The grape seed (Vitis vinifera, cv. Fernão Pires) powder (200 g) was firstly extracted using 3 l of methanol-water (80:20, v/v), followed by 3 l acetone-water (75:25, v/v) to obtain crude phenolic extract as described previously (5). After removing the organic solvents, the crude phenolic extract was chromatographed on a Lichroprep RP-18 (200×25 mm i.d.; 25–40 m particle size; Merck, Darmstadt, Germany) column to isolate F2 with the procedures similar to those already described (6). Briefly, F2 was extracted from the grape seed methanolic extract and evaporated at <30°C to dryness, and dissolved in double distilled water prior to lyophilization. The chemical and structural characterization of F2 was determined by normal-phase and reverse-phase HPLC, thioacidolysis-HPLC, ESI-MS analyses, formaldehyde-HCl precipitation and elemental analysis (7). The powder of F2 obtained was endotoxin free and stored at −20°C until used. For cells culture work in vitro, a 10 mg/ml aqueous stock solution was prepared and sterilized using the 0.22-μm filter.

Human astrocytoma U251 cell line (ATCC, Rockville, MD, USA), human hepatoma Hep3B cell line (ATCC) and human umbilical vein endothelial cells (HUVECs) (National Center for Medical Culture Collection, China) were routinely cultured in DMEM and/or RPMI-1640 supplemented with 10% FBS (Gibco-BRL, Rockville, MD, USA) and maintained at 37°C in a humidified incubator with 5% CO₂ (referred to as normoxic conditions). U251-HRE (cells stably transfected with pGL2-TK-HRE plasmid) and U251-pGL3 (cells stably transfected with pGL3-control plasmid) cells were cultured in RPMI-1640 with 5% FBS and 100 μg/ml G418 (Sigma, CA, USA). For cell culture under hypoxic conditions, cells were grown in a modular incubator chamber (CA, USA) containing 1% O₂, 5% CO₂ and 94% N₂ at 37°C. Hypoxia can also be induced using the hypoxia mimetic agent desferoxamine (Sigma).

Rapamycin, LY294002, cycloheximide (CHX), MG132 and dimethyloxalylglycine (DMOG) were purchased from Sigma. PD98059 was purchased from Beyotime (S1805, China). Recombinant human VEGF and EGF were purchased from Peprotech (Rocky Hill, NJ, USA). HIF-1α siRNA and control siRNA were purchased from Santa Cruz (CA, USA). Primary antibodies against HIF-1α and HIF-1β were purchased from BD-Transduction Laboratories (MA, USA). Antibodies specific for phosphorylated (Ser-308) or total AKT, phosphorylated (Ser-2448) or total mTOR, phosphorylated (Thr-389) or total p70S6K, phosphorylated (Ser-65) or total 4E-BP1, phosphorylated p44/42 ERK1/2 (T202/Y204) or p44/42ERK1/2 and phosphorylated (Tyr-992) or total EGFR and β-actin were purchased from Cell Signaling Technology (CA, USA).

pGL2-TK-HRE plasmid, containing three copies of the hypoxia-responsive element from the inducible nitric oxide synthase promoter upstream of the firefly luciferase reporter gene, was kindly provided by Giovanni Melillo (National Tumor Institute, Frederick, MD, USA). Hep3B cells were transiently transfected with 0.5 μg pGL2-TK-HRE plasmid. Afterward, cells were treated with different concentrations of F2 (μg/ml) followed by exposure to hypoxia for 16 h. Cell lysates were collected and luciferase levels were subsequently assayed using the Bright-Glo™ luciferase assay system (Promega, WI, USA).

Western blot analysis was performed as described previously (10). In brief, equal amounts of total protein extracts from cultured cells were fractionated by 4–15% SDS-PAGE and electrically transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, Darmstadt, Germany). Mouse or rabbit primary antibodies and appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (CST, CA, USA) were used to detect the designated proteins. Membrane-bound secondary antibodies were detected with ECL reagents (Invitrogen, CA, USA) and exposed to X-ray film. Results were normalized to the internal control (β-actin).

Total RNA was isolated using RNeasy Mini kit (Qiagen, Germany) following the manufacturer’s instructions. RNA (1 μg) was reverse transcribed with a RevertAid First Strand cDNA Sythesis kit (Toyobo, Japan). For quantitative PCR, analysis was carried out using iQ SYBR Green Supermix (Bio-Rad, CA, USA) and a CFX96 Real-Time PCR Detection System (Bio-Rad) as instructed by the manufacturer. RT-PCR reactions were performed as described previously (8).

Cells were incubated in the 6-well culture plates (Corning Costar, NY, USA) and treated with F2 (μg/ml) under normoxic and hypoxic conditions for 16 h. Cells culture medium were collected and centrifuged at 800 rpm for 5 min at 4°C to remove cellular debris. VEGF in the medium was measured by using the Quantikine human VEGF ELISA kit (R&D Systems, MN, USA) according to the manufacturer’s instructions.

Cells (10–20×10⁴) were plated in cell basal medium in the upper chamber of a Transwell (8 μm, Corning Costar), pre-coated with matrigel (BD, MA, USA). Serum-free RPMI-1640 medium with or without of F2 (μg/ml) was added to the upper and lower chamber of the Transwell. After 12 or 16 h, non-migrated cells were removed with cotton swab and migrated cells were stained and examined using ImageXpress Micro XLS Widefield High Content Screening System (Molecular Devices, CA, USA), the number of migrated cells was quantified by counting the cell number.

Matrigel (50 μl) mixed with serum-free RPMI-1640 medium (1:1) were placed into each well of the 96-well-plate (Corning Costar) and allowed to polymerize by incubation at 37°C for 30 min. A mixture of HUVECs (5×10⁴/well) (100 μl) was seeded on the gel with or without culture medium (CM) (100%), then incubated at 37°C for 12 h. Cells were stained and examined, and tube formation (branches/field) was examined and quantified.

Results are expressed as means ± SE of three independent experiments done in triplicate. One-way ANOVA followed by Dunnett’s t- and T3-test were used for statistical analysis (SPSS 16.0 software, SPSS, USA).

We used U251-HRE and U251-pGL3 cells (HIF-1α inhibitor screening model) to detect whether F2 inhibited the HIF-1α activity (16). As shown in Fig. 1A, the hypoxia-induced luciferase expression in U251-HRE cells, but not the constitutive luciferase in U251-pGL3 control cells, was significantly decreased by F2 in a concentration-dependent manner (P<0.01). Similar results were obtained in Hep3B-HRE and Hep3B-pGL3 transiently transfected cells (P<0.01). The effects of F2 on the expression of HIF-1 target genes mRNA were examined. As shown in Fig. 1B, hypoxic induction of VEGF, GLUT-1 and MMP-2 mRNA were markedly blocked by F2 in a concentration-dependent manner. In addition, HIF-1α DNA binding activity in both cell lines was also significantly decreased in the presence of F2 (P<0.01, Fig. 1C).

In order to evaluate how F2 inhibits HIF-1α activity, HIF-1α and HIF-1β in both cells were investigated at protein and mRNA levels by western blot analysis and real-time PCR, respectively. As shown in Fig. 2A–C, F2 inhibited the expressions of HIF-1α and HIF-1β protein in both cells under normoxic and hypoxic conditions in a concentration- and/or time-dependent manner without obvious changes in the cell viability (P>0.05, data not shown). For the concentration-dependent inhibition of HIF-1α protein levels, no significant difference between U251 and Hep3B cells was observed. In addition, F2 also exerted a concentration-dependent inhibition of hypoxic-induced accumulation of HIF-1α protein in other cell lines, such as A549 and MDA-MB-435S cell lines (data not shown).

For the HIF-1α and HIF-1β mRNA, no appreciable effects of F2 were found in either cell line, leaving a random distribution and an inconsistent pattern (P>0.05, Fig. 2D). Since HIF-1 activity is primarily determined by HIF-1α subunit in the nucleus, we also determined the presence of HIF-1α in the nucleus and cytoplasm, and found that the nuclear translocation of HIF-1α in both cell lines were not changed in the presence of F2 (Fig. 2E) under normoxic or hypoxic conditions.

In order to test whether the inhibitory effect of F2 on HIF-1α protein expression was associated with the degradation induction of HIF-1α via the prolyl-hydroxyltion-proteasome pathway, we assessed the impact of F2 on HIF-1α protein degradation in the presence of MG132 (a proteasome inhibitor) and DMOG (a prolyl hydroxylase inhibitor) (17). As shown in Fig. 3A, the MG132 (20 μM) and DMOG (100 μM)-induced increase in expression of HIF-1α was decreased by F2 (60 μg/ml) under normoxic condition. Immuno-fluorescence assay showed that F2 also downregulated the expression level of hydroxylation-HIF-1α (hy-HIF-1α) induced by MG132 in Hep3B cells (data not shown). In this study, HIF-1α protein was expressed as two bands and slightly dispersed in cells treated with MG132 due to its poly-ubiquitylation modification, which was consistent with previous results (18). These results indicate that F2 suppresses the accumulation of HIF-1α through a pathway independent of hydroxylation-proteasome degradation.

These results were compatible with the possibility that F2 affected the half-life of HIF-1α protein and promoted its degradation through a different pathway (17). To investigate this hypothesis, experiments were performed using cycloheximide (CHX) which could inhibit de novo protein synthesis. HIF-1α protein was measured before (time 0) and at different times after CHX was added, however, as shown in Fig. 3B, there was no change in HIF-1α half-life in U251 or Hep3B cells (P>0.05) after treatment with F2 (60 μg/ml).

A study also exists revealing that p53 might negatively regulate HIF-1α stability by interacting directly or indirectly by targeting it for HDM2-mediated degradation (19). Therefore, the effect of F2 on p53 expression was tested to see whether F2 affected the expression of HIF-1α with p53 involved. It is reported that U251 and Hep3B cells express mutant-p53 and deleted-p53, respectively (20,21). As shown in Fig. 3C, U251 cells expressed p53, which was accordance with a previous report that U251 cells expressed mutant-p53 (20), while the level of p53 was decreased in the presence of F2; p53 was not detected in Hep3B cells, which might be due to the Hep3B cells expressing the deleted p53 (21). F2 also inhibited HIF-1α expression in A549 cells which expressed wild-p53. It therefore seems that p53 is uninvolved in the regulation of HIF-1α expression by F2, because HIF-1α expressions were inhibited by F2 regardless of the type of p53 in cells or whether p53 existed or not.

Taken together, these data indicate that F2 does not influence the degradation or stability of HIF-1α, raising the possibility that it may affect its translation.

We next investigated whether F2 affected the expression of HIF-1α through the PI3K-AKT-mTOR and ERK1/2 pathways, by inhibiting the HIF-1α translational regulators P70S6K and 4E-BP1 (22–24). As shown in Fig. 4A, hypoxic induction of the p-AKT, p-mTOR, p-P70S6K, p-4E-BP1 and p-ERK1/2 in both cell lines were decreased by F2 in a concentration-dependent manner. In addition, LY294002 (PI3K inhibitor), rapamycin (mTOR inhibitor) and PD98059 (MEK1 inhibitor) could inhibited the p-AKT, p-mTOR and p-ERK1/2 induced by hypoxia followed by downregulation of the level of HIF-1α (Fig. 4B and C). These results indicate that the inhibitory effect of F2 on HIF-1α protein expression is associated with the inactivation of PI3K-AKT-mTOR and MAPK-ERK1/2 pathways.

Epidermal growth factor receptor (EGFR) is overexpressed and activated in a variety of tumors, and has been reported to regulate the translation and expression of HIF-1α through the PI3K-AKT-mTOR and MAPK-ERK1/2 pathways (25,26). In this study, we found that hypoxia promoted the phosphorylation of EGFR (p-EGFR) in both U251 and Hep3B cells, and this could be abrogated in the presence of F2 (Fig. 5A). As a ligand of EGFR and one target gene of HIF-1, EGF (10 ng/ml) also could dramatically induce the expression of HIF-1α and VEGF, both of which were decreased by F2 in a concentration-dependent manner (Fig. 5B). In addition, EGF induction of HIF-1α protein and luciferase levels (P<0.05) were both inhibited by LY294002, rapamycin and PD98059 (P<0.05, Fig. 5C and D). These results are consistent with the previous report and confirm that EGFR acts as an upstream mediator of HIF-1α in both cell lines, and probably is a potential target of F2.

The results above showed that F2 could inhibit the expression levels of HIF-1α and its target gene-VEGF mRNA. Given the key roles of VEGF and HIF-1α in the regulation of tumor angiogenesis (27,28), we investigated the effect of F2 on tumor angiogenesis using the tube formation assay and HUVECs invasion assay. As shown in Fig. 6A, conditioned media (CM) collected from both cell types exposed to hypoxia were all able to promote the invasion of HUVECs (P<0.001), whereas CM collected from cells exposed to hypoxia in the presence of F2 decreased the invasion ability of HUVECs (P<0.01). A similar result was obtained in the tube formation assay (Fig. 6B). Moreover, the levels of cellular and secreted VEGF proteins induced by hypoxia in both cell types were significantly inhibited by F2 in a concentration-dependent manner (P<0.01, Fig. 7A and B). To further confirm that VEGF level is controlled by HIF-1α in both U251 and Hep3B cells, we examined whether RNA interference of HIF-1α was able to knock down the protein level of VEGF. As shown in Fig. 7C, when HIF-1α expression was knocked down, the levels of HIF-1α and VEGF were lower under normoxic conditions and unable to be upregulated by hypoxia, confirming that HIF-1α is an upstream mediator of VEGF in U251 and Hep3B cells.

In addition, hypoxia also regulates the function of endothelial cells through the VEGFR2/HIF-1 pathway, which is also a classical pathway for endothelial angiogenesis (29,30). As shown in Fig. 7D, p-VEGFR2, p-AKT and p-ERK1/2 induced by hypoxia or VEGF (10 ng/ml) were all decreased in the presence of F2 in HUVECs. Surprisingly, HIF-1α and VEGF were also inhibited by F2 at the same time (Fig. 7D), suggesting that F2 disturbs the HIF-1/VEGF pathway not only in tumor cells, but also in endothelial cells. These results suggest that F2 may be a potential anti-angiogenesis candidate which could inhibit tumor angiogenesis by targeting the HIF-1/VEGF pathway.

Hypoxia also stimulates the invasion and migration of tumor cell through activating the HIF-1 target genes, such as MMP-2 (31). In this study, both MMP-2 mRNA and protein induced by hypoxia were decreased in the presence of F2 (Figs. 1B and 8A), indicating that F2 may influence the tumor cell invasion by regulating HIF-1α expression. As shown in Fig. 8B, Transwell assay results showed that hypoxia significantly induced the invasion of U251 and Hep3B cells (P<0.01), which could be reversed by F2 (P<0.01). To further confirm that hypoxia-induced cell invasion was mediated by HIF-1α, HIF-1α siRNA was used to knock down the HIF-1α expression in both cell types. Invasion of both types of tumor cells were impaired under normoxic and hypoxic conditions when HIF-1α expression was knocked down (P<0.001, Fig. 8C), suggesting that HIF-1α mediated both U251 and Hep3B cell invasion. These results demonstrate that F2 has strong effect upon inhibiting cell invasion through downregulation of HIF-1α expression.