Oncogene (1997) 14, 1351 ± 1359  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

In vitro inhibition of human glioblastoma cell line invasiveness by antisense uPA receptor

Sanjeeva Mohanam1*, Shravan K Chintala1*, Yoshinori Go1, Anuradha Bhattacharya1, Boyapati Venkaiah1, Douglas Boyd2, Ziya L Gokaslan1, Raymond Sawaya1 and Jasti S Rao1

1Department of Neurosurgery and 2Tumor Biology, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Houston, Texas 77030, USA

The cell surface urokinase-type plasminogen activator nogen activator (uPA)-mediated plasminogen activa- receptor (uPAR) has been shown to be a key molecule tion pathway is one of the most important cascades in regulating plasminogen-mediated extracellular pro- involved in invasion of tumor cells (Danù et al., teolysis. To investigate the role of uPAR in invasion of 1985). uPA synthesized and secreted by a variety of brain tumors, human glioblastoma cell line SNB19 was normal and malignant cells (Schmitt et al., 1992) stably transfected with a vector capable of expressing binds with high anity to a speci®c cell surface an antisense transcript complementary to the 300 base receptor, uPAR (Blasi, 1988; Moller, 1993). uPA pair of the 5' end of the uPAR mRNA. Parental and promotes tumor cell migration and invasion by stably transfected (vector, sense, and antisense) cell lines converting the proenzyme plasminogen into active were analysed for uPAR mRNA transcript by Northern plasmin, which then cleaves the extracellular matrix blot analysis, and receptor protein levels were measured components including laminin, ®bronectin and col- by radioreceptor assays and Western blotting. Signi®- lagen (Pollanen et al., 1991) and activates other cant reduction of uPAR sites was observed in the matrix-degrading enzymes such as pro-collagenases antisense transfected cell lines. The levels of uPAR (Stetler-Stevenson et al., 1993). The binding of uPA mRNA were signi®cantly decreased in antisense clones to its receptor not only increases its enzymatic compared to control, vector and sense clones. The activity but also allows a focal and directional invasive potential of the cell lines in vitro was measured proteolysis of extracellular maxtrix (Pollanen et al., by Matrigel invasion assay and migration of cells from 1991). uPARs have been found to be consistently spheroids to monolayers. The antisense transfected cells present at the invasive foci of the tumors in all types showed a markedly lower level of invasion and of human cancers (Ellis et al., 1992). Displacement of migration than the controls. The antisense clones were cell surface receptor-bound uPA with an active-site more adhesive to the ECM components compared to mutant uPA blocks spontaneous metastasis of the parental, vector and sense clones. All transfected human prostate cancer cell PC3 in nude mice (vector, sense and antisense) clones and parental cells (Crowley et al., 1993), indicating that the tumor cell produced similar levels of uPA activity without any surface is the major site for uPA-driven proteolysis of signi®cant di€erence however, MMP-2 activity was the extracellular matrix. Invasion of cancer cells in decreased in antisense clones compared to controls. vitro can be inhibited by soluble recombinant uPAR These results demonstrate that uPAR expression is (Wilhelm et al., 1994), which functions as a scavenger critical for the invasiveness of human gliomas and down for uPA. regulation of uPAR expression may be a feasible Malignant brain tumors are characterized by local approach to decrease invasiveness. invasive in®ltration and destruction of surrounding normal brain tissue; their invasive behavior seems to Keywords: plasminogen activators; receptors; glioblas- depend partly on proteolytic activities localized on toma; invasiveness cell's surface. We reported earlier that overexpression of uPAR is associated with the higher malignancy grades of astrocytoma (Yamamoto et al., 1994b). In malignant brain tumors, uPA-bound uPAR plays a Introduction major role in glioblastoma cell invasion into normal brain by virtue of its expression at the leading tumor Tissue destruction and subsequent invasion of tumor edge (Gladson et al., 1995). We found, in addition that cells into surrounding normal tissues probably result human glioblastoma cells express uPAR that could from an inappropriate control of the proteolytic contribute signi®cantly to the invasive capacity of these activity of tumor cells. Several enzyme systems cells in vitro (Mohanam et al., 1993). The selective participate in the degradation of extracellular matrix inhibition of uPA:uPAR interaction can therefore, be and basement membrane. The urokinase-type plasmi- considered a primary goal in anticancer therapy and is a therapeutically feasible approach to the treatment of malignant brain tumors. In this study, we used the Correspondence: JS Rao antisense strategy to impair the expression of uPAR on *The ®rst two authors equally contributed to the work described in the surface of a human glioblastoma cell line. Our this paper results support the notion that down-regulation of Supported in part, by NCI grant CA56792, and ACS grant EDT-91 uPAR expression by antisense uPAR vector may be an (JS Rao) and CA 58311 (D Boyd) Received 6 June 1996; revised 15 November 1996; accepted 15 important target in decreasing the invasiveness of November 1996 human gliomas. Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1352 Results analysed for receptor expression and in vitro invasive capacities. Isolation of uPAR constructs in stably transfected clones Northern blot analysis of uPAR mRNA levels among Human glioblastoma SNB19 cells were transfected with transfected glioblastoma clones the pcDNA3 vector containing sense and antisense uPAR constructs. As controls, cells were transfected To determine whether the antisense constructs reduce with the vector alone. The growth of cells in medium uPAR expression, we characterized SNB19 and its containing G418 allowed selection of drug-resistant G418-resistant clones by Northern blot analysis and colonies that had stably integrated the neo-resistant compared them with RNA from the parental cell line. gene-containing vector. Clones with vector, sense and Transfection of the vector alone and the sense construct antisense integrated construct were selected and did not alter the levels of uPAR mRNA when compared to those of parental cells. Cell lines transfected with antisense vector showed substantially reduced mRNA-encoding for uPAR in all three clones AS-1, AS-2 and AS-3 (Figure 1a). Glyceraldehyde 3- a phosphate dehydrogenase (GAPDH) mRNA was detectable in all cell lines tested and was used to correct for loading equalities of the gel. Quantitative —uPAR uPAR hybridization signals were determined after normalization with the GAPDH signal by densito- metric scanning of the autoradiograms. Figure 1b shows that uPAR mRNA levels were approximately four to ®vefold decreased in antisense clones compared to parental cell line, vector and sense clones (P50.001).

— GAPDH Measurement of uPAR by saturation binding assay To determine the e€ect of antisense uPAR cDNA transfection on uPAR, we performed binding studies 125

SNB19 on nontransfected and transfected cell lines using I- SNB19V SNB19S labeled pro-uPA. A comparison of the binding of NB19AS-1 NB19AS-2 NB19AS-3

S S S radiolabeled pro-uPA to transfected cells is shown in Table 1. All the K values obtained were in the range b d of 0.2 to 0.6 nM indicating that the cellular receptor's anity for pro-uPA is not substantially altered in all transfected cell lines. The uPAR number was reduced by 45 ± 61% in antisense transfected clones compared to parental cell line, vector and sense clones (Table 1). To further quantitate the uPAR protein content of the cell lysates of parental, vector, sense and antisense clones, we performed Western blotting with uPAR

polyclonal antibodies. Figure 2a shows that Mr 60 000 uPAR band was present in all the samples but it was

reduced in antisense clones compared to parental,

„—˜le I €roEu€e ˜inding sites in p—rent—l —nd u€e‚ ™onstru™ts

tr—nsfe™ted ƒxfIW glio˜l—stom— ™ell lines

u€e‚Ggell A effinity @u d

wA @7 redu™tionA @n

Figure 1 (a) Northern blot analysis of uPAR mRNA in SNB19 glone

ƒxfIW HFSV

and transfected cells. Total RNA was isolated and 10 mg of RNA H

ƒxfIWE†e™tor HFPQ

was electrophoresed in 1.2% agarose gels and blotted onto a S

ƒxfIWEƒense HFQP

Nylon membrane. The membrane was hybridized with 32P-labeled P

ƒxfIWEentisense I HFIW

uPAR cDNA speci®c for uPAR mRNA. After removal of the TI

ƒxfIWEentisense P HFQU

radio labeled probe, the membrane blot was rehybridized with SR

ƒxfIWEentisense Q HFPT

GAPDH cDNA probe to check the relative amounts of mRNA RS

loaded on to the gel. (b) Levels of uPAR mRNA quantitated by „he ™ells were pretre—ted with —™id ˜u€er @pr QFHA —nd then

Vg for Q h with — r—nge of ™on™entr—tions of

scanning autoradiograms with laser densitometry. Relative in™u˜—ted —t R

IPS

s“€roEu€e in the presen™e or the —˜sen™e of — SHEfold ex™ess of hybridization signal numbers were calculted by ascribing an ‘

arbitrary value of one to the least intense signal seen by Northern unl—˜eled proEu€eF efter in™u˜—tionD ™ells were w—shedD lysedD —nd

analysis after loading equalities based on the glyceraldehyde r—dio—™tivity w—s ™ountedF „he num˜er of ˜inding sites were

phosphate dehydrogenase probe were determined. In each group determined ˜y ƒ™—t™h—rd —n—lysis —nd the num˜er of re™eptors in

the uPAR mRNA band was scanned in three positions at p—rent—l ƒxfIW ™ells w—s t—ken —s IHH7F „he v—ri—tions in u€e‚

di€erent exposures, and the peak areas were averaged to give the num˜ers —mong ve™tor —nd sense ™onstru™t tr—nsfe™ted ™lones w—s

v—lues r—nged from HFP to HFT nwF „he v—lues IH7 —nd the u values presented. Data were shown as mean values+s.d. of ®ve S d di€erent experiments from each clone. *P50.001 given were me—n of three independent experiments Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1353 a (kDa)

60 — SNB19 SNB19-V SNB19-S NB19-AS1 NB19-AS2 NB19-AS3 S S S

b

Figure 3 Invasion assays of parental SNB19 and uPAR constructs transfected cell lines. Suspension of cells was layered on Matrigel-coated transwell clusters and percent of invasion was calculated as described in Materials and methods

and as well as basement membrane collagens had a marked e€ect on migration of cells from spheroids. Spheroids from parental transfected cells showed a streaming outgrowth in response to di€erent collagens. Figure 2 (a) Quantitation of uPAR protein by Western blotting. We could not ®nd migration of cells on uncoated Cell lysates (50 mg protein) were analysed by Western blotting using anti-human uPAR polyclonal antibody as described in coverslips. In contrast, migration of cells from Materials and methods. Content of uPAR protein was spheroids of antisense transfected cells AS-1, AS-2 quantitated by scanning autoradiograms (uPAR band) with laser and AS-3 was signi®cantly reduced compared to densitometry. (b) Quantitative values of uPAR protein were parental SNB19 spheroids on all the collagens tested. obtained by scanning the blot in three positions at di€erent Quantitative data showed that the cell migration from exposures and the peak areas were averaged to give the values presented. Data were shown as mean values+s.d. of three these spheroids was inhibited signi®cantly (P50.001) di€erent experiments from each clone. P50.001 in antisense clones compared to parental, vector and sense-uPAR transfected cells (Figure 5).

Adhesive potential among transfected clones vector and sense clones. Scanning of the uPAR band showed a reduction of 50 ± 70% in all antisense clones Since the antisense transfected clones exhibited reduced (AS-1, AS-2 and AS-3) parallel to receptor numbers migration and invasion we reasoned that the reduced when compared to parental, vector and sense clones invasion in antisense clones could be due to the (Figure 2b). di€erential adhesion of cells to various extracellular matrix components. The e€ect of antisense uPAR transfection on adhesion of SNB19 cells was examined Invasive potential among transfected clones by incubating the cells in 96-well plates precoated with Antisense transfectant clones (AS-1, AS-2 and AS-3) 0.25 ± 2.5 mg/ml of ®bronectin, type IV collagen, or were compared to parent SNB19 cells, vector and sense laminin. Results showed that at 1 mg/ml, adhesion of transfectants to determine the e€ect of antisense uPAR antisense uPAR-transfected AS1 cells was increased transfection on an in vitro invasion by Matrigel 1.8-fold on type IV collagen, 2.1-fold on ®bronectin invasion assay. We found no marked di€erence in and 1.9-fold on laminin (Figure 6), compared to invasion between parental (SNB19) and vector controls. Similarly adhesion of transfected AS2 and (SNB19-V) and sense (SNB19-S) construct transfected AS3 cells also increased from 1.8 to 2.3-fold in the clones. However, a signi®cant inhibition in invasive presence of type IV collagen, ®bronectin and laminin potential was noted with antisense transfected clones, (Figure 6). Cell adhesion to bovine serum albumin AS-1 (70%), AS-2 (65%) and AS-3 (60%) at P50.001 (BSA)-coated plates was minimal. Cell adhesion to (Figure 3). ECM components was speci®c, as antibodies against type IV collagen, ®bronectin, and laminin inhibited the attachment of SNB19 cells (results not shown). These E€ect of transfection on spheroid formation and cell results demonstrated that antisense uPAR transfection migration from spheroids increases cell adhesion to di€erent ECM components. Cell migration from spheroids of parental, vector, sense Testing whether an antibody against uPAR increases and antisense transfected cell lines was shown in Figure cell adhesion, we found that adhesion of parental, 4. Among the extracellular proteins both interstitial vector and sense construct transfected cells was indeed Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1354 SNB19 SNB19V SNB19S SNB19AS-1 SNB19AS-2 SNB19AS-3 Type IV coll.Type III coll. Type I coll. Type Control Figure 4 Migration of cells from parental, vector and uPAR constructs transfected SNB19 spheriods on di€erent collagen types. Spheroids were generated and plated on various types of collagen-coated coverslips as described in Materials and methods

Figure 5 Migration of cells from spheroids to monolayer on di€erent types of collagens. Spheroids from parental SNB19 cells (solid bar), sense uPAR-transfected SNB19-S cells (dotted bar), vector transfected SNB19-V cells (hatched bar), antisense uPAR- transfected SNB19-AS1 cells (open bar), SNB19-AS2 (vertical Figure 6 Adhesion of uPAR-transfected cells to di€erent hatched bar) and SNB19-AS3 (horizontal hatched bar) were extracellular matrix (ECM) proteins. Parental SNB19 cells (solid placed onto the coverslips coated with di€erent collagens. At the bar), antisense uPAR-transfected SNB19-AS1 cells (dotted bar), end of migration assay (48 h), spheroids were ®xed, stained with SNB19-AS2 (cross hatched bar), SNB19-AS3 (open bar), sense Hema3 stain and mounted onto glass microscope slides. The uPAR-transfected SNB19-S cells (vertical hatched bar) and vector transfected SNB19-V cells (horizontal hatched bar) were plated at migration of cells from the spheroids was measured using a 5 microscope calibrated with a stage and ocular micrometer a cell density of 2.5610 cells/ml in individual wells precoated with di€erent ECM proteins. After 2 h incubation, unattached cells were removed by washing with PBS, and attached cells were ®xed and stained as described in Materials and methods. Data represent the mean absorbance of triplicate determinations increased by 30 ± 35% in the presence of uPAR antibody. However, addition of uPAR antibody did not a€ect the adhesion of AS1, AS2 and AS3 cells to

di€erent ECM components (results not shown). clones and parental cells. The relative amounts of Mr Together the adhesion experiments suggested that 55 000 uPA activity, quantitated by densitometry downregulation of uPAR may modulate cell attach- showed no signi®cant di€erence among the clones ment to di€erent ECM components. and parental cell line (Figure 7b). There was no signi®cant di€erence in the cell extracts of parental

cell line and other transfected clones on the levels of Mr Measurement of uPA by ®brin zymography 55 000 uPA activity (data not shown). As the radiolabeled uPA binding was reduced in antisense transfected clones, conditioned medium and Measurement of MMP-2 by gelatin zymography cell extracts from the transfected clones and parental cells were subjected to ®brin zymography to under- A recent study (Ray and Stetler-Stevenson, 1995) stand the e€ect of transfection on uPA synthesis and showed that MMP-2 directly modulates the adhesion secretion. Figure 7a shows that similar levels of uPA of and spreading of melanoma cells. To examine whether

Mr 55 000 were present in media of all the transfected antisense transfection alters the synthesis of MMP-2, Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1355

a a

(kDa) (kDa)

70 — 55 — 72 — tPA uPA SNB19 SNB19 SNB19S SNB19V SNB19-V SNB19-S NB19-AS1 NB19-AS2 NB19-AS3 NB19-AS1 NB19-AS2 NB19-AS3 S S S S S S

b b

Figure 7 (a) Fibrin zymography of medium from SNB19 and various clones. Conditioned medium (2 mg protein) was run on Figure 8 (a) Zymographic analysis of MMP-2 levels in uPAR- 10% SDS ± PAGE containing plasminogen and ®brinogen as transfected cells. Serum-free medium was analysed on sodium described in Materials and methods. (b) Quantitative analysis of dodecyl sulfate-polyacrylamide gels impregnated with 0.1% gelatin. For details see Materials and methods. (b) Quantitative M 55 000 uPA activity by densitometry in the medium of r analysis of MMP-2 activity in the conditioned medium by parental and transfected clones. M 55 000 band was scanned in r scanning densitometry. The data are mean+s.d. of three three positions and the peak areas were averaged to obtain the values presented. The data is shown as mean values+s.d. of four independent experiments. The units of peak area are arbitrary independent experiments from di€erent clones. Units of peaks are arbitrary

location that gives the uPAR a key role in generating the cell surface-bound plasmin activity required for we grew glioma cells in monolayers and conditioned extracellular matrix destruction at sites of cell medium was collected and analysed by gelatin progression. The uPAR's location suggests further- zymography. Interestingly, the zymographic analysis more, that cell matrix interactions are necessary for showed that antisense transfected SNB19 cells cell migration and tissue invasion. As matrigel expressed signi®cantly lower levels of MMP-2 both invasion assays showed (Mohanam et al., 1993 and in the conditioned medium (Figure 8a), than parental, Stahl and Mueller, 1994), monoclonal antibodies of vector and sense controls. Quantitative levels of uPAR were ecient in impairing the cell's invasive MMP-2 were signi®cantly decreased threefold ability. Further evidence of the pivotal role of the (P50.001) in antisense clones compared to parental, uPA/uPAR system in invasive behavior of tumor cells vector, and sense clones, when estimated by densito- has been provided in in vivo and in vitro models metry (Figure 8b). (Cohen et al., 1991; Ossowski, 1992; Kobayashi et al., 1993; Kariko et al., 1993; Crowley et al., 1993; Kook et al., 1994, Stahl and Mueller, 1994; Danù et Discussion al., 1994). If the cell surface-bound uPA limits the rate of invasion, a reduction in surface uPAR should Tumor cell invasiveness is a complex multistep translate into a measurable reduction in invasiveness. process that involves cell attachment, proteolysis of Thus down regulation of uPAR could reduce uPAR- matrix components and migration of cells through mediated proteolysis in tumor cells, and expression of the disrupted matrix (Aznavoorian et al., 1993). In uPAR-cDNA construct resulting in antisense RNA is malignant tumors, the majority of uPARs are appropriate for achieving this goal. An earlier report concentrated at invasive foci (Ellis et al., 1992), a demonstrated that a 300 bp 5' fragment of the uPAR Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1356 antisense cDNA was more e€ective in reducing Andersson et al., 1993; Lampson et al., 1992). In order uPAR mRNA than using the entire uPAR coding to know whether the sense and antisense transfected region in squamous carcinoma cells (Kook et al., cells have similar migration, we used spheroid 1994). migration assay to measure the ability of cells Malignant gliomas have recently been shown to organized in a three-deminesional matrix to migrate express greater quantities of uPAR mRNA (Yama- and found that the migration of cells from tumor moto et al., 1994b and Gladson et al., 1995), and its spheroids of antisense transfected clones was signifi- level of expression in human gliomas seems to cantly reduced. Adhesion of cells to various extra- correlate with features of malignancy. We reported cellular matrix components is one of the basic earlier that human glioblastoma cell lines expressed properties of the tumor cells during invasion cascade high-anity binding sites for uPA that could in which cells initially attach to and subsequently contribute to their invasive capacities (Mohanam et degrade the extracellular matrix components. In order al., 1993). In this study we stably transfected to better understand the reduced migration and glioblastoma cell line SNB19 with expression vectors invasion properties of the transfected cells, cell containing a 300 bp 5' portion of uPAR cDNA in adhesion to various ECM components was carried sense and antisense orientations. Selected clones based out and found that adhesion of antisense uPAR on their resistance to G418 were analysed for uPAR transfected cells to di€erent ECM proteins was mRNA expression, uPAR number and invasive signi®cantly increased to that of parental, sense uPAR capabilities. A number of studies using stable construct and vector transfected cells (Figure 6). The antisense RNA expression have demonstrated reduc- increased adhesion of antisense uPAR clones could be tion in endogenous mRNA levels (Kook et al., 1994; due to the altered expression of ECM receptors, Neuenschwander et al., 1995; Saleh et al., 1996). The integrins. In support of this, a recent study by Wei et mechanism for reduction of uPAR mRNA is unclear al. (1996) showed that uPAR regulates the functions of although it may be related to interference to integrins during cell adhesion to various ECM transcription initiation or elongation, RNA proces- proteins, which is consistant with the idea that both sing or mRNA transport. Another possibility is increased uPAR expression and loss of stable cell incorporation of vector by homologous recombina- adhesion were linked to malignant transformation, tion or random integration and disruption of invasion and metastasis in various types of tumors endogenous uPAR gene. There is no notable (Pederson et al., 1994; Pepper et al., 1993; Bianchi et variation in receptor numbers among parental, vector al., 1994; Crowley et al., 1993). and sense construct transfected cell lines suggesting Gliomas do not seem capable of metastasizing out that antisense RNA is responsible for reduction in of the central nervous system (Russell and Rubinstein, receptor number in antisense transfected cells. Our 1989), although they often have many of the results showed that successful transfection of the degradative properties associated with metastatic SNB19 cells with uPAR antisense cDNA construct tumors, such as metalloproteinase and plasminogen leads to decreased uPAR mRNA transcript and activator secretion. Since studies have shown that uPA protein levels suggesting that, by whatever mechan- is involved in invasion (Testa and Quigly, 1990), we ims, the e€ect on RNA had profound biological tested whether transfection and clonal selection a€ect consequences. Other studies have shown that altered these properties. All clones used for invasion expression of uPAR may be associated with a change experiments produced uPA similar to that of parental in their anity for uPA. Treatment of human cells in conditioned medium. Glioma cells not only umbilical vein endothelial cells with phorbol-12- have high levels of uPA but also metalloproteinases myristate-13-acetate (PMA) resulted in a threefold including type IV collagenases which are capable of increase in number of uPARs and two- to threefold degrading extracellular matrix components. Gelatino- decrease in anity (Langer et al., 1993). However, we lytic activity in human glioblastomas and anaplastic observed no signi®cant alteration in receptor anity astrocytomas was much higher than in astrocytomas of SNB19 cells as a result of transfection. and normal brain tissue (Nakagawa et al., 1994). It Antisense transfected clones showed a marked has been suggested that MMP-2 is responsible for reduction in in vitro invasiveness when compared to invasion of human glioma cells in an in vitro sytem controls, suggesting that the modulation of uPAR in (Abe et al., 1994). More recently MMP-2 was shown SNB19 cells altered their invasive potential in this to reduce adhesion and promote the migration and experimental model system. In other studies Kook et invasion of melanoma cells (Ray and Stetler-Steven- al. (1994) showed that transfecting human squamous son, 1995). In our experiments, gelatinolytic examina- carcinoma HEp3 cells with antisense uPAR vector tion of conditioned medium from antisense uPAR suppressed their ability to metastasize in a chorioal- transfected cells showed that MMP-2 levels were lantoic membrane assay. Moreover, antisense oligonu- signi®cantly reduced by antisense transfection (Figure cleotides to uPAR were shown to inhibit invasion of 8) compared with MMP-2 levels of parental SNB19, transformed ®broblasts (Quattrone et al., 1995a,b). sense constructs and vector transfected cells. The Our results of the matrigel assay support the idea that underlying mechanism in the reduction of MMP-2 is receptor bound uPA activity is crucial for the invasion not known although uPAR mediated plasmin genera- through matrigel that contains a mixture of basement tion is involved in the activation of zymogens of membrane components. On the other hand, apart from MMPs. Stable transfectants of glioblastoma cells with invasion through adjacent normal brain by active antisense uPAR cDNA, as demonstrated in this study, proteolytic degradation, tumor cells also show a have not been reported previously. We believe that passive migration along the neuronal ®ber tracts and our in vitro model to down regulate uPAR is also along the blood vessels (Pedersen et al., 1993; applicable in vivo and that antisense vectors may Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1357 well be of clinical interest in the control of gliomas. uPAR mRNA were determined in all the clones by Our results suggest that signi®cant improvement in the densitometry comparison between groups was done by treatment for malignant glial neoplasms will depend standard statistical procedure (analysis of variance). on targeting molecules involved in one of the many steps of proteolytic cascade. Radiorecepetor assays Pro-uPA was labeled with [125I]iodide using Iodogen (Pierce, Rockford, IL) as described earlier (Mohanam et Materials and methods al., 1993). The cells were acid-pretreated with 50 mM glycine (pH 3.0) and 0.1 M NaCl bu€er for 3 min and incubated with binding bu€er consisting of DMEM with Cell culture 0.1% bovine serum albumin (BSA) and 20 mM 4-(2- Glioblastoma cells SNB19 were grown in Dulbecco's hydroxyethyl)-1-piperazineethanesulfonic acid (Hepes) Modi®ed Eagle Medium (DMEM) supplemented with with 0.25 to 12 nM of radioligand [125I]-Pro-uPA for 3 h 10% fetal calf serum (FCS), 100 mg/ml of streptomycin at 48C, with and without a 50-fold excess of unlabeled and 100 U/ml of penicillin in a humidi®ed atmosphere Pro-uPA. After four washes in DMEM-HEPES-BSA, the containing 5% CO2 at 378C and subcultured every 3 to 5 radioactivity was solubilized in 0.1 M Tris (pH 8.0), 10% days. Triton X-100, 10 mM EDTA, 100 mM NaCl and counted for radioactivity. Clones were characterized with respect to K and B for ligand interactions (Scatchard et al., Preparation of constructs d max 1949). A 300 base pair (bp) cDNA fragment of uPAR was polymerase chain reaction (PCR)-ampli®ed using synthetic Western blotting primers as described earlier (Kook et al., 1994) and subcloned into pcDNA3 vector (Invitrogen, San Diego, Cells were exposed to an acidic bu€er (50 mM glycine/HCl, CA) in sense and antisense orientations. Sequence analysis 100 mM NaCl pH 3.0) for 3 min at room temperature and of 300 bp of internal sequence showed 100% homology then neutralized by addition of 0.5 HEPES, 100 mM NaCl with the published sequence for uPAR cDNA (Roldan, et pH 7.5. Cells were then treated with lysis bu€er (0.1 M al., 1990). Tris-HCl, 1% Triton X-114, 10 mM EDTA, 10 mg aprotinin, pH 8.1 and freshly added 1 mM PMSF) and the lysate was clari®ed by centrifugation. The supernatant Transfection of SNB19 cells was warmed for 10 min at 378C to allow phase separation The SNB19 cells were transfected with uPAR cDNA and centrifuged. The detergent phase was collected and constructs in sense and antisense orientations, as well as electrophoresed on a 10% SDS-polyacrylamide gel with pcDNA3 vector alone using lipofectin (Life Technol- (Laemmli, 1970), followed by transfer of the proteins ogies, Gaithersburg, MD). The cells were plated at a cell onto nitrocellulose paper (Towbin et al., 1979). The blots density of 46105 cells per 60 mm dish. The following day, were incubated with anti-human uPAR polyclonal anti- 10 mg of plasmid DNA was dissolved in 50 mlofsterile body and uPAR protein bands were identi®ed using ECL water, mixed with 50 ml of lipofectin and allowed to stand Western blotting system (Amersham, Arlington Heights, at room temperature for 15 min. The cells were washed IL). The intensity of bands were scanned by densitometry with phosphate-bu€ered saline (PBS), covered with 3 ml of for quantitation. DMEM without FCS and 100 ml of DNA-lipofectin solution was added dropwise. Twelve hours later, the Invasion assay medium was replaced with DMEM containing 10% FCS. Selection was initiated 48 h after transfection by growing To determine whether antisense uPAR expression among the cells in DMEM containing 10% FCS with G418 (Life stably transfected clones a€ects glioblastoma cell line, we Technologies, Gaithersburg, MD). Stable transfectants tested vector, sense and antisense transfected cell lines and were selected with cloning cylinders after 10 to 15 days parental SNB19 cell line in an in vitro invasion assay as and weaned from the selection medium. Stable transfec- described previously (Mohanam et al., 1993). Brie¯y, tants were initially selected in 800 mg/ml of G418, and cells transwell inserts (Costar, Cambridge, MA) were coated that survived the selection were expanded in the absence of with 10 mg Matrigel (Collaborative Research, Cambridge, G418 for further studies. MA) and cells were added in 200 ml of medium in triplicate wells. After 48 h incubation, the number of cells that passed through the ®lter into the lower wells were Northern blot analysis quanti®ed and expressed as a percentage of the sum of Total cellular RNA was extracted from con¯uent cultures cells in upper and lower wells (Mohanam et al., 1993). as described earlier (Mohanam et al., 1993; Chomczynski and Sacchi, 1987). Aliquots of 10 mgofRNAwere Spheroid formation and cell migration from spheroids separated by electrophoresis on 1.2% agarose-formalde- hyde gels, capillary-transferred to a nylon membrane Parental and transfected SNB19 cells were cultured as overnight, and cross-linked with ultraviolet irradiation. spheroids in 100 mm2 tissue culture plates that had been All ®lters revealed uniform loading and intact 18S and 28S pre-coated with 0.75% agar as described (Lund-Johansen ribosomal RNA as judged from methylene-blue staining et al., 1990). Brie¯y, 36106 cells were seeded into agar- after transfer. The ®lters were hybridized at 658Covernight coated plates and cultured until spheroids formed. Glass with a uPAR cDNA probe labeled with 32P-deoxycytidine coverslips were coated with 5 mg/ml of either type I triphosphate by random primer labeling. The ®lters were collagen, type III collagen or type IV collagen (Life then washed in 0.5% standard saline citrate (SSC; 3 M Technologies, Gaithersburgh, MD); uncoated coverslips sodium chloride, 0.3 M sodium citrate) and 0.1% sodium were used as controls. Collagen-coated or uncoated dodecyl sulfate (SDS) for 20 min, then for 15 min at 658C coverslips were placed into separate wells of 24-well and exposed to X-ray ®lm at 7708C. After stripping, the culture plate, single spheroids were transferred onto membranes were rehybridized with glyceraldehyde phos- coverslips and cultured in a 5% CO2 humidi®ed incubator phate dehydrogenase (GAPDH) cDNA. The levels of at 378C for 48 h. A total of 6 ± 8 spheroids were plated for Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1358 each collagen type. At the end of the migration assay, polymerization, medium (2 mg) from the cell lines were spheroids were ®xed, stained with Hema-3 stain, mounted electrophoresed and various types of PAs separated, based onto glass microscope slides and photgraphed. The on di€erences in their molecular weight. The SDS±PAGE migration of cells from spheroid to monolayers was also gel was then washed twice with 2.5% Triton X-100 for measured using a microscope calibrated with a stage and 30 min each time, and the gel was incubated at 378C occular micrometer. overnight with glycine bu€er (pH 7.5). Upon staining with Coomassie blue and destaining, the ®nal gel has a uniformly blue background except in regions to which Cell adhesion assay PAs have migrated and activated the plasminogen to Cell adhesion to di€erent ECM components was assayed plasmin. The levels of enzymatic activity were quantitated

using 96-well plates. The plates were precoated with Mr 55 000 band in parental and transfected clones. To varying concentrations (0.25 ± 2.5 mg/ml) of either fibro- ensure that the intensity of the bands fell within a linear nectin or type IV collagen in a ®nal volume of 100 mlPBS range, the gels were incubated for di€erent time intervals (pH 7.2) overnight at 48C. The wells were blocked with before quantitative estimation by densitometry. 10% BSA (Sigma Chemical Co., St. Louis, MO) in PBS for 1 h at room temperature. Cells were trypsinized, resus- Gelatin zymography pended in DMEM containing 10% fetal calf serum, counted, and allowed to recover from the trypsinization Analysis of MMP-2 was performed with sodium dodecyl for 1 h at 378C. Cells were centrifuged, washed twice with sulfate-polyacrylamide gels impregnated with 0.1% gelatin serum-free DMEM and resuspended in serum-free DMEM (w/v) and 10% polyacrylamide (w/v) as described elsewhere to a ®nal concentration of 2.56105 cells/ml; 100 mlofcell (Nakajima et al., 1990). Parental SNB19, cells transfected suspension was added to each well. In some experiments, with antisense uPAR, cells transfected with sense uPAR, and rabbit anti-human uPAR antibody (#399R, American cells transfected with vector alone were grown in 100 mm Diagnostica, Greenwich, CT) was added during the tissue culture plates in DMEM containing 10% FCS until incubation at 20 mg/ml ®nal concentration. The cells were they reached 80% con¯uency. Cells were washed and allowedtoadherefor2±4handattheendofthe replaced with serum-free medium and allowed to grow for incubation wells were washed carefully with PBS, ®xed another 24 h. The cell-conditioned medium was collected, with Hema-3 ®xative and stained with Hema-3 solution-II and four parts of medium containing equal amounts of (CMS,Houston,TX).Thedyewasextractedin10% protein were mixed with one part of Laemmli sample bu€er methanol, and 5% acetic acid and read at 650 nm on an (minus reductant; Laemmli, 1970) prior to electrophoresis. ELISA plate reader. A linear and reproducible relationship Gels were run at constant current and then washed twice for between cell number and absorbance at 650 nm was 30 min in 50 mM Tris-HCl, pH 7.5, plus 2.5%Triton X-100, observed in all experiments. followed by overnight incubation at 378Cin50mMTris-HCl, pH 7.6, 10 mM CaCl2, 0.05% NaN3. Gels were stained with Coomassie Brilliant Blue R-250 and then destained. Fibrin zymography The enzymatic activity and molecular weight of electro- phoretically separated forms of PAs were determined in conditioned medium of all the clones and control cells by sodium dodecyl sulfate polyacrylamide gel electrophoresis Acknowledgements (SDS±PAGE) procedure as previously described (Yama- We thank Sylvia Ledesma in preparation of the manu- moto et al., 1994a). Brie¯y, the SDS±PAGE gel contains script, Leslie Wildrick for manuscript review, and Dr Jack acrylamide to which puri®ed plasminogen and ®brinogen Henkin (Abbott Laboratories, Abbott Park, IL) for was added as substrates before polymerization. After providing pro-uPA.

References

Abe T, Mori T, Kohno K, Seiki M, Hayakawa T, Welgus Ellis V, Pyke C, Eriksen J, Solgerg H and Danù K. (1992). HG, Hori S and Kuwano M. (1994). Clin. Exp. Metastasis, Ann. N.Y. Acad. Sci., 667, 13 ± 31. 12, 296 ± 304. Gladson CL, Pijuan-Thompson V, Olman MA, Gillespie GY Andersson C, Tytell M and Brunso-Bechtold S. (1993). Int. and Yacoub IZ. (1995).Am. J. Pathol. 146, 1150 ± 1160. J. Devl. Neurosci., 11, 555 ± 568. Kariko K, Kuo A, Boyd DA, Okada SS, Clines DB and Aznavoorian S, Murphy AN, Stetler-Stevenson, WG and Barnathan ES. (1993). Cancer Res., 53, 3109 ± 3117. Liotta LA (1993). Cancer, 71, 1368 ± 1383. Kobayashi H, Ohi H, Shinohara H, Fujii T, Schmitt M, Blasi F. (1988). Fibrinolysis, 2, 73 ± 84. Goretzki L, Churcholowski N, Janicke F and Grae€ H. Bianchi E, Cohen RL, Thor AT, Todd RF, Mizukami IF, (1993). Br.J. Cancer, 67, 537 ± 544. Lawrence DA, Ljung BM, Shuman MA and Smith HS. Kook YH, Adamski J, Zelent A and Ossowski L. (1994). (1994). Cancer Res., 54, 861 ± 866. EMBO J., 13, 3983 ± 3991. Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162: Laemmli UK. (1970). Nature, 227, 680 ± 685. 156 ± 159. Lampson LA, Wen P, Ramen VA, Morris JH and Sarid JA Cohen RL, Xi X-P, Crowley CW, Lucas BK, Levinson AD (1992). Cancer Res., 52, 1018 ± 1025. and Shuman MA. (1991). Blood, 78, 479 ± 487. Langer DJ, Kuo A, Kariko K, Ahuja M, Klugherz BD. Crowley CW, Cohen RL, Lucas BK, Liu G, Shuman MA Ivanics KM, Hoxie JA, Williams WV, Liang BT, Clines and Levinson AD. (1993). Proc. Natl. Acad. Sci. USA, 90, DB and Barnathan ES. (1993). Circulat Res., 730, 330 ± 5021 ± 5025. 340. Danù K, Andreasen PA, Grondahl-Hansen J, Kristensen P, Lund-Johansen M, Bjerkvig R and Rucklidge GJ. (1990) Nielsen LS and Skriver L. (1985). Adv. Cancer Res., 44, Spheriod culture in Cancer Research. Bjerkvig, R. (ed.) 139 ± 266. CRC Press, Ann Arbor, pp. 3 ± 18. Danù K, Behrendt N, Brunner N, Ellis V, Ploug M and Pyke Mohanam S, Sawaya R, McCutcheon I, Ali-Osman F, Boyd C. (1994). Fibrinolysis, 81, 189 ± 203. D and Rao JS. (1993). Cancer Res., 53, 4143 ± 4147. Inhibition of glioblastoma cell invasion by antisense uPAR S Mohanam et al 1359 Moller LB. (1993). Blood Coagul. Fibrinolysis, 4, 293 ± 303. Russell DS and Rubinstein LJ. (1989). Pathology of Tumors Nakagawa T, Kubota T, Kabuto M, Sato K, Kawano H, of the Nervous System, Edward Arnold, London, pp. 42 ± Hayakawa T and Okada Y. (1994). J. Neurosurg., 81, 69 ± 62. 77. Saleh M, Stacker SA and Wilks AF. (1996). Cancer Res., 56, Nakajima M, Morakawa K, Fabra A, Bucanna CD and 393 ± 401. Fidler IJ. (1990). J. Natl. Cancer Inst., 82, 1890 ± 1898. Schmitt M, JaÈ nicke F and Grae€ H. (1992). Fibrinolysis, 6, Neuenschwander S, Roberts Jr, CT and LeRoith D. (1995). 3 ± 26. Endocrinol., 136, 4298 ± 4303. Scatchard G. (1949). Ann. N.Y. Acad. Sci., 51, 660 ± 665. Ossowski L. (1992). Cancer Res., 52, 6754 ± 6760. Stahl A and Mueller BM. (1994). Cancer Res., 54, 3066 ± Pedersen H, Brunner N, Francis D, Osterlind K, Ronee E, 3071. Hansen HH, Danù K and Grondahl-Hansen J. (1994). Stetler-Stevenson WG, Liotta LA and Kleiner DE. (1993). Cancer Res., 54, 4671 ± 4675. FASEB J., 7, 1434 ± 1441. Pedersen PH, Marienhagen K, Mork S and Bjerkvig R. Testa JE and Quigly JP. (1990). Cancer Metastasis Rev., 9, (1993). Cancer Res., 53, 5158 ± 5165. 353 ± 367. Pepper MS, Sappino AP, Stocklin R, Montesano R, Orci L, Towbin H, Staehelin T and Gordon J. (1979). Proc. Natl. Vassalli JD. (1993). J. Cell Biol., 122, 673 ± 684. Acad. Sci. USA, 76, 4350 ± 4354. Pollanen J, Stephens RW and Vaheri A. (1991). Adv. Cancer Wei Y, Lukashev M, Simon DI, Bodary SC, Rosenberg S, Res., 57, 273 ± 328. Doyle MV and Chapman HA. (1996). Science, 273, 1551 ± Quattrone A, Fibbi G, Anichini E, Pucci M, Zamperini A, 1555. Capaccioli S and Rosso MD. (1995a). Cancer Res., 55, Wilhelm O, Weidle U and Hohl S. (1994). FEBS Lett., 337, 90 ± 95. 131 ± 134. Quattrone A, Fibbi G, Anichini E, Pucci M, Zamperini A, Yamamoto M, Sawaya R, Mohanam S, Bindal AK, Bruner Capaccioli S and Rosso MD. (1995b). Anticancer Drug JM, Oka K, Rao VH, Tomonaga M, Nicolson GL and Des., 10, 97 ± 102. Rao JS. (1994a). Cancer Res., 54, 3656 ± 3661. Ray JM and Stetler-Stevenson WG. (1995). EMBO J., 14, Yamamoto M, Sawaya R, Mohanam S, Rao VH, Bruner JM, 908 ± 917. Nicolson GL and Rao JS. (1994b). Cancer Res., 54, 5016 ± Roldan AL, Cubellis MV, Masucci MT, Behrendt N, Lund 5020. R, Danù K, Apella E and Blasi F. (1990). EMBO J., 9, 467 ± 474.