Endoglin deficiency impairs VEGFR2 but not FGFR1 or TIE2 activation and alters VEGF-mediated cellular responses in human primary endothelial cells

Abstract

Hereditary hemorrhagic telangiectasia (HHT) is a genetic disease characterized by vascular dysplasia. Mutations of the endoglin (ENG) gene that encodes a co-receptor of the transforming growth factor β1 signaling pathway cause type I HHT. ENG is primarily expressed in endothelial cells (ECs), but its interaction with other key angiogenic pathways to control angiogenesis has not been well addressed. The aim of this study is to investigate ENG interplay with VEGFR2, FGFR1 and TIE2 in primary human ECs. ENG was knocked-down with siRNA in human umbilical vein ECs (HUVECs) and human lung microvascular ECs (HMVEC-L). Gene expression was measured by RT-qPCR and Western blotting. Cell signaling pathway activation was analyzed by detecting phosphor-ERK and phosphor-AKT levels. Cell migration and apoptosis were assessed using the Boyden chamber assay and the CCK-8 Kit, respectively. Loss of ENG in HUVECs led to significantly reduced expression of VEGFR2 but not TIE2 or FGFR1, which was also confirmed in HMVEC-L. HUVECs lacking ENG had significantly lower levels of active Rac1 and a substantial reduction of the transcription factor Sp1, an activator of VEGFR2 transcription, in nuclei. Furthermore, VEGF- but not bFGF- or angiopoietin-1-induced phosphor-ERK and phosphor-AKT were suppressed in ENG deficient HUVECs. Functional analysis revealed that ENG knockdown inhibited cell migratory but enhanced anti-apoptotic activity induced by VEGF. In contrast, bFGF, angiopoietin-1 and −2 induced HUVEC migration and anti-apoptotic activities were not affected by ENG knockdown. In conclusion, ENG deficiency alters the VEGF/VEGFR2 pathway, which may play a role in HHT pathogenesis.

1. Introduction

Hereditary hemorrhagic telangiectasia (HHT), also known as Osler-Weber-Rendu syndrome, is a genetic disorder characterized by vascular dysplasia. Malformations of various blood vessels in HHT, i.e., mucosal/dermal telangiectasia, and pulmonary, cerebral and hepatic arteriovenous malformations (AVMs) can result in recurrent epistaxis, gastrointestinal bleeding and stroke.^{1, 2} Mutations in the endoglin (ENG), the activin receptor-like kinase 1 (ACVRL1 also known as ALK1) and the mothers against decapentaplegic homolog 4 (SMAD4) genes are linked to HHT.^3–6 Additionally, genetic variations of the growth differentiation factor 2 and the RAS p21 protein activator 1 have been identified in patients with a similar or overlapping phenotype to HHT.^{7, 8} The most common forms of HHT, i.e., type 1 and type 2 HHT (HHT1 and HHT2) that account for approximately 85% of all HHT cases are caused by ENG and ACVRL1/ALK1 mutations, respectively.^2–4 Both ENG and ACVRL1 encode putative receptors of the transforming growth factor-beta 1 (TGF-β1) signaling pathway that consists of two types of transmembrane serine-threonine kinase receptors, i.e., type I and type II receptors. The type II receptor binds TGF-β1 ligands, leading to the activation/phosphorylation of type I receptor. Subsequently, the type I receptor activates transcription factors SMADs to regulate gene expression, eliciting eventually a variety of biological effects including regulation of angiogenesis.^9–12

Two important type I receptors, namely ALK1 and ALK5, have long been recognized to play distinct roles in regulating cellular processes in endothelial cells (ECs).¹³ Signal transduction via ALK1 and SMAD1/5/8 induces EC proliferation and migration, while the ALK5 pathway involves the activation of SMAD2/3 leading to EC quiescence.¹³ The common mediator SMAD4 plays a central role in ALK1 and ALK5 pathways. It forms heteromeric complexes with either SMAD1/5/8 or SMAD2/3 which are translocated into the nucleus to regulate gene expression.¹⁴ ENG, a co-receptor of the TGF-β1 pathway, takes part in angiogenesis by cooperating with ALK1.^{15, 16} Eng−/− mice die in utero without proper formation of mature blood vessels in the yolk sac, while Eng+/− mice have various defects in angiogenesis exhibiting dilated and weak-walled vessels, telangiectases or recurrent nosebleeds typical of HHT.^17–19

Multiple signaling pathways are involved in angiogenesis, a complex process by which new vessels form from pre-existing vasculature. During early angiogenic sprouting, the vascular endothelial growth factor A/vascular endothelial growth factor receptor 2 (VEGF-A/VEGFR2) pathway transduces signals to promote ECs at the tip of a sprout to migrate to form filopodial extensions, and stimulate stalk cells that follow the tip cells to proliferate to form vascular lumen.^20–22 Vegfr2 knockout mice lack ECs, do not form blood vessels, and die in utero. ²³ In contrast, the EC-specific receptor tyrosine kinase TIE2, also known as TEK and its agonist angiopoietin-1 (ANGPT-1) have been shown to be required for the late stage of angiogenesis, i.e., vessel remodeling and stabilization that occur subsequent to VEGF-A/VEGFR2 actions. Mice with Angpt-1 or Tie2 knockout display a similar vascular phenotype: they have similar EC numbers compared with wildtypes but lack proper hierarchical vasculature resulting from poor vessel branching and organization.^{24, 25} The fibroblast growth factor (FGF) family consists of over 20 members,^{26, 27} with FGF2 (basic FGF, bFGF) being identified first as a potent angiogenic growth factor.²⁸ The biological effects of FGFs are mediated through four receptor tyrosine kinases, namely FGFR1, 2, 3 and 4. In endothelial cells, the predominant FGFR is FGFR1,^{29, 30} through which FGF2 induces EC migration and proliferation.^{31, 32} Mouse embryos lacking Fgfr1 or Fgf2 exhibit less vessel branching and disorganized vasculature in the yolk sac reminiscent to some extent of those seen in Angpt-1 or Tie2 knockout mice.^{32, 33} Although the causative role of ENG mutations for HHT has been established,³ ENG deficiency-mediated molecular events that contribute to vascular malformations remain to be fully elucidated, and interaction of ENG with other key signaling pathways to govern angiogenesis is not well addressed. The aim of this study is to investigate ENG interplay with VEGFR2, FGFR1 and TIE2 in primary human ECs.

2. Material and Methods

2.1. Materials

Primary human umbilical vein endothelial cells (HUVEC), primary human lung microvascular endothelial cells (HMVEC-L), endothelial basal culture medium EBM-2, fetal bovine serum (FBS), and the BulletKit containing EC growth factors VEGF, IGF1 and FGF were purchased from Lonza (Walkersville, MD, USA). Control small interfering RNA (siRNA), ENG siRNA, Lipofectamine siRNA transfection reagent RNAiMax, RPTI-MEM reduced medium and TaqMan Gene Expression PCR assays for ENG, VEGFR2, FGFR1, TIE2 and beta-actin were obtained from Thermo Fisher Scientific (Burlington, ON, Canada). miRNeasy Mini Kit and Omniscript RT Kit were purchased from Qiagen (Germantown, MD, USA). Growth factors used for cell treatment included: human VEGF₁₆₅, ANGPT-1, ANGPT-2 from R&D Systems (Minneapolis, MN, USA) and bFGF from Novus Biologicals (Centennial, CO, USA). Cell Fractionation Kit (cat# 9038) and Active Rac1 Detection Kit (cat# 8815) were obtained from Cell Signaling Technology (Danvers, MA, USA). Primary antibodies used in this study were as follows: Endoglin (cat# 14606S, Cell Signaling Technology); VEGFR2 (cat# 9698S, Cell Signaling Technology); VEGFR1 (cat# 2893, Cell Signaling Technology); phospho-ERK (cat# 9106s, Cell Signaling Technology); phospho-AKT (cat# 9271S, Cell Signaling Technology); AKT (cat# 9272S, Cell Signaling Technology); Sp1 (cat# 9389, Cell Signaling Technology); Rac1 (cat# 11894S, Cell Signaling Technology); Histone H3 (cat# 9715, Cell Signaling Technology); NF-kB p65 (cat# 6956, Cell Signaling Technology); GAPDH (cat# 60004–1-Ig, Proteintech, Rosemont, IL, USA); ERK1/2 (cat# 686902, Biolegend, San Diego, CA, USA); and beta-actin (cat# A5316, Sigma-Aldrich Canada Co., Oakville, ON, Canada). Fluorescence labeled IRDye 800CW goat anti-rat secondary antibodies, IRDye 800CW goat anti-rabbit secondary antibodies and IRDye 680CW goat anti-mouse secondary antibodies were purchased from LI-COR Corporate (Lincoln, NE, USA).

2.2. Cell culture

HUVEC and HMVEC-L were maintained in EGM-2 medium (EBM-2 medium supplemented with 5% FBS and the BulletKit) according to the manufacturer’s instructions. Cells ≤ the 5^th passage were used for all experiments.

2.3. SiRNA transfection

Control or ENG siRNA was introduced into ECs using the Lipofectamine RNAiMax reagent. Cells were seeded into 6-well plates (1.5×10⁵/well) and cultured for 24 hours. For each well of cells, siRNA/Lipofectamine complex was prepared as follows: 3 μL of 10 μM siRNA and 5 μL of Lipofectamine reagent were diluted in 250 μL RPTI-MEM reduced medium respectively; the diluted RNA and Lipofectamine reagent were mixed and incubated at room temperature for 15 min to allow the formation of siRNA/Lipofectamine complex. Subsequently, the complex was added into cells that had been replenished with 2.5 mL of fresh EGM-2 medium. Cells were cultured for another 48 hours for gene expression analysis, signaling pathway activation measurement (phosphorylated protein detection), and cell migration assay which were carried out as described below.

2.4. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted using a miRNeasy Mini Kit according to the manufacturer’s instructions, and quantified with a spectrophotometer. Reverse transcription (RT) was done using an Omniscript RT Kit. The RT reaction components in a total volume of 20 µL included: 2 µL of 5 mM dNTPs, 5 µL of random primer (300 ng/ml), 2 µL of 10x reaction buffer, 1 µL of Reverse Transcriptase (4 units), 1 µl of RNase inhibitor (1 unit), 0.5 µg of total RNA, and appropriate volume of nuclease-free ddH₂O used to adjust the volume. RT was completed by incubation at 37°C for 1 hour followed by enzyme inactivation at 65°C for 15 minutes. One µL of RT product was used for qPCR which was done in a total of 10 µL reaction volume using the TaqMan Gene Expression PCR assay kit according to the manufacturer’s instructions. qPCR was run as follows: 50 °C for 2 min, 95 °C for 10 min, and 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Normalization against β-actin (reference gene) was carried out to determine the relative levels of target genes using the formula: 2^{−(target CT- reference CT)}.

2.5. Western blotting

Cellular protein was prepared, separated by SDS-PAGE and electrically transferred onto the nitrocellulose membrane as described elsewhere.³⁴ The membranes were blocked at room temperature for 1 hour in TBS-T buffer (50 mM TrisHCl, 150 mM NaCl, pH 7.5, 0.1% Tween-20) containing 5% skimmed milk (for the detection of phosphor-proteins, TBS-T buffer containing 2% BSA was used to block non-specific binding) followed by incubation with the primary antibodies (1:1000 dilution for all antibodies) at 4 °C overnight. After 3 washes with TBS-T buffer, the membranes were incubated at room temperature in the dark for 1 hour with appropriate secondary antibodies (1:5000 dilution). The membranes were washed 3 times with TBS-T buffer, and the specific protein bands were visualized using the Odyssey fluorescence imaging system (LI-COR Corporate). Densitometry analyses were performed using the Empiria Studio Software (LI-COR Corporate). Beta actin or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control.

2.6. Growth factor treatment and measurement of phosphor-AKT (p-AKT) and phosphor-ERK (p-ERK)

After transfection was done in 6-well plates as described above, cells were cultured for 48 hours followed by 18 hours of serum starvation.^{34, 35} Cell treatment was carried out at 37 °C for 15 min with VEGF-A (100 ng/mL), bFGF (100 ng/mL) or ANGPT-1 (200 ng/mL) prepared in EBM-2 medium. EBM-2 medium alone served as control. After treatment, cells were washed once with cold PBS buffer and cellular protein was prepared as described elsewhere.³⁴ Phosphorylated AKT and p-ERK proteins were measured with Western blotting. After detection of p-AKT or p-ERK, the membrane was stripped and re-probed with anti-AKT or anti-ERK antibodies.

2.7. Cytoplasmic and nuclear protein extraction

Cytoplasmic and nuclear protein extraction was done at 4 °C using the Cell Fractionation Kit (Cell Signaling Technology) according to the manufacturer’s instructions. Briefly, HUVECs were trypsinized, washed with PBS buffer and pelleted. The cell pellet was re-suspended in Cytoplasm Isolation Buffer (CIB) (100 μL of CIB for every 1 million cells). After incubation on ice for 5 minutes, cells were centrifuged at 500 g for 5 min and the supernatant (cytoplasmic fraction) was collected. The pellet was re-suspended in Membrane Isolation Buffer (MIB) (100 μL of MIB for every 1 million cells) followed by incubation on ice for 5 minutes. After centrifugation at 8,000g for 5 minutes, the supernatant was removed, the pellet was re-suspended in Nucleus Isolation Buffer (NIB) (50 μL of NIB for every 1 million cells) and sonicated. This was the nuclear fraction. NF-kB and Sp1 in both cytoplasmic and nuclear fraction were then analyzed by Western blotting as described above. GAPDH and histone H3 were used as an internal control for cytoplasmic and nuclear fraction, respectively.

2.8. Affinity precipitation of active Rac1

Active Rac1 pull-down was performed using the Active Rac1 Detection Kit (Cell Signaling Technology) according to the manufacturer’s instructions. Five hundred µg of total protein was used for the purification of active Rac1 which was then assayed by Western blotting as describe above. Fluorescence intensity of active Rac1 was compared between ENG deficient and control HUVECs.

2.9. Boyden chamber cell migration assay

HUVEC transfection in 6-well plates was done as described above. Transfected cells were cultured for 48 hours followed by serum starvation for 16 hours. Subsequently, cells were detached and suspended in EBM-2 medium containing 0.1% BSA (migration medium) at a density of 1×10⁶ cells/mL. Cell migration was performed using the Transwell Boyden Chamber apparatus (BD Biosciences, Mississauga, ON, Canada) as described elsewhere.³⁶ Briefly, 0.5 mL of HUVEC suspension was added into the top chamber while 500 μL of migration medium containing growth factor (100 ng/mL VEGF₁₆₅, 100 ng/mL bFGF, or 200 ng/mL ANGPT-1) or 500 μL of migration medium alone as control was added into the lower chamber. After incubation at 37 °C for 4 hours, cells on the topside of the insert were removed with a cotton swab, and migrated cells present on the underside of the insert were fixed and stained using Diff Quik (Thermo Fisher Scientific) and visualized by light microscopy. The average number of migrated cells from 5 randomly selected high-power fields (HPF, 20x) were calculated and compared.

2.10. Assessment of VEGF-induced anti-apoptotic effect

HUVECs were seeded into 96-well plates (5×10³/well) and cultured for 24 hours followed by transfection that was done in triplicate with ENG or control siRNA as follows: 500 μL of the siRNA/Lipofectamine complex prepared as described above was mixed with 2.5 mL of EGM-2 medium, and then 200 μL of the mix was added into each well of cells after spent medium was removed. Cells were cultured for another 48 hours, the spent medium containing siRNA/Lipofectamine complex was removed, and 100 µL of EBM-2 medium containing growth factor (100 ng/mL VEGF₁₆₅, 100 ng/mL bFGF, 200 ng/mL ANGPT-1 or 800 ng/ mL ANGPT-2) or 100 µL of EBM-2 medium alone serving as control was added into each well. Cells were maintained in a cell culture incubator for 24 hours followed by one wash with pre-warmed (37 °C) PBS buffer. Subsequently, 50 µL of EBM-2 medium containing 5 µL of Cell Counting Kit 8 reagent (cat# 96992, Sigma-Aldrich Canada Co.) was added into each well, and incubated at 37 °C for 4 hours. OD450 was measured using a Spectramax M5 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA). The ratio of the OD450 value of VEGF-treated cells over that of non-VEGF treated cells represents the cell viability.

2.11. Statistical analyses

All data were expressed as Mean ± SD. Cell viability data and variables between two groups were analyzed with the student’s t test while the rest of the data were analyzed with the ANOVA with post-hoc Tukey test using the GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA). Analyses of supplementary data were indicated in the supplementary figures. A p-value less than 0.05 was considered statistically significant.

3. Results

Transfection with ENG siRNA but not control siRNA substantially suppressed ENG mRNA and protein expression in HUVECs and HMVEC-L (Figure 1). Silencing of ENG led to a marked reduction in mRNA levels of VEGFR2 but not FGFR1 or TIE2 in HUVECs and HMVEC-L (Figures 2). Substantially reduced VEGFR2 protein expression was also detected in ENG deficient ECs (Figure 3). Using an anti-VEGFR1 antibody that does not cross-react with VEGFR2 or VEGFR3 for Western blotting analysis, we showed that knockdown of ENG did not affect VEGFR1 expression (Supplementary Figure 1).

Figure 1.

ENG siRNA but not control siRNA inhibited ENG expression in primary human ECs. Panels A and B show the RT-qPCR results in HUVECs and HMVEC-L, respectively. C and D are Western blotting data for ENG. * p < 0.001 compared with non-transfected (Non) or control siRNA transfected (Ctrl) samples (n = 3–5 indicates the number of independent experiments).

Figure 2.

VEGFR2, FGFR1 and TIE2 mRNA expression detected by RT-qPCR. As shown in panels A and B, VEGFR2 mRNA levels in HUVECs and HMVEC-L transfected with ENG siRNA were significantly reduced compared with non-transfected (Non) or control siRNA (Ctrl) transfected cells. In contrast, FGFR1 and TIE2 mRNA levels were similar among all groups of cells (C-F). # p < 0.05 compared with non-transfected or control siRNA transfected cells.

Figure 3.

VEGFR2 protein levels measured by Western blotting analysis. Representative Western blots are shown in panels A and C as indicated. Panels B and D show the relative VEGFR2 protein levels in HUVECs and HMVEC-L respectively. ECs transfected with ENG siRNA produced significantly lower levels of VEGFR2. * p < 0.05 compared with non-transfected or control siRNA (Ctrl) transfected cells.

Signal activation testing showed that VEGF₁₆₅-induced phosphorylation of AKT and ERK was substantially repressed in ENG deficient HUVECs (Figure 4), while bFGF or ANGPT-1 mediated AKT and ERK activations were not affected after ENG knockdown (Figure 5, and ANGPT-1 data not shown). When cells were cultured in complete EGM-2 medium, phosphor-ERK levels in control and ENG siRNA transfected HUVECs were not significantly different (Supplementary Figure 2).

Figure 4.

VEGF mediated activation of ERK and AKT in HUVECs. As shown in this figure, VEGF treatment augmented p-ERK (a blot is shown in panel A) and p-AKT (a blot is shown in panel C) in non-transfected and control siRNA transfected cells but p-ERK and p-AKT levels in these 2 groups of cells are not significantly different. In contrast, p-ERK (panel B) and p-AKT (panel D) levels were markedly reduced in HUVECs lacking ENG. # p < 0.05 compared with non-transfected or control siRNA transfected cells.

Figure 5.

bFGF mediated activation of ERK and AKT in HUVECs. As shown in this figure, bFGF treatment augmented p-ERK (a blot is shown in panel A) and p-AKT (a blot is shown in panel C) in control and ENG siRNA transfected cells, but p-ERK and p-AKT levels in these 2 groups of ECs are not significantly different (panels B and D respectively).

As shown in Figure 6, ENG knockdown in HUVECs did not affect Rac1 expression as determined by Western blotting (panels A and B). However, active Rac1 levels in HUVECs transfected with ENG siRNA were significantly lower than those in control HUVECs (Figures 6C and 6D). Measurement of NF-kB and Sp1 shows that ENG deficiency did not result in marked changes in NF-kB in both cytoplasmic and nuclear fractions (Figure 7A–D as indicated). In contrast, while Sp1 in cytoplasmic fraction was undetectable (data not shown), significantly lower levels of Sp1 in nuclei of HUVECs lacking ENG were shown by Western blotting (Figure 7E–F).

Figure 6.

Total Rac1 and active Rac1 measured by Western blotting analysis. As shown in panels A and B, ENG knockdown did not affect total Rac1 levels in HUVECs. In contrast, significant reduction of active Rac1 was observed in HUVECs transfected with ENG siRNA (panels C and D).

Figure 7.

Measurement of NF-kB and Sp1 by Western blotting analysis. Cytoplasmic NF-kB levels in ENG deficient HUVECs were similar to those of control cells (panels A and B). NF-kB levels in nuclear fraction were also similar between the 2 groups of cells (panels C and D). While cytoplasmic Sp1 was not detectable (data not shown), ENG knockdown led to significant reduction of Sp1 in nuclei (panels E and F).

We also examined the cellular responses to growth factor stimulation in ENG deficient HUVECs. Cell migration was assessed with the Boyden Chamber migration assay, and HUVECs lacking ENG showed significantly lower migration capacity towards VEGF₁₆₅ compared with control cells (Figure 8). Additionally, VEGF₁₆₅ treatment resulted in a lower level of cell apoptosis under serum starvation in ENG deficient HUVECs (Figure 9). In contrast, there were no significant differences in migration towards bFGF and ANGPT-1 or anti-apoptotic activity induced by bFGF, ANGPT-1 and ANGPT-2 between control and ENG siRNA transfected HUVECs (Supplementary Figures 3 and 4).

Figure 8.

ENG deficient HUVECs exhibited reduced migration towards VEGF. Top panels show the migrated cells on the underside of the Boyden chamber insert as indicated. The number of migrated HUVECs with ENG silencing was markedly lower than that of control siRNA transfected HUVECs (lower graph). ENG = ENG siRNA transfection; and Ctrl = control siRNA transfection (data were obtained from 4 independent experiments). * p < 0.05 compared with control siRNA transfected cells treated with VEGF.

Figure 9.

ENG deficient HUVECs, under serum starvation, had enhanced cell viability in the presence of VEGF. As shown in this figure, HUVECs lacking ENG exhibited a significantly higher level of cell viability than control siRNA transfected HUVECs. ENG = ENG siRNA transfection; and Ctrl = control siRNA transfection (data were obtained from 5 independent experiments). * p < 0.05 compared with control siRNA transfected cells treated with VEGF.

4. Discussion

In the present study we examined ENG interplay with VEGFR2, FGFR1 and TIE2 in primary human ECs, and reported the following findings: 1) lack of ENG resulted in dramatic reduction in VEGFR2 but not FGFR1 or TIE2 expression in HUVECs and HMVEC-L, 2) decrease of VEGFR2 led to suppression of VEGF-induced activation of ERK and AKT in HUVECs, 3) HUVECs lacking ENG had significantly lower levels of active Rac1 and a substantial reduction of the transcription factor Sp1 in nuclei, 4) bFGF- or ANGPT-1-induced activation of ERK and AKT was not affected by ENG knockdown, 5) HUVEC migration in response to VEGF but not bFGF or ANGPT-1 was inhibited after loss of ENG, and 6) VEGF-induced anti-apoptotic effect was enhanced in ENG deficient HUVECs, while no marked differences were detected in bFGF, ANGPT-1 or ANGPT-2 induced anti-apoptotic activities between control and ENG depleted HUVECs.

TIE2 and FGFR1 are necessary for the late stage of angiogenesis.^{25, 32} ANGPT-1, the agonist of TIE2, exhibits no proliferative but migratory and anti-apoptotic effects on ECs.³⁷ ANGPT-2, another ligand for TIE2, was initially identified as an antagonist of ANGPT-1. Although ANGPT-2 at a wide dose range (0.1–1000 ng/mL) does not exert any chemotactic properties³⁷, it plays a protective role in EC apoptosis at a high concentration (800 ng/mL).³⁸ Through activating FGFR1, bFGF stimulates EC proliferation and migration.^{31, 32} In this study, we did not observe an effect of ENG on TIE2 and FGFR1 expression and activation. Furthermore, knockdown of ENG did not affect cell migration and anti-apoptotic activities induced by bFGF, ANGPT-1 or ANGPT-2 in HUVECs (see supplementary data). A few studies have explored the role of ENG on VEGFR2 expression and activity.^39–41 Using shRNA lentivirus to repress ENG expression, Tian et al. reported that ENG silencing resulted in decreased human microvascular endothelial cell (HMEC-1) sprouting in the presence of VEGF-A.³⁹ A significant reduction in the length of HUVEC sprouts formed under the influence of VEGF-A after loss of ENG was also observed by Liu et al.⁴⁰ These data are in agreement with ours, indicating that ENG is required for VEGF-induced EC motility. However, discordant results have been described regarding the role of ENG in VEGFR2 expression and activation. The binding of VEGF to VEGFR2 induces receptor dimerization and autophosphorylation, that activates several downstream signaling pathways, including the phosphatidylinositol 3-kinase/AKT pathway and the MAP kinase/ERK pathway, that are essential in EC migration and proliferation.⁴² Tian et al. showed that the VEGFR2 protein level in ENG deficient HMEC-1 was half of that in control HMEC-1, and loss of ENG suppressed VEGF-A-induced phosphorylation of VEGFR2 and ERK1/2,³⁹ supporting our findings. In contrast, Liu et al. detected similar phosphor-ERK levels in ENG deficient and control HUVECs without measuring VEGFR2 and phosphor-AKT.⁴⁰ Of note, Jin et al. demonstrated that ENG knockdown in human dermal microvascular ECs (HDMECs) did not affect VEGFR2, VEGF-A-induced phosphor-VEGFR2 or phosphor-ERK levels but enhanced activation of AKT. They did not examine VEGF-mediated cellular responses in HDMECs lacking ENG, but showed reduced sprouting of aortic explant cultures from Eng−/− mice.⁴¹ The discrepancy in VEGF-induced pathway activation reported in these studies may be attributed to: 1) different culture conditions used for different EC types, and 2) different tools used to knockdown ENG, i.e., shRNA lentivirus and chemical transfection reagent. Nevertheless, ENG appears to be necessary for VEGF-induced EC migration.

While we detected decreased p-AKT induced by VEGF in ENG-deficient primary human ECs, others revealed the elevation of p-AKT in cutaneous telangiectasia biopsies from HHT patients.^43,44 Using immunohistochemistry analysis, Alsina-Sanchís et al. found that a higher percentage of vessels were p-AKT positive in HHT2 biopsies compared with controls.⁴³ Using the same technique as Alsina-Sanchís et al., Iriarte et al. measured p-AKT in cutaneous telangiectasia biopsies of HHT1 patients, and reported that there were more p-AKT positive ECs in patients’ tissues.⁴⁴ However, no statistical significance was indicated for these data in both studies, which is possibly due to the small sample size and the data variation. Future studies should include more patients and controls as well. In view of these findings, we speculate that the enhanced AKT activation observed in cutaneous telangiectasia from HHT patients may be caused by other factors rather than VEGF.

NF-kB and Sp1 are the major transcription factors shown to up-regulate VEGFR2 expression.^45–49 Mazor et al. have reported that treatment of HUVECs with active matrix metalloproteinase-1 results in nuclear translocation of NF-kB p65 subunit and increased expression of VEGFR2, which can be blocked by CAY10512, a potent inhibitor of NF-kB.⁴⁵ They further revealed that active matrix metalloproteinase-1 promoted binding of NF-kB p65 to the first intron and the promotor region of VEGFR2 gene, leading to enhanced transcription. NF-kB mediated up-regulation of VEGFR2 is also observed in bovine aortic endothelial cells.⁴⁶ Urbich et al. reported that fluid shear stress induces VEGFR2 expression in HUVECs, which is dependent on the binding of Sp1 to the VEGFR2 promoter region.⁴⁷ Proteasome inhibitors such as lactacystin and MG-132, through blocking the binding of Sp1 to the VEGFR2 promoter region, suppress VEGFR2 expression in HUVECs.⁴⁸ Rac1, a member of the Rho-GTPase family, has been shown to regulate NF-kB and Sp1 in ECs.^50,51 In view of these findings, we hypothesized that ENG knockdown impairs Rac1 function, leading to inhibition of NF-kB and Sp1, which eventually results in VEGFR2 downregulation. Although we observed that Rac1 levels remained principally unchanged, active Rac1 was substantially decreased in HUVECs lacking ENG. Furthermore, while changes in NF-kB nuclear translocation were not noted, Sp1 in nuclei was remarkably inhibited in ENG deficient HUVECs, which might be responsible for repressed VEGFR2 expression. Supporting our findings, it has been described that the suppression of endogenous Rac1 activity by the introduction of a plasmid expressing dominant negative Rac1 into HUVECs resulted in the downregulation of VEGFR2, mediated by the inhibition of Sp1⁵¹. In the present study, we found that Sp1 reduction did not result in VEGFR1 downregulation, a phenomenon also observed by others.⁵²

Previous studies did not examine VEGF-induced human EC proliferation/survival after knockdown of ENG.^39–41 We observed enhanced cell survival induced by VEGF in ENG deficient HUVECs, although VEGFR2 and VEGF-induced AKT and ERK activation were significantly repressed. It seems that anti-apoptotic activity of VEGF in ENG deficient HUVECs is not mediated by VEGFR2. VEGFR1 has been shown to protect ECs from apoptosis. Zhang et al. reported that knockdown of VEGFR1 with shRNA lentivirus led to repressed anti-apoptotic effect of VEGF-A in primary HDMECs and human pulmonary ECs.⁵³ They further showed that loss of VEGFR1 but not VEGFR2 enhanced the pro-apoptotic molecule caspase-3 activity in serum starved endothelial cells. Placenta growth factor, a member of the VEGF family, only binds and activates VEGFR1, and addition of placenta growth factor to serum-deprived retinal microvascular EC culture prevents cells from apoptosis through the upregulation of the anti-apoptotic molecule Bcl-2.⁵⁴ It has been shown that blood outgrowth ECs from either HHT1 or HHT2 patients have a significantly lower ENG but a similar VEGFR1 protein level compared with healthy control cells,⁵⁵ in line with our finding that ENG knockdown did not affect VEGFR1 protein expression. Therefore, we speculate that VEGF-induced anti-apoptotic activity may be through VEGFR1 in ENG deficient ECs, which remains to be validated.

Several other lines of evidence have indicated that ENG deficiency or TGF-β1 signaling pathway dysfunction down-regulates VEGFR2 expression. Loss of Eng in mouse embryonic endothelial cells (Eng−/− MEECs) markedly reduced VEGFR2 protein production.³⁸ Jerkic and Letarte further revealed that the VEGFR2 protein level in Eng+/− MEECs was approximately 50% of that in wild (Eng+/+) MEECs while VEGFR2 was hardly detectable in Eng−/− MEECs as determined by Western blotting analysis.⁵⁶ Using flow cytometric analysis, Park showed that the VEGFR2 level in Eng+/− mice retinal ECs is approximately 50% of that in wildtype cells, although the authors did not perform statistical analysis.⁵⁷ More recently, Crist et al. generated a mouse model of HHT by inducible deletion of Smad4 specifically in ECs, and demonstrated that postnatal ablation of Smad4 caused various vascular defects including the formation of AVMs.⁵⁸ Of interest, lung endothelial cells from Smad4 mutants expressed significantly lower levels of ENG and VEGFR2 than cells from control littermates. A study by Fernandez-L et al. shows that human ECs from HHT patients also produce less VEGFR2.⁵⁵ Using flow cytometric analysis, they found that VEGFR2-linked fluorescence intensity in blood-derived ECs from HHT1 and HHT2 patients was 65–80% of that in normal cells.⁵⁵ In contrast, Thomas et al. reported increased VEGFR2 in HUVECs from newborns with ENG or ALK-1 mutations.⁵⁹ The reason for these discordant findings in HHT patients remains unclear. Of note, both studies did not explore VEGF-mediated cellular responses in HHT ECs, which warrants further study.

Bevacizumab, an anti-VEGF monoclonal antibody, has been used to treat HHT. Intravenous infusion of bevacizumab has been shown to be effective for severe hepatic vascular malformations and refractory nasal and gastrointestinal bleeding,^60–64 which seems contrary to our finding that loss of ENG impairs VEGF/VEGFR2 activity. However, there is evidence showing that different ENG mutants have different levels of function. Mallet et al. analyzed 15 ENG mutants identified from HHT1 patients and found that 6 maintained 50–70% of wild type activity while the rest retained only 20–30%.⁶⁵ These data suggest patients with lower ENG activity leading to significant inhibition of VEGFR2 might respond poorly to bevacizumab treatment. Indeed, it has been found 15–20% of patients did not respond to systemic application of bevacizumab,^{62, 64} while 70% partially responded.⁶² Whether there is an association between ENG mutant activity and bevacizumab treatment response remains to be explored.

It has been shown that EC-specific knockout of Smad4 or Eng in mice leads to inhibition of capillary extension in mouse retinas, while the capillary plexus exhibits higher EC density,^{58, 66} suggesting reduced migration but enhanced proliferation/survival of ECs during the development of capillary plexus, a characteristic that resembles cellular responses to VEGF in ECs lacking ENG as shown in this study.

In conclusion, loss of ENG in primary human ECs resulted in significantly reduced expression of VEGFR2 but not TIE2 or FGFR1. ENG knockdown led to the reduction of active Rac1, further inhibiting Sp1, which might be responsible for VEGFR2 downregulation. VEGF- but not bFGF- or ANGPT-1-induced phosphor-ERK and phosphor-AKT were suppressed in ENG deficient HUVECs. Furthermore, ENG knockdown inhibited cell migration but enhanced anti-apoptotic activity induced by VEGF, the latter cellular response, to our knowledge, has not been reported. Therefore, ENG deficiency impairs the VEGF/VEGFR2 pathway and alters VEGF-induced cellular responses in ECs, which may play a role in HHT pathogenesis.

Supplementary Material

2

Supplementary Figure 1. Western blotting analysis of VEGFR1. A representative blot for VEGFR1 is shown in the left panel. There were no significant differences in VEGFR1 protein levels between control siRNA and ENG siRNA transfected HUVECs as shown in the right panel (data were analyzed with the student’s t test).

Supplementary Figure 2. Measurement of p-ERK by Western blotting in HUVECs cultured in complete EGM-2 medium. A representative blot for p-ERK is shown in the left panel as indicated. Phosphor-ERK levels were not statistically different in control siRNA and ENG siRNA transfected HUVECs after cells were cultured in complete EGM for 48 hrs (data were analyzed with the student’s t test).

Supplementary Figure 3. Cell migration stimulated by bFGF and ANGPT-1. While bFGF and ANGPT-1 induced a significantly greater number of HUVECs to migrate (* p < 0.05 compared with either growth factor stimulation), migrated cell numbers in the presence of bFGF or NGPT-1 were not substantially different between control siRNA and ENG siRNA transfected HUVECs (data were analyzed with the ANOVA with post-hoc Tukey test).

Supplementary Figure 4. Anti-apoptotic effects of bFGF, ANGPT-1 and ANGPT-2. As shown in this figure, knockdown of ENG did not affect bFGF, ANGPT-1 or ANGPT-2 induced anti-apoptotic activities in HUVECs (OD450 ratios were compared between control siRNA and ENG siRNA transfected HUVECs using the student’s t test).

Abbreviations:

HHT

Hereditary hemorrhagic telangiectasia

ENG

Endoglin

EC

Endothelial cell

HUVEC

Human umbilical vein endothelial cell

HMVEC-L

Human lung microvascular endothelial cell

AVM

Arteriovenous malformation

ALK

Activin receptor-like kinase

SMAD

Mothers against decapentaplegic homolog

TGF-β

Transforming growth factor-beta

ANGPT

Angiopoietin

siRNA

Small interfering RNA

RT-qPCR

Reverse transcription-quantitative polymerase chain reaction

ERK

Extracellular signal-regulated kinase

VEGF-A

Vascular endothelial growth factor A

VEGFR2

Vascular endothelial growth factor receptor 2

TIE2 and TEK

Tyrosine-protein kinase receptor

FGF2

fibroblast growth factor 2

FGFR

Fibroblast growth factor receptor

NF-kB

Nuclear factor kappa B

Sp1

Specific protein 1

Rac1

Ras-related C3 botulinum toxin substrate 1

GAPDH

Glyceraldehyde 3-phosphate dehydrogenase