Nerve Growth Factor Induces Endothelial Cell Invasion and Cord Formation by Promoting Matrix Metalloproteinase-2 Expression through the Phosphatidylinositol 3-Kinase/Akt Signaling Pathway and AP-2 Transcription Factor*

Nerve growth factor (NGF) is a well characterized neurotrophic agonist in the nervous system that triggers angiogenesis. In this study, we investigated the signaling mechanisms involved in NGF-induced angiogenesis. NGF stimulated endothelial cell invasion and cord formation on Matrigel in vitro but had marginal effect on proliferation and migration of these cells. NGF stimulated matrix metalloproteinase (MMP)-2 mRNA expression and protein secretion in human umbilical vein endothelial cells. Using synthetic and endogenous inhibitors of MMP-2 and MMP-2 small interfering RNA suppressed NGF-induced invasion and cord formation. We demonstrated that NGF-induced MMP-2 secretion, invasion, and cord formation are regulated via activation of the NGF receptor, TrkA, phosphatidylinositol 3-kinase (PI3K), and Akt using various pharmacological inhibitors. Specifically, NGF enhanced TrkA phosphorylation, PI3K activity, and Akt phosphorylation. Introduction of NGF-neutralizing antibodies, dominant-negative Akt, or wild-type PTEN effectively inhibited NGF-induced MMP-2 secretion and cord formation. Deletion and site-directed mutagenesis analysis of the MMP-2 promoter demonstrated that the AP-2-binding site is critical for NGF-induced MMP-2 promoter activity. NGF increased the DNA binding activity of AP-2, which was suppressed by inhibitors of TrkA and PI3K. Furthermore, transfection of AP-2 small interfering RNA effectively blocked NGF-induced MMP-2 secretion and cord formation. Finally, NGF promoted neovessel formation in Matrigel plugs in vivo, which was significantly inhibited by K252a and LY294002, but it failed to promote angiogenesis using MMP-2 knock-out mice. Our data collectively suggest that NGF stimulates endothelial cell invasion and cord formation by augmenting MMP-2 via the PI3K/Akt signaling pathway and AP-2 transcription factor, which may be responsible for triggering angiogenesis.

Angiogenesis, the process by which new vascular networks are formed from pre-existing capillaries, is essential in a variety of physiological and pathological conditions, including wound healing, embryonic development, chronic inflammation, cancer, and metastasis (1). The process includes complex sequential steps that include endothelial cell proliferation and migration, extracellular matrix remodeling by proteases, and maturation of newly formed vessels (2). Angiogenesis is triggered by interactions of angiogenic soluble factors and their cognate receptors involved in the different phases of the process (3). Recent studies show that nerve growth factor (NGF), 3 one of the neurotrophins, displays potential in inducing angiogenesis in physiological and pathological conditions (4 -6).
NGF is the prototypic member of the family of neurotrophic factors that includes brain-derived neurotrophic factor, NT3 and NT4/5 (7). NGF exerts a variety of actions on the central nervous system, including regulation of proliferation, differentiation, neurite outgrowth, neurotransmission, plasticity, repair, and survival (8). Two types of membrane receptors mediate NGF signaling: the Trk family tyrosine kinase receptor, TrkA, which shows a high affinity for NGF, and p75 neurotrophin receptor (p75NR), which exhibits low affinity for NGF (9). These receptors are additionally expressed on the surface of endothelial cells (10,11), and binding of NGF to TrkA triggers endothelial cell proliferation, migration, and expression of adhesion molecules in vitro and angiogenesis in vivo (5, 6, 10 -14). In addition to its direct angiogenic activity, NGF functions as an indirect activator of angiogenesis by inducing specific molecules, such as vascular endothelial growth factor (VEGF) (4).
The cellular effects of NGF are mediated mainly by the high affinity receptor, TrkA (9,15). Upon NGF binding, TrkA dimerizes and autophosphorylates multiple tyrosines within its cytoplasmic domain (15). Signaling molecules containing Src homology 2 or phosphotyrosine-binding domains, such as phospholipase C␥ and phosphatidylinositol 3-kinase (PI3K), are essential for NGF signals (15). Ras-dependent PI3K is essential for NGF-specific protection of neuronal cells from apoptosis (16) and neuronal differentiation (17). Moreover, the Rasmediated extracellular signal-regulated kinase (ERK) 1/2 pathway appears critical for NGF-induced neuronal differentiation of PC12 cells (18). NGF also transmits its signal to endothelial cells through the TrkA receptor. Stimulation of TrkA activity by NGF leads to activation of ERK1/2, PI3K, and Akt in endothelial cells, which is possibly responsible for their proliferation and migration in vitro (10,11). However, the precise regulatory mechanisms responsible for NGF-induced angiogenesis remain to be clarified.
In this study, we investigated the mechanisms of NGF action on angiogenesis, both in vitro and in vivo. Our data show that NGF significantly stimulates endothelial cell invasion and cord formation by inducing matrix metalloproteinase (MMP)-2, and that TrkA-mediated activation of the PI3K/Akt pathway and AP-2 transcription factor are critically involved in these events.
Cell Culture-Human umbilical vein endothelial cells (HUVECs, obtained from American Type Culture Collection) were cultured in gelatin-coated plates with M199 medium containing 20% FBS, sodium heparin (100 units/ml, Sigma), endothelial cell growth supplement (50 g/ml, Sigma), and antibiotics and incubated at 37°C in 10% CO 2 in air. Human dermal microvascular endothelial cells (HDMECs, purchased from Clonetics, Walkersville, MD) were cultured in gelatin-coated plates with EBM-2 medium (Clonetics) containing 5% FBS and antibiotics and incubated at 37°C in 10% CO 2 in air.
Endothelial Cell Proliferation Assay-HUVECs and HDMECs growing in gelatin-coated plates were dispersed in 0.05% trypsin solution and resuspended with M199 containing 20% FBS and EBM-2 containing 5% FBS medium without growth factors, respectively. Approximately 3,000 cells were added to each well of gelatinized 96-well plates and incubated at 37°C under 10% CO 2 for 24 h. The medium was replaced with 100 g of fresh M199 or EBM-2 medium containing 2% FBS with or without NGF at the indicated concentrations and VEGF (10 ng/ml). After 2 days of incubation, cells were washed twice with phosphate-buffered saline, and the DNA amount was measured with CyQUANT GR reagent, according to the manufacturer's protocol (CyQUANT cell proliferation assay kit, Molecular Probes, Eugene, OR), using a fluorescence spectrometer equipped with a microplate reader (Molecular Devices, Sunnyvale, CA).
Endothelial Cell Invasion and Migration Assays-Matrigel invasion assays were performed using modified Boyden chambers with polycarbonate nucleopore membrane (Costar, Corning, NY). Precoated filters (6.5 mm in diameter, 8 m pore size, Matrigel (100 g/cm 2 )) were rehydrated with 100 l of medium. Next, 2 ϫ 10 5 cells in 100 l of serum-free M199 (for HUVECs) or EBM-2 (for HDMECs) supplemented with 0.1% bovine serum albumin were seeded into the upper part of each chamber, whereas the lower compartments were filled with 1 ml of the above medium in the presence or absence of NGF at the indicated concentrations with or without molecular specific inhibitors or TIMP-2. Following incubation for 18 h at 37°C, noninvading cells on the upper surface of the filter were wiped out with a cotton swab, and the invading cells on the lower surface of the filter were fixed and stained with the Diff-Quick kit (Fisher). Invasiveness was determined by counting cells in five microscopic fields per well, and the extent of invasion was expressed as an average number of cells per microscopic field. Transwell migration assays were performed using a similar procedure as that for the invasion assay, except that the underside of filters was coated with type I collagen (100 g/cm 2 ) without coating the upper side.
Endothelial Cell Cord Formation Assay-For the endothelial cell cord formation assay, 250 l of growth factor-reduced Matrigel was pipetted into a well of a 48-well plate and polymerized for 30 min at 37°C. HUVECs and HDMECs incubated in 1% FBS-containing M199 and EGM-2 for 12 h, respectively, were trypsinized, resuspended in the same medium, and dispersed onto the Matrigel (1 ϫ 10 5 cells/well). Cells were treated with NGF at the indicated concentration and VEGF (10 ng/ml) with or without specific kinase inhibitors and small interfering RNAs (siRNAs). After 18 h, cord formation in each well was monitored and photographed using an inverted microscope. The tubular lengths of the cells were measured using Image-Pro Plus (Media Cybermetics, Silver Spring, MD).
Gelatin Zymography-Production of MMPs by HUVECs was analyzed by gelatin zymography. HUVECs in subconfluent culture conditions (about 70 -80% confluence) were washed, replenished with serum-free M199 medium, and incubated with the indicated concentrations of NGF in the presence or absence of various kinase inhibitors for 18 h. Serum-free conditioned media (20 l) were mixed with SDS sample buffer without heating or reduction and applied to 10% polyacrylamide gels co-polymerized with 1 mg/ml gelatin. After electrophoresis, gels were washed in 2.5% (v/v) Triton X-100 for 2 h at room temperature to remove SDS, rinsed twice with water, and incubated in developing buffer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl 2 , 1 M ZnCl 2 , and 0.1 M NaN 3 ) for 18 h at 37°C. Subse-quently, gels were fixed and stained with 10% 2-propanol and 10% acetic acid containing 0.5% Coomassie Blue R-250. Protease activity was visualized as clear bands within the stained gel.
Transfection and Transduction-Dominant-negative (DN) Akt and wild-type (WT) Akt plasmids (a generous gift from Dr. Su-Jae Lee, Korea Institute of Radiological and Medical Sciences, Seoul, Korea) were transiently transfected into HUVECs using Effectene transfection reagent (Qiagen, Valencia, CA) following the supplier's instructions. The AP-2␣, AP-2␤, AP-2␥, and MMP-2 siRNAs were purchased from Santa Cruz Biotechnology. Cells were transfected with siRNA molecules using RNAiFect transfection reagent (Qiagen), as described by the manufacturer. For adenoviral transduction, cells were infected with 50 plaque-forming units per cell of control adenoviral (Ad)-LacZ, Ad-WT-PTEN, and Ad-mutant-PTEN (Ad-C124S-PTEN) (19) for 90 min at 37°C. After 48 h, transfected and transduced cells were employed for further experiments. Expression of target genes was confirmed by Western blot analysis using antibodies against target proteins or phosphorylation of downstream effector molecules.
Western Blot Analysis-HUVECs were lysed in lysis buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerol phosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). After brief sonication, lysates were clarified by centrifugation at 12,000 ϫ g for 10 min at 4°C, and protein content in the supernatant was measured using the Bradford method. An aliquot (30 -50 g of protein per lane) of total protein was separated by 10 or 12% SDS-PAGE and blotted to nitrocellulose transfer membranes (0.2 m; Amersham Biosciences). The membrane was blocked with 5% nonfat skimmed milk in TBST (20 mM Tris-HCl, pH 7.6, 137 mM NaCl, and 0.01% Tween 20) for 1 h at room temperature, followed by incubation with primary antibodies. After extensive washing with TBST, the membrane was reprobed with horseradish peroxidase-linked anti-rabbit or anti-mouse immunoglobulin (1:3,000 diluted) in 5% nonfat skimmed milk in TBST for 40 min at room temperature. Immunoblots were visualized by enhanced chemiluminescence (Amersham Biosciences), according to the manufacturer's protocol.
To measure the MMP-2, MMP-9, TIMP-1, and TIMP-2 proteins secreted, conditioned medium from each sample was subjected to protein analysis. For this purpose, the culture medium in each dish was collected and concentrated 10-fold using a Centricon 10 microconcentrator (Amicon, Beverly, MA). Concentrates (5 g/sample) were subjected to SDS-PAGE analysis.
Immunoprecipitation-HUVECs were serum-starved for 6 h in M199 containing 1% FBS. Cells were stimulated by the addition of NGF (10 ng/ml) in the presence or absence of indicated concentrations of K252a (at concentrations of 50 or 200 nM). Next, cells were lysed in 1 ml of lysis buffer (20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 137 mM NaCl, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 10% glycerol, and 1% Triton X-100). Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min, and the resulting supernatant fractions (500 g) were immunoprecipitated with 1 g of anti-TrkA antibody or anti-p85 subunit of PI3K at 4°C for 2 h. Immunoprecipitates were washed three times with lysis buffer, solubilized in SDS-PAGE sample buffer containing ␤-mercaptoethanol, and analyzed by Western blotting.
Real Time PCR-Real time PCR assays were performed using the QuantiTect SYBER Green PCR kit (Qiagen) with SYBR Green I as the fluorescent dye, enabling real time detection of PCR products, according to the manufacturer's protocol. Genespecific primers were used, and the specificities were tested under normal PCR conditions. The following oligonucleotide sequences were employed: human MMP-2 sense (5Ј-GAGAA-CCAAAGTCTGAAGAG-3Ј) and antisense (5Ј-GGAGTGAG-AATGCTGATTAG-3Ј), and human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) sense (5Ј-TGAACGGGAAG-CTCACTGG-3Ј) and antisense (5Ј-TCCACCACCCTGTTG-CTGTA-3Ј). Cycling conditions were as follows: 95°C for 15 min followed by 35 cycles of 94°C for 15 s, 60°C for 30 s, and 72°C for 30 s. For quantification, the target gene was normalized to the internal standard GAPDH gene. Results of real time PCR data are represented as C t values, whereby C t is the threshold cycle at which the amplified product is initially detected. We used the comparative C t method (20) and analyzed RNA expression in samples relative to that of the control in each experiment. Primers were constructed so that the dynamic range of both targets and GAPDH references were similar over a wide range of dilutions (1:1-10,000). PCR was performed in triplicate for each sample. Results are expressed as mean values Ϯ S.D. for the relative levels compared with the control, and minimum values from three independent experiments were obtained.
Luciferase Reporter Assays-pGL2 luciferase reporter vectors, containing various MMP-2 promoter regions, were generously provided by Dr. Etty N. Benveniste (21). pRL-CMV vector (Promega, Madison, WI) was used to evaluate transfection efficiency. Cells were seeded in 12-well plates and cultured at 37°C until they reach 50 -60% confluency. Plasmids were then transiently co-transfected into HUVECs by using Effectene. After 24 h, cells were stimulated with or without NGF (10 ng/ml), and cell lysates were prepared at 4 h of incubation using dual luciferase assay kit (Promega). The luminescence was measured by using MicroLumatPlus LB96V microplate luminometer (EG&G Berthold, Wellesley, MA) according to the manufacturer's protocol.
In Vivo Matrigel Plug Assay-Specific pathogen-free 6-week-old male C57BL/6 mice (Charles River Laboratories, Tokyo, Japan) and MMP-2 knock-out mice (generously provided from Dr. Itohara, see Ref. 22) were employed for these experiments. For the Matrigel plug assay, mice were injected subcutaneously with 0.5 ml of Matrigel containing heparin (10 units/pellet), NGF (10, 50, or 100 ng/pellet), NGF plus LY294002 (5 or 20 M), or NGF plus K252a (0.2 or 1 M). The injected Matrigel rapidly formed a single solid gel plug. After 15 days, mouse skin was easily pulled back to expose the Matrigel plug, which remained intact. To quantify the formation of functional neovessels in Matrigel, the amount of hemoglobin in each plug was assayed according to the manufacturer's protocol (Drabkin reagent kit 525, Sigma).
Statistical Analysis-The data are presented as means Ϯ S.E. Statistical comparisons between groups were performed using one-way analysis of variance, followed by the Student's t test.

NGF Induces Invasion and Cord Formation in Human
Endothelial Cells-To determine the possible mechanisms of direct angiogenic activities of NGF on endothelial cells, we initially examined the effects of the growth factor on the major steps of angiogenesis in vitro, specifically endothelial cell proliferation, migration, invasion, and cord formation. As shown in Fig. 1, A and B, NGF induced proliferation and migration of HUVECs, but not strongly, compared with VEGF. In contrast, NGF significantly stimulated HUVECs invasion through Matrigel (Fig.  1C) and cord formation on Matrigel, an in vitro function that reflects morphogenesis (Fig. 1D). NGF also displayed similar in vitro angiogenic activity in HDMECs (supplemental Fig. 1). The data suggest that NGF-induced in vitro angiogenesis is mainly because of the stimulation of invasion and cord formation of endothelial cells. Because the angiogenic activity of NGF was similar in both endothelial cell types, we used HUVECs to analyze the mechanisms of NGF-induced angiogenesis throughout the study.
Augmentation of MMP-2 Is Responsible for NGF-induced Invasion and Cord Formation in HUVECs-A crucial step in the invasion and cord formation of endothelial cells involves the degradation of extracellular matrix (ECM) components, which allows cells to efficiently traverse the basement membranes (23). Accordingly, we examined the effects of NGF on ECM-degrading protease secretion of HUVECs. Gelatin zymography of serum-free conditioned medium revealed that HUVECs constitutively secrete MMP-2, whereas NGF increased the level of pro-and activated MMP-2 ( Fig. 2A). However, MMP-9 secretion was barely detectable and not influenced by NGF (data not shown). Immunoblot analyses confirmed an increase in MMP-2 secretion by NGF (Fig. 2B). In contrast, expression of MT1-MMP, an activator of MMP-2, and secretion of TIMP-1 and TIMP-2, endogenous inhibitors of MMP-9 and MMP-2, respectively, were not significantly affected (Fig. 2B). To determine whether NGFinduced MMP-2 secretion is regulated at the transcription level, we performed real time PCR analysis and found that NGF enhanced MMP-2 mRNA levels in HUVECs in a dosedependent manner, which peaked at 12 h (Fig. 2C). Next, we examined the involvement of NGF-augmented MMP-2 on invasion and cord formation of HUVECs. Treatment with pharmacological inhibitors of MMP-2 and recombinant TIMP-2 almost completely abrogated NGF-induced invasion (Fig. 2D) and cord formation (Fig. 2E) of HUVECs. To further confirm the involvement of MMP-2 in these events, we transfected MMP-2 siRNA into HUVECs and found that MMP-2 knockdown significantly reduced NGF-induced MMP-2 secretion (Fig. 2F) and cord formation (Fig. 2G). Therefore, our results strongly suggest that NGF up-regulates MMP-2 expression at the transcriptional level, which is critical for invasion and cord formation in HUVECs.
TrkA Mediates NGF-induced MMP-2 Secretion, Invasion, and Cord Formation in HUVECs-To explore the possible signaling mechanisms involved in NGF-induced HUVECs invasion and cord formation, we initially examined the role of TrkA, a major receptor for NGF-induced angiogenesis (10). As shown in Fig. 3A, NGF rapidly phosphorylated TrkA within 5 min in HUVECs. Preincubation of these cells with the specific TrkA inhibitor, K252a, led to suppression of NGF-induced TrkA phosphorylation (Fig. 3A). In addition, pretreatment with K252a and NGF-neutralizing antibodies resulted in significant abrogation of NGF-induced MMP-2 secretion (Fig. 3B), invasion (Fig. 3C), and cord formation (Fig. 3D) of HUVECs. Therefore, phosphorylation of the TrkA receptor by NGF is critical for angiogenesis in vitro.
NGF-induced Angiogenesis in Vitro Is Mediated by Activation of the PI3K/Akt Signaling Pathway in HUVECs-To identify the downstream signaling pathway for NGF-induced MMP-2 secretion, and resultant invasion and cord formation of HUVECs, we initially treated cells with specific inhibitors of various kinases, along with NGF. Inhibitors of PI3K (LY294002) and Akt (Akt inhibitor IV), but not MAPK kinase-1 (MEK-1, PD98059), p38 (SB202190), and JNK (SP600125), significantly suppressed NGF-induced MMP-2 secretion and mRNA expression (Fig. 4, A and B, respectively). Moreover, LY294002 completely blocked NGF-induced invasion and cord formation of HUVECs (Fig. 4, C and D, respectively). In view of the above data, which strongly imply critical involvement of the PI3K/Akt signaling pathway in NGF-induced MMP-2 expression and resultant invasion and cord formation of HUVECs, we further investigated the involvement of the pathway in these events. PI3K and immunoprecipitation assays revealed that NGF stimulates PI3K activity (Fig. 5A) and phosphorylation of p85 (Fig.  5B), the regulatory domain of PI3K, within 5 min in HUVECs. In addition, NGF phosphorylated Akt, a downstream target of PI3K, within 5 min (Fig. 5C). To further confirm the involvement of the PI3K/Akt pathway in NGF-induced angiogenesis in vitro, we introduced DN-and WT-Akt, and Ad-WT-and Ad-C124-PTEN in HUVECs. As shown in Fig. 5D, left, transfection of DN-Akt significantly inhibited NGF-induced MMP-2 secretion and Akt phosphorylation, whereas WT-Akt potentiated NGF-induced MMP-2 secretion and Akt phosphorylation. Infection of HUVECs with Ad-WT-PTEN, but not Ad-C124S-PTEN, a protein and lipid phosphatase-deficient mutant, completely abrogated NGF-induced MMP-2 secretion FIGURE 1. Effects of NGF on in vitro angiogenesis of HUVECs. A, HUVEC proliferation assay. Cells were stimulated with the indicated concentrations of NGF or VEGF (10 ng/ml) and allowed to proliferate for 2 days. The DNA amount was measured using CyQUANT GR reagent. B and C, HUVEC migration (B) and invasion (C) assays. Chemotaxicells were coated with type I collagen (100 g/cm 2 ) on the underside of the filter for migration assay or Matrigel (100 g/cm 2 ) on the upper side of the filter for the invasion assay, and 2 ϫ 10 5 cells were placed in the upper well in the presence or absence of NGF at the indicated concentrations or VEGF (10 ng/ml). Photographs of HUVECs migrating and invading under the membrane after 18 h (left) are shown (ϫ40). Migration and invasion rates were determined by counting cells in four microscopic fields per sample (right). D, HUVEC cord formation assay. Cells in 1% FBS-containing medium were seeded onto Matrigel-coated 48-well plates at a density of 1 ϫ 10 5 cells/well and incubated in the presence or absence of NGF at the indicated concentrations or VEGF (10 ng/ml). After 18 h, microphotographs were obtained (ϫ40) (left). The tubular lengths of the cells were measured using Image-Pro Plus software (right). Details of all assays are described under "Experimental Procedures." All experiments were performed under at least three independent conditions, and the representative images are depicted. Scale bar ϭ 100 m. Bars Ϯ S.D. *, p Ͻ 0.05 versus nontreated control; NGF, †, p Ͻ 0.05 versus nontreated control; VEGF. and Akt phosphorylation (Fig. 5D, right). Furthermore, DN-Akt transfection and Ad-WT-PTEN infection significantly inhibited cord formation of HUVECs induced by NGF (Fig. 5E). These findings strongly suggest that the PI3K/Akt pathway mediates NGF-induced increase in MMP-2 expression and subsequent invasion and cord formation of HUVECs.

AP-2 Is Involved in NGF-induced MMP-2 Expression-
To determine which transcription factors are associated with the angiogenic effects of NGF, we measured MMP-2 promoter activity using plasmids containing various MMP-2 promoter regions conjugated with luciferase (21). NGF stimulated and sustained MMP-2 promoter activity in the constructs from WT to D6 (Ϫ139 bp) suggesting that Sp1 and AP-2 elements are important in these events (Fig. 6A). Interestingly, NGF also significantly up-regulated MMP-2 promoter activity in the D7 construct (Ϫ64, AP-2 element only) compared with the background activity of the D8 construct (Ϫ7, no element) (Fig. 6A). On the contrary, mutation of the AP-2 element in the D2 construct (Ϫ1591, mAP-2) caused a significant reduction of NGFinduced MMP-2 promoter activity (Fig. 6A), indicating the critical involvement of AP-2 element in NGF-induced MMP-2 promoter activity. To confirm this result, we performed EMSA using nuclear extracts of HUVECs treated with NGF and found that NGF increased the DNA binding activity of AP-2 oligonucleotides but not AP-1 and Sp1 in HUVECs (Fig. 6B). Because NGF-mediated MMP-2 expression is blocked by suppressing TrkA and PI3K, we investigated the inhibitory effects of K252a and LY294002 on NGF-induced up-regulation of AP-2 DNA binding activity. As shown in Fig. 6C, K252a, LY294002, and Akt inhibitor IV significantly suppressed the DNA binding activity of AP-2 induced by NGF in HUVECs. Inhibition of AP-2 using AP-2␣ siRNA dramatically inhibited NGF-induced MMP-2 secretion (Fig. 6D) and completely blocked cord formation in HUVECs (Fig. 6E). In addition, siRNA of AP-2␤ and AP-2␥, other isotypes of AP-2 that could possibly form heterodimers with AP-2␣, also significantly suppressed these events (supplemental Fig. 2). In contrast, transfection of Sp1 siRNA and mithramycin, a specific inhibitor of Sp1, reduced basal MMP-2 secretion (supplemental Fig. 3). Therefore, it might be possible that Sp1 and AP-2 are largely responsible for basal and NGF-induced MMP-2 expression, respectively. Overall, the data strongly suggest that the TrkA receptor-mediated PI3K/Akt signal pathway is critical for NGF-induced AP-2 activation, which is responsible for MMP-2 expression and subsequent cord formation of HUVECs.
Inhibition of TrkA and PI3K Attenuates NGF-induced Angiogenesis in Vivo-Finally, we investigated whether the signaling pathway that mediates the biological functions of NGF on HUVECs in vitro is actually involved in NGF-induced angiogenesis in vivo. As shown in Fig. 7A, NGF significantly induced capillary vessel formation in Matrigel plugs, displayed as a dark red color, and increased hemoglobin concentration in a dosedependent manner. The addition of K252a and LY294002 along with NGF (100 ng/pellet) significantly inhibited NGF-induced blood vessel formation and hemoglobin concentration with the Matrigel plugs (Fig. 7B). To confirm the involvement of MMP-2 on NGF-induced angiogenesis in vivo, we performed Matrigel plug assay with MMP-2 deficient mice (22). As shown in Fig.  7C, NGF did not induce angiogenesis in Matrigel plugs in MMP-2 null mice. Therefore, these results, along with the NGF-induced invasion and cord formation of HUVECs in vitro, strongly suggest that NGF promotes angiogenic effect predom-

. Effects of TrkA inhibitor and NGF-neutralizing antibodies on NGF-induced TrkA phosphorylation, MMP-2 expression, invasion, and cord formation in HUVECs.
Cells were incubated with or without the TrkA inhibitor, K252a, at the indicated concentrations or NGF-neutralizing antibodies (anti-NGF, 1 g/ml) for 1 h in the presence or absence of NGF (10 ng/ml). A, TrkA was immunoprecipitated from cell lysates, and its phosphorylation was determined by Western blotting using phosphotyrosine-specific antibodies (pTyr). Antibodies used to detect total levels of TrkA were used as a loading control. B, gelatin zymography (upper) and Western blot analysis (lower) of conditioned media from HUVECs treated with or without NGF in the presence or absence of K252a (200 nM) and anti-NGF (1 g/ml). Conditioned media were collected for analysis after 18 h. C and D, Matrigel invasion (C) and cord formation assays (D) of HUVECs cultured in the presence or absence of NGF with or without K252a and anti-NGF at the indicated concentrations. Details of all assays are described under "Experimental Procedures." All experiments were performed under at least three independent conditions, and the representative images are depicted. Bars Ϯ S.D. *, p Ͻ 0.05 versus nontreated control; NGF, **, p Ͻ 0.05 versus NGF; K252a, ***, p Ͻ 0.05 versus NGF; anti-NGF. OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42

DISCUSSION
Angiogenesis is a tightly regulated and well organized process induced by soluble angiogenic factors and their receptors. Among these factors, NGF and its receptor, TrkA, are involved in triggering angiogenesis directly or indirectly by inducing specific proteins, including VEGF (4 -6, 10). NGF stimulation of endothelial cells through TrkA directly promoted endothelial cell proliferation and migration in vitro and enhanced angiogenesis in chick chorioallantoic membrane, corneal micropocket, and ischemia hindlimb models in vivo (5, 10 -14). However, the specific mechanisms of NGF-induced angiogenesis are unknown. In this study, we attempt to identify the NGF-induced signaling pathways involved in the regulation of angiogenesis in vitro and in vivo. NGF significantly induced MMP-2 expression, invasion, and cord formation of HUVECs. NGFinduced MMP-2 expression and concomitant invasion and cord formation of HUVECs were regulated through the PI3K/Akt signaling pathway and AP-2 transcription factor. Furthermore, NGF-induced angiogenesis in vivo in a Matrigel plug model was significantly suppressed by specific inhibitors of TrkA and PI3K. In addition, NGF did not induce angiogenesis in MMP-2 knock-out mice in vivo.
Various angiogenic factors affect distinct steps in the angiogenesis process that differ from factor to factor. For example, VEGF and basic fibroblast growth factor (bFGF) stimulate all the steps of angiogenesis, including endothelial cell proliferation, migration, invasion, and cord formation (24). In contrast, insulin-like growth factor-II induces chemotactic migration, invasion, and cord formation of endothelial cells, but not proliferation (25), whereas hepatocyte growth factor/scatter factor enhances chemotactic migration and nondirectional motility (24). NGF triggers proliferation of HUVECs, HDMECs and choroidal endothelial cells (10,12,13), and migration of porcine aortic and choroidal endothelial cells in vitro (11,13). Our data show that NGF strongly enhances invasion and cord formation of HUVECs with marginal effects on proliferation and migration, compared with VEGF. Accordingly, we suggest that the main effect of NGF on angiogenesis is stimulation of invasion and cord formation of endothelial cells.
Degradation of the ECM by MMPs is required in endothelial cell migration and organization, and hence angiogenesis (23,26). MMPs are a family of structurally related Zn 2ϩ -dependent neutral metalloproteinases, which collectively degrade all the components of ECM proteins (27). The role of MMPs in angiogenesis was initially established by the finding that suppression of MMP activity by endogenous inhibitors, TIMPs, or synthetic compounds inhibits in vitro cord formation (23). Some studies demonstrate that MMP-2 or MMP-9 plays a key role in vascular endothelial cell migration and cord formation (23,28). Additionally, MT-MMPs involved in the activation of MMP-2 on the cell surface, either as activators or receptors, play critical roles in these events (29). Some of the angiogenic growth factors induce proteases that participate in angiogenesis. For instance, VEGF increases tissue factor and MMP-2 expressions in HUVECs and HDMECs, respectively (30,31). VEGF and bFGF additionally stimulate expression of urokinase plasminogen activator and its receptor (24,32,33). On the other hand, interleukin-8 up-regulates the expression of MMP-2 and MMP-9 in HDMECs, whereas hepatocyte growth factor/scatter factor enhances MT1-MMP synthesis and MMP-2 activation in HDMECs and arterial endothelial cells (34,35). In our experiments, NGF induced MMP-2 secretion and mRNA expression in HUVECs. Because MMP-2 is critically involved in the invasion and cord formation of endothelial cells (23), we investigated the effect of NGF-induced MMP-2 expression on invasion and cord formation in HUVECs. Indeed, inhibition of MMP-2 by TIMP-2 and a synthetic inhibitor and suppression of MMP-2 expression by MMP-2 siRNA significantly abrogated invasion and cord formation of HUVECs by NGF, indicating that NGF-induced up-regulation of MMP-2 is crucial for endothelial cell invasion, cord formation, and resultant angiogenesis.
MMP-2 activity is regulated by several mechanisms, including gene expression at the transcription level, proenzyme activation by coupling with MT1-MMP and TIMP-2, and inhibition by TIMP-2. Among these, transcriptional up-regulation of MMP-2 involves the activation of several well known factors, including p53, AP-1, Ets-1, C/EBP, cAMP-response elementbinding protein, PEA3, YB-1, Sp1, Sp2, and AP-2 (21). Based on real time PCR data showing that NGF-induced MMP-2 expression is regulated at the transcriptional level, we attempted to identify the transcription factor related to these events. Assays using deletion mutants of the human MMP-2 promoter revealed that the AP-2-binding site is critical for NGF-induced MMP-2 promoter activity of HUVECs, whereas the Sp1 element might be responsible for basal MMP-2 promoter activity. In addition, EMSA data also showed that the DNA binding activity of AP-2 but not AP-1 and Sp1 is significantly up-regulated by NGF in these cells. Furthermore, transfection of AP-2␣, -2␤, and -2␥ siRNA dramatically suppressed MMP-2 expression and cord formation of HUVECs. It is evident that AP-2 is predominantly involved in MMP-2 induction by NGF in these cell systems, although we cannot rule out the possible involvement of other transcription factors in NGF-induced enhancement of MMP-2 promoter activity.
The PI3K/Akt signaling pathway is one of the most important downstream targets of receptor tyrosine kinases, integrin receptors, and the Ras pathway (36). PI3K is composed of heterodimers with separate regulatory (p85) and catalytic (p110) subunits, and p85 is the substrate for many upstream regulators. Upon stimulation, the p85-p110 complex of PI3K is recruited to the membrane and phosphorylated by interactions with kinases at the Src homology 2 domain of p85. Activated PI3K phosphorylates phosphatidylinositol 4,5-bisphosphate at the D3 position to produce phosphatidylinositol 3,4,5-triphosphate that acts as lipid second messenger to stimulate downstream signaling molecules, including Akt, also designated protein kinase B. Akt, in turn, signals to various downstream effectors controlling cellular proliferation, migration, invasion, and survival (36). The PI3K/Akt signaling pathway is regulated by PTEN (also denoted MMAC1), a tumor suppressor gene located on human chromosome 10q23.3, which dephosphorylates two lipid signal transduction molecules, phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5-triphosphate (37). Numerous reports show that the PI3K/Akt signaling pathway is critically involved in endothelial cell survival, migration, and cord formation and resultant angiogenesis by specific stimuli (38). Indeed, PTEN suppression of the PI3K/Akt pathway led to inhibition of VEGF-induced angiogenesis in vitro and in vivo (39). In addition, the PI3K/Akt pathway plays an important role in endothelial cell migration and capillary-like structure formation by VEGF, bFGF, and migration inhibitory factor (40,41). In keeping with these studies, we examined the involvement of possible signaling pathways in the regulation of NGF-induced angiogenesis. Using specific kinase inhibitors, we showed that the PI3K/Akt pathway is critically involved in NGF-induced MMP-2 expression, invasion, and cord formation of HUVECs. In fact, LY294002, a specific inhibitor for PI3K, but not PD98059, SB202190, or SP600125 (inhibitors of MEK, p38, or JNK, respectively), effectively suppressed these events. Furthermore, transient transduction of adenovirus-mediated wild-type PTEN or transfection of dominant-negative Akt cord significantly abrogated MMP-2 secretion and formation induced by NGF. An EMSA showed that blockade of the PI3K/Akt pathway by LY294002 down-regulated DNA binding of AP-2 stimulated by NGF in HUVECs. Finally, LY294002, as well as K252a, a specific inhibitor of the TrkA receptor, strongly suppressed NGF-induced angiogenesis in vitro and in vivo Matrigel plug model. Thus, we conclude that the PI3K/Akt Cells were transfected with reporter vectors containing various MMP-2 promoter regions for 24 h and treated with NGF (10 ng/ml) for 3 h or left untreated. Cells were then lysed, and the extracts were analyzed for luciferase activity. Bars Ϯ S.D. *, p Ͻ 0.05 versus nontreated control. B and C, EMSA of HUVECs stimulated with NGF (10 ng/ml) for 4 h. Nuclear extracts from cells were subjected to EMSA with AP-1, AP-2, and Sp1 consensus oligonucleotides (B). Cells were stimulated with NGF (10 ng/ml) in the presence or absence of Akt inhibitor IV, LY294002, or K252a at the indicated concentrations for 4 h. Nuclear extracts from cells were subjected to the gel shift assay with AP-2 consensus oligonucleotides (C). D, Western blot analysis of cell lysates (left) or conditioned medium (right lower) and gelatin zymography of conditioned medium (right upper) of HUVECs transfected with AP-2␣ siRNA (200 nM) or control siRNA (200 nM) for detecting AP-2␣ and MMP-2 expression, respectively. E, cord formation assays of AP-2␣ siRNA-transfected cells cultured in the presence or absence of NGF (10 ng/ml). The area covered by the cord network was quantitated using Image-Pro Plus software. Details of all assays are described under "Experimental Procedures." All experiments were performed under at least three independent conditions, and the representative data are depicted. Bars Ϯ S.D. *, p Ͻ 0.05 versus nontreated control; **, p Ͻ 0.05 versus NGF in control siRNA transfected cells. pathway plays an important role in NGF-induced angiogenesis in vitro and in vivo.
Many reports, including our study, have shown that PI3K/ Akt pathway is critically involved in MMP-2 expression (19,(42)(43)(44). However, none of these studies revealed the relationship between PI3K/Akt signaling pathway and AP-2 activation. On the contrary, ERK/MAPK pathway has shown to up-regulate AP-2 DNA binding activity of the VEGF promoter (45), PAC-1 promoter (46), and MMP-2 promoter (47). Our study showed that pharmacological inhibitors of PI3K and Akt, but not ERK/MAPK inhibitor, clearly suppressed the AP-2 activation and resultant MMP-2 expression induced by NGF. Therefore, it is highly possible that the PI3K/Akt pathway can also stimulate AP-2 under certain conditions, although some extensive studies will need to confirm this phenomenon.
In summary, we have shown that NGF enhances in vitro angiogenesis mainly by inducing MMP-2 expression and subsequent invasion and cord formation of HUVECs. Mechanistic analyses disclosed that PI3K/Akt activation of NGF regulates MMP-2 expression through the AP-2 transcription factor, resulting in endothelial cell invasion and cord formation and ultimately angiogenesis.