c-Jun N-terminal Kinase Is Essential for Growth of Human T98G Glioblastoma Cells*

The c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK) pathway is activated by numerous cellular stresses. Although it has been implicated in mediating apoptosis and growth factor signaling, its role in regulating cell growth is not yet clear. Here, the influence of JNK on basal (unstimulated) growth of human tumor glioblastoma T98G cells was investigated using highly specific JNK antisense oligonucleotides to inhibitJNK expression. Transient depletion of either JNK1 or JNK2 suppressed cell growth associated with an inhibition of DNA synthesis and cell cycle arrest in S phase. The growth-inhibitory potency of JNK2 antisense (JNK2 IC50 = 0.14 μm) was greater than that of JNK1 antisense (JNK1 IC50 = 0.37 μm), suggesting that JNK2 plays a dominant role in regulating growth of T98G cells. Indeed, JNK2 antisense-treated populations exhibited greater inhibition of DNA synthesis and accumulation of S-phase cells than did the JNK1 antisense-treated cultures, with a significant proportion of these cells detaching from the tissue culture plate. JNK2 (but not JNK1) antisense-treated cultures exhibited marked elevation in the expression of the cyclin-dependent kinase inhibitor p21 cip1/waf1 accompanied by inhibition of Cdk2/Cdc2 kinase activities. Taken together, these results indicate that JNK is required for growth of T98G cells in nonstress conditions and that p21 cip1/waf1 may contribute to the sustained growth arrest of JNK2-depleted T98G cultures.

The c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK) 1 signal transduction pathway consists of a cas-cade of protein phosphorylation reactions leading to the activation of JNK. Numerous proinflammatory factors and other stressful stimuli activate JNK by dual-specificity kinases (JNK kinases), which are in turn activated via phosphorylation by upstream kinases (1)(2)(3). The function of the JNK pathway has mostly been studied within the context of cellular stress, where its activation causes both stabilization and elevated activity of targeted transcription factors via their phosphorylation by JNK (1)(2)(3)(4). Major substrates of JNK include the transcription factors c-Jun (5-7), activation transcription factor-2 (8 -10), Elk-1 (11,12), and p53 (13,14). Much less is known about the role of JNK in normally growing cells, but recent studies suggest that nonactivated JNK is important for regulating degradation of its substrates (4).
Three different JNK genes, designated JNK1, JNK2, and JNK3, have been described. JNK1 and JNK2 are ubiquitously expressed, while JNK3 is largely restricted to brain, heart, and testis. Each of the JNK genes produces several isoforms derived from alternative splicing, but the functional significance of the different isoforms is unclear. JNK1 and JNK2 display distinct substrate affinities (10,15) and may have selective or preferential roles in different biological processes. For example, JNK1 plays a role in the production of neurite-like structures in PC12 cells (16), while JNK2 has been implicated in regulating expression of E-selectin (17) and EGF-stimulated growth of A549 cells (18,19). The role of JNK in cell survival and death is complex (1)(2)20). For instance, a rapid increase in JNK activity is required for apoptosis in cells of neuronal origin following withdrawal of nerve growth factor (21), but in other instances, such as following tumor nectosis factor-␣ treatment, JNK activation appears to inhibit apoptosis (22,23).
Several lines of evidence suggest that the JNK pathway may play a role in cellular transformation and tumor cell growth. JNK activation is required for cellular transformation by certain oncogenes as well as by constitutively activated c-Ras, Rac, and protein-activated kinase-65 (24 -29), and inhibition of the JNK pathway can reverse a transformed phenotype (30,31). In addition, growth factors such as epidermal growth factor and platelet-derived growth factor can cause activation of the JNK pathway (18,(32)(33)(34), and systemic treatment of PC3 tumorbearing mice with JNK antisense oligonucleotides inhibits tumor growth and promotes regression in a high proportion of cases (35).
We have recently reported that treatment with JNK2-specific antisense oligonucleotides (JNK2AS) led to apoptosis in several cell types in a p53-dependent fashion and provided evidence that p21 cip1/waf1 , a downstream target of p53, was an important factor in the survival of cells with wild type p53 status (36). In a panel of human tumor cell lines containing null/mutant p53, the human T98G glioblastoma cell line stood out as an exception in that we failed to see any apoptosis following treatment with JNK antisense oligonucleotides (JN-KAS). The present study was initiated to further investigate the role of JNK in regulating basal growth (normal culture conditions) of human glioblastoma T98G cells. We demonstrate that targeted inhibition of JNK expression, accomplished through use of specific antisense oligonucleotides, does not result in significant apoptosis of T98G cells but rather results in marked growth suppression that is associated with inhibition of DNA synthesis, cell cycle arrest in S phase, and p53independent induction of the cyclin-dependent kinase inhibitor p21 cip1/waf1 . These findings indicate that JNK is essential for normal growth of T98G cells and suggest that the JNK pathway regulates molecules vital for replication of these tumor cells.
Treatment of Cells with Oligonucleotides-All phosphorothioate oligonucleotides used in this study were prepared and characterized by Isis Pharmaceuticals, Inc. as described previously (18,37). Two oligonucleotides, ISIS12539 (5Ј-CTCTCTGTAGGCCCGCTTGG-3Ј) and ISIS12560 (5Ј-GTCCGGGCCAGGCCAAAGTC-3Ј), termed JNK1AS and JNK2AS, respectively, specifically eliminate the expression of respective JNK proteins (17)(18)(19). To control for nonspecific events, "sense" sequence oligonucleotides (JNK1S (ISIS14320) and JNK2S (ISIS14318)), and "scrambled" sequence oligonucleotides (JNK1Scr (ISIS14321) and JNK2Scr (ISIS14319)) with the same base composition as the antisense oligonucleotides, but in arbitrary order, were employed. Oligonucleotides were delivered to cells by lipofection as described previously (18). Briefly, solutions for lipofection were made by mixing 10 g/ml Lipofectin TM (Life Technologies, Inc.) reagent in minimal essential medium (Life Technologies, Inc.) with an equal volume of oligonucleotide solution, incubating this mixture at room temperature for 15 min and diluting it with Lipofectin TM solution to a final oligonucleotide concentration of 0.4 M. Cells were incubated with the Lipofectin TM -oligonucleotide solution at 37°C in 10% CO 2 for 12-16 h, after which they were washed once with serum-free medium and cultured in complete medium.
Cell Growth and DNA Synthesis Assays-To determine the proliferation rates, cells were seeded at 25 ϫ 10 3 cells/cm 2 in six-well cluster plates, and growth was assessed by daily cell counting (Coulter counter) in triplicate. For determination of cell viability and assessment of DNA synthesis, cells were seeded at 1-5 ϫ 10 3 cells/well in 96-well tissue culture plates. The viable cell mass was determined by the addition of 3-(4,5-dimethylthiazol-2-yl-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) as described in the manufacturer's protocol (Promega). Cell viability is expressed as the ratio of viable cell mass following a given treatment to that of untreated cells ϫ 100 (Viable Cell Mass (%); see Fig. 5). Rates of DNA synthesis were measured by incorporation of [ 3 H]thymidine into polymeric DNA. Cells were treated with oligonucleotides and pulsed with 0.5 Ci of [ 3 H]thymidine/ well for 3 h at the indicated times. Harvested lysates were transferred to paper spots with a PhD-200A cell harvester (Cambridge Technologies), and the amount of radioactive DNA was quantitated by scintillation counting using Biosafe-II scintillation liquid.
Northern and Western Analyses-Total RNA was isolated from treated cells using RNA Stat-60 (Tel-Test B) according to the manufacturer's instructions. RNA (30 g/lane) was size-separated in agaroseformaldehyde gels and transferred to GeneScreen Plus nylon membranes (NEN Life Science Products). JNK1 and JNK2 cDNAs, isolated from pBSJNK2-1ϫHA and 3ϫHA-JNK1SR␣3, respectively, were labeled with [␣-32 P]dATP using a random primer labeling kit (Roche Molecular Biochemicals). Hybridization and washes were performed by the method of Church and Gilbert (38), and hybridization signals were visualized and quantified using a PhosphorImager (Molecular Dynamics. To monitor the quality of sample loading and transfer, membranes were hybridized to a 24-base oligonucleotide 3Ј-ACGGTATCT-GATCGTCTTCGAACC-5Ј complementary to 18 S RNA.
For Western analyses, 50 g of cell whole-cell lysates (18) were size-fractionated in 12% SDS-PAGE and transferred onto polyvinylidene difluoride membranes. Proteins were detected using an enhanced chemiluminescence system (Amersham Pharmacia Biotech) following incubation of the polyvinylidene difluoride membranes with specific antibodies (Santa Cruz Biotechnology. Flow Cytometric Analyses-Cell cycle distribution was analyzed using propidium iodide (PI)-stained cells. Briefly, 2-5 ϫ 10 6 cells were fixed in 70% ethanol, incubated with 1 g/ml RNase A, stained with 1 g/ml propidium iodide (Roche Molecular Biochemicals), and analyzed on a FACScalibur flow cytometer (Becton Dickinson). The percentages of cells in the various stages of the cell cycle were determined using the ModFitLT software program.
S-phase cells were detected using 5-bromo-2Ј-deoxyuridine (BrdUrd) Labeling and Detection Kit I (Roche Molecular Biochemicals) as described by the manufacturer with some modifications. Briefly, the cells were treated with oligonucleotides and 24 h later pulsed for 1 h with 10 M of BrdUrd labeling reagent to allow incorporation of BrdUrd into DNA in place of thymidine. The cells were fixed for 24 h with 70% ethanol diluted in glycine buffer (50 mM glycine, 0.01% Triton X-100, 0.15 M NaCl, pH 2.0) and incubated with anti-BrdUrd monoclonal antibodies. After incubation with anti-mouse Ig-fluorescein, bound anti-BrdUrd monoclonal antibodies were visualized by fluorescence-activated cell sorting analysis.
The number of apoptotic cells following treatment with JNKAS was determined using an APO-BRDU TM kit (Pharmingen) following the manufacturer's protocol. Briefly, 1-2 ϫ 10 6 cells were fixed in 1% methanol-free formaldehyde following incubation in 70% ethanol at Ϫ20°C. To detect genomic DNA degradation, bromodeoxyuridine triphosphate was incorporated into DNA breaks by terminal deoxynucleotidyl transferase enzyme. Finally, cells were stained with fluorescein-labeled anti-BrdUrd antibody to detect DNA breaks and with PI/RNase A solution for counterstaining of the total DNA. Dual parameter display was used to assess the number of apoptotic cells following analysis on FACScalibur flow cytometer.
In Vitro Kinase Activity Assays-To detect JNK activation following UV-C exposure, JNKAS-treated and control cultures were irradiated with 40 J/m 2 UV-C 24 h postlipofection, washed with cold phosphatebuffered saline 30 min later, and suspended in whole-cell extract buffer (18). An in vitro JNK kinase assay was performed using a fusion protein containing glutathione linked to the 1-222 fragment of human c-Jun (GST-c-Jun) as substrate, as described (6). Briefly, 50 g of cell lysate was incubated for 3 h at 4°C with 10 g of GST-c-Jun bound to glutathione-Sepaharose-4B (Amersham Pharmacia Biotech). After washing three times in whole-cell extract buffer and once in kinase reaction buffer (20 mM HEPES, pH 7.7, 20 mM MgCl 2 , 20 mM ␤-glycerophosphate, 20 mM p-nitrophenyl phosphate, 0.1 mM sodium vanadate, 2 mM dithiothreitol, the beads were incubated with 30 l of kinase reaction buffer containing 20 M ATP and 5 Ci of [␣-32 P]ATP for 30 min at 30°C. The reaction was stopped by the addition of Laemmli sample buffer. Samples were boiled for 5 min and resolved by electrophoresis through 12% polyacrylamide-SDS gels (SDS-PAGE).
To detect Cdk2 and Cdc2 kinase activities following JNKAS treatment, cells were harvested in Cdk2/Cdc2 lysis buffer (50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 5 mM NaF, 0.1% Triton X-100, 1 mM dithiothreitol, 0.1 mM sodium orthovanadate) supplemented with protease inhibitors, and cellular debris were removed from soluble extracts by centrifugation at 16,000 ϫ g for 10 min at 4°C. Cdk2 and Cdc2 were immunoprecipitated from 100-g aliquots after a 16-h incubation at 4°C in the presence of either anti-Cdk2 (Pharmingen), or anti-cyclin B1 antibodies (Santa Cruz Biotechnology), respectively. Immunoprecipitates were washed with Cdk2/Cdc2 lysis buffer and with H1 kinase buffer (50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , and 1 mM dithiothreitol). Kinase activities associated with anti-Cdk2 and anti-cyclin B1 immunoprecipitates were assayed in 50 l of H1 kinase buffer containing 10 g of histone H1 (Ambion) supplemented with 2 mM EGTA and 10 Ci of [␥-32 P]ATP. Reactions were carried out for 30 min at 37°C and stopped by the addition of Laemmli sample buffer. Reaction products were electrophoresed in 15% SDS-polyacrylamide gels. Results were visualized with a PhosphorImager (Molecular Dynamics).

Specific Inhibition of JNK1 and JNK2 Expression by JN-KAS-
To examine a role for the JNK pathway in mediating basal growth of T98G cells and to test whether there is a preferential role for JNK1 or JNK2, we employed specific JNK1 and JNK2 antisense oligonucleotides (18 -20). Preliminary experiments using a control fluorescein isothiocyanate-labeled oligonucleotide (c-raf ISIS13153 ) revealed that the efficiency of uptake of the phosphorothioate oligonucleotide, delivered as described under "Experimental Procedures," was nearly 100% (Fig. 1). The specificity of the elimination of JNK1 and JNK2 mRNAs by JNKAS in T98G cells is shown in Fig. 2. While neither mock treatment nor treatment with control "scrambled" (JNKScr) oligonucleotides had any effect on JNK mRNA levels in T98G cells, treatment with JNK1AS and JNK2AS resulted in Ͼ95% elimination of the respective mRNAs 24 h postlipofection ( Fig. 2A). Although the JNK1 and JNK2 mRNA levels have largely recovered by 72 h, an effective "window of observation" was provided by the antisense treatment as indicated in Fig. 2B (shaded area). The efficiency of inhibition of JNK expression was confirmed by Western analysis (Fig. 3A). Major JNK isoforms migrate with apparent molecular masses of 46 and 54 kDa. The slower migrating proteins (p54 JNK ) are largely, but not exclusively, composed of the JNK2 isoforms, whereas the faster migrating proteins (p46 JNK ) are largely, but not exclusively, composed of JNK1 isoforms (1). Treatment with JNK1AS and JNK2AS markedly attenuated the 46-and 54-kDa components, respectively (Fig. 3A), consistent with the known distribution of JNK isoforms. Time course studies of 35 S-metabolically labeled T98G cells indicated that the halflives for degradation of JNK1 and JNK2 proteins are 14 and 3.4 h, respectively (Fig. 3B). This is consistent with the marked reduction in steady-state levels of both JNK proteins (a reflection of both reduced synthesis and degradation of existing proteins) between 30 and 50 h post-treatment (data not shown). Determination of kinase activity in cells irradiated with UV-C (40 J/m 2 ) 24 h postlipofection revealed a markedly lower level of JNK activation in JNKAS-treated cells relative to control cells (Fig. 3C).
Growth Inhibition of T98G Cells Treated with JNKAS-To determine whether inhibition of JNK has an effect on the growth of T98G cells, we assessed the viability of T98G cells over a 24 -120-h time period following JNKAS treatment. Viable cell mass was measured both using an MTS-based colorimetric assay and by direct cell counts. The viability of mockand JNKScr-treated cultures was very similar to that of untreated cells over the entire time course (Fig. 4, A and B). In contrast, treatment with either JNK1AS or JNK2AS led to a marked reduction in viable cell mass. Growth suppression following treatment with JNK2AS was more pronounced than seen with JNK1AS treatment (ϳ40% versus 65% that of controls at 24 h postlipofection, respectively), especially at later time points (ϳ15% versus 75%, respectively, at 120 h postlipofection), although statistically significant differences in growth inhibition by both JNKAS were observed as early as 24 h postlipofection. As determined by direct counting from 24 -72 h postlipofection, the number of viable cells in attached JNK1AStreated cultures was significantly diminished relative to that of control cultures (Fig. 4C). Consistent with the transient loss in JNK expression, JNK1AS-treated cells resumed growth at later times with cell numbers recovering to 70% of that seen in control cultures by day 5 (Fig. 4B). JNK2AS-treated cells exhibited a greater loss in cell mass than their JNK1AS-treated counterparts at all times, and, unlike that seen with JNK1AS, their growth did not recover with time. Five days post-lipofection, the viable cell mass of the JNK2AS-treated population was 15% of that seen for control groups (Fig. 4B). In summary, although the magnitude of the growth inhibition achieved by JNKAS differed somewhat depending on the method used to assess growth (MTS assay versus cell counting), the general observation of a growth inhibition in response to JNKAS treatment was apparent regardless of the method employed. The Growth Inhibition in JNK-depleted T98G Cultures Is Associated with Cell Cycle Arrest and Inhibition of DNA Synthesis-At least two possible cellular responses could account for the reduction in viable cell mass seen following treatment with JNKAS: decreased DNA synthesis and/or selective death of JNKAS-treated cells. First, we examined the treated cells for evidence of apoptosis. JNKAS-treated cultures, particularly JNK2AS-treated cultures, exhibited some morphological alterations frequently seen in cells undergoing apoptosis including fewer mitoses, rounded up cells (Fig. 6), and detachment of many cells (up to 30%) from the plate surface (Fig. 4C). Detached cells were examined for viability. Up to 90% of detached cells in JNKAS-treated cultures excluded trypan blue and were therefore living cells, although they did not reattach or grow when placed in fresh medium in new dishes (data not shown).
Four additional methods were used to assess apoptosis in JNKAS-treated cells including detection of apoptotic bodies in DAPI-stained cells and various types of flow cytometric analy- sis: i.e. cell cycle distribution of PI-stained cells, detection of BrdUrd incorporation into cellular DNA breaks (APO-BR-DU TM ), and detection of annexin V-stained populations. Detection of DNA fragmentation by DAPI staining revealed no evidence of apoptotic cells in any treatment groups through 72 h postlipofection (Figs. 7 and 8, data not shown). Interestingly, the structure of nuclei in JNKAS-treated cells differed from that of either mock-treated cultures or cells treated with JNK-Scr oligonucleotides (Fig. 7A). The cause of these alterations is not known, but they may reflect the presence of cells with doubled DNA content that are unable to finish progression through the S phase and mitosis following JNKAS treatment.
No cells with condensed or fragmented nuclei indicative of apoptosis were evident in any treatment group. More sensitive methods for assessing apoptosis based on detection of BrdUrd incorporation into DNA breaks (APO-BRDU) (Fig. 7B) and detection of specific membrane alterations using annexin V (data not shown) also showed no evidence of apoptosis in any of the treatment groups up to 48 h. Although a small fraction of JNK2AS-treated cells (2.9%) did appear to undergo apoptosis at 48 -72 h (data not shown), this percentage of apoptotic cells is similar to that seen when assaying the negative control provided with the kit (3.3%). Thus, while JNK2-depletion might result in a low level of apoptosis in T98G cells, our results indicate that apoptosis cannot account for the magnitude of the reduction in viable cell mass seen in JNK2AStreated cultures.
While no evidence for apoptosis of JNKAS-treated cultures was found, JNKAS treatment resulted in profound alterations in the distribution of cells throughout the cell cycle, indicative of cell cycle arrest (Fig. 8). The percentages of cells in G 1 , S, and G 2 /M phases for untreated and control cultures subjected to mock-lipofection or treatment with control oligonucleotides (JNKScr) were similar: G 1 ϭ 63 Ϯ 7%; S ϭ 23 Ϯ 6%; G 2 /M ϭ 15 Ϯ 6% (average data from three independent experiments; Fig. 8, A and C). In contrast, lipofection with either JNK1AS or JNK2AS resulted in a 2-fold reduction in the number of cells in the G 1 compartment and an increase in the number of cells in other cell cycle compartments. Interestingly, while JNK1AStreated cells accumulated in both S and G 2 /M phases (35 Ϯ 5.5%; 31 Ϯ 4%, respectively), JNK2AS-treated cultures displayed an accumulation of cells exclusively in S phase (50 Ϯ 2%). Additional analysis of distribution of cells through the cell cycle was done using a BrdUrd pulse incorporation assay, which allows detection of S-phase cells by the presence of BrdUrd in their cellular DNA (Fig. 8B). For control populations (mock-and JNKScr-treated cells) and for JNK1AS-treated cultures, the percentage of S-phase cells determined with this method (23 and 35%, respectively) were similar to those obtained with PI staining (22 and 34.5%, respectively; data were analyzed using ModFitLT software). However, in the case of JNK2AS-treated cells, BrdUrd incorporation analysis revealed much lower number of S-phase cells than PI staining (28 versus 50%). This discrepancy is likely to reflect the fact that T98G cells are not actively progressing though S phase and thus are not incorporating BrdUrd due to JNK2AS inhibitory influence To further explore the nature of the growth arrest by JN-KAS, we directly examined DNA synthesis in JNKAS-treated cultures. As assessed by uptake of [ 3 H]thymidine, DNA synthesis was greatly inhibited in both JNK1AS-and JNK2AStreated populations (Fig. 9). The effect was greater for JNK2AS-treated cultures, which exhibited nearly complete inhibition of DNA synthesis by 24 h postlipofection (Fig. 9A). Interestingly, while DNA synthesis in JNK1AS-treated cells started to recover by 72 h postlipofection (consistent with the return of JNK1 expression and transient reduction in viable cell mass; Figs. 1 and 4), no such recovery was observed in JNK2AS-treated cultures (Fig. 9B). This is consistent with the lack of recovery in viable cell mass of JNK2AS-treated cultures noted above (Fig. 4). Thus, while transient depletion of JNK1 led to a transient inhibition of DNA synthesis and reduction in growth of T98G cells, similar elimination of JNK2 expression led to sustained inhibition of DNA synthesis and permanent growth arrest of T98G cells.  Ͻ 0.001). B, recovery of DNA synthesis following treatment with JNK1AS but not with JNK2AS. Relative DNA synthesis rates were determined as described above at different time points after lipofection. Values in mock-treated cultures were set to 100%. The value labeled Control oligonucleotides represents the mean of values received from two sense and two scrambled sequence oligonucleotide treatments. Oligonucleotide treatments were performed in triplicate, and duplicate samples from each well were assayed for [ 3 H]thymidine incorporation. duction of p21 cip1/waf1 Expression and Inhibition of Cyclin-dependent Kinase Activities-The cyclin-dependent kinase inhibitor p21 cip1/waf1 is believed to play an important role in mediating cell cycle arrest in response to a variety of treatments. Therefore, we investigated its expression in T98G cells following treatment with JNK1AS and JNK2AS. Marked elevation of both p21 cip1/waf1 mRNA (Fig. 10A) and protein (Fig.  10B) was detected in JNK2AS-treated cultures. Interestingly, this effect was not shared by JNK1AS-treated cells, suggesting that induction of p21 cip1/waf1 expression may contribute to the specific features of growth arrest observed in JNK2-depleted cultures such as permanent inhibition of DNA synthesis and profound sustained S-phase arrest. No changes in the expression of other cell cycle regulatory proteins were observed following any treatment, except for a slight elevation of p27 kip1 protein levels in JNK2AS-treated cultures (Fig. 10B). As expected, elevation of p21 cip1/waf1 levels in JNK2-depleted cells was correlated with a marked inhibition of cyclin-dependent kinase (Cdk2 and Cdc2) activities in JNK2AS-treated cells (Fig. 10C). DISCUSSION The role of the JNK pathway in mediating cellular responses to extracellular stimuli has been studied extensively over the past several years, and the proapoptotic function of activated JNK is supported by numerous studies (1)(2)(3)39). However, evidence supportive of a prosurvival role for the JNK pathway has also been documented (21,30,40). One plausible explanation for these seemingly disparate effects is that JNK serves different functions under normal growth conditions and during stress. Support for this view has come from findings that, in contrast to activated (phosphorylated) JNK, which activates and stabilizes its substrates during stress, nonactivated (nonphosphorylated) JNK targets them to degradation (reviewed in Ref. 4). Regulation of normal cell growth may also require basal JNK kinase activity. Indeed, N-terminal phosphorylation of c-Jun has been implicated in the regulation of cell growth both in vitro and in vivo (41).
Using high affinity and high specificity phosphorothioate antisense oligonucleotides targeting JNK1 and JNK2 (18,19), we have demonstrated that inhibition of JNK1 and JNK2 expression in otherwise unstressed human T98G glioblastoma cells results in marked suppression of growth. The inhibitory effect seen with JNK2AS was much greater than that observed with JNK1AS (Figs. 4 and 5), possibly reflecting specific functions of JNK2 not shared by JNK1. However, physical properties of JNK proteins such as their half-lives could also influence the outcome. The half-life of JNK1 is considerably longer than that of JNK2 (Fig. 3B), and significant elimination of JNK1 protein (Ͻ20% of steady-state level) is not achieved until approximately 40 h after JNKAS treatment. Thus, we cannot completely exclude the possibility that a more prolonged depletion of JNK1 in T98G cells would lead to effects comparable with those observed here with depletion of JNK2. In any event, examination of several other human tumor lines has revealed a similarly critical role of JNK2 for cell growth. For instance, JNK2 is required for epidermal growth factor-stimulated growth of A549 cells in soft agar (18,19), and JNK2AS, but not JNK1AS, treatment of mice bearing established xenografts of human PC3 cells was found to inhibit tumor growth (35).
Our findings reported here with T98G cells differ from our recent observations in several other tumor cell models in which we found that cells deficient of p53 function undergo apoptosis in response to JNK2AS treatment (36). The survival of cells with wild type p53 status was associated with elevated levels of p21 cip1/waf1 expression, which was not seen in p53-deficient cells. The HCT116 cells lacking p21 cip1/waf1 (51) underwent extensive apoptosis in response to JNK2AS treatment (36). These findings support the hypothesis that p21 cip1/waf1 plays an important role in p53-mediated protection of tumor cells from JNK2AS effects. Although exhibiting marked growth arrest, T98G did not undergo apoptosis following JNK2AS treatment (Figs. 7 and 8) despite the absence of functional p53 in these cells. Moreover, only JNK2AS-treated T98G cells were found to express elevated levels of p21 cip1/waf1 (Fig. 10), suggesting that induction of p21 cip1/waf1 may be executed by a p53-independent mechanism resulting in cell cycle arrest and protection of T98G cells from apoptosis. While the mechanism(s) contributing to the induction of p21 cip1/waf1 and downstream effects remains to be determined, it is clear that these effects are p53-independent.
Although we found that about 10% of JNK2AS-treated T98G cells detached from the tissue culture plate 24h postlipofection (Fig. 4C), no apoptosis was detected in the detached populations. The vast majority of these cells were found to be viable based on the criterion of trypan blue dye exclusion (data not shown). The percentage of detached cells further increased FIG. 10. Induction of p21 cip1/waf1 expression in JNK2AS-treated cells and associated inhibition of cyclindependent kinase activities. A, Northern analysis of p21 cip1/waf1 gene expression 24 h post-treatment with 0.4 M of either JNK1AS, JNK2AS, JNK1Scr, or JNK2Scr oligonucleotides. RNA was prepared using Stat-60 reagent as described under "Experimental Procedures." Hybridization with a probe recognizing 18 S RNA was used to assess the quality of loading and transfer among samples. B, Western blot analysis for expression of cell cycle regulatory proteins in JNK1ASand JNK2AS-treated cultures. Cells were harvested 24 h after lipofection, and 100 g of whole-cell extracts were analyzed. C, inhibition of Cdk2/Cdc2 kinase activities in JNK1AS-and JNK2AS-treated cultures. In vitro kinase assays were performed as described under "Experimental Procedures" after immunoprecipitating complexes containing Cdk2 and Cdc2 and using Histone H1 as a substrate. The sample labeled ϪH1 depicts a control reaction without substrate. with time (30% by 72 h postlipofection), suggesting a possible role for JNK2 in regulating cell adhesion. Indeed, using the same JNKAS, others have reported JNK2AS-specific inhibition of tumor nectosis factor-␣-mediated induction in the expression of E-selectin (17), a protein involved in cell adhesion. The role of the JNK pathway in anoikis, apoptotic cell death resulting from detachment, has been postulated but remains controversial (42,43). Although we cannot exclude the possibility that the majority of JNK2AS-treated cells may eventually die by apoptosis subsequent to their detachment from the extracellular matrix, this effect was not detected at the times examined in this study.
We have shown that JNK depletion in T98G cells results in a dramatic inhibition of DNA synthesis (Fig. 9) associated with accumulation of cells in the S and G 2 phases of the cell cycle (Fig. 8). In previous studies, we observed that inhibition of c-Jun N-terminal phosphorylation in T98G cells results in impaired DNA repair (40), suggesting that an intact JNK pathway is required for normal DNA repair. Together with our current findings, these observations indicate a possible role for JNK in DNA repair during progression of cells through late S phase. Other components of the DNA replication and/or cell cycle machinery could also be downstream targets of JNK. The Cdk inhibitor p21 cip1/waf1 , which was specifically elevated only in JNK2AS-treated cells, is an attractive target of JNK2 signaling. However, it is likely to be an indirect target, since we have noted that p21 cip1/waf1 expression remains elevated up to 72 h after JNKAS treatment associated with sustained growth arrest, although JNK expression has nearly returned to pretreatment levels (data not shown; Fig. 2). It is worth noting that a p21 cip1/waf1 expression has been linked to S-phase arrest in several other model systems including hyperoxia (44) and treatment of cells with the retinoid CD437 (45).
The phenotype of JNK2AS-treated cells shares certain characteristics with the "permanent" cycle arrest observed following ␥-irradiation of human fibroblasts (46,47). Like JNK2AStreated cells, these fibroblasts express elevated levels of p21 cip1/waf1 and are unable to synthesize DNA or divide for extended time, but they remain viable by the criterion of trypan blue exclusion. Induction of p21 cip1/waf1 plays an important role in growth arrest of p53-deficient cells such as T98G (48), as was previously observed in human astrocytoma cells (49). Whether p21 cip1/waf1 induction is required for the growth-inhibitory effects seen with JNK2AS remains to be proven. We have found that transient overexpression of p21 cip1/waf1 in T98G cells in the absence of JNKAS treatment results in G 1phase arrest rather than S-phase arrest (data not shown). This observation strongly suggests that the growth-inhibitory effects of JNK2AS treatment are not limited to induction of p21 cip1/waf1 but must act in collaboration with other JNK2ASregulated factors to cause S-phase arrest. Further studies are under way to identify these components.
Unlike T98G cells and several other human cancer cell lines we have examined, JNKAS treatment is not growth-inhibitory for normal prostate epithelial cells (data not shown). Likewise, although double knockouts of JNK1 and JNK2 are not viable, cells derived from either JNK1 or JNK2 knockout mice grow normally (50,51). These observations suggest that JNK1 and JNK2 possess redundant functions (and can therefore compensate for each other) and that neither is crucial for the growth of nontransformed cells. Thus, the requirement of JNK2 for growth may be accentuated in tumor cells. The recent development of procedures to generate somatic gene knockouts in human tumor cells (52,53) provides the tools to investigate the effects of abrogating JNK expression in human cancer cells. While further investigations are necessary to better under-stand the role of JNK in regulating tumor cell growth and to identify the specific JNK targets involved, our findings highlight the importance of JNK for maintaining tumor cell homeostasis in the absence of overt stress and suggest that strategies aimed at eliminating JNK could have a therapeutic benefit.