Vascular endothelial growth factor receptor 3 regulates endothelial function through β-arrestin 1

Abstract

Background--

Receptor signaling is central to vascular endothelial function and is dysregulated in vascular diseases such as atherosclerosis and pulmonary arterial hypertension (PAH). Signaling pathways involved in endothelial function include vascular endothelial growth factor receptors (VEGFRs) and G protein-coupled receptors (GPCRs), which classically activate distinct intracellular signaling pathways and responses. The mechanisms that regulate these signaling pathways have not been fully elucidated and it is unclear what nodes for crosstalk exist between these diverse signaling pathways. For example, multifunctional β-arrestin (ARRB) adapter proteins are best known as regulators of GPCR signaling but their role at other receptors and their physiological importance in the setting of vascular disease is unclear.

Methods--

We used a combination of human samples from PAH, human microvascular endothelial cells from lung, and Arrb knockout mice to determine the role of ARRB1 in endothelial VEGFR3 signaling. Additionally, a number of biochemical analyses were performed to determine the interaction between ARRB1 and VEGFR3, signaling mediators downstream of VEGFR3, and the internalization of VEGFR3.

Results—

Expression of ARRB1 and VEGFR3 was reduced in human PAH and deletion of Arrb1 in mice exposed to hypoxia led to worse PAH with a loss of VEGFR3 signaling. Knockdown of ARRB1 inhibited VEGF-C-induced endothelial cell proliferation, migration and tube formation, along with reduced VEGFR3, Akt and eNOS phosphorylation. This regulation was mediated by direct ARRB1 binding to the VEGFR3 kinase domain and resulted in decreased VEGFR3 internalization.

Conclusions--

Our results demonstrate a novel role for ARRB1 in VEGF receptor regulation and suggests a mechanism for cross-talk between GPCRs and VEGFRs in PAH. These findings also suggest that strategies to promote ARRB1-mediated VEGFR3 signaling could be useful in the treatment of pulmonary hypertension and other vascular disease.

Introduction

The endothelium, a monolayer of endothelial cells lining the inside of blood and lymphatic vessels, is constantly exposed to a variety of stimuli and insults from circulation, and plays a central role in vascular health and disease. Endothelial dysfunction, characterized by an imbalance in signaling between endothelium-derived vasodilators, e.g. nitric oxide (NO) and prostacyclin, and vasoconstrictors, e.g. thromboxane A2, endothelin-1 and angiotensin II, has been associated with many cardiovascular diseases¹. Endothelial dysfunction occurs both in diseases of large vessels, such as atherosclerosis, and of small vessels, such as pulmonary arterial hypertension (PAH)². PAH is a progressive disorder affecting pulmonary arterioles (50–100 μm in size) where dysfunction of endothelial cells, proliferation of smooth muscle cells, and obliteration of pulmonary arterioles lead to high pulmonary vascular resistance, right ventricular hypertrophy and failure, and ultimately death³.

The signaling mechanisms underlying the development of PAH remain to be fully elucidated. One set of critical signals that regulates endothelial function in PAH are transduced by vascular endothelial growth factor receptors (VEGFRs). VEGFRs belong to the family of receptor tyrosine kinases (RTKs) and play a central role in endothelial function, including cell proliferation and survival, angiogenesis and lymphangiogenesis⁴. VEGFRs are activated by several closely related vascular endothelial growth factors (VEGFs)^{5, 6}. Among them, VEGF/VEGFA is the ligand of both VEGFR1 (also known as Flt-1) and VEGFR2 (KDR/Flk-1), which are the predominant isoforms for blood endothelial cells, while VEGF-C and VEGF-D preferentially recognize and activate VEGFR3 (Flt-4), whose expression is largely restricted to lymphatic endothelial cells after embryonic development^{5, 6}. VEGFR3 was also recently found to be expressed in the endothelial tip cells during angiogenesis and tumor vasculature^{7, 8}. In rodent models of PAH, the VEGFR2 inhibitor SU-5416 has been used in combination with chronic hypoxia to induce PAH⁹. Furthermore, endothelial cell-specific knockout of Vefgr3 in mice causes exacerbation of chronic hypoxia-induced PAH¹⁰, suggesting a role for VEGFR3 signaling in endothelial cells in the pathogenesis of PAH.

Another group of other receptors that play important roles in endothelial dysfunction in PAH is G protein-coupled receptors (GPCRs)^{11, 12}. GPCR signaling is transduced either classically through G proteins, or through the multifunctional adapter proteins β-arrestin 1 (ARRB1) and 2 (ARRB2)¹³. On smooth muscle cells, activation of prostacyclin receptors promotes vasodilation and antiproliferative signaling while activation of type A endothelin receptors leads to vasoconstriction and cell proliferation^{14, 15}. Pharmacologically targeting G protein-mediated signaling of the prostacyclin receptor with agonists, or the type A endothelin receptors with antagonists has proven to be effective in PAH, with benefits in exercise capacity and clinical endpoints^{16, 17}.

At this time, it is unclear how signaling between canonically distinct signaling pathways, such as GPCRs and VEGFRs, are coordinated to regulate endothelial function. Based on their well-characterized role as acting as signaling hubs¹³, we hypothesized that ARRBs would play an important role in regulating endothelial function in PAH. Here we identified that Arrb1 expression was reduced in human idiopathic PAH (IPAH), Arrb1 knockout (KO) resulted in increased severity of hypoxia-induced PAH and identified a novel role for ARRB1 in the regulation of VEGFR3 signaling to regulate endothelial function.

Methods

A full description of the methods is presented in the online-only Data Supplement. The data, analytic methods, and study materials will be made available to other researchers for purpose of reproducing or replicating the procedure (available at the corresponding author’s laboratory).

Cell Culture

Human microvascular endothelial cells from lung (HMVEC-L), human umbilical endothelial cells (HUVEC) and HEK 293T cells were obtained from Duke Cell Culture Facility. HMVEC-L cells were cultured in EGM-2MV (Lonza). HUVEC cells were cultured in EGM-2 (Lonza). HEK293T were cultured in MEM supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin. All cell lines have been tested for mycoplasma contamination. All cells were maintained at 37 °C with 5% CO2-humidified atmosphere. For the analysis of receptor degradation, HMVEC-L cells were serum starved for 4 hours in EBM-2 and then incubated for 30 minutes with 100 ug/ml of cycloheximide (CHX, Sigma) before VEGF-C (100 ng/ml) stimulation.

Immunofluorescence and Morphometric Analysis

Lung tissues were paraffin-embedded and sectioned at 5 μm. Sections were deparaffinized and rehydrated in a graded ethanol series to water. Antigen retrieval was carried out in citrate buffer (pH 6.0) for 20 min by microwave oven, followed by cooling at room temperature for 20 min. The mice lung sections were then blocked with 10% goat norm serum in PBS for an hour, and incubated at 4°C with anti-VWF (1:200, Dako) and anti-α-smooth muscle actin (1:100, Sigma) antibodies in 10% goat norm serum in PBS overnight, followed by incubation with AlexaFluor 488 goat anti-mouse or AlexaFluor 594 goat anti-rabbit secondary antibody for 1 h in the dark. For human lung sections, sections were incubated with anti-LYVE1 (1:100, Abcam) and anti-VEGFR3 (1:25, R&D Systems). DNA was stained with Hoechst. Coverslips were mounted with Prolong Antifade Reagent (Molecular Probes). Images were acquired using a LSM upright 780 confocal microscope.

To assess pulmonary vascular remodeling in mice under chronic hypoxia, lung sections were stained for α-smooth muscle actin and VWF, and ten random fields were examined for 20–80 μm muscular arteries. The external and internal media perimeters of 70–90 muscular arteries were measured using ImageJ and external and internal media radii were calculated using r = perimeter/2π. The medial wall thickness was expressed as (external media radii – internal media radii)/external media radii or as a ratio of medial area to cross sectional area (CSA) using (total vascular area − lumen area)/total vascular area¹⁸. Quantifications were performed by investigators blinded to the experimental groups.

Next-generation sequencing (RNA-seq)

Total RNA was extracted from ~30 mg of lung tissue by using the RNeasy Mini plus Kit (Qiagen) according to the manufacturer’s instructions. RNA quality was analyzed with an Agilent 2100 Bioanalyzer (Agilent Technologies). All RNA samples had RNA integrity numbers (RINs) greater than 9. Enrichment of mRNA was performed by NEBNext® Poly(A) mRNA Magnetic Isolation Module. cDNA libraries were constructed using NEBNext® Ultra™ RNA Library Prep Kit for Illumina®. The cDNA library quality and size distribution were assessed by an Agilent 2100 bioanalyzer. Six libraries with different indices were pooled in a lane and sequenced (50 bp single end read) by Illumina HiSeq 2000/2500 platform at Duke University Genome Sequencing & Analysis Core Facility. Index sequences from raw reads of RNA-seq were trimmed and clean reads were aligned to the mouse genome (mm10) by web-based Galaxy platform using TopHat¹⁹. Differential gene expression was analyzed by CuffDiff and Data were visualized using the R package CummeRbund. The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE110131.

Human samples

Deidentified tissue (whole lung and formalin-fixed, paraffin-embedded) samples were provided by the Pulmonary Hypertension Breakthrough Initiative (PHBI) with local IRB approval. Pulmonary arterial endothelial cells (PAECs) for VEGFR3 and ARRB1 expression analysis were obtained from human lung tissues from either unused, explanted normal donor lungs or explanted lungs from confirmed subjects with PAH undergoing lung transplantation at the Cleveland Clinic²⁰.

Animals

All animal experiments were conducted in compliance with institutional guidelines and were approved by Duke University Institutional Animal Care and Use Committee (Protocol: A175–16-08). Arrb1and Arrb2 KO mice were provided by Dr. Robert J. Lefkowitz and have been described previously^{21, 22}. All animals were maintained on a C57Bl/6j background. For inducing pulmonary hypertension, 8 to 12 weeks old male mice were kept at normal altitude (normoxia, FiO₂ ~ 150 mmHg) or a pressure corresponding to 18000 feet (hypoxia, FiO₂ ~ 75 mmHg) for 4 weeks. . For MAZ51 treatment, mice were intraperitoneally injected 6 days/week for the duration of hypoxia with 150 μl of MAZ51 in 8.5% DMSO/0.5% Carboxymethylcellulose sodium/0.4% Polysorbate-80 at a dose of 10 mg/kg/day.

Hemodynamic Analysis

Right heart catheterization was performed as described²³. Mice were anesthetized with xylazine (2.5 mg/kg) and ketamine (10 mg/kg). A needle is inserted into the trachea to serve as endotracheal intubation, the cannula connected to a volume cycled rodent ventilator on normal air with a tidal volume of 0.25 ml and respiratory rate of 110/min. The right ventricular pressure was recorded using a Millar PE10 catheter via the jugular vein. Mice were then euthanized and the hearts and lungs were flushed with PBS. The right lung was snap frozen in liquid nitrogen for protein and RNA analysis. The left lung was inflated and fixed with 4% paraformaldehyde for immunohistochemistry. Right ventricular hypertrophy was quantified as the ratio of right ventricular to left ventricular and septal weight (RV/LV+S). Measurements were performed by investigators blinded to the experimental groups.

4D Cardiac Micro-CT

We performed 4D high-resolution micro-CT with cardiorespiratory gating as described²⁴. A 25 g C57BL/6 mouse was anesthetized with ketamine (115 mg/kg, 20 mg/ml, 0.17 ml) and diazepam (27 mg/kg, 0.5 mg/ml 0.16 ml). A solid-state pressure transducer on the breathing valve measured airway pressure and electrodes (Blue Sensor, Medicotest, UK) taped to the animal footpads acquired the ECG signal. Both signals were processed with Coulbourn modules (Coulbourn Instruments, Allentown, PA) and displayed on a computer monitor using a custom-written LabVIEW application (National Instruments, Austin, TX). Body temperature was recorded using a rectal thermistor and maintained at 36.5 °C by an infrared lamp and feedback controller system (Digi-Sense®, Cole Parmer, Chicago, IL). A catheter was inserted following a tail vein cut down and used for injection of the contrast agent. Animals were placed in a cradle and scanned in a vertical position. During imaging, anesthesia was maintained with ketamine (0.04 ml) delivered IP every 30 min. A blood pool contrast agent for micro-CT with prolonged vascular residence time, Fenestra™ VC²⁵ (ART Advanced Research Technologies, Saint-Laurent, Quebec, Canada), containing iodine at 50 mg/ml concentration, was used.

Statistics

All graphs and data generated in this study were analyzed using GraphPad Prism 7.04 Software. All the quantitative data is presented as means ± SEM. The statistical significance of differences was determined using Student’s two-tailed t test in two groups, and one-way or two-way ANOVA along with Tukey’s or Sidak’s multiple comparison test in multiple groups. A P value ≤ 0.05 was considered statistically significant.

Results

ARRB1 expression is reduced in human PAH and ARRB1 deletion in mice increases susceptibility to hypoxia-induced PAH.

Given that GPCR signaling is tightly regulated by ARRBs and that alterations in GPCR signaling have been implicated in PAH^{11, 12}, we tested for ARRB expression in PAH. Human IPAH and control lung samples and PAECs isolated from IPAH and controls were tested for ARRB1 and 2 expression at the protein level. This demonstrated that both in lung lysates and in cultured PAECs that ARRB1 levels in IPAH were lower than those in control samples (Figure 1a–d), while levels of ARRB2 were not altered (Figure 1a, b). To test whether such a decrease could contribute to the pathogenesis of PAH, we next tested the effects of Arrb1 knockout on the development of hypoxia-induced PAH in mice. Hypoxia-induced pulmonary vascular remodeling resembles changes in early human PAH, characterized by precapillary artery muscularization with smooth muscle cell hypertrophy and proliferation²⁶. Wide type (WT), Arrb1 and Arrb2 KO male mice were exposed to hypobaric hypoxia (corresponding to 18,000 feet altitude, PAO₂ ~ 75 mmHg) for 4 weeks while littermate controls were maintained in normoxia (FiO₂ = 0.21, PAO₂ ~ 150 mmHg). We found that Arrb1 KO male mice were more sensitive to hypoxia-induced PAH with increased right ventricle (RV) hypertrophy (Figure 1e) and higher RV systolic pressures (RVSPs) (Figure 1f), while Arrb2 KO male mice developed RV hypertrophy comparable to WT mice (Figure 1c, d). Interesting, Arrb1 KO female mice were not more susceptible to hypoxia-induced PAH compared to WT female mice (Supplemental figure 1), which has been in other genetic models of PAH²⁷. Due to this difference, we used male mice for further study. Arrb1 KO mice had worse RV function than WT mice in response to chronic hypoxia (Figure 1g, h), whereas Arrb2 KO mice relatively preserved RV function compared to WT mice (Figure 1g, h). Of note, there were no differences in left ventricular filling pressures between WT and Arrbs KO mice, consistent with intact left ventricular function in all genotypes (Supplemental figure 2). Consistent with the measured hemodynamic changes, hypoxia exposure of Arrb1 KO mice caused a significant increase in the medial wall thickness of small pulmonary arteries compared to WT mice (Figure 1i, j). Taken together, these data implicate ARRB1 as a mediator in the development of PAH.

Figure 1. ARRB1 is downregulated in human IPAH and loss of ARRB1 in mice results in worsened hypoxia-induced PAH.

(a) Western blot of ARRB1 and 2 expression in lungs from patients with IPAH and controls (n = 5 per group) using an anti-ARRB1/2 antibody (A-1, Santa Cruz). (b) ARRB1, but not ARRB2, expression is significantly decreased in human IPAH. Signal is normalized to GAPDH control. Statistical analysis was performed by two-tailed unpaired t test. (c) Western blot of ARRB1 and 2 expression in pulmonary arterial endothelial cells (PAECs) isolated from patients with IPAH and controls (Control: n = 3; IPAH: n = 8) using an anti-ARRB1/2 antibody (A-1, Santa Cruz). (d) ARRB1 expression is significantly downregulated in PAECs from human IPAH. Signal is normalized to GAPDH control. Statistical analysis was performed by two-tailed unpaired t test. (e) Arrb1 KO male mice develop more severe PAH induced by hypobaric hypoxia compared to WT and Arrb2 KO male mice as assessed by (e) RV hypertrophy and (f) RV systolic pressure at cardiac catheterization (n = 15–20 mice per group). (g) Micro-CT images demonstrate RV enlargement in Arrb1 KO male mice exposed to hypoxia with (h) decreased RV fractional shortening compared to WT and Arrb2 KO male mice (n = 3–6 mice per group). (i) Representative images of immunofluorescence for smooth muscle α-actin (α-SMA, red), von Willebrand’s factor (VWF, green) and nuclei (blue) in lung sections from WT, Arrb1 KO and Arrb2 KO mice exposed to hypoxia (scale bar = 50 μm). (i-j) Arrb1 KO mice demonstrate increased vascular medial thickness and media area/cross-sectional area (CSA) compared to WT and Arrb2 KO mice (n = 5 mice per group). Statistical analysis was performed by one-way ANOVA and Tukey’s multiple comparison test. Error bars indicate SEM. *, p < 0.05; #, p < 0.05

The VEGF-C/VEGFR3 signaling pathway is down-regulated in Arrb1 KO mice with hypoxia-induced PAH.

To investigate the mechanisms responsible for the severe PAH in Arrb1 KO mice, we performed lung transcriptome analysis using RNA-seq (Supplemental figure 3). Approximately 90 genes were differentially expressed between WT and Arrb1 KO mice under chronic hypoxia at false discovery rate (FDR) less than 0.01 (Figure 2a), and about 150 differentially expressed genes at FDR less than 0.05 (Supplemental Data 2). Ingenuity Pathway Analysis identified that the pathway that was most significantly reduced in the Arrb1 KO mice was VEGF signaling; common pathways known to be regulated by GPCRs were also reduced (Supplemental Table 1). Within VEGF signaling, transcripts that failed to increase in response to hypoxia in Arrb1 KO mice were primarily involved in VEGFR3 signaling, such as Vegfc (Figure 2b). Similarly, associated target molecules in VEGFR3 signaling, such as lymphatic vessel endothelial hyaluronan receptor 1 (Lyve1), von Willebrand’s factor (Vwf), FK506 binding protein 5 (Fkbp5) and integrin subunit alpha 6 (Itga6) were reduced in Arrb1 KO mice exposed to hypoxia compared to WT mice (Figure 2b and Supplemental Figure 4). To verify the results, we performed RT-qPCR and confirmed that the expression of Vegfc was reduced in Arrb1 KO mice compared with WT mice, but not Vegfb (Figure 2c). As VEGF-C acts primarily through VEGFR3, we next examined whether VEGFR3 signaling was disrupted at the protein level in Arrb1 KO mice under chronic hypoxia compared to WT mice. We found that VEGF-C-induced phosphorylation of VEGFR3 at tyrosine residues 1221 and 1222 was reduced, which has been reported to be important for stimulation of cell proliferation and cell migration^{28, 29}. VEGFR3 also activates downstream kinases Akt and extracellular signal-regulated kinases (Erk)²⁸, while c-Src mediates phosphorylation of VEGFR3 induced by extracellular matrix³⁰. We found that the phosphorylation of Akt at S473 was significantly inhibited in Arrb1 KO mice, but there were no changes in the phosphorylation of Erk, p38 MAPK and Src (Figure 2d, e, Supplemental Figure 5a, b). Accordingly, VEGFR3 and LYVE1 expression were reduced in human IPAH compared to controls (Figure 2f), as shown in a recent study¹⁰.

Figure 2. Deletion of Arrb1 attenuates the VEGF-C/VEGFR3 signaling in mice exposed to hypoxia.

(a) Heat map showing differential expressed genes at false discovery rate (FDR, < 0.01) between WT and Arrb1 KO mice subjected to hypoxia (n = 3 mice per group). Vegfc expression is pointed by a red arrow. (b) Decreased mRNA expression by RNA-seq of Vegfc, Lyve1 and vwf in the lung of Arrb1 KO compared to WT mice exposed to hypoxia. Statistical analysis was performed by CuffDiff. (c) RT-qPCR mRNA expression of Vegfc, Lyve1 and Vegfb in the lungs of WT, Arrb KO, and Arrb2 KO mice subjected to normoxia or hypoxia (n = 3 mice per group). (d) Protein expression in lung lystates from WT, Arrb1 KO, and Arrb2 KO mice (n = 3 mice per group). Quantification and normalization to WT exposed to normoxia demonstrates (e) a significant decrease in VEGFR3 and Akt phosphorylation in Arrb1 KO mice. Statistical analysis was performed by one-way ANOVA and Tukey’s multiple comparison test. (f) Loss of VEGFR3 in IPAH lung sections stained for LYVE1 (green), VEGFR3 (red), and nuclei (Hoechst; blue). Scale bar = 50 μm. (g) Inhibiting VEGFR3 signaling by the VEGFR3 inhibitor MAZ51 exacerbates PAH in WT mice (n = 8 mice per group), but not Arrb1 KO mice (n = 9 mice per group). Statistical analysis was performed by two-tailed unpaired t test. *, p < 0.05; ***, p < 0.001.

To further test if ARRB1 was required for VEGFR3 signaling in vivo, WT and Arrb1 KO mice were injected with VEGFR3 selective agonist VEGF-C (C156S)³¹, phosphorylation of VEGFR3 in lungs was induced in WT mice, but not Arrb1 KO mice (Supplemental figure 6). Moreover, to further confirm that ARRB1-mediated VEGFR3 signaling was important in the development of PAH, WT and Arrb1 KO mice subjected to hypoxia were injected with VEGFR3 inhibitor MAZ51, which blocks the activation of VEGFR3 without blocking VEGF-C-mediated stimulation of VEGFR2³². WT mice injected with MAZ51 developed worse PAH than mice with vehicle (Figure 2g), consistent with the phenotype of mice with genetic deletion of Vegfr3¹⁰. However, this effect was abolished in Arrb1 KO mice (Figure 2g), consistent with MAZ 51 targeting an ARRB1-regulated pathway, i.e., VEGFR3. These findings suggest that ARRB1 plays a role in regulating VEGFR3 signaling and expression in the pathogenesis of human IPAH.

ARRBs are required for VEGFR3-mediated endothelial cell proliferation, migration and angiogenesis.

VEGFR3 is primarily expressed in endothelial cells, where it regulates cell proliferation, angiogenesis and lymphoangiogenesis⁴. Since VEGF-C/VEGFR3 signaling was disrupted in Arrb1 KO mice, we next examined if ARRB1 regulates endothelial cell function in HMVEC-L, which mainly consists of mixed endothelial cells from blood and lymphatic vessels. We found that HMVEC-L cells had a higher expression level of VEGFR3 compared with HUVEC and HEK 293T cells and that ARRB1 was the main isoform expressed in endothelial cells (Figure 3a). Next, the level of ARRB1 or 2 was decreased by siRNAs and confirmed by immunoblotting and qPCR (Figure 3b, c). ARRB1 knockdown resulted in decreased expression of VEGFR3 in HMVEC-L cells at baseline (Supplemental Figure 7a). Knockdown of ARRB1 or 2 inhibited HMVEC-L cell proliferation in complete media (Supplemental Figure 7b) or proliferation induced by VEGF-C after serum starvation (Figure 3d). Similarly, HMVEC-L cell migration in response to VEGF-C was significantly attenuated by suppression of ARRB1 or 2 (Figure 3e, f). Cell migration was also reduced in complete media by ARRB1 or 2 knockdown (Figure 3, E and F and Supplemental Figure 7c, d). Moreover, VEGF-C-induced tube formation was disrupted in HMVEC-L cells transfected with siRNAs against ARRB1 or 2 (Figure 3g, h). These results demonstrate an essential role for ARRBs in VEGF-C-induced endothelial function in vitro.

Figure 3. Knockdown of ARRB1 or 2 inhibits VEGF-C-induced cell proliferation, migration and angiogenesis in HMVEC-L cells.

(a) ARRB1 is the predominant isoform in human microvascular endothelial cells from lung (HMVEC-L) and human umbilical vein endothelial cells (HUVEC). Confirmation of ARRB1 and 2 siRNA knockdown in HMVEC-L by (b) protein expression and (c) mRNA expression. (d) Loss of VEGF-C-induced cell proliferation as quantified in a BrdU assay in HMVEC-L cells transfected with either ARRB1 or 2 siRNA compared to control siRNA-treated cells. (e-f) Loss of VEGF-C-induced cell migration in a scratch assay in HMVEC-L cells transfected with either βarr1 or 2 siRNA compared to control siRNA-treated cells. (g-h) Loss of VEGF-C-induced tube formation in HMVEC-L cells transfected with either ARRB1 or 2 siRNA compared to control siRNA-treated cells. Images were taken 4 hours after seeding and quantified. Representative images are shown. Data represents mean ± SEM of at least 3 independent experiments. Statistical analysis was performed by two-way ANOVA and Sidak’s multiple comparisons test. *, p < 0.05.

ARRB1 is required for VEGF-C-induced phosphorylation of VEGFR3 and downstream signaling to Akt.

To determine if loss of ARRB1 or 2 affects VEGF-C/VEGFR3 signaling, we transfected HMVEC-L cells with siRNAs targeting ARRB1 or 2. Loss of ARRB1 or 2 significantly decreased VEGF-C-induced VEGFR3 phosphorylation (Figure 4a, b). Consistent with this, VEGF-C-induced Akt activation was only impaired by ARRB1 knockdown, but not ARRB2 knockdown (Figure 4c, d), similar to the phenotypes of the Arrb1 and 2 KO mice under chronic hypoxia. Furthermore, we found that knockdown of either ARRB1 or 2 reduced the activating phosphorylation at S1177 on eNOS (Figure 4a, e), which has been shown to be important for endothelial function³³. Notably, not all signaling pathways downstream of VEGFR3 demonstrated this pattern, as ERK phosphorylation was not affected by ARRB knockdown (Figure 4f).

Figure 4. ARRB1 is required for VEGF-C/VEGFR3 signaling.

(a) Representative immunoblot demonstrating loss of VEGF-C-induced VEGFR3 activation in HMVEC-L cells transfected with ARRB1 siRNAs (U: untreated and unstarved.) Quantitation of VEGF-C-induced phosphorylation of (b) VEGFR3, (c-d) Akt, (e) eNOS, and (f) Erk phosphorylation. No significant difference was observed in (f). Data represents mean ± SEM of at least 3 independent experiments. Statistical analysis was performed using two-way ANOVA and Tukey’s multiple comparisons test. (g) Overexpression of adenoviral vectors expressing a constitutively active Akt (myr)) or GFP in HMVEC-L. (h) Representative image demonstrating partial rescue of HMVEC-L tube formation by constitutively active Akt, with quantification (i). Representative images are shown. Data represents mean ± SEM of at least 3 independent experiments. Statistical analysis was performed using one-way ANOVA and Tukey’s multiple comparisons test. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

As loss of ARRB1 disrupted VEGF-C-induced tube formation and downstream Akt activation, we next examined whether activation of Akt by expressing a constitutively active Akt mutant could rescue tube formation. HMVEC-L cells were infected with adenovirus expressing myr-Akt and then transfected with siRNA targeting ARRB1 (Figure 4g). Overexpression of constitutive Akt partially rescued the tube formation defect induced by Arrb1 knockdown (Figure 4g, h). These data are consistent with ARRB1 being essential for VEGF-C-induced VEGFR3 signaling via Akt, contributing to angiogenesis.

ARRB1 interacts with the kinase domain of VEGFR3 and VEGFR3 kinase activity is required for the interaction.

Since ARRBs are multifunctional adapter proteins¹³ and ARRB1 has been reported to bind to the RTK insulin-like growth factor 1 receptor (IGF-1R)³⁴, we considered that ARRB1 might regulate VEGFR3 through a direct interaction. To test this, we performed reciprocal immunoprecipitation using exogenously expressed VEGFR3, ARRB1 and 2. We found that VEGFR3 binds to both ARRB1 and 2 (Figure 5a, b). To confirm the interaction between VEGFR3 and ARRB1 at endogenous levels, we immunoprecipitated VEGFR3 in HMVEC-L cells stimulated with VEGF-C, and found that VEGFR3 bound to ARRB1 after VEGF-C stimulation (Figure 5c). Similarly, we detected the interaction between VEGFR3 and ARRB1 in HUVEC cells (Figure 5d). Next, we tested whether VEGFR3 kinase activity was required for the VEGFR3 and ARRB1 interaction. We constructed two kinase-dead mutants, VEGFR3 Y1068F²⁸ and VEGFR3 R1046P²⁹ (Figure 5e). VEGF-C treatment increased the interaction of WT VEGFR3 with ARRB1 and this interaction was significantly impaired in both kinase-dead mutants (Figure 5f). To identify which region of VEGFR3 is required for the binding to ARRB1, we constructed a series of C-terminal truncation mutants: VEGFR3 (1–1297, 1–1173, 1–844) which correspond to the VEGFR3 short isoform, a loss of the C-terminal tail distal to the kinase domain, and a loss of both the kinase domain and the distal C-terminal tail, respectively (Figure 5e). Only VEGFR3 (1–844) markedly reduced the binding to ARRB1 compared to WT (Figure 5g), consistent with the kinase domain of VEGFR3 being required for VEGFR3 binding to ARRB1. These results demonstrate that ARRB1 regulates VEGFR3 signaling through direct binding to VEGFR3 that requires kinase activity.

Figure 5. ARRB1 binds to the kinase domain of VEGFR3 and the interaction is kinase activity dependent.

(a) HA-tagged ARRB1 and 2 interact with FLAG-tagged VEGFR3 when overexpressed in HEK 293T cells. (b) Similarly, both FLAG-tagged ARRB1 and ARRB2 interact with unlabeled VEGFR3 when overexpressed in HEK293T cells. (c) ARRB1 interacts with VEGFR3 at baseline, with an increase in the association forty minutes after VEGF-C stimulation at endogenous levels in HMVEC-L and at baseline in (d) HUVEC cells. (e) Schematic representation of VEGFR3. Amino acid positions in the intracellular domain are indicated by numbers. Short splicing isoform is indicated by an arrow. Mutations for kinase dead mutants are labeled. (f) The ligand-dependent interaction of VEGFR3 and ARRB1 is lost in kinase-dead VEGFR3 mutants R1046P and Y1068F when overexpressed in HEK293T cells. (g) Overexpression of FLAG-tagged VEGFR3 truncation mutants with HA-tagged ARRB1 in HEK293T cells demonstrates that the ARRB1:VEGFR3 interaction requires the kinase domain. Representative immunoblots are shown from at least three independent experiments.

ARRB1 limits VEGF-C-induced VEGFR3 internalization and degradation.

ARRBs have been shown to act as endocytic adaptors and mediate trafficking of a number of surface receptors³⁵, including the RTK IGF-1R³⁶. Since ARRB1 interacts with VEGFR3, we examined whether ARRB1 regulates VEGFR3 internalization. As assessed by receptor biotinylation, surface VEGFR3 decreased significantly faster in HMVEC-L cells with ARRB1 knockdown (Figure 6a, b). Consistent with this, receptor internalization monitored by a bystander BRET-based membrane dissociation assay, where VEGFR3-RlucII internalization leads to a reduction of the net BRET ratio from myr-palm labeled mVenus (Figure 6c), demonstrated that overexpression of ARRB1 decreased VEGFR3 internalization (Figure 6d). Moreover, HMVEC-L cells treated with siARRB1 resulted in a significantly faster reduction of VEGFR3 levels than in control cells at 30 or 45 minutes after VEGF-C treatment in the presence of the protein translation inhibitor cycloheximide (Figure 6e, f). These findings are consistent with ARRB1 inhibiting VEGF-C-induced VEGFR3 internalization and degradation.

Figure 6. ARRB1 inhibits VEGF-C-induced VEGFR3 internalization and degradation.

(a) Representative immunoblot of surface VEGFR3 as probed in a receptor biotinylation assay in HMVEC-L cells. (b) Quantification demonstrating that βarr1 siRNA knockdown significantly enhances receptor internalization (surface VEGFR3 expression normalized to 0 time point for each condition (n = 3)). (c) Illustration of a bystander BRET-based sensor configurations for monitoring VEGFR3 internalization from the plasma membrane (PM). (d) ARRB1 overexpression inhibits VEGFR3 internalization in HEK 293T cells 60 minutes after stimulation with VEGF-C. Net BRET ratios were the means ± SEM. representative of at least three independent experiments. (e) VEGFR3 degrades faster in HMVEC-L cells transfected with siRNA to ARRB1 compared to control siRNA-treated cells in response to VEGF-C stimulation in the presence of CHX. (f) Densitometric quantification of VEGFR3 expression normalized to 0 time point for each condition (n = 5). Statistical analysis was performed by two-way ANOVA and Sidak’s multiple comparisons test. *, p < 0.05. (g) Model for ARRB1 regulation of VEGFR3-mediated endothelial function. Upon VEGF-C binding to VEGFR3, ARRB1 is recruited to VEGFR3 kinase domain, which is required for receptor activation and signaling to Akt and eNOS. ARRB1 also inhibits the endocytosis and subsequent degradation of VEGFR3.

Discussion

We report a novel role for ARRB1 in mediating VEGFR3 signaling through a direct interaction that is required for downstream Akt activation. In human IPAH, we found that ARRB1 expression was reduced and was associated with a loss of VEGFR3 expression. In mice, the ablation of Arrb1 led to the development of worse PAH, reduced VEGFR3 phosphorylation and impaired downstream Akt activity. Moreover, knockdown of ARRB inhibited VEGF-C-induced cell proliferation, migration, in vitro tube formation, and the activating phosphorylation of VEGFR3, Akt, and eNOS. Loss of ARRB1 also promoted VEGF-C-induced internalization and degradation. Together, our findings are consistent with a model in which ARRB1 inhibits VEGF-C-induced VEGFR3 internalization and degradation while promoting VEGFR3 signaling (Figure 6g). This model is consistent with our observations in human IPAH and Arrb1 KO mice that loss of Arrb1 is associated with PAH.

ARRB1 and 2 are ubiquitously expressed multifunctional proteins best known for regulating GPCRs by acting as scaffolds that mediate receptor desensitization, internalization, trafficking, and signaling to kinases and other cascades³⁵. In addition to regulating GPCR signaling, emerging evidence indicates ARRB can interact and mediate signaling of RTKs, such as IGF-1R^{34, 36}. At IGF-1R, binding of ARRB1 causes ligand-dependent degradation of the receptor and generates prolonged ERK signaling. Conversely, ARRB2 counteracts β-arrestin1 by promoting degradation of a ligand-unbound IGF-1R and generates transient ERK activation, while protecting the receptor against agonist-induced degradation³⁷. Similar to the role of ARRB2 at the IGF-1R, we found that ARRB1 inhibited VEGF-C-induced internalization and degradation of VEGFR3 in HMVEC-L cells. Also, ARRB1 prevents insulin-induced insulin receptor substrate 1 (IRS-1) ubiquitination and degradation by competing with an endogenous E3 ligase Mdm2 for IRS-1³⁸. Therefore, we speculate that ARRB1 could compete with an E3 ligase for VEGFR3, thereby delaying VEGF-C-dependent receptor degradation through recycling pathways.

In HMVEC-L cells, knockdown of either ARRB1 or ARRB2 suppressed VEGF-C-induced VEGFR3 phosphorylation, but phosphorylation of Akt was only impaired in cells with ARRB1 knockdown. Due to the high sequence and structural similarity between the two ARRBs, they are likely partially functionally redundant. Nonetheless, distinct functional specialization of the two ARRB isoforms in the regulation of GPCRs³⁹ and IGF-1R³⁷ in terms of desensitization, internalization, signaling and their physiological effects. For example, the niacin receptor GPR109A-induced ERK activation is impaired by knockdown of either of the ARRB isoforms in cultured cells, whereas only ARRB1 is required for phosphorylation of the downstream effector cytosolic phospholipase A2 and activation of arachidonic acid release⁴⁰. Moreover, Arrb1 KO mice exhibit reduced nicotinic acid-induced flushing response, while Arrb2 KO mice show a similar level of response as the WT mice⁴⁰. Also, the functional divergence of ARRB isoforms could occur in different cell types, as we found that ARRB1 is the predominant isoform in endothelial cells. Consequently, VEGFR3 might preferentially recruit ARRB1 for signaling to Akt. Overall, the functional divergence of ARRB isoforms adds another layer in the regulation of VEGF-C/VEGFR3 signaling that requires further study.

Endothelial dysfunction is central to the pathogenesis of PAH, through complex mechanisms that include abnormal endothelial cell proliferation, concurrent neoangiogenesis and the unbalanced production of various endothelial vasoactive mediators, such as NO, prostacyclin and endothelin-1². While our findings are consistent with VEGFR3 in vascular endothelial cells contributed to the development of PAH, we cannot rule out a potential contribution from VEGFR3 signaling in the lymphatics. The VEGFR3 signaling pathway is one important regulator of endothelial function, as endothelial-specific deletion of VEGFR3 results in worse PAH¹⁰. Similarly, endothelial NO synthase (eNOS), which produces NO in endothelium, has a crucial role in the pathogenesis of PAH in animal models³³. While overexpression of eNOS in mice prevents the development of PAH in hypoxia⁴¹, genetic deletion of eNOS leads to severe PAH in mice under mild hypoxia⁴². Moreover, Akt phosphorylates eNOS at serine 1177, resulting in increased NO production⁴³. Here, Akt activation was impaired in Arrb1 KO mice and phosphorylation of eNOS at serine 1177 was diminished in response to VEGF-C in HMVEC-L cells treated with ARRB1 siRNA. Interestingly, phosphorylation of eNOS at serine 1177 was also attenuated in ARRB2 siRNA HMVEC-L in response to VEGF-C, suggesting loss of ARRB2 impacts eNOS activation through an Akt-independent manner. We also found that VEGF-C/VEGFR3 signaling pathway was impaired by genetic deletion of ARRB1 in mice with worse PAH. Moreover, ARRB1 knockdown inhibited HMVEC-L cell proliferation, migration and in vitro tube formation. Collectively, these data indicate ARRB1 regulate endothelial function through the VEGF-C/VEGFR3 signaling pathway. Our findings suggest that ARRB1 mediates VEGFR3-mediated eNOS activation through activation of Akt and loss of ARRB1 results in endothelial dysfunction due to inactivation of eNOS.

In summary, we have identified a novel role for ARRB1 in endothelial function through VEGFR3 signaling that partially explains the ARRB1 deficiency we found in IPAH patients. These results demonstrate that dysregulation of ARRB1 has an important role in endothelial function through VEGFR3 in PAH. Notably, abnormal ARRB expression has been implicated in other diseases and animal models, for example, cancer^{44, 45}, diabetes mellitus⁴⁶, and colitis⁴⁷. Therefore, therapies that increase ARRB1-mediated VEGFR3 signaling could be useful in the treatment of PAH and other vascular diseases.

Clinical Perspective.

What Is New?

• We discovered a novel mechanism for regulation of the vascular endothelial growth factor receptor 3 (VEGFR3) by the adapter protein β-arrestin 1 (ARRB1).

• ARRB1 was required for VEGFR3 signaling to downstream effectors, such as Akt and eNOS that promote endothelial function as assessed by proliferation and tube formation of vascular endothelial cells.

• ARRB1 expression was reduced in endothelial cells from patients with IPAH and Arrb1 KO mice developed severe pulmonary hypertension that was associated with a loss of VEGFR3 signaling.

What Are the Clinical Implications?

• Our findings indicate that common therapeutic targets, such as G-protein coupled receptors (GPCRs) and VEGFRs, can potentially coordinate their signaling through ARRBs to regulate endothelial function.

• ARRB1-mediated VEGFR3 signaling may play an important role in the development of pulmonary vascular disease and may also be important in other vascular diseases associated with endothelial dysfunction, such as coronary artery disease and peripheral vascular disease.