Role of Mitogen-activated Protein Kinases and c-Jun/AP-1 trans-Activating Activity in the Regulation of Protease mRNAs and the Malignant Phenotype in NIH 3T3 Fibroblasts*

Ras activates a multitude of downstream activities with roles in cellular proliferation, invasion and metastasis, differentiation, and programmed cell death. In this work we have evaluated the requirement of extracellular signal-regulated protein kinase (ERK), c-Jun NH2-terminal kinase kinase (JNKK), and c-Jun/AP-1 activities in transformation and extracellular matrix invasion ofras oncogene expressing NIH 3T3 fibroblasts by expressing stable mutant genes that constitutively inhibit these activities. Whereas the inhibition of ERK activity reverts the transformed and invasive phenotype, the inhibition of the JNK pathway and AP-1trans-activating activities by JNKK[K129R] and c-Jun(TAM67) had no effect on the ability of the rasoncogene-expressing cells to grow in soft agar or invade Matrigel basement membrane. Thus an elevated JNK activity and/or c-Jun/AP-1trans-activating activity are not absolute requirements forras transformation or invasion through basement membrane, and the dependence on AP-1 activity for transformation is cell-specific. However, inhibition of JNK kinase (JNKK) inras-transformed cells with normally elevated JNK activity switches the protease-dependent invasive phenotype from a urokinase plasminogen activator (uPA)-dependent to a cathepsin L (CL)-dependent invasive phenotype. Conversely, treatment of ras-transformed cells of low constitutive JNK activity with the JNK stimulator, anisomycin, converts the protease mRNA levels from those characteristic of a CL-dependent to a uPA-dependent phenotype. These protease phenotypes can be duplicated in untransformed NIH 3T3 cells that express platelet-derived growth factor receptors and m1 muscarinic receptors that selectively stimulate the ERK or JNK pathways, respectively. It is concluded that high ERK activity is required for both protease phenotypes, whereas the JNK pathway and c-Jun/AP-1 activity are not required for transformation but regulate a switch between uPA and CL protease phenotypes in both transformed and untransformed cells. Inras-transformed NIH 3T3 fibroblasts, the uPA- and CL-dependent protease phenotypes are redundant in their ability to invade through basement membrane.

The Ras protein is at the center of a web of multiple downstream pathways that are differentially regulated to produce signals that control processes as disparate as cell division, cell differentiation, and programmed cell death (reviewed in Refs. [1][2][3]. Two of the better characterized pathways downstream from Ras are from Ras to ERK-1 1 and -2 through the sequential activation of Raf and MEK1, and the pathway from Ras leading to the activation of JNK1 through the activation of Cdc42 and Rac 1 which then sequentially activate PAK, MEK kinase-2 and/or -3, JNKK/SEK/MKK4, and JNK1. The ERKs and JNK1 are mitogen-activated protein kinases that translocate to the nucleus and activate transcription factors involved in Rasregulated processes. The protein c-Jun is a nuclear substrate of JNK1 (4 -7), and c-Jun is a prominent component of many AP-1 transcription factors that bind to AP-1/TRE enhancer elements in gene regulatory sites (8,9). Such an AP-1 enhancer element is a part of the ras response element present in many protease genes with roles in cellular migration and invasion. These protease gene products are secreted to degrade the proteins of the extracellular matrix through which the cells invade during cell migration (10). In addition, the secreted proteolytic activities activate growth and migratory factors that regulate cellular invasion (10). Proteases implicated in invasion processes include the serine-type protease uPA, the cysteine-type proteases cathepsins B and L, the aspartate-type protease cathepsin D, and the matrix metallo-type proteases (MMPs) (10 -12). Specific protease genes up-regulated by AP-1 induction include the type I collagenases (MMP-1,-13), type IV collagenase (MMP-9), stromelysin-1 (MMP-3), matrilysin (MMP-7), and urokinase plasminogen activator (13)(14)(15)(16)(17)(18)(19)(20)(21). Other genes up-regulated by a ras-responsive AP-1 regulatory element with possible roles in generating the properties exhibited by transformed cells are cyclin D1 and uPA-receptor (22,23).
The introduction of an activated ras gene into NIH 3T3 fibroblasts results in transformed cells capable of metastasizing to the nude mouse lung following tail vein injection (24,25). We have previously characterized protease gene expressions in ras-transformed NIH 3T3 cells and have shown that at least two distinct protease phenotypes are generated, apparently dependent on the isoform of ras oncogene expressed by the cell (25). Cells transformed by the EJ/vHa-ras oncogene had elevated levels of uPA mRNA and secreted uPA protease activity but basal expression of CL. Inhibition of uPA and CL protease gene expressions with antisense oligonucleotides showed that the EJ/vHa-ras-transformed cells metastasized to the nude mouse lung after tail vein injection in a uPA-dependent and CL-independent manner. In contrast to the EJ/vHa-ras cells, NIH 3T3 cells transformed by the RAS1 Leu del oncogene showed elevated levels of CL mRNA and secreted CL protease activity but low basal expression of uPA. Selective inhibition of CL and uPA gene expressions with antisense oligonucleotides showed the RAS1 Leu del-transformed cells colonized the lung in a CL-dependent and uPA-independent manner. The EJ/vHa-rastransformed cells were designated a ras uPAϩ/CLϪ phenotype and the RAS1 Leu del-transformed cells a reciprocal ras CLϩ/uPAϪ phenotype.
Based on the observation of the two different protease-dependent metastatic phenotypes (ras uPAϩ/CLϪ and ras CLϩ/uPAϪ ) generated by the EJ/vHa-ras and RAS1 Leu del oncogenes, it was hypothesized that the different ras isoforms activated different combinations of Ras downstream activities. Analysis of the ERK and JNK pathway activities downstream from Ras showed that in the EJ/vHa-ras-transformed cells of phenotype ras uPAϩ/CLϪ both ERK and JNK activities were elevated (21). Correlated with the elevated JNK activity, the EJ/vHa-rastransformed cells showed high levels of c-Jun and hyperphosphorylated c-Jun. The RAS1 Leu del-transformed cells had a constitutively elevated ERK activity, identical in magnitude to the ERK activity in the EJ/vHa-ras-transformed cells. However, in the RAS1 Leu del-transformed cells of phenotype ras CLϩ/uPAϪ , the JNK activity and c-Jun levels appeared inhibited (JNK activity was ϳ30% of the activity in untransformed NIH 3T3 cells) (21). When a dominant negative c-Jun, TAM67, was expressed in EJ/vHa-ras-transformed cells of phenotype ras uPAϩ/CLϪ , the uPA mRNA levels were decreased to basal, and the CL mRNA levels were elevated to that identical to the level of CL mRNA in RAS1 Leu del of phenotype ras CLϩ/uPAϪ (21). This indicated that c-Jun levels regulate a switch between the uPA-and CL-dependent protease phenotypes, with high levels of c-Jun generating a uPAϩ/CLϪ protease phenotype and low levels of c-Jun a CLϩ/uPAϪ protease phenotype in the ras-transformed NIH 3T3 cells.
In this paper we show that expression of TAM67 inhibits both wild-type c-Jun expression and total AP-1 trans-activating activity in the ras-transformed NIH 3T3 cells. However, the cells remain transformed and invasive, as shown by growth in soft agar and invasion through Matrigel basement membrane. This result is surprising, since prior reports in other cell lines showed a requirement for AP-1 activity to maintain the transformed state induced by a ras oncogene (26 -31). We also show that inhibition of JNK1 in ras-transformed cells by expression of a dominant negative JNKK/SEK similarly has no effect on the ability of the cell to grow in soft agar or invade basement membrane. However, as with the inhibition of c-Jun/AP-1 through TAM67, expression of the mutant JNKK[K129R] in EJ/vHa-ras-transformed cells switches the protease phenotype for invasion from ras uPAϩ/CLϪ to ras CLϩ/uPAϪ . The protease phenotype could be switched in reverse, from ras CLϩ/uPAϪ to ras uPAϩ/CLϪ , by stimulating the intrinsically low JNK activity in RAS1 Leu del-transformed cells with the JNK stimulator anisomycin. Finally, the mRNA levels characteristic of the ras CLϩ/uPAϪ or ras uPAϩ/CLϪ phenotypes can be generated in non-transformed NIH 3T3 fibroblasts with selective inducers that activate either ERK alone or both ERK and JNK activities, respectively, indicating that a ras oncogene is not required to regulate the switch between CL and uPA expression. A switch in protease expressions in cellular invasions through regulation of JNK or c-Jun activities allows an invasive cell to overcome inhibitors of one class of protease (i.e. serine-or cysteine-type proteases) by switching to the expression of another class of protease.

EXPERIMENTAL PROCEDURES
Plasmids and Cell Lines-NIH 3T3 cells and cells transformed with EJ/vHa-ras and RAS1 Leu del were gifts from Merck (West Point, PA) and cultured at low passage number as described (24,25). These cells have previously been characterized with regard to Ras activity and metastatic potential (24,25). The cells expressed either the chimeric EJ/vHa-ras gene constructed by Gibbs et al. (32) or the RAS1 Leu del gene constructed by DeFoe-Jones et al. (33).
The c-Jun(TAM67) genes 297-CMV-c-TAM and 296-CMV-vector (empty vector) were a gift from Michael Birrer (National Cancer Institute, Biomarkers and Prevention Research Branch, Rockville, MD). c-Jun(TAM67) codes for a c-Jun protein lacking the trans-activation domain (residues 2-122) (30,34). The plasmids pCEP4 ERK1-K71R and pCEP4-ERK2-K52R (5,35) containing an NH 2 -terminal 6ϫHis motif were gifts of Melanie Cobb (University of Texas Southwestern Medical Center, Dallas). The dominant negative JNKK/SEK1 plasmid pEBG-SEK1(K129R) (6) was a gift from Leonard Zon (Dana Farber Cancer Institute, Boston). The pGEX-c-Jun-(1-135) plasmid containing an NH 2 -terminal GST utilized in the JNK assay was synthesized by Woodgett and co-workers (7) and was obtained from John Kyriakis (Massachusetts General Hospital, Boston). The plasmid pBL-CAT2 (36) containing a 3ϫAP1 consensus sequence from human collagenase upstream of the thymidine kinase minimal promoter and the chloramphenicol acetyltransferase (CAT) gene was a gift from Michael Karin (University of California, San Diego). The pRSVZ plasmid containing the ␤-galactosidase gene under control of a Rous sarcoma virus long terminal repeat promoter was obtained from the American Tissue Culture Collection (ATCC).
Introduction of Stably Expressing Genes into Cells by Calcium Phosphate Transfection-Genes were introduced into NIH 3T3 cells by the calcium phosphate precipitation method of Wigler et al. (37). The experimental plasmid DNA (1-2 g) was transfected along with 1 g of pcDNA3 (neomycin resistance expression plasmid, Invitrogen, San Diego, CA) and made up to 10 g of total DNA with puc19 DNA. Cells were exposed to the DNA/calcium phosphate (10 g DNA, 125 mM CaCl 2 , 0.75 mM Na 2 HPO 4 , 5 mM KCl, 140 mM NaCl, 6 mM glucose, 25 mM HEPES (pH 7.0) solution) for 4 h, rinsed with phosphate-buffered saline, and allowed to grow overnight in complete media. On the following day cells were divided into three 100-mm plates, and media were changed to DMEM, 10% FBS containing 200 g/ml G418. Twenty-four hours later the media were changed to complete media containing 600 g/ml G418, and the cells were selected for 14 -21 days. G418-resistant colonies were isolated for expansion using cloning cylinders, and cells were maintained in DMEM, 10% FBS containing 200 g/ml G418 until analysis.
Electrophoresis and Western Blot Analysis-Protein concentrations were determined with the Bio-Rad Protein Assay Kit, and the samples normalized, the samples boiled for 2 min in 6ϫ PAGE buffer, and subjected to PAGE electrophoresis. Silver staining of PAGE gels was performed with the Bio-Rad Silver Stain Kit (Bio-Rad).
For Western blotting, cells were grown to confluence in 100-mm dishes; the cell lysate was isolated as above, and equal amounts of protein were electrophoresed through a 12% SDS-PAGE gel at 150 V for 90 min; the gels washed for 15 min in 100 ml of transfer buffer (20% methanol, 0.05% SDS, 25 mM Tris, 192 mM glycine) and transferred to nitrocellulose membranes (Nytran) in a Bio-Rad transblot apparatus at 200 mA for 90 min or overnight at 20 mA in ice-cold transfer buffer. Blots were blocked for at least 2 h in 5% (w/v) non-fat dried milk in TBST buffer (20 mM Tris-HCl (pH 7.0), 137 mM NaCl and containing 0.1% Tween 20); primary antibody was added to the blocking solution at a concentration recommended by the manufacturer and allowed to incubate overnight. Blots were rinsed with 10 ml of TBST and incubated with a 1:2000 dilution of secondary antibody (goat anti-rabbit peroxidase conjugate) for 2 h. The blots were washed 3 times with 10 ml of TBST, blotted dry, immersed for 30 s in 5 ml of LumiGLO chemifluorescence substrate (Kirkegaard & Perry, Gaithersburg, MD), briefly allowed to drip dry, placed between plastic sheets, and exposed to x-ray film (Sterling x-ray Film, Dublin, OH).
Verification of c-Jun(TAM67) Expression-c-Jun(TAM67) mRNA was verified by Northern blot analysis using the 820-base pair BamHI fragment of the pCMV-TAM67 plasmid which binds to both the native c-Jun message (2.7-3.0 kilobase pairs) and the truncated c-Jun(TAM67) message (866 base pairs). Cell lines that showed c-Jun(TAM67) mRNA were further analyzed by Western blot analysis using a c-Jun antibody (sc-44-G, Santa Cruz Biotechnology Inc., Santa Cruz, CA) which recognizes the COOH-terminal DNA-binding region (residues 247-263) of c-Jun. Clones expressing TAM67 showed a primary band at 29 kDa and the absence of a native c-Jun band at 39 kDa.
Verification of Dominant Negative ERK Expression-Dominant negative plasmids pCEP4-ERK1[K71R] and pCEP4-ERK2[K52R] express catalytically inert ERKs with an NH 2 -terminal His 6 -tag (5,35). Transfected cells showing G418 resistance were grown to confluence in 100-mm plates and treated with denaturing lysis buffer (DLB) (8 M urea, 0.1 M sodium phosphate, 0.01 M Tris-HCl (pH 8.0)), and the clarified cell lysates (see above) were purified over Ni 2ϩ spin columns (Quiagen, Chatsworth, CA) that had been pre-equilibrated with 600 l of DLB (pH 6.3). Loaded columns were washed twice with 600 l of DLB containing 20 mM imidazole to remove non-binding proteins. Bound proteins were then eluted in 2 washes of 75 l each of DLB (pH 4.5) containing 200 mM imidazole. The Ni 2ϩ column eluents were analyzed by SDS-PAGE and Western blotting to show the expression of mutant ERK1 and ERK2 in transfected cells. Western analyses utilized anti-ERK antibody SC-94 (Santa Cruz Biotechnology, Santa Cruz, CA) which recognizes both ERK1 and ERK2. In addition, ras-transformed cells expressing the mutant ERKs genes showed an inhibition of the normally elevated ERK kinase activity to basal levels, identical to the levels of ERK activity present in untransformed NIH 3T3 cells.
Kinase Assays-The JNK kinase assay was carried out as described previously (21).
The ERK assays were performed on immunoprecipitated ERK1/2 from cell lysates normalized to a concentration of 100 g of total protein using 1 l of anti-ERK1 antibody (SC-94-G, Santa Cruz Biotechnology, Santa Cruz, CA) which recognizes both ERK1 and ERK2. The immunoprecipitate was washed 3 times with 500 l of kinase buffer (50 mM HEPES (pH 7.4), 10 mM MgCl 2 , 5 mM mercaptoethanol, 2 g/ml aprotinin, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin A, 20 mM ␤-glycerophosphate, 50 M Na 3 VO 4 , and 10 mM p-nitrophenol phosphate) followed by incubation for 30 min at room temperature in 30 l of kinase buffer containing 0.25 mg/ml myelin basic protein and 50 M ATP of 10 Ci/ml [ 32 P]ATP. The reaction was terminated by addition of 10 l of 6ϫ PAGE loading buffer, boiled 2 min, PAGE electrophoresed through a 15% acrylamide gel, and the gel dried and autoradiographed. To quantitate the incorporation of 32 P into myelin basic protein, the bands determined by autoradiography were cut out, put into 1 ml of scintillation fluid (Bio-Scint II, Research Products International, Mount Prospect, IL), and counted in a scintillation counter.
Measurement of AP-1 Activity with 3ϫAP1-pBL2-CAT Vector-Cells lines were plated in triplicate in six-well plates at 25 ϫ 10 3 cells/well in DMEM containing 10% FBS 1 day prior to transfection. Cells were transfected by treatment with a DNA/FuGENE solution containing 1 g of 3ϫAP-1 pBL2-CAT DNA, 1 g of ␤-galactosidase pRSVZ DNA, and 3 l of FuGENE reagent (Boehringer Mannheim) mixed according to the manufacturer's instructions. The FuGENE DNA reagent was added dropwise to cells and the plates returned to the incubator for 48 h. The media were removed, and the cells were washed with phosphate-buffered saline, and 350 l of TEN buffer (40 mM Tris-HCl (pH 7.5), 1 mM EDTA (pH 7.8), and 150 mM NaCl) was added, and the cells were scraped with a rubber policeman. Following centrifugation in the cold, the pellet was resuspended in 100 l of 0.25 M Tris-Cl (pH 7.5), and subjected to three freeze-thaw cycles. The cell lysate was centrifuged and the supernatant frozen at Ϫ80°C. A 20-l aliquot of each cell lysate was assayed for ␤-galactosidase activity against the substrate 2-nitrophenyl-␤-galactopyranoside as described by Hauser et al. (38). A 5-l aliquot of each cell lysate was analyzed for CAT activity by the addition to the aliquot 2 l of 100 Ci/ml [ 14 C]chloramphenicol (40 -60 mCi/ mmol), 20 l of 4 mM acetyl-CoA, 25 l of 1 M Tris-Cl (pH 7.5) and H 2 O to a final volume of 150 l, and incubated at 37°C for 1.5 h. The reaction was extracted with 1 ml of ethyl acetate and the ethyl acetate evaporated in a Speedvac evaporator. Samples were resuspended in 30 l of ethyl acetate and spotted 2 cm above the edge of a plastic-backed TLC sheet. Thin layer chromatography was run in 19:1 chloroform/ methanol, and the sheet air-dried and placed on film for autoradiography. The film was aligned with the TLC sheet, and the radioactive spots were cut out and added to scintillation fluid and counted in a scintillation counter. CAT activity was calculated as percent acetylated product that was normalized by the ␤-galactosidase values for each sample to correct for transcription efficiency.
Growth in Soft Agar-The procedure utilized was similar to Clark et al. (39). Cells were grown in 6-well plates in complete media, lifted by incubation in trypsin/EDTA (Sigma), counted in a hemocytometer, and mixed at a concentration of 10,000 cells/2 ml of DMEM, 0.33% agar (Difco). The cells in 0.33% agar were plated over 2 ml of DMEM, 0.5% agar that had been allowed to harden in a 6-well dish. Cells were fed 0.5 ml of DMEM, 0.33% agar twice a week and allowed to grow for 14 days at which time colony formation was determined under low magnification (ϫ 10) in an inverted microscope.
Matrigel Invasion Assays-The Matrigel invasion chambers were prepared at a 1:20 dilution of Matrigel (Becton Dickinson) with cold serum-free DMEM by the procedure described by Repesh (40). Conditioned media were prepared from untransformed NIH 3T3 cells and stored in aliquots at Ϫ70°C prior to use. The conditioned media were allowed to thaw, and 750 l was added to the bottom of each well of a 24-well plate. Cells (50,000 cells in 0.5 ml of DMEM/BSA (0.1% BSA)) were added to the top of Matrigel inserts in invasion chambers (Corning Costar, Kennebunk, ME), and the invasion chambers were incubated for 18 -20 h at 37°C and 5% CO 2 . Non-invading cells were removed from the upper surface of the membrane by "scrubbing" with a cottontipped swab moved gently over the membrane surface. The membrane was fixed in methanol (1-2 min) and stained with Diff-Quik (Scientific Products, McGaw Park, IL). Dyed inserts were dried and the membranes removed, placed on top of a drop of immersion oil on a slide, and covered with a coverslip. Invading cells were counted under the microscope at ϫ 100, with at least 15 fields counted and the average number of cells per field calculated. In some experiments the total number of invading cells was determined with similar results.
Invasions in the presence of antibody were carried out as follows. Polyclonal anti-CL antibody (Athens Research and Technology, Athens, GA) was diluted to a stock concentration of 80 g/ml, and polyclonal anti-uPA (mouse) antibody (American Diagnostica, Greenwich, CT) was diluted to a stock concentration of 37.5 g/l. Antibody solution (15 l) was added to a Matrigel insert (CoStar, Cambridge, MA), and the treated insert was incubated for 15 min. Cells (50,000 in 200 l of DMEM/BSA) were added and the inserts incubated for a further 15 min. The inserts were placed into an invasion chamber over conditioned media (750 l) prepared from untransformed NIH 3T3 fibroblasts. The chambers were incubated for 24 h at 37°C in 5% CO 2 .
Treatment of Cells with JNK and ERK Pathway Stimulators-M1-3T3 cells expressing the acetylcholine G q -linked m1 muscarinic receptor have been previously characterized by Coso et al. (41). The m1-3T3 cells were grown to confluency in 100-mm culture plates in complete media containing 10% FBS. Sixteen hours prior to treatment, the media were changed to DMEM containing 0.1% FBS. Immediately prior to treatment the media were changed to serum-free DMEM. The stimulators utilized were PDGF (type B, B), obtained from Upstate Biotechnology (Lake Placid, NY), and carbachol and anisomycin (Sigma). The concentrations utilized and the time course of treatment are described in the figure legends.
Northern Blot Analyses for uPA and CL-The mouse CL cDNA plasmid was a gift from Dr. G. Gary Sahagian (Tufts University School of Medicine, Boston), and mouse uPA cDNA was from Dominique Belin (Center Medical Universitaire, University of Geneva). The Northern blots were standardized for equal application of RNA by comparing the UV fluorescence of the 18 S and 28 S rRNA bands in the agarose gels according to the method of Bonini and Hofmann (42). Several blots were reprobed with glyceraldehyde-3-phosphate dehydrogenase, according to methods previously described (21), to verify the equal loading of RNA in the Northern analyses.
Quantification of Autoradiograms-Autoradiograms were scanned into Photoshop 3.05 for the Macintosh, and the relative intensity of the bands was analyzed using the public domain NIH image software. 2 Statistical analyses were carried out utilizing analysis of variance, and the reported values represent the mean Ϯ S.D. Fig. 1 shows that in agreement with a previous report (21), parental EJ/vHa-ras-transformed cells have a 6-fold higher JNK activity than untransformed NIH 3T3 cells, while in RAS1 Leu del-transformed cells JNK activity is inhibited to approximately 30% that observed in NIH 3T3 cell controls (lanes 1-3).  5) showing the inhibition of JNK activity to a level similar to the RAS1 Leu del-transformed cells. While other EJ/vHa-ras[K129R] clones tested showed inhibition of JNK activity (data not shown), clone 12 was neomycin-resistant but showed only partial inhibition of its JNK activity (Fig. 1, lane 6). This cell line was retained in subsequent experiments as a negative control. RAS1 Leu del-transformed cells were similarly transfected with the JNKK[K129R] and the neomycin resistance plasmids, and clones were selected for neomycin resistance. The RAS1 Leu del cells have an intrinsically low JNK activity (Fig. 1), and therefore assays of the JNK activity in RAS1 Leu del cells expressing JNKK[K129R] are not informative.

Characterization of Cells Expressing Dominant Negative Mutant Genes-
TAM67 lacks the NH 2 -terminal activation domain (amino acid residues 3-122) of c-Jun and acts to down-regulate c-Jun/AP-1 trans-activating enhancer activity (30). EJ/vHa-ras-transformed cells expressing the dominant negative c-Jun, TAM67, have been previously reported (21). However, the evidence for the expression of TAM67 in the EJ/vHa-ras-TAM clones has not been presented. In addition, we have overexpressed TAM67 in RAS1 Leu del-transformed cells, even though c-Jun levels are nor-mally low in these cells (21), to prevent the possible induction of c-Jun by environmental stimuli. Expression of TAM67 was verified in transfected cells by the appearance of the truncated message (866 nucleotides) in Northern analyses. In addition, Western blotting showed the appearance of a 29-kDa protein due to the expression of TAM67 and the disappearance of the band due to native c-Jun of 39 kDa in c-Jun-expressing cells (Fig. 2). Levels of native wild-type c-Jun are decreased in TAM67-expressing cells as c-Jun gene expression is auto-regulated by a c-Jun/AP-1 enhancer site (43). The observation of c-Jun downregulation supports the ability of TAM67 to act as a dominant negative gene in our stably transfected cells. Clone EJ/vHa-ras-TAM5 was neomycin-resistant but showed no TAM67 expression (Fig. 2, lane 12). This clone was incorporated into subsequent experiments as a negative control to test for possible biases introduced by the selection process.
It has been previously reported that TAM67 not only represses c-Jun activity but also quenches total AP-1 trans-activating activity through its association with other Jun and Fos family members to form inactive complexes at AP-1 gene enhancer sites (34). Accordingly, the expression of TAM67 and similar c-Jun mutants lacking a functional activation domain have been previously used to inhibit total AP-1 trans-activating activity in designated cells (26,27,34). In order to determine if AP-1 trans-activating activity was repressed in EJ/vHa-ras-TAM cells, parental cells and cells stably expressing TAM67 were transiently transfected with a 3ϫAP-1-tk-CAT reporter plasmid (pBL-CAT2) to assay AP-1 trans-activation activity (Fig. 3). Parental EJ/vHa-ras-transformed cells showed a 3.5fold higher AP-1 trans-activating activity than untransformed NIH 3T3 cells and a 6-fold higher activity than RAS1 Leu deltransformed cells. EJ/vHa-ras-TAM4 cells showed a 5-fold decrease in AP-1 activity from that of parental EJ/vHa-ras cells (significant at p Ͻ 0.001) to levels identical to the activity of RAS1 Leu del cells (not significant at p Ͼ 0.05). EJ/vHa-ras-TAM5 cells, which were previously shown not to express observable TAM67 (Fig. 2) and were incorporated as a negative control, showed an elevated AP-1 activity (2-fold greater than NIH 3T3 controls and 3-fold greater than EJ/vHa-ras-TAM4 cells).
EJ/vHa-ras-and RAS1 Leu del-transformed cells were prepared expressing the catalytically inert mutant ERK genes, (His) 6 -ERK1[K71R] and (His) 6 -ERK2[K52R], which act as dominant negatives genes for ERK-1 and ERK-2, respectively, and also express an NH 2 -terminal 6ϫHis tag (5,35). Expression of mutant ERK1 and ERK2 genes was verified by passing the cell lysates from transfected cells over a Ni 2ϩ affinity column which binds the mutant proteins through their 6ϫHis motif and then eluting the mutant proteins in imidazole buffer. The eluted mutant ERK-1 and ERK-2 protein bands were observed in SDS-PAGE and identified by Western analysis with anti-ERK antibody. Control cells transfected with mock plasmid showed no protein bands in the same analysis. In addition, measurement of ERK activity in mutant ERK-expressing cells against the ERK-specific substrate myelin basic protein showed levels of ERK activity at or below that present in untransformed NIH 3T3 control cells, whereas the parental EJ/vHa-ras-and RAS1 Leu del-transformed cells had an elevated ERK activity (3-fold higher than untransformed NIH 3T3 cells). The level of activated JNK1 was measured in EJ/vHa-ras clones expressing dominant negative ERK1/2 and found to be identical to the level in the parental EJ/vHa-ras-transformed cells (data not shown), demonstrating that the inhibition of ERK1/2 activities had no effect on the high activity of JNK1 in EJ/vHa-ras cells.
Neither JNK Activity nor c-Jun/AP-1 Are Required for the Maintenance of the Transformed and Invasive Phenotype of EJ/vHa-ras-and RAS1 Leu del-transformed Cells-EJ/vHa-ras-JNKK clones 7-12 and RAS1 Leu del-JNKK clones 1-3 expressing dominant negative JNKK were tested for growth in soft agar. Similarly, EJ/vHa-ras-TAM clones 1-6 and RAS1 Leu del-TAM clones 1-3 expressing TAM67 were also tested for growth in soft agar. All clones expressing JNKK[K129R] or TAM67 formed greater than 100 non-adhering colonies (foci) visible under low magnification. The number and types of non-adhering colonies were identical to the number of colonies formed in soft agar by the parental EJ/vHa-ras-transformed and RAS1 Leu del-transformed cells. The untransformed NIH 3T3 cells formed no foci in soft agar in the same assay. EJ/vHa-ras (5 clones tested) cells and RAS1 Leu del (4 clones tested) cells expressing dominant negative ERK1 and ERK2 also did not form foci in the soft agar assay. The results showed that elevated JNK or c-Jun/AP-1 activities are not required for growth in soft agar, even by the EJ/vHa-ras cells which normally have a constitutively elevated JNK and c-Jun/AP-1 activities. However, elevated ERK activity is required for growth in soft agar.
Selected clones of EJ/vHa-ras-and RAS1 Leu del-transformed cells expressing dominant negative JNKK[K129R] or TAM67 were tested in a Matrigel basement membrane invasion assay. In this assay cells are placed on top of a Matrigel basement membrane in an invasion chamber, and conditioned media were placed below the membrane in the chamber. The chamber is incubated for 18 h at 37°C and 5% CO 2 , and the number of cells that invade through to the bottom side of the basement membrane are counted. Matrigel invasion has been found to generally correlate with the metastatic potential of the tumor cell (40,44). In support of its applicability as an assay of the invasive phenotype, inhibitors of in vivo invasion and metastasis are found to similarly inhibit invasion of transformed cells through Matrigel (45,46). In this assay, both the EJ/vHa-ras and RAS1 Leu del cells expressing JNKK[K129R] or TAM67 showed an identical number of invading cells as did parental EJ/vHa-ras-transformed cells and RAS1 Leu del-transformed cells, whereas untransformed NIH 3T3 cells showed a low level of invasion (3-20-fold lower, depending on the experiment) (Fig. 4). Cells expressing the dominant negative ERK1 and ERK2 also showed a low invasion cell number, identical to that observed for untransformed NIH 3T3 cells (Fig. 4C). Thus, elevated JNK activity or c-Jun and AP-1 trans-activating activity are not required for invasion through the Matrigel basement membrane by the EJ/vHa-ras-and RAS1 Leu del-transformed cells, whereas elevated ERK activity is required.

. AP-1 trans-activating enhancer activity assayed with a 3؋AP1-tk-CAT reporter plasmid in untransformed NIH 3T3, RAS1 Leu del-transformed, EJ/vHa-ras-transformed, and EJ/vHa-ras cells expressing c-Jun(TAM67).
Cells were co-transfected with the pBL-CAT2 (3ϫAP1-tk-CAT) and pRSVZ (␤-galactosidase) plasmids, and the CAT activity was assayed after 48 h. Results were normalized to ␤-galactosidase activity to control for transfection efficiency. Three independent transfections were performed for each data point, with the mean transcriptional activity plotted. The error bars show the standard deviation from the mean. The experiment was carried out two additional times with similar results. Asterisk shows cell lines significantly different from EJ/vHa cells at p Ͻ 0.001. Thus, EJ/vHa-ras-transformed cells are significantly different from RAS1 Leu del-transformed (RAS1del)-transformed, EJ/vHa-ras-TAM4-transformed, and NIH 3T3 cells at p Ͻ 0.001, whereas NIH 3T3, RAS1 Leu del-transformed, and EJ/vHa-ras-TAM4-transformed cells showed no significant difference in their AP-1 activity (not significant at p Ͼ 0.05). cells of phenotype ras CLϩ/uPAϪ expressed a 4-fold elevated level of CL mRNA and a low basal concentration of uPA mRNA (lane 3 versus lane 1). These levels of uPA and CL mRNA are similar to those previously reported by EJ/vHa-ras-and RAS1 Leu deltransformed cells of phenotype ras uPAϩ/CLϪ and ras CLϩ/uPAϪ , respectively (21,25). Fig. 5 then shows that the expression of dominant negative JNKK in the EJ/vHa-ras cells causes a switch in the protease mRNA levels from those characteristic of the parental EJ/vHa-ras cells (phenotype ras uPAϩ/CLϪ ) to the phenotype ras CLϩ/uPAϪ characteristic of the RAS1 Leu del-transformed cells. This switch involves a 2.5-3.5-fold increase in the level of CL mRNA and the 3-fold decrease in the level of uPA mRNA. The negative control Clone 12 cells, which we have previously shown to express neomycin resistance but lack detectable expression of JNKK[K129R], showed uPA and CL mRNA levels identical to that of the parental EJ/vHa-rastransformed cells and indicate that the selection procedure did JNKK were not significantly different at p Ͼ 0.05 from parental EJ/ vHa-ras-or RAS1 Leu   not preferentially select for a CLϩ/uPA-phenotype. Other EJ/ vHa-JNKK1 clones showed a similar switch in protease mRNA levels from ras uPAϩ/CLϪ to ras CLϩ/uPAϪ as shown for clone 7 and clone 8.
RAS1 Leu del-transformed cells were also transfected with the JNKK[K129R] plasmid, and isolated neomycin-resistant cells showed no change in protease phenotype from parental RAS1 Leu del-transformed cells of phenotype ras CLϩ/uPAϪ (data not shown). Similarly, transfection of untransformed NIH 3T3 cells with JNKK[K129R] yielded neomycin-resistant clones with low basal levels of uPA and CL mRNA, similar to the level of parental NIH 3T3 cells. These results were expected, as the RAS1 Leu del-transformed cells and untransformed NIH 3T3 cells normally show low JNK activity, and the expression of JNKK[K129R] would not be predicted to change the protease phenotype of the cells. Fig. 6 shows the number of EJ/vHa-ras-transformed cells and JNKK[K129R]-expressing EJ/vHa-ras cells that invaded the Matrigel membrane in the presence of inhibiting antibodies against uPA or CL. Invasion of EJ/vHa-rastransformed cells through Matrigel was inhibited by 59% (significant at p ϭ 0.0049) with anti-uPA antibody and was not significantly inhibited by anti-CL antibody, as expected for cells of phenotype ras uPAϩ/CLϪ . In contrast, invasion of the RAS1 Leu del-transformed cells was inhibited by anti-CL antibody by 41% (significant at p ϭ 0.01) and was not inhibited by anti-uPA antibody, as expected for cells of phenotype ras CLϩ/uPAϪ . The invasion of the EJ/vHa-ras-transformed cells expressing JNKK[K129R] was not inhibited by anti-uPA antibody, but its invasion was inhibited by anti-CL antibody by 34% (significant at p ϭ 0.0006), demonstrating a switch from a ras uPAϩ/CLϪinvasive phenotype to a ras CLϩ/uPAϪ -invasive phenotype upon expression of JNKK [K129R]. Fig. 7 shows the expression of the dominant negative ERK-1 and -2 decreased the concentration of CL mRNA in RAS1 Leu del cells and of uPA mRNA in EJ/vHaras cells to levels identical to those in untransformed NIH 3T3 cells. EJ/vHa-dnERK-clone 5 cells have been shown to express neomycin resistance but not to express dominant negative ERK activity. It was included as a control cell line and shows the selection procedure did not preferentially select for a low protease phenotype. Accordingly, a stimulated ERK activity is required for the elevated expression of uPA or CL mRNA characteristic of the EJ/vHa-ras-or RAS1 Leu del-transformed phenotypes, respectively.

Decrease of Elevated uPA and CL mRNA Levels by Expression of Dominant Negative ERK1[K71R] and ERK2[K52R] in Ras-transformed Cells-
Switch of a ras CLϩ/uPAϪ to a ras uPAϩ/CLϪ Phenotype by Treatment of RAS1 Leu del-Transformed Cells with the JNK Pathway Stimulator Anisomycin-Treatment of cells with low concentrations of anisomycin has been shown to selectively stimulate c-Jun phosphorylation in the absence of ERK activation (47). The RAS1 Leu del-transformed cells of phenotype ras CLϩ/uPAϪ have an intrinsically low JNK activity (21). Accordingly, we determined whether treatment of RAS1 Leu del cells with anisomycin would be sufficient to switch its phenotype (ras CLϩ/uPAϪ ) to the ras uPAϩ/CLϪ phenotype. Fig. 8 shows that treatment of RAS1 Leu del-transformed cells with anisomycin increased JNK activity by 2.5-4-fold from 1 to 8 h after initiation of treatment. Parallel with the rise in JNK activity is a 2.5-4-fold increase in uPA mRNA levels at 2 and 4 h (Fig. 8).
Corresponding to the increase in uPA mRNA levels, we observed a decrease in CL mRNA to levels in untransformed NIH

and parental EJ/vHa-ras-and RAS1 Leu del-transformed cells co-transfected with empty and neomycin-expressing vectors (first three cell lines) followed by selected clones of EJ/vHa-ras-transformed cells co-transfected with the dominant negative ERK1[K71R] and ERK2[K52R]
and the neomycin-resistant vector. All cells were selected for neomycin resistance, and the relative levels of CL and uPA mRNA were determined with respect to NIH 3T3 controls. The Northern blots were standardized for equal application of RNA by comparing the UV fluorescence of the 18 S and 28 S rRNA bands in the agarose gels (see "Experimental Procedures"). Error bars represent the S.D. of multiple determinations. Clone EJ/vHa-ERK5 was neomycin-resistant but did not express dominant negative ERK activity. Bottom, CL and uPA mRNA levels in NIH 3T3 cells, and parental EJ/vHa-ras-and RAS1 Leu del-transformed cells co-transfected with empty and neomycin-expressing vectors (first three cell lines) followed by selected clones of RAS1 Leu del-transformed cells co-transfected with the dominant negative ERK1[K71R] and ERK2[K52R] and the neomycin-resistant vector. All cells were selected for neomycin resistance and the relative levels of CL and uPA mRNA determined relative to NIH 3T3 control. The Northern blots were standardized for equal application of RNA by comparing the UV fluorescence of the 18 S and 28 S rRNA bands in the agarose gels (see "Experimental Procedures"). Error bars represent the S.D. of multiple determinations.

3T3 cells (data not shown).
Treatment of Non-transformed m1-NIH 3T3 Cells with PDGF Stimulates ERK Activity Alone and Leads to an Elevation of CL mRNA Levels-We determined whether the mRNA levels characteristic of the ras uPAϩ/CLϪ and ras CLϩ/uPAϪ phenotypes could be reproduced in non-transformed cells by stimulating ERK activity or ERK and JNK activities in the absence of constitutively activated ras. m1-3T3 cells contain the G q trimeric protein-linked acetylcholine receptors (AChRs) as well as PDGF receptors (41). Cells treated with PDGF showed a sustained 2-3-fold stimulation of ERK activity at 4, 6, and 8 h after PDGF treatment (Fig. 9B). PDGF treatment had no effect on the low basal activity of JNK observed in these cells. If the same cells were treated with the m1-AChR receptor agonist, carbachol, ERK activity was stimulated only briefly (at 5-10 min, see Ref. 41 and data not shown) and otherwise remained at low basal levels at 2-8 h after initiation of carbachol treatment (Fig. 9A). However, carbachol treatment increased JNK activity by 4-fold in a sustained manner at 4 -8 h (Fig. 9A). Similar results were previously reported by Coso et al. (41). Fig.  9D shows that the PDGF-induced increase in ERK but not JNK activity correlated with a 3-fold increase in the levels of CL mRNA, whereas the levels of uPA mRNA remained at basal levels. This mRNA profile is similar to the uPA and CL mRNA levels observed in the RAS1 Leu del-transformed cells (Fig. 5).
The elevation of JNK activity alone had no effect on the level of either uPA mRNA or CL mRNA (Fig. 9C). This is in agreement with the findings that high levels of ERK are required for elevated expression of CL or uPA mRNA in RAS1 Leu del and EJ/vHa-ras oncogene-containing cells (Fig. 7). Fig. 10A shows that the combined treatment of cells with PDGF and carbachol increased both ERK and JNK activity 2-3-fold by 6 and 8 h. Anisomycin activates c-Jun by a complementary pathway not utilizing the G q protein (47), and the combined treatment of m1-3T3 cells with PDGF and anisomycin similarly increased both ERK and JNK activities 2-3-fold above basal levels at 4 -8 h (Fig. 10B). Treatment of the m1-3T3 cells with both the PDGF and m1 agonist carbachol or PDGF and anisomycin increased by 2-3fold the uPA mRNA levels at 6 -8 h after initiation of treatment but not the CL mRNA levels (Fig. 10, C and D). These stimulated levels are similar to the uPA and CL mRNA levels observed in the EJ/vHa-ras-transformed cells (Fig. 5). DISCUSSION The EJ/vHa-ras-and RAS1 Leu del Oncogene-transformed Cells Do Not Require Elevated JNK or AP-1 Activities to Maintain the Transformed State or for Invasion through Basement FIG. 9. Time course of changes in JNK and ERK activities and the corresponding CL and uPA mRNA levels on treatment of m1-3T3 cells with carbachol or PDGF. m1-3T3 cells were treated with PDGF or carbachol for the times indicated, and their JNK and ERK activities were determined using GST-c-Jun-(1-135) and myelin basic protein as substrates. The same cells were analyzed for uPA and CL mRNA levels by Northern blotting (see "Experimental Procedures"). Experiments were repeated at least three times, and the means Ϯ S.D. for each value are shown. A, m1-3T3 cells were treated with 100 M carbachol, and the JNK activity was significantly increased (p Ͻ 0.001, n ϭ 3) at 4, 6, and 8 h after treatment, whereas ERK activity showed no significant change (p ϭ not significant, n ϭ 3). B, m1-3T3 cells were treated with 10 ng/ml PDGF(B,B), and the JNK activity showed no significant increase (p ϭ not significant, n ϭ 6), but ERK activity was significantly increased (p Ͻ 0.05, n ϭ 4) at 4, 6, and 8 h following treatment. C, treatment with 100 M carbachol led to no change in levels of CL and uPA mRNA (p ϭ not significant, n ϭ 3). D, treatment with 10 ng/ml PDGF led to increased levels of CL mRNA at 4,6 and 8 h (p Ͻ 0.01, n ϭ 3), and no significant change in the level of uPA mRNA (p ϭ not significant, n ϭ 3).

Treatment with Both PDGF and Carbachol or PDGF and Anisomycin Stimulates both ERK and JNK Activity and Increases uPA mRNA Levels-
Membrane-The expression of JNKK[K129R] and c-Jun-(TAM67) activity did not inhibit the ability of cells containing the EJ/vHa-ras or RAS1 Leu del oncogenes to form colonies in soft agar or to invade Matrigel. These results indicate a lack of the requirement of JNK activity and Jun/AP-1 activity in transformation, even in the EJ/vHa-ras-transformed cells which normally have a high JNK activity and a corresponding high activity of c-Jun/AP-1. This is the first report showing (i) constitutive inhibition of the JNK pathway is compatible with ras transformation, and (ii) that expression of a dominant negative c-Jun which acts to quench AP-1 activity is compatible with ras transformation.
Previous experiments indicated that c-Jun activity is required for the transformed state induced by activated ras (9, 26, 27, 48 -52). Furthermore, the lack of reversion of the transformed state on expression of dominant negative c-Jun-(TAM67) is in conflict with previous reports of reversion of transformation in other cell types (28 -31, 53). However, there are two reports in conflict with the concept that c-Jun expression is essential for the initiation and/or maintenance of the ras-transformed state. Whereas Johnson et al. (26) showed that mutant Ha-ras could not initially transform c-jun Ϫ/Ϫ mouse embryo fibroblasts unless a c-Jun plasmid was also co-expressed, after a period of time c-jun null fibroblasts containing a ras oncogene spontaneously developed an ability to form tumors and acquired other properties of the transformed phenotype. Similarly, Marshall-Heyman et al. (27) reported two ras oncogene-transformed rat fibrosarcoma cells lines had spontaneously acquired a deficiency in c-jun expression, suggesting that c-Jun is not essential for the maintenance of transformation. However, both authors showed that in the c-jundeficient ras-transformed cells other Jun family AP-1 proteins had increased their activity, presumably compensating for the loss of c-Jun in the AP-1 complex. The requirement of AP-1 activity in these c-jun-deficient cells was demonstrated by the expression of dominant negative c-jun genes, such as TAM67, that act as general quenchers of AP-1 trans-activation (34). For both cases, expression of the mutant c-Jun down-regulated AP-1 activity and reverted the transformed cells to a normal phenotype (26,27). Thus although c-Jun was not specifically required, AP-1 activity appeared to be required for transformation in these c-Jun-deficient cells. However in our cells, AP-1 activity is not required for the maintenance of transformation.
The apparent conflict between our observations and previous reports may possibly be explained by a difference in genetic background among cell lines. The requirement of multiple mutational events involving the activation of oncogenes and loss of tumor suppressor genes leading to the transformation of a primary cell to a neoplastic cell is a generally accepted principle of carcinogenesis (54). As a hypothetical example we consider the opposing effects of c-Jun and tumor suppressor genes in the regulation of cyclin D1/Cdk4 activity. c-Jun/AP-1 activity increases the expression of the cyclin D1 gene (22), which when complexed with Cdk4 has a kinase activity required for the progression of cells through the critical control point between the G 1 and S phases in the cell cycle (55,56). However, counteracting cyclin D1 activity are tumor suppressor proteins that act to inhibit the activity of the cyclin D1⅐Cdk4 complex (55,56). If the tumor suppressor activity is lost, as occurs in many tumor cells, this may reduce the requirement for AP-1 stimulation of cyclin D1 expression. In carcinogenesis, the progression of genetic changes leading to a highly malignant cell can involve the loss of a requirement for AP-1 activity.
Several AP-1-responsive genes with roles in facilitating the invasion of malignant cells have been described including uPA, the uPA receptor, as well as the metalloproteinases MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9 (13)(14)(15)(16)(17)(18)(19)(20)(21). Thus, when AP-1 activity is inhibited and the invasive malignant phenotype is maintained, new genes may be expressed to compensate for the loss of these AP-1-dependent genes. An example of such a compensatory change, shown in this work, is the substitution of CL protease gene expression for uPA protease gene expression in the context of suppressed c-Jun/AP-1 activity.
ERK activities are constitutively elevated in both EJ/vHaand RAS1 Leu del-transformed NIH 3T3 cells. This is in agreement with prior reports indicating that ERK activity is required for cell proliferation and the transformed phenotype (57)(58)(59). It was therefore not surprising that the expression of dominant negative ERK1[K71R] and ERK2[K52R] in the rastransformed cells inhibited the ability of these cells to grow in soft agar and to invade Matrigel basement membrane. We were unable to isolate a single clone that expressed detectable levels of dominant negative ERK1 and ERK2 and was still transformed.
The JNK Pathway Regulates a Switch between uPA-and CL-dependent Protease Phenotypes That Are Redundant in Their Ability to Invade Basement Membrane-In this paper we show the JNK pathway can act as a switch between uPA-and CL-dependent protease-invasive phenotypes that are redundant in their ability to invade basement membrane. The EJ/ vHa-ras-transformed cells have an intrinsically high activity of JNK and high levels of c-Jun and phosphorylated c-Jun (21). The inhibition of JNKK activity in EJ/vHa-ras-transformed cells decreased its JNK activity to basal levels. Correlated with a decrease in JNK activity was a decreased uPA mRNA concentration and increased CL mRNA concentration, resulting in an apparent switch of protease mRNA levels from those characteristic of the EJ/vHa-ras of phenotype ras uPAϩ/CLϪ to those characteristic of RAS1 Leu del-transformed cells of phenotype ras CLϩ/uPAϪ (Fig. 5). This switch in phenotype is similar to that observed previously on introduction of the dominant negative c-Jun(TAM67) into EJ/vHa-ras-transformed cells (21). As the expression of both dominant negative JNKK and c-Jun change the protease phenotype from ras uPAϩ/CLϪ to ras CLϩ/uPAϪ in EJ/vHa-ras-transformed cells, it is suggested that the JNK pathway in these cells is required for production of the ras uPAϩ/CLϪ protease phenotype through activation of c-Jun.
In agreement with the hypothesis that JNK activity acts as a switch between transformed protease phenotypes, the stimulation of JNK activity by anisomycin in RAS1 Leu del-transformed cells, which normally contain low JNK and AP-1 activities and low basal levels of uPA mRNA, was sufficient to transiently increase uPA mRNA levels to the levels observed in EJ/vHa-ras-transformed cells (Fig. 8). Corresponding with the increase in JNK activity and the increase in uPA mRNA levels was a decrease in CL mRNA levels. This represents an apparent switch of phenotype from ras CLϩ/uPAϪ to ras uPAϩ/CLϪ in RAS1 Leu del-transformed cells.
Whereas the JNK pathway acts as a switch between protease phenotypes, ERK pathway activity is required for the elevated levels of either uPA or CL mRNA. Expression of dominant negative ERK1[K71R] and ERK2[K52R] both in EJ/vHa-ras cells and in RAS1 Leu del cells resulted in a decrease in elevated uPA mRNA and CL mRNA levels, respectively, to basal levels (Fig. 7). The requirement for ERK activity for uPA expression found in these experiments is in agreement with prior data which showed ERK pathway activity was required for the enhanced expression of uPA in a squamous cell carcinoma cell line (UM-SCC-1) and in an ovarian carcinoma cell line (OVCAR-3) (20, 60, 61).
The uPA gene contains the PEA3-AP-1-like ras-responsive enhancer element (20,21,62). The PEA3 portion of the element binds transcription factors of the Ets family that have been shown to be activated by the ERK pathway, whereas the AP-1-like portion of the enhancer binds c-Jun activated by the JNK pathway (20,21,63). Accordingly, it is not surprising that the activation of both the ERK and JNK pathways are required to stimulate uPA gene expression, and thus the inhibition of JNK activity and/or c-Jun/AP-1 activity in EJ/vHa-ras-transformed cells is sufficient to inhibit ras-stimulated uPA gene expression. Although the loss of elevated uPA expression on inhibition of JNK and/or c-Jun/AP-1 is understood, it is not known why the inhibition of JNK activity leads to elevated CL mRNA levels or the expression of high JNK activity decreases elevated CL mRNA levels in the ras-transformed cells.
Cells of the different protease mRNA phenotypes were tested for their ability to invade through Matrigel basement membrane in the presence of inactivating antibodies directed against secreted uPA or CL. Invasion of the EJ/vHa-ras-transformed cells with uPA and CL mRNA levels characteristic of the ras uPAϩ/CLϪ phenotype was significantly inhibited by anti-uPA antibody and not by anti-CL antibody. Expression of JNKK[K129R] in the EJ/vHa-ras-transformed cells reversed its protease dependence for invasion through Matrigel from FIG. 10. Time course of change in JNK and ERK activities and the corresponding changes in the levels of CL or uPA mRNA on treatment of m1-3T3 cells with PDGF and carbachol or PDGF and anisomycin. M1-3T3 cells were treated with PDGF plus carbachol or PDGF plus anisomycin, and their JNK and ERK activities were determined using GST-c-Jun-(1-135) and myelin basic protein as substrates. The same cells were analyzed for uPA and CL mRNA levels by Northern blotting (see "Experimental Procedures"). Experiments were carried out at least 3 times, and the means Ϯ S.D. for each value are shown. A, treatment of m1-3T3 cells with 10 ng/ml PDGF and 100 M carbachol increased both JNK and ERK activities significantly (p Ͻ 0.05, n ϭ 4) at 4, 6, and 8 h. B, treatment of m1-3T3 cells with 10 ng/ml PDGF and 25 ng/ml anisomycin increased both JNK and ERK activities significantly (p Ͻ 0.05, n ϭ 3) at 4, 6, and 8 h. C, m1-3T3 cells treated with 10 ng/ml PDGF and 100 M carbachol led to a significant increase in uPA mRNA levels (p Ͻ 0.05, n ϭ 3) at 8 h but not CL mRNA levels (p ϭ not significant, n ϭ 3). D, m1-3T3 cells treated with 10 ng/ml of PDGF and 25 ng/ml anisomycin led to a significant increase in uPA mRNA levels at 6 and 8 h (p Ͻ 0.05, n ϭ 3) but not CL mRNA levels (p ϭ not significant, n ϭ 3).
FIG. 11. Conditions for generation of the ras uPA؉/CL؊ or ras CL؉/uPA؊ protease phenotype. All uPAϩ/CLϪ phenotypes showed elevated ERK and JNK and/or AP-1 activities, whereas all CLϩ/ uPAϪ phenotypes show only the elevation of ERK activity, with inhibition of basal activities of JNK and/or AP-1. In boxes, first line describes cell type and second line the treatment or dominant negative gene expression. AChR refers to the presence of acetylcholine m1 receptor.
uPA-dependent to CL-dependent, as shown by inhibition of Matrigel invasion by anti-CL antibody instead of anti-uPA antibody. Accordingly, the switch in the relative levels of CL and uPA mRNA correlated with a switch in protease dependence for basement membrane invasion.
Regulation of CL and uPA mRNA Profiles in Untransformed Cells by Stimulation of ERK Activity and by Stimulation of ERK Plus JNK Activities-The activated ras oncogene can activate additional pathways to those that lead to the activation of ERK and JNK (1)(2)(3). To determine whether constitutively activated ras is necessary for the generation of the CLϩ/ uPAϪ and uPAϩ/CLϪ protease phenotypes, we investigated the role of ERK and JNK activities in inducing uPA and CL mRNA levels in untransformed m1-3T3 cells. These cells contain the G q -coupled m1 muscarinic acetylcholine receptor (AChR) which is stimulated by carbachol to induce a sustained high JNK activity with only a brief stimulation of ERK activity ( Fig. 9 and Ref. (41). Anisomycin also stimulates the activation of c-Jun by promotion of the phosphorylation of its NH 2 -terminal trans-activation domain but by a different pathway than does the carbachol-stimulated AChR (47). Treatment of untransformed m1-3T3 cells with PDGF alone was found to stimulate ERK and CL mRNA levels, whereas treatment with PDGF and carbachol or with PDGF and anisomycin increased both ERK and JNK activities and increased uPA mRNA levels but not CL mRNA levels (Figs. 9 and 10). Thus stimulation of ERK activity in the absence of elevated JNK activation generates a protease phenotype in untransformed m1-3T3 cells similar to the ras CLϩ/uPAϪ phenotype formed by RAS1 Leu del-transformed cells, whereas the combined stimulation of ERK and JNK activities generate a protease mRNA profile similar to that observed in the ras uPAϩ/CLϪ phenotype of EJ/vHa-rastransformed cells. An activated ras is not required to generate these protease phenotypes.
The various ways that the ras uPAϩ/CLϪ or ras CLϩ/uPAϪ mRNA phenotypes were generated in this work are summarized in Fig. 11. The ras uPAϩ/CLϪ phenotype was formed in NIH 3T3 cells by (a) transformation with the EJ/vHa-ras oncogene, (b) treatment of RAS1 Leu del-transformed cells with anisomycin, (c) treatment of untransformed m1-3T3 cells with both PDGF and carbachol, or (d) treatment of untransformed m1-3T3 cells with PDGF and anisomycin. The ras CLϩ/uPAϪ phenotype was generated by (a) expression of the RAS1 Leu del oncogene, (b) expression of the dominant negative JNKK[K129R] gene in EJ/vHa-ras-transformed cells, (c) expression of the dominant negative c-Jun(TAM67) in EJ/vHa-ras-transformed cells, or (d) treatment of untransformed m1-3T3 cells with PDGF. In the ras uPAϩ/CLϪ phenotype both the ERK and JNK activities are elevated, whereas in the ras CLϩ/uPAϪ phenotype the ERK activity is elevated but the JNK activity and/or c-Jun and AP-1 activity is inhibited.
In summary, we have shown that JNK and c-Jun/AP-1 activities are not necessarily required to maintain a transformed and invasive phenotype. Rather in our cells, they regulate a switch between uPA-and CL-dependent protease invasive phenotypes. In ras-transformed NIH 3T3 cells these phenotypes are redundant in their ability to facilitate invasion through basement membrane. Although we have analyzed the expression of only two genes, other genes that contribute to the phenotypic properties of these cells most likely depend on whether the JNK pathway is activated or inhibited. Future work will be required to identify these accompanying genes.