Glycation of fibronectin inhibits VEGF‐induced angiogenesis by uncoupling VEGF receptor‐2‐c‐Src crosstalk

Abstract Glycation of extracellular matrix proteins has been demonstrated to contribute to the pathogenesis of vascular complications. However, no previous report has shown the role of glycated fibronectin (FN) in vascular endothelial growth factor (VEGF)‐induced angiogenesis. Thus, this study aimed to investigate the effects of glycated FN on VEGF signalling and to clarify the molecular mechanisms involved. FN was incubated with methylglyoxal (MGO) in vitro to synthesize glycated FN, and human umbilical vein endothelial cells (HUVECs) were seeded onto unmodified and MGO‐glycated FN. Then, VEGF‐induced angiogenesis and VEGF‐induced VEGF receptor‐2 (VEGFR‐2) signalling activation were measured. The results demonstrated that normal FN‐positive bands (260 kD) vanished and advanced glycation end products (AGEs) appeared in MGO‐glycated FN and glycated FN clearly changed to a higher molecular mass. The glycation of FN inhibited VEGF‐induced VEGF receptor‐2 (VEGFR‐2), Akt and ERK1/2 activation and VEGF‐induced cell migration, proliferation and tube formation. The glycation of FN also inhibited the recruitment of c‐Src to VEGFR‐2 by sequestering c‐Src through receptor for AGEs (RAGE) and the anti‐RAGE antibody restored VEGF‐induced VEGFR‐2, Akt and ERK1/2 phosphorylation, endothelial cell migration, proliferation and tube formation. Furthermore, the glycation of FN significantly inhibited VEGF‐induced neovascularization in the Matrigel plugs implanted into subcutaneous tissue of mice. Taken together, these data suggest that the glycation of FN may inhibit VEGF signalling and VEGF‐induced angiogenesis by uncoupling VEGFR‐2‐c‐Src interaction. This may provide a novel mechanism for the impaired angiogenesis in diabetic ischaemic diseases.


| INTRODUC TI ON
Diabetes mellitus, a progressive, chronic disease, is characterized by hyperglycaemia and can cause both microvascular and macrovascular complications. 1,2 Incompetent angiogenesis generally exists in diabetic vascular complications, and a lot of abnormalities associated with both excessive and defective angiogenesis have been observed in people with diabetes. 3,4 Angiogenesis is driven by a multiplicity of molecular mechanisms, and vascular endothelial growth factor (VEGF) is well known to be one of the most important angiogenic mediators under both physiological and pathophysiological conditions. 5,6 In the haemangioblast, VEGF receptor-2 (VEGFR-2) is the firstly expressed endothelial tyrosine kinase receptor and targeted blockade of the VEGFR-2 in mice inhibits endothelial differentiation and blood-island formation. 7,8 However, diabetes mellitus is associated with impaired activation of VEGF/VEGFR-2-induced signalling pathway, which we now call VEGF resistance. 9, 10 VEGF resistance has been demonstrated to be one of the molecular bases for the abnormal angiogenesis in diabetes mellitus. 11 However, no previous study has fully illuminated the mechanisms involved in VEGF resistance.
Advanced glycation end products (AGEs), which are formed on macromolecules, in particular, proteins, by non-enzymatic glycoxidation in hyperglycaemia environment, are well known to contribute to the pathogenesis of vascular complications in diabetes. 12,13 The extracellular matrix (ECM) proteins can be directly modified by AGEs, and the accumulation of glycated ECM proteins in the vessel wall may promote the vasculature stiffness and impair vascular cell structure and function. 14,15 Furthermore, the important arginine residues, such as arg-gly-asp (RGD) motifs, in ECM proteins may be directly modified by AGEs, causing loss of charge and structural distortion, which is associated with decreased binding affinity of integrins and cell detachment. 16 Additionally, AGE-modified ECM can directly bind to receptor for AGEs (RAGE) and activate intracellular pathways. 17,18 The ECM proteins have been demonstrated to play important roles in regulating VEGF signalling by interaction with integrins. Fibronectin (FN) is one of the components of the ECM proteins and is almost localized into all tissues. In the developing embryo, both FN and its major receptor, integrin α5β1, play a crucial role in vasculogenesis and angiogenesis. [19][20][21] FN also contributes to the regulation of VEGF signalling and VEGF-induced angiogenesis.
VEGF can directly bind specifically to FN, and the extracellular interaction between FN and VEGF is of crucial importance in VEGF signalling activation. 22,23 It has been shown that glycation may alter the functions of ECM protein, including vitronectin, 24 laminin, 16,25 collagen 26,27 and FN. 16,17,28 However, no previous studies have investigated the effects of the glycated FN on VEGF signalling and VEGF-induced angiogenesis.
It has been demonstrated that c-Src also plays a crucial role in regulating VEGF signalling. With VEGF stimulation, Src, but not Yes or Fyn, directly interacts with VEGFR-2, indicating that c-Src is the main Src family member in the regulation of endothelial cell functions. 29 VEGF-induced recruitment of c-Src to VEGFR-2 is critical for VEGF angiogenic signalling pathway activation. 30 On the other hand, the binding of AGEs to RAGE induces a cooperative binding interaction between c-Src and RAGE. 31 Therefore, in this study we hypothesized that MGO-glycated FN may directly bind to RAGE and impair VEGF angiogenic signalling pathway by sequestering c-Src.

| Cell culture
HUVECs were cultured in Medium 200 (Cascade Biologics) with lowserum growth supplement, and cells used were passaged 3-7 times.

| Glycation of FN in vitro
As previously described, 16,18 FN was glycated by incubating the protein (1 mg/mL) with MGO (0.1, 1, 10 and 50 mM) in 100 mM sodium phosphate buffer, at 37°C for 7 days. Unmodified FN was subjected to the same conditions without MGO. To identify the characterization of glycation of FN by MGO, AGE formation was detected by Western blotting and FN was also immunoblotted in the same blots after stripping. Furthermore, AGE-specific fluorescence at emission of 440 nm and excitation of 370 nm was measured using a fluorescence spectrophotometer.

| Cell migration assay
Transwell migration chambers with an 8.0-μm-size porous membrane were used to perform HUVEC migration. HUVECs (2 × 10 4 ) were added to the upper chambers, which were pre-coated with FN or MGO-FN overnight at 4°C. Then, the cells were stimulated with VEGF (50 ng/mL) or vehicle control for 24 hours. If there is a need, cells would be pre-treated with RAGE function-blocking antibody (10 μg/mL) or Src inhibitor, su6656 (5 μM) for 30 minutes before VEGF (50 ng/mL) stimulation. Then, cells remaining in the upper chamber were removed and cells that had migrated to the lower chamber were stained with crystal violet (0.5%) and counted.

| Tube formation assay
HUVECs (1 × 10 5 ) were seeded onto growth-factor-reduced Matrigel (250 μL) with FN or MGO-glycated FN in a 24-well plate and stimulated by VEGF (50 ng/mL) or vehicle control for 24 hours. In some experiments, cells were pre-treated with RAGE function-blocking antibody (10 μg/mL) or Src inhibitor, su6656 (5 μM) for 30 minutes, followed by seeding onto Matrigel. Then, tube formation was photographed by an EVOS microscope (Electron Microscopy Sciences) and the total length of the tube was quantified using Image-Pro Plus software.

| Cell proliferation assay
HUVECs (1 × 10 5 ) were cultured on 96-well plates, which were precoated with FN or MGO-FN overnight at 4°C, and grown to 50% confluence. Then, the cells were stimulated with VEGF or vehicle control for 24 hours. In some experiments, cells were pre-treated with RAGE function-blocking antibody (10 μg/mL) for 30 minutes before VEGF (50 ng/mL) stimulation. Then, a 10-μL aliquot of Cell Counting Kit 8 (CCK8) solution was added to each well for 4 hours.
The absorbance was measured at 450 nm using a Spectra Max M5 microplate reader (Molecular Devices). HUVEC proliferation was evaluated based on optical density (OD) values.

| Western blotting
HUVECs were cultured on FN or MGO-glycated FN and grown to confluence, followed by stimulation with VEGF (50 ng/mL) for 10 minutes.
Then, the cells were lysed in RIPA lysis buffer supplemented with 1% (v/v) protease and phosphatase inhibitor cocktail (Thermo Scientific).
Equal amounts of proteins were subjected to SDS-PAGE and transferred to polyvinylidene fluoride membranes (Bio-Rad Laboratories).

| Immunoprecipitation
HUVECs cultured on FN or MGO-FN were exposed to VEGF (50 ng/ mL) for 10 minutes. Classic IP kit was used to perform protein immunoprecipitation. The samples were incubated with protein A/G plus agarose beads and primary antibodies against c-Src, VEGFR-2 or RAGE overnight at 4°C, with gentle rotation. According to the IP kit instruction, the immune complexes were captured and solubilized by 50 μL SDS sample buffer. Then, the immune complexes were immunoblotted with primary antibodies against VEGFR-2 (1:500), c-Src (1:250) and RAGE (1:250).

| Solid-phase binding assay
Solid-phase binding assay was used to investigate the potential binding of unmodified FN and glycated FN to recombinant RAGE or recombinant integrin α5β1 in vitro. Microliter plate wells were coated with unmodified or glycated FN at 4°C overnight. After washing and blocking, recombinant RAGE or recombinant integrin α5β1 was added and incubated at room temperature for 1 hour, after which biotinylated anti-RAGE antibody or biotinylated anti-β1 antibody, streptavidin-HRP conjugate and HRP substrate were sequentially added and OD450 was measured.

| Statistical analysis
All of the data are shown as mean ± standard deviation (SD).
Results were analysed by one-way ANOVA followed by post hoc comparison. A P value of <.05 was considered as statistically significant.

F I G U R E 2 Glycation of FN inhibits VEGF signalling and VEGF-induced angiogenesis. A, MGO-FN inhibits VEGF-induced activation
of VEGFR-2. HUVECs were cultured on FN or MGO-FN and stimulated with VEGF (50 ng/mL) or vehicle control for 10 minutes. Phosphorylation (p) of VEGFR-2, Akt and ERK1/2, and total VEGFR-2, Akt and ERK1/2 were analysed by Western blotting in total cell lysates. Representative images of three independent experiments and densitometric analysis of phosphorylated VEGFR-2, Akt and ERK1/2 normalized to total VEGFR-2, Akt and ERK1/2 are shown. All data shown are mean ± SD for triplicate experiments and are expressed as fold changes. **P < .01. B, MGO-FN inhibits VEGF-induced cell migration. HUVECs were added to the upper chambers containing FN or MGO-FN-coated porous filters and then exposed to VEGF (50 ng/mL) or vehicle control. After 24 hours, the cells were fixed, stained with crystal violet and the cells that had migrated to the lower chambers were counted. Representative images of cell migration are shown. Scale bar is 200 μm. Quantitative assessment of triplicate cell migration experiments was performed. All data shown are mean ± SD and are expressed as % of control. **P < .01. C, MGO-FN inhibits VEGF-induced tube formation. HUVECs were seeded onto Matrigel with FN or MGO-FN in the presence of VEGF (50 ng/mL) or vehicle control for 16 hours. Representative images of tube formation are shown. Scale bar is 500 μm. Quantitative assessment of triplicate tube formation experiments was performed. All data shown are mean ± SD and are expressed as % of control. **P < .01. D, MGO-FN inhibits VEGF-induced endothelial cell proliferation. HUVECs were cultured on FN-or MGO-FN-coated 96-well plates and grown to 50% confluence, followed by stimulation with VEGF or vehicle control for 24 hours. Then, cell proliferation was measured using the CCK8 assay. All data shown are mean ± SD for triplicate experiments and are expressed as % of control. **P < .01   Figure S1). Furthermore, with VEGF (50 ng/mL) stimulation for 24 and 48 hours, the phosphorylation of VEGFR-2, Akt and ERK1/2 has not been activated and showed no significant difference among the six groups ( Figure S2). This most probably because VEGF rapidly induced activation of VEGFR-2 and the phosphoryla- In endothelial cells, VEGF also induces formation of cordlike structures that mimic blood vessels. To investigate the effects of glycated FN on tube formation, HUVECs were added to 24-well plates precoated with Matrigel in the presence of control FN or glycated FN and were exposed to VEGF. The results showed the glycation of FN also significantly inhibited the augmentation of tube formation induced by VEGF ( Figure 2C). Furthermore, VEGF-induced endothelial cell proliferation was also significantly inhibited by glycation of FN ( Figure 2D). Taken together, these data suggested that MGOglycated FN impairs activation of VEGF pro-angiogenic signalling in endothelial cells.

| Glycated FN directly binds to RAGE
To further investigate how glycation of FN prevents the binding of c-Src to VEGFR-2, we hypothesized that glycated FN directly

F I G U R E 5 A pool of c-Src is sequestered by RAGE upon glycated FN stimulation. A and B, Glycation of FN induces RAGE-c-Src complex formation. MGO-FN induces RAGE-c-Src interaction.
HUVECs were cultured on FN or MGO-FN and stimulated with VEGF (50 ng/mL) or vehicle control for 10 minutes. Cell lysates were prepared and immunoprecipitated with anti-RAGE or anti-c-Src antibodies and protein A/G plus agarose beads. Then, captured proteins were analysed by Western blotting with anti-c-Src and anti-RAGE antibodies as indicated. Representative images of three independent experiments and densitometric analysis are shown. All data shown are mean ± SD for triplicate experiments and are expressed as fold changes. **P < .01. C, The blockade of RAGE restores activation of VEGF signalling. HUVECs were pre-treated with anti-RAGE antibody (10 μg/mL) or control IgG for 30 minutes. Then, cells were seeded onto MGO-FN-coated 6-well plates, followed by stimulation with VEGF (50 ng/mL) or vehicle control for 10 minutes. Phosphorylation (p) of VEGFR-2, Akt and ERK1/2, and total VEGFR-2, Akt and ERK1/2 were analysed by Western blotting in total cell lysates. Representative images of three independent experiments and densitometric analysis of phosphorylated VEGFR-2, Akt and ERK1/2 normalized to total VEGFR-2, Akt and ERK1/2 are shown. All data shown are mean ± SD for triplicate experiments and are expressed as fold changes. **P < .01. D, The blockade of RAGE restores VEGF-induced cell migration. HUVECs were pre-treated with anti-RAGE antibody (10 μg/mL) or control IgG for 30 minutes. Then, cells were added to the upper chambers containing MGO-FN-coated porous filters, followed by stimulation with VEGF (50 ng/mL) or vehicle control. After 24 hours, the cells were fixed, stained with crystal violet and the cells that had migrated to the lower chambers were counted. Representative images of cell migration are shown. Scale bar is 200 μm. Quantitative assessment of triplicate cell migration experiments was performed. All data shown are mean ± SD and are expressed as % of control. **P < .01. E, The blockade of RAGE restores VEGF-induced tube formation. HUVECs, pre-treated with anti-RAGE antibody (10 μg/mL) or control IgG for 30 minutes, were seeded onto Matrigel with MGO-FN in the presence of VEGF (50 ng/mL) or vehicle control for 24 hours. Representative images of tube formation are shown. Scale bar is 500 μm. Quantitative assessment of triplicate tube formation experiments was performed. All data shown are mean ± SD and are expressed as % of control. **P < .01. F, The blockade of RAGE restores VEGF-induced cell proliferation. HUVECs, pre-treated with anti-RAGE antibody (10 μg/mL) or control IgG for 30 minutes, were seeded onto MGO-FN-coated 96-well plates and grown to 50% confluence, followed by stimulation with VEGF or vehicle control for 24 hours. Then, cell proliferation was measured using the CCK8 assay. All data shown are mean ± SD for triplicate experiments and are expressed as % of control. ***P < .001 binds to receptor for AGEs (RAGE), as AGEs can induce formation of RAGE-c-Src complexes. 31 Solid-phase binding assay was used to detect the potential binding of glycated FN to recombinant RAGE in vitro. The results showed that unmodified FN did not bind to RAGE, but glycated FN bound to RAGE in a concentrationdependent manner ( Figure 4A). However in the meanwhile, the normal binding between FN and integrin α5β1 was significantly inhibited by the glycation (Figure S4). To further confirm the directly binding of glycated FN to RAGE, the downstream signalling pathway of RAGE, such as NF-κB, was determined in endothelial cells. The results demonstrated that the NF-κB phosphorylation was significantly increased in HUVECs plated onto glycated FN ( Figure 4B). All of these results suggested glycation of FN may directly bind to RAGE in endothelial cells.

| A pool of c-Src is sequestered by RAGE upon glycated FN stimulation
To investigate the effects of binding of glycated FN to RAGE on c-Src, the interaction between RAGE and c-Src was measured. RAGE-c-Src complexes were captured by an immobilized anti-RAGE antibody ( Figure 5A) or an anti-c-Src antibody ( Figure 5B), respectively, and detected by immunoblotting. The results demonstrated that no matter with or without VEGF stimulation, the formation of RAGE-c-Src complexes was significantly increased in HUVECs grown on glycated FN ( Figure 5A,B). Concomitant with this increase, the accumulation of VEGFR-2-c-Src complexes upon VEGF stimulation was significantly decreased in cells grown on glycated FN (Figure 3). This suggests that glycated FN-induced RAGE-c-Src interaction can compete for binding to c-Src with VEGF-stimulated VEGFR-2 and therefore that RAGE may limit the pool of Src available for interaction with VEGFR-2.
Remarkably, the effects of blockade of RAGE on VEGF signalling activation were further determined. HUVECs plated onto glycated FN were exposed to the anti-RAGE blocking antibody for 60 minutes, followed by VEGF stimulation. The results showed that the pre-treatment with anti-RAGE antibody restored phosphorylation of VEGFR-2, Akt and ERK1/2 ( Figure 5C) and VEGF-induced angiogenesis in vitro ( Figure 5D-F), indicating that the blockade of RAGE eliminates the inhibitory effects of MGO-glycated FN on VEGF angiogenic signalling. As a whole, these data indicate that glycation of FN can directly bind to RAGE and result in the interaction of RAGE to c-Src, thereby preventing the recruitment of c-Src to VEGFR-2 and the consequent activation of VEGF-VEGFR-2 signalling pathway that involves Akt and ERK1/2 that induces endothelial cell migration and tube formation.

| D ISCUSS I ON
During diabetes mellitus, AGEs, which are well known to contribute to vascular complications, accumulate in the vessel wall and increase ECM protein glycation and cross-linking. Modification of proteins by AGEs constitutes one of the key pathogenic events in diabetes and accumulation of these adducts with FN can be about as much as four times as other ECM proteins. 32 Previous studies indicate that the glycation of FN has been proposed to contribute to diabetic microangiopathy evolution. 28 Although it is well known diabetes-induced modification of FN leads to impaired endothelial function and differentiation, 33,34 it is not fully clarified what results in glycated FN accumulation and how glycated FN exerts its antiangiogenic effects.
MGO is significantly increased in diabetic serum and as a precursor for AGE formation, can lead to rapid modification of protein amino acid residues. 35,36 Previous studies have demonstrated that MGO-derived AGEs develop rapidly on the diabetic mice vessels. 37 In the present report, we synthesized MGO-glycated FN and evaluated the conformational and structural properties of MGO modified FN. Consistent with our findings, a previous report also showed the conformational changes in MGO-glycated FN which existed as high-molecular-weight products. 38 Using tandem mass spectrometry, previous studies have also demonstrated that after treatment with MGO, three lysine and eleven arginine residues in the III 9-10 sequence of FN were glycated by AGE modification, indicating the domains which bind to integrin and heparin were AGE modification Although few previous studies have focused the role of RAGE in VEGF angiogenic function, a lot of reports have shown that knockout of RAGE or using soluble RAGE significantly increased hindlimb ischaemia-induced angiogenesis in diabetes animal model, [41][42][43] which were at least partly consistent with the findings in this study. As a whole, the present results indicate that a pool of c-Src is sequestered by RAGE upon MGO-glycated FN stimulation, consequently uncoupling VEGFR-2-c-Src interaction and inhibiting VEGF signalling.
Consistent with the cell culture results, the Matrigel plug experiments also showed significant anti-angiogenesis effects of glycated FN in vivo. However, a limitation of our Matrigel plug assay is that we cannot definitively proved the inhibitory effects of glycated FN on F I G U R E 7 Schematic diagrams representing the molecular mechanism underlying the glycation of FN-regulated VEGF signalling and VEGF-induced angiogenesis. A, Under the physiology condition, the binding of FN to VEGF enhances VEGF signalling and VEGFinduced angiogenesis and the recruitment of c-Src to VEGFR-2 is critical for VEGFR-2 phosphorylation and the downstream angiogenic signalling activation. B, Glycation of FN directly binds to RAGE and results in the interaction of RAGE to c-Src, thereby preventing the binding of c-Src to VEGFR-2 and the consequent activation of VEGF-VEGFR-2 signalling pathway that leads to angiogenesis In summary, the present study showed that MGO-glycated FN impairs VEGF angiogenic effects with mechanisms involving the inactivation of the VEGF/VEGFR-2 angiogenic signalling pathway and disruption of the VEGFR-2-c-Src crosstalk (Figure 7). The findings suggest that the formation of glycated FN may be involved in the underlying molecular mechanism of diabetic microangiopathy.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.