Inhibition of Angiogenesis and Invasion by 3,3′-Diindolylmethane Is Mediated by the Nuclear Factor–κB Downstream Target Genes MMP-9 and uPA that Regulated Bioavailability of Vascular Endothelial Growth Factor in Prostate Cancer

The growth and metastasis of prostate cancer and other tumors is dependent on the induction of new blood vessels from preexisting ones through angiogenesis (1). Growing evidence shows that prostate cancer cells secrete high levels of growth factors and matrix-degrading proteases, including fibroblast growth factor (FGF), vascular endothelial growth factor (VEGF), platelet-derived growth factor, matrix metalloproteinases (MMP; ref. 2), and serine proteases, in particular, the urokinase-type plasminogen activator (uPA; ref. 3). Many studies have documented that VEGF is a critical mediator of angiogenesis and regulates most of the steps in the angiogenic cascade, including proliferation, migration, and tube formation of endothelial cells (4, 5). VEGF exists in at least six isoforms with variable amino acid residues produced through alternative splicing: VEGF121, VEGF145, VEGF165, VEGF183, VEGF189, and VEGF206 (6). VEGF121, VEGF165, and VEGF189 are the major forms secreted by most cell types. After secretion, VEGF121 diffuses relatively freely in tissues, whereas approximately half of the secreted VEGF165 binds to heparin sulfate proteoglycans (HSPG) on the cell surface and matrix basement membrane. VEGF189 remains almost completely sequestered by HSPGs in the extracellular matrix (ECM), making the HSPGs a reservoir of VEGF that can be released by proteolysis (6). VEGF is subjected to multilevel regulation to ensure proper expression under physiologic and pathologic conditions. A wide range of cytokines, oncogenic proteins, or transcription factors, such as insulin-like growth factor 1, c-Src, Ras, nuclear factor-κB (NF-κB), and SP1, have been shown to regulate VEGF gene transcription (7–9). Hypoxic condition can also affect VEGF mRNA stability (10) and translation initiation (11). Extracellular cleavage of VEGF by uPA (12, 13) or plasmin (14) and processing by MMPs can regulate its mitogenic effect, bioavailability, and vascular patterning in tumors (15). These multilevel regulated patterns of VEGF provide various targets for the inhibition of angiogenesis induced by tumors.

Epidemiologic studies have shown a strong inverse correlation between consumption of cruciferous vegetables and a risk of prostate cancer (16). One of the bioactive components of these vegetables is indole-3-carbinol. Indole-3-carbinol, in a low-pH environment, is easily converted to its dimeric product 3,3′-diindolylmethane (DIM). DIM can inhibit the growth of tumors through induction of apoptosis and G₁ cell-cycle arrest in human prostate cancer cells (17, 18). Recent studies have shown that DIM inhibits cell proliferation, migration, invasion, and capillary tube formation of cultured human umbilical vascular endothelial cell (HUVEC) and represses vascularization in a Matrigel plug assay and tumorigenesis in human breast cancer cell xenograft model in mice (19). Another study showed that DIM inhibits angiogenesis via inhibition of growth factor–induced Ras signaling in HUVECs (20). However, the molecular mechanism by which DIM inhibits angiogenesis and invasion has not been fully elucidated. Tumor angiogenesis involves degradation of the basement membrane of the original vessel, endothelial cell activation, migration, proliferation, and formation of new capillaries. All these processes are controlled by angiogenic factors secreted either by the tumor or the surrounding stroma. During the process of tumor angiogenesis, tumor cells express high levels of angiogenic factors, such as VEGF, MMP-9, and uPA, which are important for angiogenesis, invasion, and metastasis. Little is known about whether DIM is able to inhibit angiogenesis in prostate tumor cells by regulating these angiogenic factors. In this report, we show that BioResponse DIM (B-DIM), a formulated DIM with higher bioavailability, inhibits angiogenesis and invasion by reducing the bioavailability of VEGF due to decreased expression of uPA and MMP-9, resulting from the inhibition of NF-κB DNA binding activity in human prostate cancer cells.

Cell lines and culture. Prostate cancer cell lines LNCaP (AR⁺ and responsive to androgen) and C4-2B (AR⁺ but nonresponsive to androgen) cells were maintained in RPMI 1640 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS), 10 μmol/L HEPES, 100 units/mL penicillin, and 100 μg/mL streptomycin. HUVECs were purchased from American Type Culture Collection (Manassas, VA) and cultured in F12K medium (American Type Culture Collection) supplemented with 10% FBS, 0.1 mg/mL heparin sulfate, 0.05 mg/mL endothelial cell growth factor supplement (BD Bioscience, San Jose, CA), 100 units/mL penicillin, and 100 μg/mL streptomycin. All cells were cultured in a 5% CO₂–humidified atmosphere at 37°C.

Reagent and antibody. B-DIM, a formulated DIM with higher bioavailability, was kindly provided by Dr. Michael Zeligs (BioResponse, Boulder, CO) and was dissolved in DMSO to make 50 mmol/L stock solutions and stored at −20°C in multiple aliquots. Antibody against human VEGF (SC-152, A-20) and anti-uPA antibody were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti–MMP-9 monoclonal antibody was obtained from R&D Systems (Minneapolis, MN). The monoclonal antibody to β-actin was purchased from Sigma-Aldrich (St. Louis, MO). Matrigel, growth factor–reduced Matrigel, and recombinant human VEGF165 were obtained from BD Bioscience (Bedford, MA). VEGF neutralizing antibody was obtained from Lab Vision (Fremont, CA). Anti-CD31 antibody was purchased from Ventana Medical Systems, Inc. (Tucson, AZ). VEGF quantikine ELISA kits and MMP-9 fluorescent assay kits were purchased from R&D Systems.

Matrigel in vitro HUVECs tube formation assay. LNCaP and C4-2B cells cultured in serum-free RPMI 1640 were treated with B-DIM for 24 h. The conditioned media were collected, centrifuged, transferred to fresh tubes, and stored at −20°C. Growth factor–reduced Matrigel (125 μL), after being thawed on ice, was plated in an 8-well chamber (Fisher, Hanover Park, IL). The chamber was then incubated at 37°C for 30 min to allow the Matrigel to polymerize. HUVECs maintained in F12K complete medium and starved for 4 h were trypsinized and seeded (5 × 10⁴ cells per well) in each well with 250 μL of conditioned medium from LNCaP or C4-2B cells treated with DMSO or 10 μmol/L of B-DIM in serum-free RPMI 1640 for 24 h. VEGF165 (50 ng/mL) was used as a positive control and medium only was used as a negative control. Before being added into the HUVECs, conditioned media from LNCaP and C4-2B cells were preincubated with anti-VEGF neutralizing antibody at 37°C for 1 h to confirm that the tube formation induced by conditioned medium is related to VEGF secreted by prostate cancer cells. HUVECs grown in growth factor–reduced Matrigel were treated with DMSO or 10 μmol/L B-DIM for 6 h to observe whether B-DIM directly affects tube formation of HUVECs, The chamber was incubated for 6 h. Each well was photographed using an inverted microscope with digital camera. Image analysis of tubule/capillary length was carried out using the software image analysis program Scion Image downloaded from the NIH website.¹

ELISA assay for VEGF and MMP-9. LNCaP and C4-2B cells were seeded in six-well plates. After 24 h, the cells were incubated with serum-free medium for another 24 h and then treated with B-DIM or DMSO as a vehicle control in 1% FBS for 24 h. The culture media were collected, centrifuged to remove cellular debris, and stored at −70°C until assay for VEGF and MMP-9. In addition, LNCaP and C4-2B cells were seeded in six-well plates. After 24 h, the cells were transfected with MMP-9 plasmid or pcDNA3 control plasmid using effectene transfection reagent (Qiagen, Valencia, CA). The media were removed after 18 h of transfection, and the cells were treated with 10 μmol/L B-DIM in serum-free media and incubated for 24 h. To investigate the effects of uPA on the release of VEGF, LNCaP and C4-2B cells were transfected with uPA small interfering RNA (siRNA; Santa Cruz) or control siRNA using DharmaFECT3 siRNA transfection reagent (Dharmacon, Lafayette, CO). The media were removed after 18 h of transfection, and the cells were incubated in serum-free media for 24 h. The culture media were collected, centrifuged to remove cellular debris, and stored at −70°C until assay for VEGF. The cells in the plate were trypsinizied, and the total number of cells was determined by cell counting. The assay was done using a commercially available VEGF and MMP-9 ELISA kit (R&D System) according to the manufacturer's instruction. For MMP-9 assay, the culture media were concentrated using the microcon concentrator (Millipore, Billerica, MA). Results were normalized to the cell number.

Real-time reverse transcription–PCR. LNCaP and C4-2B cells grown in 1% FBS were treated with 10 μmol/L B-DIM or DMSO as a vehicle control for 24 h. The total RNA was isolated using the Trizol reagent. One microgram of RNA was reverse transcribed using a reverse transcription system (Invitrogen, Carlsbad, CA) according to the manufacturer's instruction. Real-time PCR was used to quantify mRNA expression. Sequence of oligonucleotide primers and probes for total VEGF, VEGF121, VEGF165, and VEGF189 used in this study was described by Dr. Gustafsson (21). β-Actin was used for internal control to correct the potential variation in RNA loading. Targets and β-actin gene were run under the same conditions. All reactions were done in a 25-μL volume containing the sample cDNA, TaqMan fast universal PCR mastermix, primers, and probes. Before the PCR cycles, samples were incubated for 2 min at 50°C and 10 min at 95°C. Thermal cycles consisted of 45 cycles at 95°C for 15 s and 65°C for 1 min.

Western blot assay. LNCaP and C4-2B cells were treated as above. The cells were harvested by scraping from the wells and washed twice with cold PBS. The cell pellet was suspended in 125 mmol/L Tris-HCl (pH 6.8), sonicated for 10 s, and an equal volume of 4% SDS was added. The lysates were boiled for 10 min. Protein concentration was determined using bicinchoninic acid protein assay (Pierce, Rockford, IL). The proteins were separated by SDS-PAGE, transferred to nitrocellulose, blocked, and incubated with the following primary antibodies: VEGF (1:1,000, Santa Cruz), MMP-9 (2 μg/mL, R&D Systems), uPA (1:500, Santa Cruz), and β-actin (1:10,000, Sigma). The membrane was washed and incubated with the respective secondary antibodies conjugated with peroxidase. Protein detection was done with chemiluminescence detection system (Pierce).

Electrophoretic mobility shift assay for NF-κB DNA binding activity. LNCaP and C4-2B cells were seeded in a six-well plate. After 24 h, the cells were treated with 10 μmol/L of B-DIM or DMSO as a vehicle control and incubated for another 24 h. Nuclear extracts were prepared according to the method described by Wang et al. (22). For detecting NF-κB DNA binding activity, electrophoretic mobility shift assay (EMSA) was done by incubating 3 μg of nuclear protein of each sample with IRDye 700–labeled NF-κB oligonucleotide (LI-COR, Lincoln, NE) as described (22). Retinoblastoma protein was used as a protein loading control.

Invasion assay. The effects of B-DIM treatment on cell invasion were determined using BD BioCoat Tumor Invasion Assay System (BD Bioscience, Bedford, MA) according to the instruction of the manufacturer. Briefly, LNCaP and C4-2B cells with serum-free media containing 10 and 25 μmol/L of B-DIM or DMSO were seeded into the upper chamber of the system. Bottom wells were filled with complete media. After 24 h incubation, the cells in the upper chamber were removed and the cells invaded through the Matrigel membrane were stained with 4 μg/mL Calcein AM in PBS at 37°C for 1 h. The fluorescently labeled cells were photographed under a fluorescence microscope. The fluorescence of the invaded cells was read in ULTRA Multifunctional Microplate Reader (TECAN, Durham, NC) at excitation/emission wavelengths of 530/590 nm.

Transient transfections and reporter gene assay. The plasmid pGL3-V2274 containing the VEGF promoter, enhancer, and firefly luciferase reporter gene was kindly provided by Dr. Keping Xie (University of Texas M. D. Anderson Cancer Center). LNCaP and C4-2B cells were seeded at a density of 8 × 10³ per well in 96-well plate and incubated for 24 h. The cells were cotransfected with pGL3-V2274 or pGL3-basic plasmid and cytomegalovirus (CMV)-β-galactosidase plasmid. After 24 h of transfection, the cells were treated with 10 μmol/L of B-DIM or DMSO as vehicle control for another 24 h. Luciferase activity was assayed using Steady-Glo Luciferase Assay System (Promega, Madison, WI). β-Galactosidase activity was used as a control for transfection efficiency.

In vivo Matrigel plug angiogenesis assay. Our data showed that the condition medium from LNCaP and C4-2B cells treated with B-DIM reduced the tube formation of HUVECs in vitro through reduction in bioavailability of VEGF. To assess whether B-DIM inhibits angiogenesis in vivo, angiogenesis assay was done using male CB17 SCID mice. The mice were randomly segregated into two groups (four mice per group). C4-2B cells (4 × 10⁶ in 50 μL PBS admixed with 0.25 mL Matrigel) were injected s.c. into the bilateral flanks of each mouse (eight lesions per group). The control group of mice was gavaged with the vehicle (sesame oil), whereas the other group of mice was gavaged with B-DIM (5 mg/kg body weight, daily) for 4 weeks. On termination, Matrigel plugs with tumor cells were harvested and fixed in 10% formalin dehydrated and embedded in paraffin. Matrigel plug slices were stained with H&E and anti-CD31 antibody. SPOT Advanced Imaging software was used to capture bright-field images. Image analyses of the area of CD31-positive blood vessels, vessel number, and length of vessel perimeter in each of the entire field were carried out using software image analysis program (Scion Image downloaded from NIH website).¹

Data analysis. Experiments presented in the figures are representative of three or more different repetitions. The data are presented as the mean values ± SE. Comparisons between groups were evaluated by a two-tailed Student's t test. Values of P < 0.05 were considered to be statistically significant.

Reduced tube formation of HUVECs induced by conditioned media from LNCaP and C4-2B cells treated with B-DIM. Tumor cells can induce the formation of new blood vessels by secreting endothelial cell specific growth factors including VEGF and FGF. To determine whether conditioned media from LNCaP and C4-2B cells could induce the tube formation of HUVECs and whether conditioned media from LNCaP and C4-2B cells treated with B-DIM could reduce the tube formation, we did the tube formation assay in growth factor–reduced Matrigel in vitro. Conditioned media from LNCaP and C4-2B cells (Fig. 1A and B) as well as VEGF (Fig. 1C) were able to significantly induce the tube formation of HUVECs in 6-h incubation (compared with Fig. 1C, medium only as negative control). To further explore whether VEGF in conditioned media may be the inducer of the tube formation of HUVECs, 10 μg/mL of anti-VEGF neutralizing antibody was preincubated with conditioned media from LNCaP and C4-2B cells at 37°C for 1 h and then added into HUVECs. We found that anti-VEGF neutralizing antibody reduced the tube formation, suggesting that VEGF is the inducer of the tube formation induced by conditioned medium from LNCaP and C4-2B cells (Fig. 1A). Importantly, we found that conditioned media from LNCaP and C4-2B cells treated with 10 μmol/L of B-DIM inhibited tube formation (Fig. 1A). To rule out the possibility that inhibition of tube formation by conditioned media from LNCaP and C4-2B cells treated with 10 μmol/L of B-DIM was mediated through direct effect of B-DIM on tube formation of HUVECs, HUVECs grown in growth factor–reduced Matrigel were treated with 10 μmol/L of B-DIM for 6 h (Fig. 1C). B-DIM had no effect on tube formation within 6 h of incubation (compared with Fig. 1C, DMSO control). Figure 1B and D showed quantification of cumulative tube length of endothelial cells. These results suggest that B-DIM treatment inhibits tube formation by inhibiting secretion of proangiogenic factor VEGF in LNCaP and C4-2B cells.

B-DIM reduced cellular VEGF and the secretion of VEGF. To directly determine whether the reduced tube formation seen earlier was a consequence of the decreased secretion of VEGF, we analyzed the levels of VEGF in conditioned media from LNCaP and C4-2B cells treated with B-DIM. As shown in Fig. 2A and B, 1 and 10 μmol/L of B-DIM treatment markedly decreased the VEGF secretion from LNCaP and C4-2B cells compared with DMSO control (P < 0.05 or 0.01). Western blot analysis showed that B-DIM treatment reduced cellular VEGF in LNCaP cells but had no significant effect on cellular VEGF in C4-2B cells (Fig. 2C and D). The band with 57 kDa molecular weight corresponded with the long intact isoforms of VEGF189 predominantly associated with cell surface and ECM. The long VEGF189 may be the storage form, which is released and activated by two serine proteases, uPA and plasmin (13).

B-DIM down-regulated VEGF at the level of transcription in LNCaP but not in C4-2B cells. Houck et al. reported that VEGF bioavailability could be dually regulated at the genetic level by alternative splicing of VEGF mRNA that determines whether VEGF will be soluble or incorporated into a biological reservoir and later activated through proteolysis after plasminogen activation (14). VEGF121, VEGF165, and VEGF189 are the major forms secreted by most cell types. To determine whether the decreased secretion of VEGF from LNCaP and C4-2B cells treated with B-DIM was due to transcription down-regulation of soluble isoforms of VEGF, such as VEGF121 and VEGF165, we conducted real-time reverse transcription–PCR (RT-PCR) assay using isoform-specific VEGF primers and probes and tested the relative distribution of mRNA levels of major isoforms in LNCaP and C4-2B cells, respectively. We found that VEGF121 is the most abundant in prostate cancer cell line LNCaP and C4-2B cells; and VEGF165 is the lowest (Fig. 3A). Importantly, we found that total VEGF and VEGF121 was significantly reduced by 10 μmol/L of B-DIM treatment, but VEGF165 and VEGF189 showed less reduction at the RNA level in LNCaP cells. However, 10 μmol/L of B-DIM treatment had no significant effect on all isoforms of VEGF mRNA in C4-2B cells (Fig. 3B). To further test whether B-DIM reduced the VEGF expression at the transcriptional level, we used a reporter construct, pGL3-V2274 (VEGF 5′ flanking region from +50 to −2274), which contains a luciferase gene driven by the VEGF promoter. This reporter plasmid was transiently transfected into LNCaP and C4-2B cells. After 24 h of transfection, the cells were treated with 10 μmol/L of B-DIM for another 24 h. The luciferase activity was determined. B-DIM treatment induced significant decrease in luciferase activity of LNCaP but not C4-2B cells (Fig. 3C). Therefore, we speculated that decreased soluble VEGF in culture media from C4-2B cells treated with B-DIM may result from regulation of posttranscription or the reduced bioavailability of bound-VEGF, such as VEGF189 and VEGF165. Many studies indicated that serine protease and matrix-degrading proteases were able to release bound-VEGF into tissue culture media (13–15). uPA and MMP-9 are respectively the key serine protease and matrix-degrading protease involved in the regulation of bioavailability of VEGF; thus, we have investigated whether expression of MMP-9 could regulate the bioavailability of VEGF.

B-DIM reduced bioavailability of VEGF through repressing MMP-9 expression. MMP-9 has been shown to be expressed at higher levels in prostate cancer cell lines and prostate cancer tissue. It is known to specially cleave the type IV collagen, the major ECM, and is able to regulate bioavailability of ECM-binding VEGF via degrading the ECM (2, 23). In this study, we determined MMP-9 levels from conditioned media. As shown in Fig. 4A and B, LNCaP and C4-2B cells treated with 1, 10, and 25 μmol/L of B-DIM for 24 h resulted in significant decrease in the MMP-9 levels in culture supernatants compared with the control (treated with DMSO). In addition, cellular MMP-9 levels in the treated cells were examined by Western blot analysis. As depicted in Fig. 4A, and B (bottom), B-DIM dramatically repressed the MMP-9 expression. To explore whether reduced MMP-9 expression by B-DIM was responsible for the decreased bioavailability of VEGF, we assayed the VEGF levels in conditioned media from LNCaP and C4-2B cells transfected with MMP-9 plasmid and treated with B-DIM. The results showed that MMP-9 transfection significantly increases VEGF levels from conditioned media compared with transfection of control plasmid in LNCaP and C4-2B cells. Importantly, 10 μmol/L of B-DIM treatment for LNCaP and C4-2B cells transfected with MMP-9 plasmid dramatically reduced the release of VEGF (Fig. 4C and D). Western blot analysis showed that B-DIM repressed MMP-9 expression in LNCaP and C4-2B cells transfected with MMP-9 plasmid. These results revealed that B-DIM could reduce bioavailability of VEGF through repressing MMP-9 expression. Because uPA also plays an important role in the regulation of VEGF bioavailability, we next examined the effect of B-DIM on uPA and its consequence in VEGF bioavailability.

uPA regulated the bioavailability of the VEGF and B-DIM repressed uPA expression in LNCaP and C4-2B cells. Many studies have shown that uPA is involved in the angiogenesis induced by tumors (24). uPA, directly or via plasmin formation, leads to the release or activation of many angiogenic growth factors, such as basic FGF (bFGF) and VEGF. In this study, down-regulation of uPA by siRNA significantly repressed the release of VEGF in LNCaP and C4-2B cells (Fig. 5A). This result suggests that B-DIM may decrease the bioavailability of VEGF via repressing the uPA expression. In this study, we found that 10 μmol/L of B-DIM treatment significantly decreased the uPA expression in LNCaP and C4-2B cells compared with DMSO control (Fig. 5B). These results suggest that B-DIM may regulate the bioavailability of bound-VEGF by suppressing MMP-9 and uPA expression to inhibit angiogenesis induced by prostate cancer cells. It is known that VEGF, uPA, and MMP-9 are transcriptionally regulated by NF-κB; thus, we have tested whether B-DIM regulated expression of VEGF, uPA, and MMP-9 is mediated through NF-κB.

B-DIM inhibited the NF-κB DNA binding activity. NF-κB activation mediates expression of a large number of target genes, such as VEGF, MMP-9, and uPA, which are involved in angiogenesis and invasion (25). To further explore the mechanism by which B-DIM regulates the expression of VEGF, MMP-9, and uPA, we detected NF-κB DNA binding activity. Nuclear extract from cells treated with B-DIM or DMSO control were subjected to EMSA. Figure 5C showed that treatment with 10 μmol/L of B-DIM for 24 h strongly inhibited NF-κB DNA binding activity in both LNCaP and C4-2B cells, suggesting that down-regulation of VEGF, MMP-9, and uPA by B-DIM could be partly due to the inhibition of NF-κB DNA binding activity, especially in LNCaP cells. However, the regulatory mechanism of VEGF, MMP-9, and uPA by B-DIM in C4-2B cells could be more complex compared with those in LNCaP cells.

B-DIM inhibited LNCaP and C4-2B cell invasion. The levels of VEGF, MMP-9, and uPA are known to correlate with prostate cancer invasion and metastasis. Our previous report showed that down-regulation of VEGF and MMP-9 by siRNA inhibited the invasion of human pancreatic cancer cells (22). In this study, we found that B-DIM treatment was able to lower the bioavailability of VEGF by inhibiting MMP-9 and uPA expression and NF-κB DNA binding activity. We have speculated that B-DIM may inhibit cell invasion of LNCaP and C4-2B due to inactivation of NF-κB, MMP-9, and uPA. Therefore, we did the Matrigel invasion assay. The results showed that 10 and 25 μmol/L of B-DIM significantly inhibited LNCaP (Fig. 6A) and C4-2B (Fig. 6B) cell invasion. These results suggest that B-DIM inhibits LNCaP and C4-2B cell invasion by decreasing expression or bioavailability of VEGF via inhibiting matrix-degrading proteases, such as uPA and MMP-9.

B-DIM inhibits in vivo angiogenesis. The effect of B-DIM on angiogenesis was investigated in vivo using a s.c. implanted Matrigel plug assay. Histologic analysis showed that a larger number of endothelial cells were well organized into capillary or large vascular cavities engorged with RBC in the control group, whereas B-DIM treatment strongly inhibited the formation of large vessels (Fig. 6C,, middle and right show histologic pictures of peripheral and deeper areas of Matrigel plugs, respectively). In addition, B-DIM treatment also significantly reduced the vessel number (Fig. 6D , left), vascular area (middle) and total length of vascular perimeter (right) relative to the control group. These results provide strong evidence in support of the antiangiogenic activity of B-DIM in vivo, which is believed to be partially responsible for the antitumor activity of B-DIM as observed previously by our laboratory in a prostate cancer animal model (26).

Angiogenesis is a tightly regulated event critical for tumor growth beyond 2 to 3 mm in diameter and is an essential component of tumor metastasis (1). Tumor angiogenesis involves degradation of the basement membrane of the original vessel, endothelial cell activation, migration, proliferation and the formation of new capillaries. All of these processes are controlled by angiogenic factors secreted either by the tumor or the surrounding stroma. VEGF, known as a potential angiogenic factor, is coded by a single gene and coding region that contains eight exons separated by seven introns (27). VEGF exists in at least six isoforms produced through alternative splicing (6). VEGF121, VEGF165, and VEGF189 are the major forms secreted by most cell types. Exons 1, 2, 3, 4, 5, and 8 are common to all isoforms. In addition, VEGF165 contains exon 7 and VEGF189 contains exons 6 and 7. Exons 6 and 7 have been shown to encode the ECM-binding domain of the VEGF. This domain is able to bind HSPGs and other matrix proteins (6). Thus, after secretion, VEGF189 and part of VEGF165 bind to ECM and cell membrane. Extracellular enzymes, such as heparinases, plasmin, uPA, and MMP are able to release bound-VEGF and regulate the VEGF bioavailability.

In the present study, we showed that conditioned media from LNCaP and C4-2B cells induced tube formation of HUVECs. Conditioned media from LNCaP and C4-2B cells preincubated with anti-VEGF neutralizing antibody decreased tube formation of HUVECs. These results suggest that VEGF secreted by prostate cancer cells is a potential inducer of HUVEC tube formation. Interestingly, conditioned media from LNCaP and C4-2B cells treated with 10 μmol/L of B-DIM induced less tube formation of HUVECs, and only the culture media containing 10 μmol/L of B-DIM had no such effects on tube formation within 6 h. Therefore, we speculate that decreased VEGF levels in conditioned media by B-DIM treatment are responsible for reduced tube formation. As predicted, we found that B-DIM decreased VEGF levels in conditioned media from LNCaP and C4-2B cells. Because VEGF is tightly regulated through transcriptional and posttranscriptional levels, we explore the transcriptional mechanism by which B-DIM may regulate VEGF. We did real-time RT-PCR using isoform-specific primers and probes and VEGF reporter gene assay. We showed that B-DIM inhibited VEGF transcription in LNCaP cells but not in C4-2B cells. Houck et al. indicated that VEGF bioavailability could be dually regulated at the genetic level by alternative splicing of VEGF mRNA that determines whether VEGF will be soluble or incorporated into a biological reservoir and later activated through proteolysis after plasminogen activation (14). It is well known that uPA and MMP-9 can release bound-VEGF and regulate VEGF bioavailability. In this study, we found that B-DIM repressed the secretion of MMP-9 and protein synthesis in both cell lines and reduced the active uPA levels in LNCaP and C4-2B cells. These results suggest that B-DIM might inhibit tube formation of HUVECs induced by VEGF secreted from prostate cancer cells. In LNCaP cells, B-DIM decreased the bioactive VEGF by repressing VEGF gene transcription and reducing the bioavailability of VEGF bound to HSPGs in the ECM and cell membrane, which is released due to degradation of ECM by MMP-9 and cleavage by uPA. However, in C4-2B cells, B-DIM suppressed the bioactive VEGF mainly by reducing the bioavailability of VEGF via decreasing uPA and MMP-9 activity.

Increasing evidence suggests that MMP-9 contributes to the formation of microenvironment conducive to promote tumor growth and angiogenesis through increasing the association of VEGF and VEGFR2 (23, 28). MMP-9 can release bound VEGF from ECM via its proteolytic activity. Bergers et al. reported that the regulation of VEGF in the extracellular environment is implicated in the angiogenic switch. They found that MMP-9 could render normal islets angiogenic releasing VEGF, whereas MMP-9 inhibitor could reduce angiogenic switching and tumor growth (29). A recent study has shown that VEGF bioavailability is regulated by MMPs through intramolecular processing, in which a subset of MMPs including MMP-9 could cleave matrix-bound isoforms of VEGF that release soluble fragments. Furthermore, MMP-cleaved VEGF has a distinct role in vascular patterning in tumors (15). In the current study, we found that B-DIM inhibited the secretion of MMP-9 and protein synthesis in both cell lines. MMP-9 transfections significantly increase VEGF levels from conditioned media compared with transfection of control plasmid in LNCaP and C4-2B cells. Importantly, 10 μmol/L of B-DIM treatment for LNCaP and C4-2B cells transfected with MMP-9 plasmid dramatically reduced the release of VEGF. Western blot analysis showed that B-DIM repressed MMP-9 expression in LNCaP and C4-2B cells transfected with MMP-9 plasmid, which indicated that B-DIM may regulate the MMP-9 expression through posttranscriptional mechanism. In addition, MMP-9 plasmid was constructed by inserting MMP-9 ORF into pcDNA3.1 vector. MMP-9 expression is driven by the CMV immediate/early promoter and enhancer containing many transcription factor binding sites, such as AP2, NF-κB, and cAMP-responsive element binding protein (30). In this study, we found that B-DIM inhibited NF-κB DNA binding activity. Thus, B-DIM may repress the MMP-9 expression in cells transfected with MMP-9 plasmid via inhibiting NF-κB DNA binding activity. Taken together, these results suggest that B-DIM decreases the secretion of VEGF and reduces the tube formation induced by tumor cells via inhibiting MMP-9 expression, leading to decreased bioavailability of the VEGF.

uPA is a highly specific serine protease which catalyzes plasmin formation from plasminogen. uPA, directly or via plasmin formation, leads to the release or activation of many angiogenic growth factors, such as bFGF and VEGF. Plouet et al. showed that uPA was able to cleave the VEGF189 without interaction with heparansulfate or sulfate modifications of the core proteoglycan into a 40 kDa isoform with more active mitogen (13). Therefore, VEGF189 released by MMP-9 via degradation of the ECM could be a substrate for uPA, and as such uPA, together with MMP-9, plays important roles in regulating the bioavailability of the VEGF. Plouet et al. also found that maturation of VEGF189 by uPA cleavage acquires the ability to activate KDR/FlK-1 (VEGFR-2). Native VEGF189 binds to Flt-1 (VEGFR-1) but not to VEGFR-2 (13). VEGFR-1 does not seem to transduce mitogenic signals in endothelial cells (31). uPA cleaves native VEGF189 into a diffusible isoform of 40 kDa with strong mitogenic activity which is able to activate VEGFR-2 and induce proliferation (13). In this study, we found that down-regulation of uPA by siRNA could reduce the release of VEGF. We also found that B-DIM inhibited uPA and MMP-9 expression, which protected native VEGF189 bound to HSPGs from cleavage by uPA, thereby inhibiting the mitogenic activity of VEGF for the proliferation of HUVECs.

NF-κB transcription factor is constitutively activated in prostate cancer cells and has been shown to contribute to the development and/or progression of prostate cancer (25). Moreover, angiogenesis, invasion, and metastasis are necessary for the development and progression of prostate cancer. Increasing evidence suggests that inhibition of NF-κB activity could suppress the angiogenesis, invasion, and metastasis by down-regulating the expression of NF-κB downstream target genes, such as VEGF, uPA, and MMP-9 (32, 33). NF-κB is normally retained in the cytoplasm in an inactive form, bound by inhibitory proteins called inhibitors of κB (IκB) in the most resting cells. NF-κB activation involves its release from its inhibitor and its subsequent translocation from the cytoplasma to the nucleus, where it binds to cognate sequences in the promoter region of many target genes. Therefore, NF-κB DNA binding activity is a hallmark for its activation. In this study, we found that B-DIM treatment inhibited the NF-κB DNA binding activity and decreased the expression of NF-κB downstream target genes, such as VEGF, MMP-9, and uPA, leading to the reduction of HUVEC tube formation in vitro and in vivo angiogenesis induced by the decreased VEGF secreted from prostate cancer cells and thereby inhibited LNCaP and C4-2B cell invasion. In a previous study, we found that DIM reduced phosphorylation of IκBα and blocked the translocation of p65, a subunit of NF-κB to the nucleus. DNA binding analysis and transfection studies with IκB kinase cDNA revealed that overexpression of IκB kinase mediates IκBα phosphorylation, which activates NF-κB and this activation was completely abrogated by DIM treatment in breast cancer cells (34). However, the exact mechanism by which B-DIM regulates the NF-κB activity, resulting in the inhibition of angiogenesis in prostate cancer, remains to be further investigated.

In summary, our results suggest that B-DIM reduces the bioavailability of VEGF by inhibiting NF-κB DNA binding activity leading to decreased MMP-9 and uPA expression, resulting in the inhibition of angiogenesis and invasion of prostate cancer cells. Compared with the inhibition of tube formation of HUVECs, B-DIM exhibits more inhibitory activity on angiogenesis in vivo, suggesting that B-DIM not only reduces the bioavailability of VEGF but also inhibits paracrine signaling of VEGF on endothelial cells in vivo. These results provide strong evidence in support of the antiangiogenic activity of B-DIM in vivo, which could contribute to the antitumor activity of B-DIM as previously supported by our laboratory in an animal model of prostate cancer (26).

Grant support: Department of Defense Prostate Cancer Research Program grant DAMD17-03-1-0042 and the National Cancer Institute, NIH grant 5R01CA108535-03 (F.H. Sarkar).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Keping Xie (University of Texas M. D. Anderson Cancer Center) for providing the plasmid pGL3-V2274 containing the VEGF promoter, enhancer, and firefly luciferase reporter gene and Dr. Rafael Fridman (Department of Pathology, Wayne State University School of Medicine, Detroit, Michigan) for providing the MMP-9 plasmid.