EMCN depletion did not alter dextran internalization in the presence of BSA or VEGF stimulation. C) Transferrin internalization was visualized by intracellular fluorescence intensity (red). VEGFR2 internalization was visualized by intracellular fluorescence intensity (green). Early endosome marker Rab5 (white) was covisualized in each condition. D) VEGF stimulation significantly increased transferrin internalization in siNT cells. Transferrin internalization was significantly reduced in siEMCN compared with siNT-transfected HRECs following BSA and VEGF stimulation. E) In both siNT- and siEMCN-transfected cells, Pearson’s colocalization coefficient for transferrin and VEGFR2 was significantly higher than dextran and VEGFR2 following both BSA and VEGF stimulation. N/s, nonsignificant; n = 12–18. Scale bars, 20 μm. ***P < 0.001, siNT BSA vs. siNT VEGF or siEMCN BSA vs. siEMCN VEGF.

Loss of EMCN increases total and p-(Y1175) VEGFR2

Because VEGF-induced VEGFR2 activation normally leads to VEGFR2 endocytosis and a resultant decline in total VEGFR2 following prolonged stimulation, we determined whether EMCN knockdown influenced VEGFR2 receptor expression and activation. HRECs were transfected with siEMCN or siNT, followed by stimulation with VEGF (10 ng/ml) at 37°C for 0, 5, 10, 30, or 60 min. Depletion of EMCN led to an increase in total VEGFR2 protein expression without VEGF at baseline (1 vs. 1.924 ± 0.138) as well as 5 min (0.986 ± 0.067 vs. 1.848 ± 0.245) and 10 min (1.251 ± 0.113 vs. 1.904 ± 0.18) after VEGF stimulation. There was no significant difference in total VEGFR2 levels in siEMCN- or siNT-treated HRECs at 30 min (1.253 ± 0.232 vs. 1.28 ± 0.194) or 60 min after VEGF stimulation (0.889 ± 0.07 vs. 0.751 ± 0.124) (Fig. 6A, C).

Figure 6.

EMCN depletion leads to increased VEGFR2 phosphorylation with a reduction in VEGFR2 mRNA following extended VEGF stimulation. A) Immunoblot and quantitative analysis of B–D of HRECs stimulated with VEGF (10 ng/ml) at 37°C for the indicated time points in the presence of siNT or siEMCN and subjected to Western blot. B) VEGF stimulation for 5 min increased VEGFR2 phosphorylation (Y1175) in siEMCN-transfected HRECs compared with siNT controls. Data were normalized to siNT control. C) siEMCN increased total VEGFR2 protein at 0, 5, and 10 min after VEGF stimulation. After 30 and 60 min of VEGF stimulation, there was no detectable difference in total VEGFR2 protein expression in siEMCN vs. siNT controls. Data were normalized to non–VEGF-stimulated siNT controls. Data were normalized to time-matched siNT controls in B and C. For each time point: siNT vs. siEMCN. D) Quantitative analysis of HRECs stimulated with VEGF (10 ng/ml) at 37°C for the indicated time points. Quantitative PCR results show that knockdown of EMCN did not affect VEGFR2 mRNA at baseline and after 5 and 10 min of VEGF stimulation. Depletion of EMCN significantly decreased VEGFR2 mRNA after 30 and 60 min of VEGF stimulation. Data were normalized to non–VEGF-stimulated siNT controls. For each time point: siNT vs. siEMCN. Ctrl, control; norm, normalized. *P < 0.05, **P < 0.01, ***P < 0.001.

Because of the known involvement of Y1175 phosphorylation of the VEGFR2 in VEGF-induced endothelial proliferation and migration through the activation of the PLC-Raf-MEK-ERK pathway (3), we investigated Y1175 phosphorylation in EMCN-depleted HRECs stimulated with VEGF. Results revealed a significant increase in VEGFR2 phosphorylation at the Y1175 residue in siEMCN- but not siNT-treated HRECs following 5 min of VEGF stimulation (10.03 ± 1.82 vs. 19.12 ± 3.304, P < 0.05). Although not statistically significant, a trend toward an increase in p-VEGFR2 Y1175 was noted after 10 min (13.65 ± 2.792 vs. 15.28 ± 4.167) and 30 min (7.591 ± 1.38 vs. 11.02 ± 2.381) of VEGF stimulation. Neither siNT nor siEMCN showed a difference in phosphorylation of VEGFR2 (Y1175) at baseline (1 ± 0 vs. 1.81 ± 0.136) or following 60 min of VEGF stimulation (6.591 ± 0.977 vs. 7.358 ± 1.025) (Fig. 6B).

To understand the effect of EMCN knockdown on VEGFR2 levels, we examined whether depletion of EMCN affected VEGFR2 transcriptional regulation or VEGFR2 mRNA stability. HRECs were transduced with siEMCN or siNT followed by stimulation with VEGF (10 ng/ml) at 37°C for 0, 5, 10, 30, and 60 min. Quantitative PCR results show that EMCN depletion did not impact the levels of VEGFR2 mRNA at baseline or following 5 and 10 min of VEGF (10 ng/ml) stimulation at 37°C. However, VEGFR2 mRNA was significantly reduced following 30 min (0.737 ± 0.04 vs. 0.077 ± 0.013) and 60 min (1.45 ± 0.14 vs. 0.06 ± 0.003) of VEGF (10 ng/ml) stimulation in the cells depleted of EMCN (Fig. 6D).

DISCUSSION

Our data demonstrate a novel role for EMCN in VEGF-induced VEGFR2 endocytosis. EMCN was previously identified by our lab as a mediator of VEGF-driven endothelial activities in vitro and angiogenesis in vivo (18). This study expands upon these findings and demonstrates an interaction between EMCN and VEGFR2. It has been shown that, following internalization, VEGFR2 continues to signal within the endosomes (13). Consequently, lack of internalization has a significant impact on receptor-driven signaling and EC biology (7). In this work, we show that EMCN is required for VEGF-stimulated but not basal VEGFR2 internalization and that, although loss of EMCN prevented receptor internalization, retention on the membrane resulted in heightened Y1175 phosphorylation. Taken together, these results suggest that inhibition of VEGFR2 internalization upon EMCN depletion is sufficient to prevent VEGF-induced angiogenesis despite persistent activation of cell surface VEGFR2 Y1175.

Upon VEGF stimulation, VEGFR2 can form multiprotein complexes with coreceptors such as NRP1, NRP2, heparan sulfate proteoglycans, endoglin (3, 4), and other cell surface proteins with varying biologic effects. For example, NRP1 binding to VEGFR2 in the presence of VEGF favors EC chemotaxis, proliferation, and permeability (22). Following internalization, the NRP1-VEGFR2 complex directs VEGFR2 toward membrane recycling, thus avoiding degradation (12). Unlike NRP1, which is internalized and has been shown to colocalize with VEGFR2 in the early endosomes (23), EMCN was not internalized with VEGFR2 following VEGF stimulation. These results suggest that EMCN-VEGFR2 interactions are transient and may only serve to facilitate VEGF-induced VEGFR2 endocytosis.

VEGFR2 is an integral component of the endothelial junctional complex (24, 25). In this study, ICC results show well-defined junctional colocalization between VEGFR2 and EMCN. Although this interaction was not dependent upon VEGF stimulation, it is unclear whether the co-IP between EMCN and VEGFR2 is direct or indirect. Previous work has demonstrated that tyrosine kinase receptor activity can be regulated through associations with neighboring EC junctional proteins such as vascular endothelial protein tyrosine phosphatase. Vascular endothelial protein tyrosine phosphatase dephosphorylates receptors, such as VEGFR2 and Tie2, and dampens downstream functional activity (26, 27). Interactions between vascular endothelial cadherin and VEGFR2 negatively regulate receptor signaling by modulating VEGFR2 endocytosis (13). Vascular endothelial cadherin expression prevents VEGFR2 endocytosis, whereas its absence enhances the efficiency and speed of VEGFR2 internalization. Increased VEGFR2 internalization does not terminate receptor activation but leads to heightened VEGFR2 signaling within the early endosomes. Given that EMCN depletion prevents ligand-induced VEGFR2 internalization, we predict that EMCN interactions with VEGFR2 are an early and necessary step for receptor endocytosis.

Endocytosis of plasma membrane receptors can occur through clathrin-dependent or -independent pathways. Clathrin-dependent pathways are named because of the high expression of clathrin in the endocytic vesical coat, whereas cargo endocytosis through clathrin-independent pathways involves cholesterol-rich membrane domains. Clathrin-independent pathways consist of caveolae-dependent endocytosis or micropinocytosis (28). Although previous groups have demonstrated clathrin-dependent pathways as the preferred route for VEGFR2 endocytosis (13), this preference can shift in the presence of changes in ligand concentration, interactions with coreceptors (28), EC confluency (13), or following the blockade of 1 endocytic pathway (28). Upon VEGF stimulation, the route of VEGFR2 endocytosis may also shift from clathrin-dependent endocytosis toward CDC42-dependent micropinocytosis (7). Our results show that EMCN is not only required for VEGF-induced VEGFR2 endocytosis but that loss of EMCN selectively impacts transferrin uptake through clathrin-mediated endocytosis. No effect of EMCN inhibition was noted for dextran internalization through micropinocytosis. This suggests a role for EMCN in modulating VEGFR2 internalization through clathrin-dependent pathways. We might expect that context would have an effect on the modes of receptor internalization; therefore, we cannot exclude the possibility that, under some experimental conditions, VEGFR2 endocytosis may shift away from clathrin pathways and toward micropinocytosis.

A distinct set of adaptor and cell surface binding proteins are required for the internalization of VEGF-bound activated VEGFR2 vs. unbound VEGFR2. These interactions not only control receptor endocytosis but also modulate signaling output. Prior to internalization, VEGF-bound VEGFR2 undergoes ubiquitination followed by a ubiquitin-dependent interaction with intracellular adaptor proteins, such as epsin1 and 2 (14). At the cell surface, VEGF-bound VEGFR2 interacts with NRP1 and facilitates VEGFR2 trafficking into early endosome Rab5- or antigen-1–positive endosomes that favor receptor recycling and enhance signal output vs. trafficking to the lysosome for degradation (29). Interactions between activated VEGFR2 and guanine exchange factors ARNO and GEP100, which modulate GTPase ARF6, promote receptor internalization and enhance recycling as well as signaling (12). We speculate that EMCN may recruit additional cell surface or intracellular proteins required for VEGF-induced VEGFR2 endocytosis. This is supported by our results, which show that although EMCN knockdown significantly prevented endocytosis of VEGF-bound VEGFR2, there was no impact on basal VEGFR2 endocytosis, which occurs independent of ligand binding.

Previous findings from our lab show that depletion of EMCN led to dampened VEGFR2 Y951 phosphorylation (18), whereas our current work demonstrated heightened Y1175 phosphorylation in the absence of EMCN due to VEGFR2 retention on the EC surface following VEGF stimulation. VEGFR2 phosphorylation is regulated by several factors, including receptor internalization, ligand binding affinities, coreceptors, tyrosine phosphorylation sites, receptor uptake, rate of degradation, speed of recycling, and exposure to phosphatases (3). Phosphatase localization dictates where and when active tyrosine residues are dephosphorylated (5). Slow trafficking through the early endosomes has been correlated with attenuated Y1175 phosphorylation, presumably because of extended exposure to endosome-localized protein tyrosine phosphatase 1b (30). Conversely, the T-cell protein tyrosine phosphatase, which is located to both the plasma membrane and cytosol, preferentially dephosphorylates Y951 (31). Thus, we speculate that VEGFR2 Y1175 remains phosphorylated because of its persistent localization to the EC surface, resulting in lack of interaction with the Y1175-selective phosphatase, phosphatase protein tyrosine phosphatase 1b, which is localized to the early endosomes.

The manipulation of intracellular adaptor proteins or small GTPase ARF6, such as epsin 1 and 2, has been shown to inhibit VEGFR2 endocytosis and result in increased VEGFR2 protein levels (12, 14). We observed that EMCN depletion, which blocked VEGFR2 endocytosis, led to increased total VEGFR2 protein levels at baseline and following <30 min of VEGF stimulation. We suspect this effect reflects reduced VEGFR2 degradation that would normally result from its internalization.

VEGF signaling is central to a number of developmental, homeostatic, and pathologic conditions, and inhibition of this pathway has significant therapeutic applications. Targeting VEGFR2 activity selectively in the endothelium represents a valuable approach to modulate VEGF-driven angiogenesis. Our findings illustrate a central role for the glycocalyx, namely EMCN, in modulating VEGF-induced VEGFR2 trafficking and point to EMCN as a potential EC-specific therapeutic target.

Glossary

ARF6

ADP-ribosylation factor 6

BSA

bovine serum albumin

EBM

endothelial basal medium

EC

endothelial cell

EGM

EC growth medium

EMCN

endomucin

FBS

fetal bovine serum

HREC

human retinal microvascular EC

ICC

immunocytochemistry

IP

immunoprecipitation

NRP

neuropilin

PFA

paraformaldehyde

PQB

primaquine bisphosphate

siCtrl

nontargeting control siRNA

siEMCN

siRNA directed against EMCN

siNT

nontargeting siRNA

siRNA

small interfering RNA

VEGFR2

VEGF receptor 2

AUTHOR CONTRIBUTIONS

M. E. LeBlanc, K. L. Saez-Torres, and I. Cano performed the research; Z. Hu, M. Saint-Geniez, and Y.-S. Ng provided critical discussion and technical help for the experiments; M. E. LeBlanc, M. Saint-Geniez, Y.-S. Ng, and P. A. D’Amore designed the research; M. E. LeBlanc analyzed the data and wrote the first draft of the manuscript; K. L. Saez-Torres, I. Cano, Z. Hu, M. Saint-Geniez, and Y.-S. Ng reviewed the manuscript; and P. A. D’Amore directed the study, analyzed and interpreted results with M. E. LeBlanc, and did the major editing of the manuscript.