Adenovirus-mediated Delivery of Antisense Gene to Urokinase-type Plasminogen Activator Receptor Suppresses Glioma Invasion and Tumor Growth¹

Tumor cell invasion requires the participation of proteolytic events, of which the uPA³ and uPAR are key components. uPA is a serine protease that catalyzes the conversion of inactive plasminogen into plasmin, which then degrades a variety of extracellular matrix proteins and activates metalloproteinases and growth factors (1, 2). uPA binds to its cell surface receptor (uPAR), thereby localizing its proteolytic activity at the cell surface (3). The binding of uPA to uPAR not only increases the proteolytic activity of uPA but also favors a focal and directional proteolysis of extracellular matrix molecules (4). The uPAR protein is a 313-amino acid polypeptide with a 22-amino acid signal peptide encoded by a 1.4-kb mRNA transcript (5). A single-chain, highly glycosylated protein with a heterogeneous molecular weight of M_r 50,000–60,000, uPAR is anchored on the cell membrane by a glycosylphosphatidylinositol moiety (6). Both uPA and uPAR are expressed by normal cells; their overexpression has been observed in many invasive tumor cell lines and during tumor growth (7). Conversely, decreasing uPAR levels by using antisense techniques has been shown to inhibit tumor cell invasiveness and metastasis (8, 9). In our earlier studies of human gliomas, expression of uPAR was greater in the high-grade tumors than in the low-grade gliomas (10). Furthermore, the binding between uPA and uPAR is thought to play a major role in the invasion of glioblastoma cells into normal brain by virtue of concentrating the proteolytic activity at the leading edge of the tumor (11, 12). The expression of uPAR by human glioblastoma cells could significantly contribute to the invasive capacity of these cells (13). Moreover, stable transfectants of the human glioblastoma cell line SNB19 with antisense uPAR-expressing plasmid DNA have been shown to alter the invasive properties of these cells both in vitro (14) and in vivo (15). Conversely, transfecting the low-grade neuroglioma cell line H4, which normally expresses low levels of uPAR, with full-length uPAR cDNA increased uPAR expression and rendered these cells highly invasive compared to the parental cells (16).

In the present study, we constructed a replication-deficient recombinant Ad (Ad-uPAR) with the intent of effectively delivering an antisense uPAR gene that would down-regulate uPAR levels in human gliomas. We examined the biological activity of the Ad-uPAR construct in human gliomas in both in vitro and in vivo models.

The miniexpression cassette containing a 300-bp DNA fragment of the 5′ end of the uPAR gene with a CMV promoter and the polyadenylation signal of BGH was extracted from a pcDNA-uPAR-AS clone (14) by digestion with PvuII. The resulting 3.5-kb DNA fragment containing the 300-bp antisense expression cassette for uPAR was purified from agarose gels and partially digested with NruI to yield a 1.5-kb DNA fragment, which was blunt-end ligated into the EcoRV site of the adeno-shuttle vector pAdΔE1sp1A. The sequence of the resulting clone, pAd-uPAR was confirmed, and this plasmid construct was cotransfected with the 40.5-kb pJM17 vector into human embryonic kidney 293 cells to isolate recombinant Ads (17). Recombinant virus plaques were identified and amplified by PCR using primers specific for the 300-bp uPAR and for the CMV and BGH polyadenylation signal. The recombinant virus was purified by ultracentrifugation in cesium chloride step gradients (17). The viral DNA was also sequenced to confirm the configuration of the expression cassette. The purified viral preparations were verified to be free of wild-type Ad by PCR with E1-specific primers.

Human glioblastoma cell lines were obtained from the American Type Culture Collection (Manassas, VA) and grown in DMEM/F12 medium (1:1, v/v) supplemented with 10% FCS in a humidified atmosphere containing 5% CO₂ at 37°C. Viral stocks were suitably diluted in serum-free medium to obtain the desired MOI or PFU and added to cell monolayers or tumor cell spheroids (1 ml/60-mm dish or 3 ml/100-mm dish) and incubated at 37°C for 30 min. The necessary amount of culture medium with 10% FCS was then added, and the cells were incubated for the desired time periods.

Total cell lysates were prepared in extraction buffer containing Tris [0.1 m (pH 7.5)], Triton-X114 (1.0%), EDTA (10 mm), aprotinin, and phenylmethylsulfonyl fluoride. The extracts were incubated at 37°C for 5 min and centrifuged to separate the lower (detergent) phase that contains mostly hydrophobic membrane proteins including the glycosylphosphatidylinositol-anchored uPAR. Subsequently 20 μg of protein from these samples were separated under nonreducing conditions by 15% SDS-PAGE and transferred to nitrocellulose membranes (Schleicher & Schuell, Keene, NH). The membranes were probed with polyclonal antibodies to uPAR (#399 American Diagnostics, Inc.) and secondary antibodies (anti-rabbit-horseradish peroxidase) as required and developed according to enhanced chemiluminescence protocol (Amersham). For loading control, samples were SDS-PAGE-separated under reducing conditions and probed with monoclonal antibodies for β-actin. Alternatively, proteins were stained with Coomassie Blue to correct for loading controls.

Pro-uPA was labeled with [¹²⁵I]iodide by using iodogen as described previously (13). SNB19 cells in 24-well culture plates (2 × 10⁵ cells) were infected with Ad-uPAR or Ad-CMV (100 MOI for 3 days) or mock-infected, and receptor binding was performed as described previously (14).

SNB19 cells were infected with Ad-uPAR or Ad-CMV in serum-free medium, conditioned medium was collected, and the receptor-bound uPA was extracted with acid buffer 50 mm glycine-HC1 (pH 3.0) with 0.1 m NaCl, and cell extracts were prepared. Fibrin zymography was performed to detect uPA in these extracts as described previously (12).

Invasion of SNB19 human glioblastoma cells in vitro was assessed by measuring invasion through Matrigel-coated (Collaborative Research, Inc., Boston, MA) transwell inserts (Costar, Cambridge, MA) according to a previously described procedure (13). Briefly, transwell membranes with a 8-μm pore size were coated with a final concentration of 0.78 mg/ml Matrigel in cold serum-free DMEM. Cells were trypsinized, and 200 μl of cell suspension (1 × 10⁶ cells/ml) from each treatment were added in triplicate wells. After a 48-h incubation, the number of cells that passed through the filter into the lower wells was quantified, and the values were expressed as a percentage of the sum of the cells in the upper and lower wells (13).

Glioma spheroids of SNB19 cells were cultured in 100-mm tissue culture plates (Corning, Corning, NY) precoated with 0.75% agar prepared in DMEM. Briefly, 3 × 10⁶ cells were suspended in DMEM, seeded onto 0.5% agar-coated plates, and cultured until spheroid aggregates were formed. Spheroids of 100–200 μm in diameter were selected and infected with increasing PFUs of an Ad vector expressing green fluorescent protein (Ad-GFP) for 2 days, after which infectivity was determined by fluorescence microscopy. On the basis of these results, suitable MOI of Ad-CMV or Ad-uPAR (5 × 10⁸ PFUs) were used to infect SNB19 spheroids or fetal rat brain aggregates. After 3 days, progressive destruction of fetal rat brain aggregates and invasion of SNB19 cells were observed by phase-contrast microscopy and photographed as described previously (15).

For the tumor regression experiments, U87-MG cells (5 × 10⁶) were s.c. injected into nude mice. After 8–10 days, when tumor size had reached 4–5 mm, the mice were injected with Ad-CMV or Ad-uPAR (5 × 10⁹ PFUs) every other day for a total of five times. Tumor size was measured every second day, and tumor volume was calculated from the formula

We constructed a replication-defective recombinant Ad (Ad-uPAR) with a 1.5-kb miniexpression cassette that expresses a truncated 300-bp antisense message to the 5′ end of the uPAR gene, spanning 46 nucleotides of the noncoding region and rest in the coding region. The uPAR expression cassette is driven by a CMV promoter and a BGH polyadenylation signal, in the E1-deleted region of adenovirus type 5. Western blot analysis was performed to examine the effect of Ad-uPAR infection on uPAR protein levels in SNB19 cells. The uPAR protein band (M_r 60,000) was decreased in a dose-dependent fashion with increase in MOI (Fig. 1,A). Quantitation of the uPAR protein bands on Western blots by densitometry showed significantly (P < 0.001) decreased levels of protein in cells that were infected with Ad-uPAR above 20 MOI; this decrease reached 90% in cells infected with 200 MOI of Ad-uPAR, as compared to Ad-CMV-infected control cells. The decrease in uPAR protein level also was related to time; SNB19 cells infected with 100 MOI of Ad-uPAR showed a 90% decrease in uPAR level by day 7 (Fig. 1,B). A similar decrease in uPAR level over 5 days was observed in Ad-uPAR-infected U87-MG cells, another human glioblastoma cell line (Fig. 1,C). We then used a receptor binding assay to compare ¹²⁵I-pro-uPA binding to cell surface uPARs in Ad-uPAR and Ad-CMV control cells infected with different MOI. Results in Fig. 2 show that the ligand-receptor binding on Ad-uPAR-infected SNB19 cells infected with 50, 100, and 200 MOI was significantly reduced [70%, 85%, and 90%, respectively (p < 0.001)] compared to Ad-CMV control cells (Fig. 2). Finally, we used fibrin zymography to assess uPA activity in conditioned medium, cellular fraction, and receptor-bound fraction of Ad-uPAR-infected SNB19 cells. In the receptor-bound fraction of SNB19 cells, the intensity of the uPA band (M_r 55,000) decreased as the MOI increased (Fig. 3, upper panel). The relative amounts of receptor-bound uPA activity quantified by densitometry decreased by 3–8-fold (Fig. 3, lower panel) as the Ad-uPAR MOI was increased, compared to the activity in the Ad-CMV vector control (200 MOI). No significant changes were noted in uPA activity in cell extracts and secreted fractions (data not shown). Thus, a decrease in receptor-bound uPA activity corresponded to the observed decrease in uPAR protein level.

The effect of down-regulation of uPAR levels by Ad-uPAR infection on the invasiveness of SNB19 cells was studied using two different invasion models in vitro: namely, Matrigel and invasion of cells into fetal rat brain aggregates (spheroid model). When cells were placed at a density of 0.5 × 10⁶ cells/ml in the upper chamber, staining of Ad-uPAR-transfected cells that invaded through the Matrigel was significantly less compared to mock-infected and Ad-CMV-infected cells (Fig. 4,A). Quantitative analysis of the number of cells by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (13) showed that 42% mock control cells and 39% Ad-CMV-transfected cells invaded to the lower side of the membrane at 48 h (Fig. 4,B). In contrast, only 11% of Ad-uPAR-transfected SNB19 cells invaded, with a significant reduction (P < 0.001) compared to mock-infected and Ad-CMV controls. In the coculture assays, glioma spheroids infected with Ad-CMV and Ad-uPAR initially attached to fetal rat brain aggregates, and by 72 h, tumor spheroids from Ad-CMV control-infected SNB19 cells progressively invaded (Fig. 4,C, upper panel) into the aggregate. In contrast, Ad-uPAR-infected SNB19 spheroids failed to invade the rat brain aggregates (Fig. 4 C, lower panel).

Finally, we investigated the effect of Ad-uPAR infection on two glioblastoma cell lines on tumor growth in nude mice. U87-MG cells and a SNB19 variant that expresses β-gal (to facilitate detection and spread of tumor growth) were infected with Ad-uPAR or Ad-CMV and then injected either s.c. (U87-MG) or intracranially (SNB19) into nude mice. All six of the mice injected with the Ad-CMV-infected cells developed tumors; none of those animals injected with Ad-uPAR-infected cells developed tumors over a 4-week follow-up period (Table 1). In a separate experiment, 4–5 mm s.c. tumors were developed in nude mice by injecting mice with U87-MG cells. The tumors were then injected every second day with Ad-CMV or Ad-uPAR (5 × 10⁹ PFUs) for a total of five doses. All six Ad-uPAR-injected mice showed tumor regression starting on the ninth day after the first injection and continuing for 21 days, at which time, about an 85% inhibition was seen relative to the Ad-CMV-injected tumors (Fig. 5,A). Fig. 5 (B and C) shows that Ad-uPAR-treated animals showed a reduced tumor size compared to Ad-CMV-treated animals.

Malignant gliomas are the most common primary brain tumors in adults and children and are refractory to conventional forms of therapy (18). Current gene therapies for cancer involve attempts to kill cancerous cells by delivering conditionally toxic genes, cytokines, or costimulatory molecules that enhance cell-mediated killing of malignant cells or by delivering tumor suppressor genes that drive cells into apoptosis. Controlling the invasion process is a better approach that needs to be investigated for its therapeutic potential against gliomas. The present work involved targeting the uPA-uPAR system, which is critical for glioma invasion, by using an antisense gene delivery strategy with a recombinant Ad vector.

Substantial evidence proves the involvement of the proteolytic cascade triggered by the uPA-uPAR system in the process of invasion (7, 19, 20). We have previously observed that uPAR protein expression in high-grade glioma cell lines and tumors was higher than that in low-grade and anaplastic astrocytomas (10). Down-regulation of uPAR has been shown to decrease the invasiveness of glioma cells both in vitro and in vivo (14, 15), whereas overexpression of uPAR in a relatively noninvasive neuroglioma cell line rendered these cells more invasive (16). In the present study, a replication-defective recombinant Ad containing a 300-bp antisense gene to human uPAR was constructed and used to effectively down-regulate the uPAR levels in human glioma cells. The decrease in uPAR levels correlated with a corresponding decrease in uPA activity in the receptor-bound fraction. However, uPA levels in cellular and secreted fractions did not change, indicating that down-regulation of uPAR had no influence on the expression of its ligand, uPA. The decrease in uPAR protein by Ad-uPAR-infected glioma cells inhibited the invasion of these cells in Matrigel and spheroid models of invasion (Fig. 4, A and C). uPA and uPAR are known to localize at the leading edge of invading tumors and to facilitate the directional proteolysis of extracellular matrix components (4). In human glioblastoma cells (U251-MG) propagated intracerebrally in immunodeficient mice, uPA mRNA was shown to be present at the leading edge of the tumor margin, whereas uPAR mRNA was present throughout the tumor (11). The same cells, in vitro, expressed uPA and uPAR proteins that were localized to the integrin (α_vβ₃) cell-matrix contacts. We have shown the localization of uPAR protein at the leading edge of the tumors and near sites of vascular proliferation in human gliomas (10). Finally, transfecting SNB19 cells with the antisense uPAR vector inhibited tumorigenicity and invasiveness in vivo (15). From these reports, which implicate the uPA-uPAR system in invasion, it is logical to expect that down-regulation of uPAR levels, which in turn decrease the levels of bound uPA, should result in an overall inhibition of invasion.

The uPA-uPAR system may have other roles in addition to plasminogen activation (19). Although we saw no significant growth inhibition after infecting glioma cells with Ad-uPAR, the number of cells in superconfluent cultures at the end of the experiment (12 days) was 20–30% less than that in control vector-infected cultures. The additional observation that Ad-uPAR-infected cells were swollen and did not compact as closely as Ad-CMV-infected control cells implies the existence of other cell-cell contact effects that could be influenced by the decrease in uPAR. Indeed, adhesive events in human 293 embryonic kidney cells were shown be modulated by uPAR overexpression (21). Human monocytic cells rendered deficient in uPAR by antisense oligonucleotides show defective Mac-1-dependent adhesion to fibrinogen and migration on plastic (22). The regulation of cell adhesion and migration is thought to be coordinated by signals mediated through uPA-uPAR interactions with integrins (23). The s.c. U87-MG tumor regression observed in the present study might be due to a possible change in adhesive properties that leads to tumor cell death after Ad-uPAR injections. These findings suggest that uPAR has a role well beyond that of simply degrading ECM components through the uPA-mediated activation of plasminogen. The observation that tumors did not form after Ad-uPAR-infected U87-MG or SNB19 cells in nude mice supports our previous findings in which stably transfected antisense-uPAR SNB19 glioma cells also failed to form brain tumors in nude mice (15).

The uPA-uPAR system has been targeted in several anticancer therapies. uPA binds to its receptor by its amino-terminal fragment; the ligand binding domain is similar in sequence to the growth factor domain of epidermal growth factor and exhibits species restriction between mice and humans (20). Blocking the uPA-uPAR interaction confers the same effect on cell surface plasminogen activation as does inhibition of uPAR expression (24). Several antagonistic peptides identified by bacteriophage display as blocking uPA binding have been shown to inhibit angiogenesis and primary tumor growth in syngeneic mice (25). In another study, the Ad-mediated expression of a secreted antagonist of murine uPA/uPAR was shown to suppress angiogenesis-dependent tumor growth and dissemination in mice (26). Conversely, reducing uPAR levels by using antisense oligonucleotides was also found to inhibit tumor growth, invasion, and metastasis in some cancers (8, 9, 27). Down-regulating uPAR expression with an antisense strategy produced a protracted period of dormancy in human epidermoid carcinoma cells (28). With regard to safety, administering an antisense gene to uPAR may not have serious toxic consequences; transgenic mice lacking uPAR had no developmental defects (29). Down-regulating uPAR could have an added advantage over simply blocking the uPA-uPAR interaction because uPAR participates in cell adhesion and transmission of extracellular signals across membranes independently of uPA (19). Taken together, our present data on Ad-mediated antisense therapy, strongly support the therapeutic value of down-regulating the overexpression of uPAR in gliomas.