Synergistic Down-Regulation of Urokinase Plasminogen Activator Receptor and Matrix Metalloproteinase-9 in SNB19 Glioblastoma Cells Efficiently Inhibits Glioma Cell Invasion, Angiogenesis, and Tumor Growth¹

Tumor progression involves modulation of tumor cell adhesion during migration and the degradation of ECM³ during invasion. An intricate balance of proteases, their activators, and their inhibitors regulate both these processes during tumor invasion. ECM-degrading proteinases can be divided into three classes: (a) serine proteinases; (b) MMPs; and (c) cysteine proteinases. The urokinase plasminogen activator system is known to play a pivotal role in both tumor cell adhesion and matrix degradation. The serine protease uPA binds to its cell surface receptor uPAR and converts plasminogen to plasmin, a broad specificity serine protease with potent ECM-degrading properties. On its activation, plasmin degrades various ECM components, including fibrin, fibronectin, VN, laminin, and gelatins (1). In addition, plasmin can activate latent elastase and the MMPs, which are potent enzymes that can digest a variety of ECM components, including collagens, proteoglycans, elastin, laminin, and fibronectin.

Up-regulation of uPA (or its receptor) is associated with increased aggressiveness in diverse tumor types (2). In murine tumor models, expression or administration of uPAR antagonists has a marked inhibitory effect on the metastatic ability of cancers cells (3), primary tumor growth (4), and the down-regulation of uPAR, which leads to dormancy of carcinoma cells in vivo (5). In our earlier studies on human gliomas, expression of uPAR was greater in the high-grade tumors than in low-grade gliomas (6). The binding of uPA to uPAR is thought to play a major role in the invasion of glioblastoma cells into normal brain by concentrating the proteolytic activity at the leading edge of the tumor (6, 7). Stable transfection of a human glioblastoma cell line with antisense uPAR-expressing plasmid DNA has been shown to reduce the invasive properties of these cells both in vitro (8) and in vivo (9). Conversely, transfection of H4, a low-grade neuroglioma cell line which normally expresses low levels of uPAR, with full-length uPAR cDNA increased uPAR expression; the transfected cells showed a high invasive capability compared with the parental cells (10). Our studies on adenovirus-mediated delivery of the antisense uPAR gene showed suppression of glioma invasion and tumor growth (11). Taken together, these findings suggest that uPAR could be an important target for preventing cancer proliferation.

In addition, ligand binding of uPA to uPAR initiates interaction between a number of cell surface proteins, e.g., VN, integrin receptors, and caveolin at focal adhesion sites (12). This process leads to intracellular phosphorylation of kinases, including cytokeratins, Src kinases, activation of the Jak/Stat pathway, and subsequently, kinases of the mitogen-activated protein kinase signaling pathways (13). There is evidence that uPA binding to uPAR-mediated signaling events results in the expression of cathepsin B and M_r 92,000 gelatinase (MMP-9) in monocytic cells (14). The ability of tumor cells to migrate into ECM in vitro is known to be mediated by the cooperative expression of a family of adhesion receptors called integrins and cell surface proteinases, such as uPA and MMPs (15). MMPs are proteolytic enzymes and their basic mechanism of action, the degradation of proteins, regulates various cell behaviors relevant to cancer biology. These include cancer cell growth, differentiation, apoptosis, migration, and invasion, as well as the regulation of tumor angiogenesis and immune surveillance. MMP-2 and MMP-9 are the two most abundant MMPs found in gliomas (16). We have shown previously that MMP-2, MMP-9, and MT-MMP are found in greater amounts during the progression of human gliomas and in glioma cell lines in culture (17, 18, 19). MMP-9 was also shown to be important in endothelial cell morphogenesis and the formation of capillaries in glial/endothelial cocultures in vitro (20). Furthermore, antisense oligonucleotides that blocked MMP-9 gene expression in SNB19 glioblastoma cells inhibited tumor formation in nude mice, thereby providing evidence that MMP-9 expression facilitates glioma invasion in vivo (21, 22).

Gliomas are characterized by a high proliferation rate, extensive angiogenesis, and marked local invasion that makes these tumors resistant to the conventional treatment modalities of surgery, chemotherapy, and radiotherapy (23). There are great hopes that gene therapy can manage malignant tumors, including gliomas, and various approaches to gene therapy are being tested currently (24). On the basis of these previous observations, we hypothesized that manipulation of the expression of these two proteases in a single step may have a synergistic effect in inhibiting glioma tumor growth. In the present study, we constructed a replication-deficient recombinant adenovirus (Ad-uPAR-MMP-9) to efficiently deliver antisense uPAR and antisense MMP-9 genes to down-regulate uPAR and MMP-9 levels in human gliomas. We show that down-regulation of these two protease molecules has an additive effect in inhibiting glioma invasion and angiogenesis.

We have constructed previously an adenovirus expressing an antisense message for the uPAR gene using a pAd-uPAR vector containing the 300-bp DNA fragment of the 5′ end of the uPAR gene in the antisense orientation with a CMV promoter and bovine growth hormone polyadenylation signal (11). We then subcloned a 528-kb fragment of the 5′ end of the MMP-9 cDNA in an antisense orientation driven with its own independent promoter elements CMV and SV40 polyadenylation signal into pAd-uPAR. The sequence of the resulting clone pAd-uPAR-MMP-9 was confirmed, and the plasmid construct was cotransfected with 40–45-kb pJM17 vector into human embryonic kidney 293 cells to isolate recombinant Ads (25). Recombinant viral plaques were identified and amplified by PCR using primers specific for CMV and the SV40 polyadenylation signal. The virus particles were propagated in 293 cells and purified by ultracentrifugation in cesium chloride step gradients (25). The viral DNA was also sequenced to confirm the configuration of the expression cassette. A standard plaque assay was performed to determine the infectious particles. The control virus Ad-CMV has a CMV promoter and bovine growth hormone poly (A) signal but no gene insert in the E1-deleted region.

We used the established human glioma cell line SNB19 for this study. The 293 human transformed embryonal kidney cell line was purchased from American Type Culture Collection (Manassas, VA). All cell lines were cultured at 37°C in media recommended by suppliers in a humidified atmosphere with 5% CO₂. Viral stocks were suitably diluted in serum-free medium to obtain the desired MOI or PFU and added to cell monolayers and tumor cell spheroids (1 ml/60-mm dish or 3 ml/100-mm dish), and the cells were then incubated at 37°C for 30 min. The necessary amount of culture medium with 10% FCS was added, and cells were incubated for the desired time periods.

SNB19 cells (1 × 10⁴) were seeded on VN-coated, eight-well chamber slides, incubated for 24 h, and infected with 100 MOI of Ad-CMV, Ad-uPAR, or Ad-uPAR-MMP-9. After another 72 h, cells were fixed with 3.7% formaldehyde and incubated with 1% BSA in PBS at room temperature for 1 h for blocking. After the slides were washed with PBS, either IgG anti-uPAR (rabbit) or IgG anti-MMP-9 (mouse) was added at a concentration of 1:500. The slides were incubated at room temperature for 1 h and washed three times with PBS to remove excess primary antibody. Cells were then incubated with antimouse Texas red conjugate or antirabbit FITC conjugate IgG (1:500 dilution) for 1 h at room temperature. The slides were then washed three times and covered with glass cover slips, and fluorescent photomicrographs were obtained.

Total cell lysates were prepared in extraction buffer containing 0.1 m Tris (pH 7.5), Triton X-114 (1.0%), EDTA (10 mm), aprotinin, and phenylmethylsulfonyl fluoride. Protein (20 μg) from these samples were separated under nonreducing conditions and transferred onto nylon membranes. The membranes were probed with rabbit antihuman uPAR polyclonal antibody (#399R; American Diagnostics, Inc., Greenwich, CT). Antirabbit horseradish peroxidase was used as the secondary antibody, and membranes were developed according to an enhanced chemiluminescence protocol (Amersham, Arlington Heights, IL).

Spheroids of SNB19 cells were prepared by suspending 2 × 10⁶ SNB19 cells in DMEM, seeding cells onto 100-mm tissue culture plates coated with 0.75% agar, and culturing cells until spheroid aggregates formed. Spheroids measuring ∼150 μm in diameter (∼4 × 10⁴ cells/spheroid) were selected and infected with adenovirus vectors (Ad-CMV, Ad-uPAR, or Ad-uPAR-MMP-9) at indicated MOI. Three days after infection, a single glioma spheroid was placed in the center of each well in VN-coated, 96-well microplates, and 200 μl of serum-free medium were added to each well. Spheroids were cultured at 37°C for 72 h, after which the spheroids were fixed and stained with Hema-3, and migration from the spheroids was assessed under light microscopy.

SNB19 cells were infected with the indicated MOI of Ad CMV, Ad uPAR or Ad uPAR-MMP-9. The conditioned media and cell extracts were collected, and gelatin zymography was performed to detect MMP-9 levels as described previously (26).

SNB19 cells were infected with 100 MOI of Ad-CMV, Ad-uPAR, or Ad-uPAR-MMP-9 for 4 days, and in vitro invasion of glioma cells was measured by the invasion of cells through Matrigel-coated (Collaborative Research, Inc., Boston, MA) transwell inserts (Costar, Cambridge, MA). Briefly, transwell inserts with an 8-μm pore size were coated with a final concentration of 0.7 mg/ml Matrigel, cells were trypsinized, and 200 μl of cell suspension (1 × 10⁶ cells/ml) were added in triplicate wells. After a 24-h incubation period, cells that passed through the filter into the bottom wells were quantitated as described earlier (10, 11) and expressed as a percentage of the sum of cells in the top and bottom wells. Cells on the bottom side of the membrane were fixed, stained with Hema-3, and photographed.

Multicellular glioma spheroids produced from SNB19 cells were cultured in 35-mm Petri dishes coated with 0.75% Noble agar prepared in DMEM. Briefly, 3 × 10⁶ cells were suspended in 10 ml of medium, seeded onto 0.75% agar plates, and cultured until spheroids were formed. Spheroids of 100–200 μm in diameter were selected and infected with the indicated MOI of Ad-CMV and Ad-uPAR-MMP-9. After 3 days of infection, tumor spheroids were stained with the fluorescent dye 1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanineperchlorate and confronted with fetal rat brain aggregates that were stained with 3,3′-dioctadecyloxacarbocyanine perchlorate. Progressive destruction of fetal rat brain aggregates and invasion of SNB19 cells were observed by phase contrast microscopy and photographed as described previously (9).

SNB19 cells (2 × 10⁴) were seeded in eight-well chamber slides and infected with the indicated MOIs of Ad-CMV, Ad-uPAR, and Ad-uPAR-MMP-9. After a 24-h incubation period, the medium was removed, and 4 × 10⁴ human dermal endothelial cells were seeded and allowed to coculture for 72 h. Endothelial cells were stained for factor VIII antigen for 1 h after fixing the cells in 3.7% formaldehyde and blocking with 2% BSA. Factor VIII antibody was purchased from the DAKO Corp. (Carpinteria, CA). The cells were washed with PBS and incubated with FITC-conjugated secondary antibody for 1 h. The specimens were then washed and examined with confocal scanning laser microscopy (20).

For the intracranial tumor model, SNB19 cells that express GFP were infected in culture with 100 MOI of Ad-CMV or 50 MOI of Ad uPAR-MMP-9 for 5 days, trypsinized, counted, and injected intracranially into nude mice. At the end of the 4-week follow-up period (e.g., when the control mice start showing symptoms), the mice were killed via cardiac perfusion with formaldehyde, the brains were removed, and frozen sections were prepared and observed for GFP fluorescence. The 3–5-μm sections were blindly reviewed and scored semiquantitatively for tumor size in each case. The average cross-sectional diameter was used to calculate tumor size and compared with the controls and treated group. The variation between the sections in each group was <10%. For the tumor regression experiments, U87-MG cells (5 × 10⁶) were s.c. injected into nude mice, and tumor formation was followed for ≤4 weeks. After 8–10 days, when tumor size had reached 4–5 mm, the mice were injected with 5 × 10⁸ PFUs of either Ad-CMV or Ad-uPAR-MMP-9 every other day for a total of five times. Tumor size was measured every 2^nd day, and tumor volume was calculated with the formula 1/6Π (R_max)² × (R_min)², where R_max and R_min are the maximum and minimum tumor radii, respectively.

Western blot analysis was performed to examine the effect of the replication-deficient Ad-CMV, Ad-uPAR, and Ad-uPAR-MMP-9 infections on uPAR protein levels in SNB19 cells. Fig. 1,A shows a dose-dependent decrease in uPAR protein levels. There was a significant decrease in cells infected with 25 MOI; cells infected with 50 MOI of Ad-uPAR-MMP-9 had a 90% reduction in uPAR protein levels. The decrease in uPAR protein was also related to time (Fig. 1 B). SNB19 cells infected at 50 MOI showed a significant decrease by day 3, and the decrease reached 90% by day 4. A similar decrease in uPAR levels by the single gene construct Ad-uPAR from the standpoint of dose and time was seen at a much higher dose (100 MOI) and later time interval (day 6).

Gelatin zymography was performed to examine the effect of Ad-uPAR-MMP-9 infection on MMP-9 enzymatic activity in SNB19 cells. Fig. 1,C shows that the MMP-9 enzymatic activity (M_R 92,000) decreased in a dose-dependent fashion with increasing MOI. Densitometric quantification of MMP-9 enzymatic activity showed a significant decrease in the cells infected with Ad-uPAR-MMP-9 at 50 MOI. This decrease reached >90% of the cells infected with Ad-uPAR-MMP-9 at 100 MOI (with the percentage decrease determined relative to those in Ad-CMV-infected control cells). The decrease in MMP-9 enzymatic activity was also related to time, because SNB19 cells infected with 50 MOI of Ad-uPAR-MMP-9 showed a 90% decrease in MMP-9 activity by day 5 (Fig. 1 D).

Immunohistochemical analyses of SNB19 cells infected with Ad-uPAR-MMP-9 showed intense staining of uPAR and MMP-9 in mock-infected cells and was comparable with that of the Ad-CMV-infected cells (Fig. 2,A). Very weak signals were detected in Ad-uPAR-MMP-9-infected cells, and Ad-uPAR-infected cells showed intermediate staining when compared with mock- or Ad-CMV-infected cells (Fig. 2,A). A tube formation assay was used to determine the capability of the bicistronic virus to regulate angiogenesis in tumors. In the presence of SNB19 cell, the endothelial cells formed capillary-like structures within 18 h. As seen in Fig. 2 B, 100 MOI of Ad-uPAR could repress capillary formation by ∼50% when compared with mock- or Ad-CMV-infected cocultures. Endothelial cells cocultured with only 10 MOI of Ad-uPAR-MMP-9 could inhibit capillary formation by ∼80%. Increasing the MOI of the bicistronic vector progressively inhibited capillary-like structures in a dose-dependent manner.

The control spheroids showed significant cell migration and were comparable with the spheroids infected with Ad-CMV (Fig. 3 A). The spheroids infected with 100 MOI of Ad-uPAR showed significant reduction in cell migration. Spheroids infected with 10 MOI of Ad-uPAR-MMP-9 showed significantly less migration compared with that of the control and Ad-uPAR-infected spheroids. SNB19 spheroids infected with 25 and 50 MOI of Ad-uPAR-MMP-9 did not attach to the VN-coated chamber slides in all four replicates, and, no cell migration was detected.

We used two models (Matrigel invasion and spheroid) to study the effect of the Ad-uPAR-MMP-9 bicistronic vector on glioma invasion. Cells infected with Ad-CMV, Ad-uPAR, and Ad-uPAR-MMP-9 were allowed to invade through Matrigel for 24 h. The staining of Ad-uPAR- or Ad-uPAR-MMP-9-transfected cells that invaded through the Matrigel was significantly less compared with mock- and Ad-CMV-infected cells (Fig. 3,B). Only 25% of Ad-uPAR-infected cells invaded compared with the Ad-CMV-infected cells. In the Ad-uPAR-MMP-9-infected cells, a 10-fold decrease in the invasive potential of the glioma cells was seen when compared with the Ad-CMV-infected cells (Fig. 3 C).

To study the invasion of SNB19 cells in a more three-dimensional system, tumor spheroids and fetal rat brain aggregates stained with the fluorescent dyes 1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanineperchlorate or 3,3′-dioctadecyloxacarbocyanine perchlorate were cocultured. Staining of the cells with these dyes allows better visualization and characterization of the invasive pattern via confocal laser scanning microscopy than other in vitro invasion assays. Spheroids of multicellular SNB19 cells infected with mock, 25, 50, or 100 MOI of Ad-uPAR-MMP-9 were confronted with fetal rat brain aggregates, and images were obtained using confocal microscopy. In the Ad-CMV-infected spheroids, complete invasion of the fetal rat brain aggregate was observed after 72 h. In the case of spheroids infected with 25 MOI of the bicistronic adenovirus, only a small region of cell colocalization was seen, and no significant invasion was noticed at 72 h. In the case of spheroids infected with 50 MOI of the same virus, colocalization was seen but significantly less than that seen in the 25 MOI-infected spheroids. In the case of the spheroids infected with 100 MOI of the virus, no visible colocalization was seen, and the glioma cell spheroids completely failed to invade the rat brain aggregates (Fig. 4,A). Glioma spheroids infected with Ad-CMV progressively invaded by ∼90% at 72 h. However, glioma spheroids infected with Ad-uPAR-MMP-9 at 25, 50, and 100 MOI in cocultures with fetal rat brain aggregates invaded only 2–15% at 72 h. (Fig. 4 B).

SNB19 variants that express GFP were infected with mock, Ad-CMV, or Ad-uPAR-MMP-9 (50 MOI) and then injected intracerebrally into nude mice. All 10 of the mice in each group injected with mock- and Ad-CMV-infected cells developed tumors, whereas none of the animals injected with Ad-uPAR-MMP-9-infected cells developed any significant tumors over a 4–5 week follow-up period (Fig. 4,C). Quantitation of tumor size showed a significant reduction (P < 0.001) in Ad-uPAR-MMP-9-infected cells compared with mock and Ad-CMV-infected cells (Fig. 4,D). In a separate experiment, the in vivo intratumoral effect of Ad-uPAR-MMP-9 injection was analyzed in nude mice bearing U-87 s.c. tumors. When tumor size reached ∼5–6 mm in diameter, 10 mice in each group received three intratumoral injections of either Ad-CMV or Ad-uPAR-MMP-9 at 5 × 10⁸ PFU every 48 h. Tumor growth was recorded from the first injection (day 0). All 10 mice injected with the Ad-uPAR-MMP-9 vector showed tumor regression beginning on day 5 after the third injection, and regression continued for 20 days, at which ∼80% of inhibition was observed in the treated group (Fig. 5,A). More than 90–95% inhibition of tumor growth in the Ad-uPAR-MMP-9-treated group was observed compared with the Ad-CMV-injected tumors (Fig. 5,A). Fig. 5 B shows that Ad-uPAR-MMP-9-treated animals showed reduced tumor size compared with Ad-CMV-treated animals.

The study of glioma invasion includes various aspects, such as glioma cell genotype, its ability to secrete proteases, which can break down extracellular barriers thereby facilitating motility, the relationship between the tumor cells and surrounding stroma, and finally, the mechanics of the cytoskeleton that lead to the highly invasive phenotype. As gene therapy technology improves, there are an increasing number of therapeutic agents that permit manipulation of cancerous cells. Targeting specific molecules unique to tumor cells by antisense inhibition was shown to have potential effectiveness in cancer therapy. Because many clinical studies are already based on protease activity inhibition and it is known that serine proteinases, such as uPA, and metalloproteinases, such as MMPs, are essential for cell invasion, we simultaneously targeted the uPA receptor and a metalloprotease (uPAR and MMP-9) by using an antisense gene delivery strategy with a replication-deficient recombinant Ad vector. The anticancer effect of Ad-uPAR-MMP-9 was demonstrated in in vitro and in vivo studies using the SNB19 and U87 MG glioma cell lines. We demonstrated that adenoviral delivery of antisense uPAR and MMP-9 genes has an additive effect in tumor regression as compared with our previous studies on the expression of the single Ad-mediated antisense uPAR and antisense MMP-9 constructs in SNB19 cells.

Ad-uPAR-MMP-9 infection significantly reduced uPAR and MMP-9 levels in SNB19 cells as determined by immunohistochemical analysis and Western blotting. uPAR and MMP-9 protein expression decreased in a dose-dependent fashion with a >90% decrease in cells infected at 50 MOI. This decrease was more significant when compared with the Ad-uPAR infection where a less similar decrease in the protein levels was observed when the cells were infected with 100 MOI of Ad-uPAR. A similar decrease in MMP-9 enzyme activity was observed as determined by gelatin zymography. We then determined the invasive potential of SNB19 cells after infection with Ad-uPAR-MMP-9 and compared it with the vector controls and Ad-uPAR infection in Matrigel and spheroid assays. Ad-uPAR-MMP-9 infection had a more superior effect in inhibiting glioma invasion in both the model systems compared with Ad-uPAR infection. Several reports have demonstrated the role of uPAR and MMP-9 in ECM degradation, which is a key step for cancer invasion. uPAR levels correlated strongly with metastatic potential in human cancer cell lines of melanoma, breast, lung, and colon (26, 27, 28). Previous studies have shown that the uPAR antibody reduces tumor cell invasion in colon cancers and glioblastomas (29, 30). Our previous work with stably transfected antisense uPAR SNB19 glioma cells failed to form tumors in nude mice (9), and the adenovirus-mediated delivery of antisense uPAR gene suppressed glioma invasion and growth (11). Similarly, reducing uPAR levels by using antisense oligonucleotides was also found to inhibit tumor growth, invasion, and metastasis in some cancers (3, 31). However, other studies on colon cancer metastasis have shown that antisense uPAR mRNA could only partially control metastasis. This may be attributable to several factors in the uPA system, including uPAR binding capacity, the amount of uPA available for binding, presence of the endogenous uPA inhibitor PAI-1, as well as the ratio of potent active uPA and inactive ligands occupied on uPAR. All these factors are involved in determining the capacity of tumor invasion and metastasis. In addition, other proteases, including matrix metalloproteases, may compensate for the lack of uPA activity and uPAR protein. Therefore, apart from the reduction of uPAR expression, we simultaneously targeted MMP-9 to obtain maximum inhibition of glioma invasion.

We also assessed the effect of the bicistronic construct in migration because it is one of the first steps in invasion. Our results show that SNB19 spheroids infected with Ad-uPAR-MMP-9 have significantly reduced migration as compared with parental and vector controls, as well as more distinct inhibition than the Ad-uPAR-infected SNB19 spheroids. In addition to regulating plasminogen activation at the cell surface, recent studies indicate that uPAR possesses other cellular activities, including an ability to generate intracellular signal transduction, regulate integrin function, and directly bind VN (12). It has also been shown that disruption of the caveolin-integrin-uPAR complex inactivates signaling through extracellular signal-regulated kinase and adhesion and migration in kidney epithelial and smooth muscle cells (12). Studies on leukocyte migration to the sites of inflammation using uPAR^−/− demonstrated that an intact uPAR molecule, probably by activating β2 integrins, is required for in vivo transendothelial migration (32). In the present study, SNB19 spheroids infected with Ad-uPAR-MMP-9 failed to adhere to the VN-coated plates at 50 MOI. Recent studies document that MMP-9 is involved in several steps of cancer development including tumor cell migration. The localization of MMP-9 to specialized surface protrusions called invadopodia on the cell membrane by binding to CD44, a receptor for hyaluronan, is required to promote tumor cell migration (31). MMP-9 is also produced by epithelial cells and actively involved in the migration of repairing bronchial epithelial cells (33). Using inhibitory monoclonal antibodies and unselective MMP inhibitors, it was found that gelatinase B is crucial for the migration of Langerhans and dermal dendrite cells from human and murine skin (34).

We then tested the effect of the bicistronic construct on tumor growth. SNB19 cells infected with Ad-uPAR-MMP-9 (50 MOI) did not form any detectable tumors in nude mice. We have reported previously that apoptosis measured by DNA fragmentation was higher in the brains of animals injected with the antisense stable transfectants of antisense uPAR than in those injected with the parental cells (35). Alternatively, Ad-uPAR-MMP-9 was injected directly into pre-established U87 s.c. tumors. Despite low doses (5 × 10⁸ PFU × 3) compared with our previous work with Ad-uPAR (5 × 10⁹ PFU × 5; Ref. 11), there was significant tumor regression for ≤12 weeks. This effect could be a result of simultaneously targeting uPAR and MMP-9, which have many biological roles, including cell adhesion and migration, resulting in decreased invasion/angiogenesis, which in turn could lead to tumor growth suppression.

To understand the underlying mechanism of tumor growth inhibition, we analyzed the effect of this bicistronic construct on angiogenesis. The importance of the interaction between uPA and uPAR during angiogenesis has also been demonstrated in a number of in vivo systems. Adenovirally delivered NH₂-terminal fragments specifically inhibited tumor angiogenesis in syngeneic and xenograft murine tumor models (36). Similarly, a fusion protein consisting of the receptor-binding, NH₂-terminal fragment of uPA and Fc portion of human IgG, which functions as a high-affinity uPAR antagonist, inhibited basic fibroblast growth factor-induced angiogenesis in s.c. injected Matrigel (4). Microvessel density was markedly decreased in a homologous immunocompetent model of tumor cells transfected with a mutant murine uPA that retains receptor binding but not proteolytic activity (37). An octamer peptide derived from the nonreceptor binding region of uPA was shown to inhibit tumor growth and angiogenesis in breast cancer (38) and in combination with cisplatin in glioblastomas (39). In the present study, although Ad-uPAR (100 MOI) infection inhibited capillary-like network formation when SNB19 cells were cocultured with human dermal endothelial cells, Ad-uPAR-MMP-9 was very effective in inhibiting capillary formation even at very low concentrations (10 MOI). With respect to MMPs, potential targets for inhibiting angiogenesis include the gelatinases and MT-MMPs. MMP-9 has been shown to be important for angiogenesis in two transgenic models of tumor progression, the K14-HPV16 skin cancer model (40) and the RIP1-Tag2 insulinoma model (41). Genetic studies with MMP-9 knockout mice showed impaired angiogenesis during development, thereby supporting the role of MMP-9 in angiogenesis (42).

Extracellular proteases have been shown to cooperatively influence matrix degradation and tumor cell invasion through proteolytic cascades, with individual proteases having distinct roles in tumor growth, invasion, migration, and angiogenesis. This proteolytic cascade process releases various growth and differentiation factors that are sequestered on the cell surface or within the ECM, which contribute to the evolution of a migratory or invasive cell phenotype (43). Results accumulated over the last decade suggest that uPAR is also involved in motility control through other mechanisms. These include induction of signal transduction events after ligation with uPA, binding to the ECM molecule VN, and association with integrins and other transmembrane partners (44). Binding of uPAR to integrins may enrich integrin clusters with signaling molecules, such as Src-family kinases that localize to rafts and are important to integrin function. Signals derived from integrin/uPAR complexes promote the function of other integrins, regulating cell matrix interactions (45). Integrins have also been found to cooperate with metalloproteases to promote tumor cell invasion and angiogenesis (46). Lund et al. (47) demonstrated the requirement for both plasminogen deficiency and metalloprotease inhibition for complete inhibition of the healing process, indicating that there is a functional overlap between the two classes of matrix-degrading proteases. These findings suggest that effective arrest of cancer progression will require the combined use of inhibitors of MMPs and inhibitors of the plasmin/Plg activation system.

In the clinical setting, tumors exhibit a diverse array of mutated signaling pathways. In light of our understanding of the complex nature of uPAR and MMP-9 in multiple biological events, such as migration, adhesion, angiogenesis, tumorigenicity, and differentiation, our attempt to simultaneously target both uPAR and MMP-9 has clear clinical implications with a broad field of action. Although recombinant adenoviruses are excellent gene delivery vehicles for cancer gene therapy, toxicity caused by high vector doses and therapeutic effectiveness of Ad-mediated gene transfer could be limiting factors. It has been reported recently that preimmunity can inhibit systemic expression from adenoviral vectors, but at high vector doses, it may potentiate hepatotoxicity (48). Our present work addresses the toxicity issue to some extent, because our results clearly demonstrate that the bicistronic construct (Ad-uPAR-MMP-9) had a more prominent effect in inhibiting tumor invasion and growth at much lower doses compared with a single gene construct (Ad-uPAR). Consequently, these findings emphasize the potential of this protocol, although adenoviral delivery of genes is known to be transient and to last for only 4–6 weeks. Nonetheless, the use of novel vectors with long-term expression of genes at therapeutic levels or with modified fibers to specifically infect gliomas is necessary to treat intracranial tumors.