Activation of c-Jun-NH2-kinase by UV irradiation is dependent on p21ras.

We have demonstrated previously that Jun-NH2-kinase (JNK) activation in vitro is potentiated by association with the p21(ras) protein. To determine if in vivo activation of JNK also depends on p21(ras), we have used M1311 cells that carry the cDNA for the neutralizing antibody to p21(ras), Y13-259, under a dexamethasone-inducible promoter. The ability of UV to activate JNK gradually decreased over a 4-day period of cell growth in dexamethasone. This decrease coincides with weaker transcriptional activation measured via gel shift and chloramphenicol acetyltransferase assays. Peptides corresponding to amino acids 96-110 on p21(ras), which were shown to block Ras-JNK association, inhibited UV-mediated JNK activation in mouse fibroblast 3T3-4A cells as well as in M1311 cells, further supporting the role of p21(ras) in UV-mediated JNK activation. Overall, the present studies provide in vivo confirmation of the role p21(ras) plays in JNK activation by UV irradiation.

Jun-NH 2 -kinase (JNK) 1 represents a family of stress-activated protein kinases that phosphorylate serine/threonine in the NH 2 -terminal domain of transcription factors c-Jun, ATF2, ELK1, and p53 (1)(2)(3)(4). JNK activation has been shown to occur in response to various types of external stress such as UV, x-rays, heat shock, and inflammatory cytokines (1,5,6). Alternate cellular pathways are involved in JNK activation, as demonstrated for heat shock and UV irradiation (7). The ability of UV irradiation to activate signal transduction components requires cell surface receptors, as shown for epidermal growth factor receptor and insulin receptor (8), followed by the activation of Src-related tyrosine kinases (9). Several G proteins, including growth factor receptor binding protein 2, SOS Ras and Rac, also play an important role in transmitting the proper signal to protein kinases, such as MEKK and JNKK, which, in turn, activate JNK (10,11). A focal point in the activation of diverse signal transduction pathways is p21 ras , which appears to serve as a docking site for the binding of Raf-1 (12) JNK and its substrate c-Jun (13). Although the Raf-1-mitogen-activated protein kinase pathway results in the activation of transcription factors other than those activated by the MEKK and JNKK pathway (i.e. c-Fos and c-Jun, respectively), certain cross-talk between the two pathways appears to exist (14).
We have recently found that the p21 ras protein stimulates phosphorylation of JNK and enhances JNK-catalyzed phosphorylation of c-Jun (13). We obtained additional evidence that p21 ras interacts directly with both c-Jun and JNK proteins. This interaction is inhibited by specific peptides, corresponding to the effector domains of p21 ras (residues 35-47, 96 -110, and 115-126) identified in molecular modeling studies (13,(15)(16)(17). All of these peptides block oncogenic p21-induced oocyte maturation (18).
Since p21 ras appears to be involved in the JNK-Jun signaling pathway, we have undertaken a study to determine whether in vivo activation of these two proteins is p21 ras -dependent. For this purpose, we have used the cell line M1311, NIH-3T3 cells that carry the cDNA for the neutralizing antibodies to p21 ras , Y13-259. Although the heavy chain of these antibodies is constitutively expressed in M1311 cells, the light chain cDNA is under a dexamethasone-inducible promoter; thus, upon exposure to dexamethasone, there is expression of functional Y13-259 antibodies that are capable of reverting the transformed phenotype in these cells (19). Using the M1311 cell system as a model, we demonstrate the contribution of p21 ras to UV-mediated JNK activation.

EXPERIMENTAL PROCEDURES
Cell Lines-M1311 cells are NIH-3T3 derivatives that were transfected with cDNA for the heavy and light chains of Y13-259 antibodies against the p21 ras protein (19). While the heavy chain cDNA is constitutively expressed, the light chain is under an inducible promoter, which is responsive to dexamethasone. The DL6373 cells contain, in addition to the cDNA of Y13-259, a cDNA for Ha-ras that has been deleted in the binding site of the Y13-259 antibodies (amino acids 63-73; Ref. 19). Both cultures were maintained in DMEM supplemented with 10% charcoal-deactivated fetal calf serum (Colorado Corp., Denver, CO), 10 mg/ml insulin, and 10 units/ml each of penicillin and streptomycin. Cells (1 ϫ 10 5 ) were plated in 150-mm plates and were incubated either in the presence or absence of 100 M dexamethasone, as indicated under "Results." NIH-3T3-4A cells are mouse fibroblasts that carry one copy of ts polyoma virus 3T3-4A (20). These cells were kindly provided by Dr. C. Basilico and maintained at 37°C in an incubator in DMEM supplemented with 10% calf serum.
UV Irradiation-Cells were exposed to UV irradiation as indicated previously (7). Briefly, prior to irradiation, the cells were washed with phosphate-buffered saline and, with the lids off, placed in marked areas in the tissue culture hood, which had been precalibrated for the required dose of UV using the germicidal lamp (254 nm) with the aid of a UV-C probe (UVP, San Diego, CA). The media that were removed prior to irradiation were added again after UV exposure, and the cells were harvested at the indicated time points.
Oligonucleotide and Peptide Synthesis-Oligonucleotides representing the dimer of the UV response element (URE) (ACTATGACAA-CAGCTATGACAACAGT) or the AP1 (CTGACTCATCCGTGACTA-ACT) target sequences were synthesized in-house with the aid of a Cyclone Plus DNA synthesizer (Milligen Biosearch, Bedford, MA). Complementary DNA strands were purified and annealed by standard procedures. Peptides synthesized in this study were purified to Ͼ99% as verified by high-performance liquid chromatography analysis, as described previously (18).

SDS-Polyacrylamide Gel Electrophoresis and Western
Blotting-Whole-cell extracts (100 g) were incubated at 100°C for 15 min in the presence of Laemmli sample buffer prior to separation on 15% SDSpolyacrylamide gel electrophoresis. The gels were transferred to a polyvinylidene difluoride membrane (Millipore) in Tris-glycine buffer with 90 volts for 1 h at 4°C using an electroblotter (Bio-Rad). Membranes were then blocked with Triton X-100 (0.5%) and nonfat milk (5%) and blotted with antibodies either to p21 (Y13-259; Oncogene Science; 1:3000 dilution), to JNK (PharMingen; 1:3000 dilution), or to c-Jun (Santa Cruz Biotechnology; 1:2000 dilution). Binding of each antibody was detected by a chemiluminescence detection kit (ECL) according to the manufacturer's recommendations (Amersham Corp.). The secondary antibody provided with the kit was diluted 1:15,000. In all cases, protein concentration was carefully evaluated prior to loading and by Ponceau staining after blotting.
Electrophoretic Mobility Shift Assay (EMSA)-The synthetic URE or AP1 target sequence was labeled with [␥-32 P]dATP (3000 Ci/mol; DuPont NEN) using polynucleotide kinase (Promega Corp., Madison, WI) according to standard procedures. The labeled DNA (0.4 g, 4400 cpm) was incubated with nuclear proteins (0.8 g, as specified under "Results") for 20 min at room temperature in the presence of 100 g of poly(dI⅐dC) oligomer (Boehringer Mannheim) and DNA binding buffer, as described previously (7). The complexes were then separated on 7.5% polyacrylamide gel and autoradiographed. Each of these experiments was reproduced three times, using independently prepared protein extracts.
Electroporations and CAT Assays-Thirty g of the mammalian expression vector, as indicated under "Results," were cotransfected with 3 g of ␤-galactosidase construct by electroporation into 5 ϫ 10 6 cells using 270 volts, 1050 microfarads, 25 mM HEPES, pH 7.1, 0.25 mM NaHPO 4 , and 140 mM NaCl. Correction for transfection efficiency was based on normalization of values obtained for the pURECAT relative to the values obtained for ␤-galactosidase activity. Levels of CAT activity were determined through standard thin-layer chromatography, and the percentage of conversion of chloramphenicol to the acetylated form was quantitated with the aid of a radioimaging blot analyzer (AMBIS, San Diego, CA).
Protein Kinase Assays-To assay JNK activity, the fusion protein glutathione S-transferase-Jun beads were used as the substrate for JNK as described previously (7,13). Briefly, aliquots of 50 g were incubated with [␥-32 P]ATP (50 cpm/fmol; DuPont NEN) for 15 min. Following extensive washing, the phosphorylated glutathione S-transferase-Jun-bound beads were boiled in SDS sample buffer, and the eluted proteins were separated on a 15% SDS-polyacrylamide gel. The gel was dried, and phosphorylation of the Jun substrate was determined by quantitative autoradiography using a computerized radioimaging blot analyzer.

Interaction between Ras and JNK and c-Jun in Cellular
Extracts-To demonstrate direct interaction of p21 ras and JNK, we have immunoprecipitated whole-cell extracts from 3T3-4A cells with the anti-Ras antibody, Y13-259. The immunoprecipitated protein was subjected to Western blotting with either anti-JNK (Fig. 1, lane 3) or anti-Jun antibody (Fig. 1, lane 6), revealing that both JNK and Jun proteins coprecipitated with p21 ras . Immunoprecipitation of cellular proteins with antibod-ies to p21 ras , followed by immunoblotting using antibodies to JNK, revealed a noticeable increase in p21 ras -bound JNK after UV irradiation ( Fig. 2A). Control reactions in which the same membranes were incubated with antibodies to p21 ras revealed equal amounts of p21 ras in the precipitable complex before and after UV irradiation (Fig. 2B). Further support for this interaction comes from analysis of JNK activities in protein extracts that were immunodepleted with antibodies (Y13-259) to p21 ras or to JNK. As shown in Fig. 3, the supernatant that was depleted of Ras and Ras-associated proteins contain a noticeable decrease in JNK activity as measured by solid-phase kinase reactions using pGEX-Jun as a substrate. Such a decrease was seen using two other antibodies to p21 ras (Y13-238 and F235-1.7.1; data not shown). The degree of decreased JNK activity was higher when antibodies to JNK were used for immunodepletion, indicating that about 50% of JNK is bound to p21 ras (Fig. 3B). Nonrelevant antibodies used as a control had no effect (Fig. 3). These results demonstrate that a significant fraction of the UV-activated JNK is physically associated with p21 ras and thus corroborate our previous findings in which both proteins were found to bind to beads containing p21 ras (13).
Peptide Inhibition of Ras-JNK Interaction-To test the specificity of Ras-JNK association, we have tested peptides corresponding to different regions within the p21 ras protein for their ability to affect this interaction. Such peptides were selected based on the identification of potential effector domains from molecular mechanics and dynamics studies (15)(16)(17). Fig. 4 demonstrates binding of p21 ras to JNK protein, when both components are purified from bacterially expressed cDNAs. Such binding can be further potentiated if Ras is charged with GTP or sodium vanadate (21), possibly due to its altered conformation. Ras binding to pGEX-JNK was strongly inhibited by the peptide that contains amino acids 96 -110 of the p21 ras protein. This peptide was previously shown potent in inhibiting the interaction between p21 ras and c-Jun (13). In addition, we find that a peptide corresponding to amino acids 115-126 in p21 ras is also capable of mediating strong inhibition of JNK-Ras interaction (Fig. 4). The 115-126 peptide was not efficient in inhibiting Ras-Jun interactions (13). The association between JNK and p21 ras was also inhibited by c-Jun as well as by an excess of free JNK protein, which competed for the binding of soluble p21 ras to pGEX-JNK (Fig. 4). Reactions performed in the presence of kinase buffer without any peptide or with nonrelevant peptides did not have any effect on JNK-Ras association (data not shown).
UV Irradiation-mediated JNK Activation Requires p21 ras -Based on the observations that p21 ras interacts with JNK both in vitro and in cellular protein extracts, it was of interest to elucidate whether this interaction is in fact necessary for JNK activation. Some support for the role of this interaction in JNK activity came from our previous studies in which we have shown that the presence of p21 ras is sufficient to potentiate JNK activation in vitro (13). To this end, we have used M1311 cells that were maintained under normal growth conditions with or without dexamethasone. In the presence of dexamethasone, the expression of Y13-259 light chain cDNA is induced to enable the formation of functional antibodies that neutralize p21 ras (19). UV irradiation was chosen as an external source of stress to induce JNK since it has been well studied with respect to cellular requirements to achieve proper activation of the kinase, including membrane components and nuclear DNA lesions (22). Exposure of M1311 cells to UV irradiation led to a noticeable increase in JNK activation (Fig. 5A), which resembled the pattern seen with other cell types (7,13,22). However, JNK activity progressively decreased as a function of the time of exposure of the cells to dexamethasone over a 5-day period (Fig. 5A). After 5 days of exposure, UV irradiation caused a very weak activation of JNK. Unlike its effect on UV-mediated JNK activation, neutralizing p21 ras by dexamethasone-induced expression of anti-Ras antibodies did not impair the ability of heat shock to activate JNK (data not shown). Heat shock is an alternate form of stress shown to use cellular pathways other than UV, which are not dependent on membrane integrity and thus considered Ras-independent (7). As a control for the possible effect of dexamethasone itself, we have also maintained 3T3-4A cells, which are well characterized for JNK inducibility, in dexamethasone. As shown in Fig. 5B, dexamethasone did not affect the ability of UV to induce JNK activation, even after 5 days of exposure. As an additional control, we examined a variant of M1311 cells, DL6373 cells (19), that express a mutant p21 protein which lacks the Y13-259 binding domain (residues 63-73; Ref. 23) and is, therefore, not inactivated by expression of this antibody (19). When DL6373 cells were subjected to UV irradiation, no changes in the degree of UVmediated JNK activation were noticed over the 5-day period (Fig. 5B). These results further confirm that the observed changes in M1311 cells are due to p21 ras inactivation and indicate a requirement for p21 ras in UV-mediated activation of JNK.
Transcriptional Activities of JNK Substrates in M1311 Cells-To measure transcriptional activities of the JNK substrates Jun and ATF2 in M1311 cells, we used the URE (TGA-CAACA), which has been shown to serve as a target for binding of c-Jun and ATF2 (24) as well as the AP1 target sequence. Previous studies demonstrate correlation between c-Jun phosphorylation by JNK and its DNA binding activities in gel shifts (25), as well as transcriptional activities in CAT assays using c-Jun promoter/target sequences (26). Gel shift assays, with proteins prepared 30 min after UV treatment, demonstrated

FIG. 2. Western blot of Ras-bound JNK in protein extracts.
Proteins prepared before (Ϫ) and 30 min after (ϩ) UV irradiation were immunoprecipitated with antibodies to p21 ras followed by Western blot analysis with antibodies to JNK (A) or with antibodies to p21 ras (B). As a positive control, the enriched fraction of JNK was loaded (P), whereas a negative control consists of protein extracts depleted of JNK (N).

FIG. 3. JNK activity in Ras-immunodepleted extracts.
Protein extracts prepared before (C) or after UV irradiation (UV) were analyzed for JNK activity using pGEX-Jun as a substrate in a solid-phase kinase reaction. Indicated antibodies (numbers in superscript indicate regions to which they were developed; poly reflects polyclonal antibodies) were used to immunodeplete the protein extracts, in which case the supernatant was used for the kinase reactions. B, quantification of the kinase reaction shown in A via radioimaging. Column 1, UV; column 2, antibody to JNK 133-350 ; column 3, antibody Y13-259 to Ras; column 4, antibody to JNK 186 -615 ; column 5, polyclonal antibody to JNK; column 6, mouse IgG; column 7, rabbit IgG ; column 8, antibody to Rb. Error bars were calculated based on three independent experiments. FIG. 4. Western blot of ras-p21 from incubation of ras-p21 with pGEX-JNK beads in the presence of different competitors as indicated in the figure. pGEX-Jun 5-89 represents the NH 2 domain, whereas c-Jun is the full-length protein. pGEX2T is the parent pGEX construct. Analysis of p21 ras bound to pGEX-JNK was performed after washing the glutathione S-transferase beads-bound complex by Western blotting using antibodies to p21 ras . The first lane depicts pGEX-JNK-Ras association using the bacterially expressed and subsequently purified proteins p21 ras and pGEX-JNK. The amount of p21 ras bound to pGEX-JNK decreases if NH 2 -terminal c-Jun or full-length c-Jun were added, in soluble form, to the reaction mixture (lanes 2 and 3). Similarly, excess soluble JNK can compete with pGEX-JNK association with p21 ras (lane 5). A control construct, pGEX-2T, added to the pGEX-JNK-p21 ras binding reaction did not inhibit JNK-Ras association (lane 4). Competition with various peptides representing different regions on p21 ras enabled us to identify domains of p21 that are required for interaction with JNK. Respective peptide competitors were added to the reaction in 10 ϫ excess (4 g peptide competitor versus 0.4 g of p21 ras ). that binding to the URE is inhibited in the M1311 cells that were maintained under dexamethasone (Fig. 6A, compare lanes D (Ϫ) and C (Ϫ) under the M1311 panel). In control DL6373 cells, dexamethasone treatment led to increased binding to the URE target sequence. UV treatment, however, decreased binding to the URE, in both the DL6373 and M1311 cells, a phenomenon we have observed previously in other transformed cells and which is attributed to the induction of a UV-inducible transcriptional inhibitor (27). In all cases, the complex formed was inhibited by an excess of cold URE or AP1 target sequences (Fig. 6A).
As an independent measure for transcriptional activities, we have transfected CAT vectors driven by either URE or AP1 target sequences into M1311 and DL6373 cells. As shown in Fig. 6B, whereas URE-CAT and AP1-CAT activity measured 32 h after UV treatment was clearly induced after UV irradiation (6 -12-fold), dexamethasone did not affect these activities in the DL6373 cells, although it greatly reduced URE-CAT and AP1-CAT transcription in the M1311 cells (Fig. 6B). When tested at earlier times (i.e. 24 h), no transcriptional activity was observed, in agreement with the pattern seen in the EMSA, which was performed on proteins prepared 30 min after UV treatment (Fig. 6A).
Analysis of JNK activities in protein preparations used for EMSA confirmed that dexamethasone-maintained M1311 cells had lost the ability of UV irradiation to properly activate JNK (Fig. 6C). Changes in binding to the URE thus indicate that transcriptional activities in UV-irradiated M1311 cells correlate with alteration by impaired Ras-dependent JNK activities.
Peptide Inhibition of UV-induced JNK Activation-Since the p21 ras peptide corresponding to amino acids 96 -110 was found to inhibit the binding of p21 to JNK (Fig. 1), we performed experiments in which 3T3-4A and M1311 cells were exposed to UV light in the presence and absence of selected peptides that P-labeled URE target sequence. When competition assays were performed, excess cold sequences (ϩURE or ϩAP1, respectively) were added 10 min before incubation with the labeled URE. Shown is an autoradiograph of the reaction mixture, which was separated on 7.5% polyacrylamide gel electrophoresis. Bars, SD. B, CAT assays; CAT constructs driven by the URE sequence were transfected together with ␤-galactosidase construct into the DL6373 or M1311 cells as indicated, and cells were maintained with (ϩdex) or without the hormone. Proteins prepared 32 h after UV treatment were used for analysis of CAT (see "Experimental Procedures"). Values shown as fold increase in URE-CAT activity were calculated based on the ratio between UV-to sham-treated cells after normalization for transfection efficiencies (based on ␤-galactosidase values). Quantitation of CAT activities was performed with the aid of a radioimaging blot analyzer (AMBIS). Bars, SD. C, JNK activities. Analysis of JNK activities in the proteins used for EMSA assays (see A above) was performed by measuring the phosphorylation of pGEX-Jun 5-89 . were added to the incubation medium. To insure proper introduction of peptides into the cells, we have used the scrapeloading technique (28); this approach has been shown previously to successfully alter Ras activities (29). As shown in Fig.  7, the presence of the peptide significantly reduced the level of JNK activation in UV-treated 3T3-4A cells. For non-dexamethasone-treated M1311 cells, a significant (50%) reduction in the degree of JNK activation was also observed (Fig. 7). Dexamethasone-maintained M1311 cells (for 1 day, yielding a limited inhibition on UV-mediated activation; see Fig. 2) exhibited the most striking decrease in JNK activation of almost 90%, indicating that this peptide had a synergistic effect on the anti-Ras antibody induced by dexamethasone. Control reactions were performed by incubating these cells with different peptides, including epidermal growth factor receptor, Src, mitogen-activated protein kinase, and a mutant form of the Ras peptide corresponding to amino acids 96 -110. None of these peptides was able to elicit inhibition of UV-mediated JNK activation in M1311 cells, nor could they affect the degree of JNK activation in the cells that were maintained in dexamethasone (data not shown). DISCUSSION We previously found that p21 ras binds to purified bacterially expressed JNK and Jun proteins (13). The present studies further confirm this interaction both in vitro and in functional in vivo-based assays. These studies demonstrate that p21 rasbound JNK is increased after UV treatment, and that depletion of p21 ras from protein extracts results in a marked decrease in JNK activity. That the amount of p21 ras -bound JNK depends on cellular stress and increases significantly after UV irradiation points to a role for p21 ras in UV-mediated JNK activation. Although immunodepletion with antibodies to p21 ras resulted in about a 50% decrease in JNK activity, antibodies to JNK reduced this activity twice as efficiently, indicating that about one-half of the cellular JNK is bound to p21 ras .
To probe the physiological significance of the direct interaction between p21 ras and JNK and Jun, we have used a model cell system in which p21 ras activation is blocked by an endogenously produced anti-p21 ras neutralizing antibody. Induction of Y13-259 antibody expression in the NIH-3T3 (M1311) cell line blocked the activation of JNK since JNK-induced phosphorylation of Jun diminished over a 5-day period, corresponding to increased expression of the anti-p21 ras antibody (Fig. 2). Expression of ras cDNA, which lacks the recognition site for these antibodies, as seen in the DL6373 cells, restored the ability of UV to activate JNK, further supporting the conclusion that it is Ras inactivation that led to impaired JNK activation in M1311 cells. An independent support for the role of Ras-JNK interaction in JNK activation by UV comes from experiments in which a peptide corresponding to a Ras domain required for interaction with the kinase (amino acids 96 -110) was introduced into M1311 cells, where it inhibited JNK activation after UV irradiation (Fig. 7). When dexamethasone was added to the medium allowing expression of Y13-259 antibodies, the inhibition mediated by this peptide was found to be synergistic with that induced by the antibody itself. Further support for p21 ras -mediated JNK activation upon UV irradiation comes from the observation that JNK activation by heat shock, an alternate form of stress using different cellular pathways that do not require membrane integrity (7), was not inhibited in dexamethasone-maintained M1311 cells.
Our findings that membrane-bound p21 ras binds to JNK and is required for its activation imply that JNK (and Jun) may become activated at or near the cell membrane. Interestingly, protein extracts prepared in the presence of Triton X-114, which was shown to extract p21 ras from the membrane, led to the identification of JNK as a p21 ras -bound protein. 2 Recent studies suggest that JNK binds to growth factor receptor binding protein 2 (11), the adapter protein that links tyrosine kinase receptors to the SOS protein, which, in turn, activates p21 ras by promoting GTP/GDP exchange. Although these findings suggest that JNK may be activated in the cytosol near the cell membrane, different JNK isozymes may be activated by alternate pathways, which include nuclear DNA lesions formed after UV irradiation (21). Nuclear localization of JNK was demonstrated recently to occur in UV-treated cells (30).
Expression of the anti-p21 ras antibody Y13-259 in the NIH-3T3 (M1311) cells was also found to strongly inhibit transcriptional activation, as evidenced by weaker complexes and decreasing transcriptional activities measured by gel shift and CAT assays using the URE as a target sequence. The latter indicates that p21 ras -associated JNK may play an important role in mediating the transcriptional activation of the JNK substrates tested here, viz c-Jun and ATF2.
Because other kinases (i.e. p54 and p38) were also shown capable of phosphorylating ATF2, which forms a heterodimer with c-Jun for binding to the URE (24), we cannot exclude the possibility that the mechanism of inhibition of URE-DNA binding activity, although Ras-dependent, is not solely related to JNK. Similarly, dexamethasone inhibition of URE-CAT activities in M1311 cells may not be confined to JNK inhibition. Further elucidating JNK-related transcriptional activities in Ras-dependent and -independent pathways will allow us to clarify the complex regulation of AP1 and ATF2 activities. Changes in JNK activity are likely to also affect its substrate stability, as shown for c-Jun, which is targeted by JNK for ubiquitination in a phosphorylation-dependent manner (31). In all, the functional significance of UV-mediated Ras-dependent transcriptional activities is expected to affect cell ability to cope with the UV effect via changes at the level of DNA repair, cell cycle, and possibly apoptosis.