CdGAP/ARHGAP31, a Cdc42/Rac1 GTPase regulator, is critical for vascular development and VEGF-mediated angiogenesis

Mutations in the CdGAP/ARHGAP31 gene, which encodes a GTPase-activating protein for Rac1 and Cdc42, have been reported causative in the Adams-Oliver developmental syndrome often associated with vascular defects. However, despite its abundant expression in endothelial cells, CdGAP function in the vasculature remains unknown. Here, we show that vascular development is impaired in CdGAP-deficient mouse embryos at E15.5. This is associated with superficial vessel defects and subcutaneous edema, resulting in 44% embryonic/perinatal lethality. VEGF-driven angiogenesis is defective in CdGAP−/− mice, showing reduced capillary sprouting from aortic ring explants. Similarly, VEGF-dependent endothelial cell migration and capillary formation are inhibited upon CdGAP knockdown. Mechanistically, CdGAP associates with VEGF receptor-2 and controls VEGF-dependent signaling. Consequently, CdGAP depletion results in impaired VEGF-mediated Rac1 activation and reduced phosphorylation of critical intracellular mediators including Gab1, Akt, PLCγ and SHP2. These findings are the first to demonstrate the importance of CdGAP in embryonic vascular development and VEGF-induced signaling, and highlight CdGAP as a potential therapeutic target to treat pathological angiogenesis and vascular dysfunction.


Results
CdGAP depletion causes incompletely penetrant embryonic lethality. To explore the role of CdGAP during embryonic development, we generated a CdGAP-deficient mouse model. First, we designed a conditional floxed exon 1 allele to remove the ATG start codon of the CdGAP gene ( Supplementary Fig. 1a). The conditional CdGAP fl/fl mice were then crossed with mice expressing Cre recombinase under the Meox2 promoter, which is active as early as E5 in epiblast-derived tissues 37 . Next, to eliminate the possibility of non-specific effects caused by Meox2-Cre recombinase expression, heterozygous CdGAP fl/− ; Meox2-Cre − mice were intercrossed to generate CdGAP −/− mice, as assessed by PCR ( Supplementary Fig. 1b). The complete absence of CdGAP expression in CdGAP −/− mice was confirmed by western blotting of protein lysates of lung, brain, and heart tissues compared to those isolated from CdGAP fl/fl mice ( Supplementary Fig. 1c). CdGAP −/− pups initially seemed morphologically indistinguishable from control CdGAP fl/fl pups. However, they were not born at the expected Mendelian ratio, and showed 44% lethality by post-natal day (P) 21 (Fig. 1a). Thus, the complete loss of CdGAP expression leads to incompletely penetrant embryonic/perinatal lethality.

CdGAP-deficient embryos display vascular defects and edema.
To better evaluate the potential cause of lethality of CdGAP-deficient mice, we examined E15.5 CdGAP −/− whole embryos, which were obtained at the expected Mendelian ratio (Fig. 1a). Intriguingly, hypovascularization was apparent in 89% of CdGAP-depleted embryos, which were paler in appearance, and in 20% of heterozygous CdGAP fl/− embryos (Fig. 1b,c). This was accompanied by progressive superficial vessel defects of varying severity as defined by the presence of hemorrhages (white asterisks) in 73% of CdGAP-deficient embryos and 3% of CdGAP fl/− embryos (Fig. 1b,c). This was further evidenced by the development of prominent subcutaneous edema (black asterisks) in 77% of CdGAP-deficient embryos ( Fig. 1b-d) and the infiltration of red blood cells into the subcutaneous regions of CdGAP −/− embryos (Fig. 1e). Furthermore, hypovascularization was also evident in the meninges surrounding the brains dissected from CdGAP −/− embryos (Fig. 1c,f) and the noticeably paler livers (white arrowhead) of CdGAP-deficient embryos (Fig. 1b,c). Taken together, these results indicate that CdGAP plays a key role in vascular development during embryogenesis.
CdGAP is required for VEGF-mediated angiogenesis. Due to the pronounced vascular deficits observed in CdGAP-null embryos, we next examined whether CdGAP was involved in the promotion of VEGF-induced angiogenesis. To test this, aortas from surviving 6 week-old CdGAP −/− or control mice were harvested and embedded in Matrigel to analyze VEGF-dependent angiogenic sprouting. In this context, VEGF induced a two-fold increase in the growth of capillary sprouts from the aortic rings of control mice compared to the unstimulated condition (Fig. 2a,b). In striking contrast, VEGF failed to significantly induce the growth of capillary sprouts from CdGAP-null aortic rings (Fig. 2a,b). These results demonstrate that CdGAP is essential for VEGF-mediated capillary protrusion and elongation in the adult mouse ex vivo.
CdGAP is required for VEGF-induced endothelial cell migration and capillary formation. We next investigated if endothelial CdGAP was involved in the promotion of cell migration and capillary formation in response to VEGF in vitro. CdGAP expression was downregulated in HUVECs electroporated with siRNAs targeting the 5′ UTR of human CdGAP 29 (Fig. 2c), and Boyden chamber migration assays were first performed. VEGF induced a 3-fold increase in the migration of control HUVECs relative to unstimulated cells (Fig. 2d,e). Depletion of CdGAP resulted in a non-significant reduction in basal migration, but led to a 40% decrease in VEGF-stimulated migration when compared to control cells (Fig. 2d,e). Since migration of endothelial cells is essential for angiogenesis, we also evaluated the contribution of CdGAP to the formation of capillaries on Matrigel in response to VEGF (Fig. 2f,g). CdGAP-depleted HUVECs were impaired in their ability to form an interconnected capillary network (Fig. 2f), and a 50% decrease in overall tube length was observed compared to the control condition (Fig. 2g). We further assessed the ability of CdGAP to promote capillary sprouting from collagen-embedded cell spheroids. While VEGF stimulated a robust 3-fold induction of capillary sprouting and extension relative to the unstimulated condition, it was reduced by half in CdGAP-depleted HUVECs (Fig. 2h-j).
Overall, these results demonstrate that CdGAP is critical for VEGF-dependent endothelial cell protrusion, migration, and capillary formation.
CdGAP constitutively interacts with VEGFR2. Since CdGAP was required for VEGF-induced cell migration and capillary formation, we next examined if CdGAP could associate with and regulate VEGFR2 activation, the main VEGF receptor required for the promotion of these biological processes. HUVECs were stimulated with VEGF for 10 or 20 minutes and VEGFR2 was isolated by immunoprecipitation (Fig. 3a). Subsequent western blotting analysis revealed a basal association of CdGAP with VEGFR2, which was further increased after  10 and 20 minutes of VEGF stimulation (Fig. 3a,b). However, this association did not affect VEGFR2 activation in response to VEGF stimulation, and robust autophosphorylation at Y1175 and Y951 was observed in both control and CdGAP-depleted endothelial cells (Fig. 3c-e). Altogether, these results indicate that CdGAP is part of the VEGFR2 signaling complex in endothelial cells, likely acting downstream of VEGFR2 activation to promote the angiogenic response (Fig. 3c-e).

VEGF-induced Rac1 activation is impaired in CdGAP-depleted endothelial cells. VEGF-induced
endothelial cell motility and capillary formation are in part regulated by the small GTPase Rac1 18,38 , and CdGAP has been described as a regulator of Rac1 activity 25 . Thus, we next assessed whether loss of CdGAP affected VEGF-induced Rac1 activation. To achieve this, pull-down assays using the Cdc42/Rac interactive binding (CRIB) domain of PAK fused to GST were performed to assess the levels of active GTP-Rac1 in control and CdGAP-depleted endothelial cells stimulated with VEGF. We observed that VEGF induced a significant 1.8-fold increase in GTP-Rac1 after 20 minutes in control cells (Fig. 3f,g). While CdGAP-depletion in endothelial cells resulted in no significant change in basal levels of active Rac1, VEGF treatment failed to induce a significant Rac1 on total VEGFR2 protein levels normalized to control, unstimulated cells by densitometric analysis. (n > 3, * P ≤ 0.05, * * P ≤ 0.01, * * * P ≤ 0.001, n.s., P > 0.05, One-way ANOVA, Bonferroni's multiple comparisons posttest). (f) Control and CdGAP-depleted HUVECs were stimulated with VEGF for the indicated times. GTPloaded Rac1 was pulled down from protein lysates with GST-CRIB. Rac1-GTP (Rac GTPase assay) and total cell lysates (TCL) were resolved by SDS-PAGE and immunoblotted with the indicated antibodies. Akt was detected as a protein loading control. (g) Quantification of the ratios of GTP-bound Rac1 on total Rac1 levels normalized to those of unstimulated control cells (siCTL) by densitometric analysis. Error bars indicate SEM (n = 5, * * P ≤ 0.01, n.s., P > 0.05, Unpaired student's t-test). activation after 10 and 20 minutes (Fig. 3f,g). Therefore, these data reveal that CdGAP is required for the optimal activation of Rac1 during VEGF-mediated signaling in endothelial cells.
CdGAP depletion impairs VEGF-mediated signaling in endothelial cells. We next explored the contribution of CdGAP to VEGF-induced signaling pathways in endothelial cells. VEGFR2-dependent tyrosine phosphorylation of the scaffolding adapter protein Gab1 mediates its association with signaling proteins, including PI3K, SHP2, and PLCγ 12,13 , and this contributes to their optimal activation/phosphorylation, as well as to the downstream activation of the pro-angiogenic mediators Akt and ERK1/2 [12][13][14][15][16] . To determine if CdGAP was involved in the regulation of these VEGF-induced signaling pathways, control and CdGAP-depleted HUVECs were stimulated with VEGF for 5 to 20 minutes, and the tyrosine phosphorylation level of immunoprecipitated Gab1 was assessed by western blotting. In control HUVECs, tyrosine phosphorylation of Gab1 was maximally increased after 20 minutes of VEGF stimulation, while it was significantly reduced in CdGAP-depleted cells (Fig. 4a,b). Consistent with this, the VEGF-dependent phospho-activation of Akt was strongly inhibited in CdGAP-depleted cells (Fig. 4c,d), and the phosphorylation of SHP2 and PLCγ was also significantly decreased (Fig. 4c,e,f). In contrast, the downstream activation of ERK1/2 was not affected in these conditions (Fig. 4c,g). Taken together, these results demonstrate that CdGAP is essential for VEGF-mediated phosphorylation of Gab1, a central intracellular signaling intermediate, and for the optimal downtream activation and/or phosphorylation of Akt and SHP2, and to a lesser extent of PLCγ , during endothelial cell migration and angiogenesis.

Discussion
The CdGAP gene was reported as one of the 17 RhoGAPs highly expressed in HUVECs and in the developing vascular system of zebrafish 24,27 , but until now, its contribution to vascular endothelial cell biology remained unexplored. In this study we demonstrate for the first time that CdGAP is critical for VEGF-induced angiogenic were resolved by SDS-PAGE and immunoblotted with the indicated antibodies. Quantification of the ratios of phosphoAkt/total Akt (d), phosphoPLCγ /total PLCγ (e), phosphoSHP2/total SHP2 (f), and phosphoERK1/2/ total ERK1/2 (g) normalized to control unstimulated cells by densitometric analysis. Error bars indicate SEM (n = 5, * P ≤ 0.05, * * P ≤ 0.01, * * * P ≤ 0.001, n.s., P > 0.05, One-way ANOVA, Newman-Keuls post-test).
Scientific RepoRts | 6:27485 | DOI: 10.1038/srep27485 signaling in human endothelial cells, and correlate this with a function for CdGAP in mammalian vascular development in vivo. CdGAP-deficient embryos are hypovascularized and present vascular defects such as hemorrhages, which lead to edema and may be in part responsible for the partial embryonic/perinatal lethality observed in CdGAP-deficient mice. CdGAP is crucial for sprouting angiogenesis from aortic explants and from endothelial cell spheroids, for in vitro capillary formation, and for endothelial cell migration in response to VEGF. Surprisingly, depletion of CdGAP does not lead to an increase in Rac1 activation, and this may be explained by impaired VEGF signaling in CdGAP-depleted cells. Overall, our work describes an essential regulation of angiogenesis by CdGAP, as evidenced by impaired sprouting and attenuated regulation of endothelial cell motility and capillary formation in response to VEGF.
We show that CdGAP associates with VEGFR2 and promotes VEGF-triggered activation of downstream signaling pathways essential for optimal Rac1 activation and angiogenesis. Remarkably, CdGAP tightly controls the tyrosine phosphorylation of Gab1, an important mediator of actin reorganization and an enhancer of the PI3K-Akt and SHP2-dependent signaling pathways essential for endothelial cell migration and capillary formation in response to VEGF 12,13,16,39 . Similarly to CdGAP, Gab1 is highly expressed in blood vessels and limbs at mid-gestation, and Gab1-deficient mice also display vascular defects and hemorrhages 40 . Moreover, Gab1 and Akt1 are critical for VEGF-and ischemia-induced angiogenesis in vivo [15][16][17]41,42 , emphasizing the molecular link between CdGAP, VEGF intracellular mediators, and vascular functions. The observed genetic association between the CdGAP gene and an increased risk for coronary artery disease also supports a role for CdGAP in endothelial cell biology 43 . Collectively, these data strongly suggest that the altered function of CdGAP may potentially contribute to the development of a wide range of vascular and cardiac dysfunctions.
Recently, mutations in the CdGAP gene have been identified in families of patients with autosomal dominant AOS 33,34,44,45 . AOS is a heterogeneous disorder, typically characterized by the presence of both aplasia cutis congenita of the scalp vertex and terminal limb defects 35,36 . Additionally, vascular and cardiac anomalies, including cutis marmorata telangiectatica and pulmonary hypertension, have been observed in AOS-affected patients 35,46 . Although no cardiovascular malformations were observed in the identified AOS patients carrying heterozygous gain-of-function mutations in the CdGAP gene 33,34,46 , our results demonstrate here that the complete loss of CdGAP expression in mouse embryos leads to hypovascularization, vascular defects and incomplete embryonic/perinatal lethality. Interestingly, mutations in the DOCK6 gene encoding a Rac1/Cdc42 guanine nucleotide exchange factor responsible for an autosomal-recessive variant of AOS were recently associated with impaired vascular functions, supporting the importance of Rac1/Cdc42 signaling processes in vascular development 45,47 . Furthermore, mutations in several genes of the Notch signaling pathway, including EOGT, RBPJ, Notch, and the Notch ligand Dll4, that encode important regulators of vascular development [48][49][50] were recently identified in AOS patients [51][52][53][54] . Together with our findings, these recent genetic studies converge towards two major altered signaling pathways in the etiology of the disorder, the Notch and Cdc42/Rac1 pathways, which may be interconnected to control the development of the embryonic vasculature 55 . Notably, it was reported that VEGF modulates Dll4 expression via the ERK/Akt-dependent inactivation of GSK3β 50 , and CdGAP, which is also phosphorylated and inactivated by GSK3β 56 , interacts with VEGFR2 to regulate VEGF-induced Akt activation. Consequently, CdGAP might indirectly support the Dll4/Notch axis to control vascular development, providing a molecular link between the altered Notch and CdGAP pathways in the pathogenesis of AOS. In summary, these findings illustrate a critical role for CdGAP in angiogenesis and VEGF signaling, and provide novel mechanistic insights into the molecular causes of AOS.
Generation of CdGAP fl/fl and CdGAP-deficient mice. The F1 CdGAP fl/+ mice were generated by InGenious Targeting Laboratory, inc. For targeting CdGAP, a single loxP site containing an engineered EcoRV sequence was inserted 177bp 5′ of exon 1, and a loxP/FRT flanked Neo cassette containing an engineered Spe1sequence was inserted 270bp 3′ of exon 1. F1 CdGAP fl/+ mice were in a mixed (C57BL/6 × 129/SvEv) hybrid genetic background. F1 CdGAP fl/+ mice were crossed with mice expressing FLP recombinase (Jackson Laboratories) to delete the Neo cassette from the germline. CdGAP −/− mice were generated by crossing F4 CdGAP fl/fl mice with mice expressing Cre recombinase under the Meox2 promoter (a gift from N. Saidah, Institut de Recherches Cliniques de Montreal), which is active as early as embryonic day 5 in epiblast-derived tissues 37 . CdGAP fl/− ; Meox2-Cre − were intercrossed to generate CdGAP −/− ; Meox2-Cre − used in the study. CdGAP fl/fl ; Meox2-Cre − were used as controls ( Supplementary Fig. 2). All experiments were performed using age, sex and littermate genetic controls. Genotyping was performed using DNA isolated from mouse tails collected upon weaning or from mouse embryos at E15.5 using the following locus-specific primer sets to detect the exon 1 deleted allele: 5′ CCTGCGCTGTGCAAAGAGCCT3′ ; 5′ CCCAAGTTTAAGACCCGAGCTCC3′ ; and to detect the floxed allele: 5′ GAGCAATTCCATGAGCACCCATC3′ ; 5′ AGGTTGGAAATTTTTGGCAGCTGT3′ . The presence of the Cre allele under the meox2 promoter was assessed by using the primers and protocol provided by Jackson Laboratories. All animals were cared for in accordance with guidelines established by the McGill Migration Assay. Nucleofected HUVECs (1 × 10 5 cells in 200 μ l of serum-free M200 medium) were seeded on Transwell filters previously coated for 2 h with 0.2% Gelatin, and inserted in 24-well plates containing 500 μ l of serum-free M200 medium with or without VEGF(10 ng/ml). At the end of the assay (6 h), cells were fixed with phosphate-buffered formalin for 20 min and stained with crystal violet (0.1% in 20% methanol). Cells remaining on the upper surface of the filter were wiped off and those that migrated through the filter were visualized and counted (a minimum of 6 fields/insert at 10x magnification) with a Nikon eclipse TE300 microscope and photographed with a Nikon digital sight DS FI2 camera. Spheroid assay. This assay was based on previously described protocols 58 . Briefly, cells (1 × 10 3 ) nucleofected with control or CdGAP siRNAs were immediately cultured in supplemented M200 medium and 0.5% methyl cellulose as hanging drops (15 μ l) for 36 h. Spheroids were next collected by centrifugation (200 g, 3 min) and 50-75 spheroids were resuspended in 500 μ l of complete medium containing 0.5% of methyl cellulose. The suspension was added to an equal volume of a collagen solution (250 μ l of home-made rat tail collagen 59 , 100 μ l NaOH 0.1 M, 50 μ l PBS 10X, 100 μ l Dulbecco's modified Eagle medium (DMEM) and 50 ng/ml of VEGF or PBS as a control) and mixed thoroughly on ice. One ml of this mix was deposited per well in a pre-heated 24-well plate and collagen was allowed to solidify at 37 °C for 10 minutes. DMEM (30 μ l) was added to each well to prevent drying. Spheroids were incubated for 6-8 h and then imaged using the Zeiss Axio Observer.Z1 microscope. Photographs were taken with the AxioCam HRm (Zeiss) camera using a 20x objective lens. Images were captured with the Axiovison 4.8.2 software (Zeiss). The number and total length of all sprouts were determined for at least 10 spheroids per condition using the Metamorph software (Molecular Devices).
Capillary formation assay. This assay was performed as described in 60 . Briefly, 48 h after nucleofection of siRNAs, HUVECs (2 × 10 4 cells/wells) were seeded in duplicates on solidified Matrigel (BD Biosciences) in flat-bottom 96-well plates (50 μ l/well) and incubated for 4-6 h at 37 °C in non-supplemented M200 medium containing VEGF (50 ng/ml). Pictures were taken using a Nikon eclipse TE300 microscope and photographed with a Nikon digital sight DS FI2 camera using a 20x objective lens. The length of assembled capillaries was quantified using the Metamorph Software.
Cell stimulation, lysis, immunoprecipitation, immunoblotting and Rho GTPase Assays. Following starvation in M200 medium for 6 h, HUVECs were stimulated with VEGF (50 ng/ml) at 37 °C for the indicated times and incubated in PBS containing 1 mM Na 3 VO 4 for 30 min on ice. Cells were next lysed in a 50 mM HEPES (pH 7.5) lysis buffer containing 0.5% Nonidet P-40, 0.5% Triton X-100, 10% glycerol, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 10 μ g/ml aprotinin, 10 μ g/ml leupeptin, 5 mM NaF and 1 mM Na 3 VO 4 for 20 minutes. The quantification of Rac1 activity was performed as previously described 61 . Briefly, equal amounts of protein lysates (120 μ g) were incubated for 1 h at 4 °C with 15 μ g of the PAK CRIB domain fused to GST (GST-CRIB) and coupled to glutathione-Sepharose beads. Beads were washed 3 times with binding buffer containing 1% NP-40, followed by 2 washes with binding buffer without NP-40. Gab1 and VEGFR2 were immunoprecipitated from HUVEC protein lysates (300 μ g and 1 mg, respectively) O/N at 4 °C with 2 μ g Scientific RepoRts | 6:27485 | DOI: 10.1038/srep27485 of anti-Gab1 and 4 μ g of anti-FLK-1 rabbit antibodies, and further incubated for 3h with 30 μ l of a 50% suspension of Protein A-coupled Sepharose beads (Amersham Biosciences/GE Healthcare). Immunoprecipitated/ pull-downed proteins or total protein extracts (50 μ g) were subjected to SDS-PAGE and transferred onto nitrocellulose membranes (0.45 μ m) (Bio-Rad). Western blotting was carried out using appropriate HRP-conjugated secondary antibodies and chemiluminescence-based detection systems according to the manufacturer's recommendations (ECL detection reagent from Amersham, or Visualizer kit from Millipore). Data Analysis. Statistical analysis was performed with GraphPad Prism 6 software. P-values of less than 0.05 were considered to be significant (* P ≤ 0.05 * * P ≤ 0.01 * * * P ≤ 0.001). Densitometric analyses were performed with the Quantity One 4.6.3 software (Bio-Rad) and are respresented as the mean of at least 3 independent experiments. Error bars indicate the SEM.