ATF1 phosphorylation by the ERK MAPK pathway is required for epidermal growth factor-induced c-jun expression.

Epidermal growth factor induction of c-jun expression requires ATF1 and MEF2 sites in the c-jun promoter. We find that activation of the c-jun promoter through the ATF1 site requires phosphorylation of ATF1 at serine 63. A serine 63 to alanine mutation of ATF1 acts to block epidermal growth factor (EGF) induction of a transfected c-jun gene. ATF1 can be phosphorylated by mitogen- and stress-activated protein kinase 1 (MSK1), which is activated by EGF and ERK1/2. Kinase-dead MSK1 mutants blocked EGF induction of a transfected c-jun gene suggesting that MSK1 or a similar family member is required for induced c-jun expression. Use of the MEK1 inhibitor U0126 and dominant negative MEK1 further showed that MSK1 activation and c-jun induction require the ERK pathway. In contrast, a JNK inhibitor blocked EGF induction of c-jun expression but not ATF1 phosphorylation. These results show that the two MAPK pathways, ERK and JNK, are required for EGF-induced c-jun expression and that the ERK pathway acts through downstream phosphorylation of ATF1.

The proto-oncogene c-jun plays a central role in cellular signal transduction and regulation of proliferation, differentiation, and neoplastic transformation (1). The products of the c-jun gene family (c-Jun, Jun B, and Jun D) either homodimerize or form heterodimers with members of the c-fos gene family (c-Fos, Fos B, Fra1, and Fra2) or ATF family of transcription factors (2)(3)(4). The Jun homodimers and Jun/Fos heterodimers bind specifically to AP1 sites (consensus sequence 5Ј-TGAGTCA-3Ј) also known as 12-O-tetradecanoylphorbol-13acetate-responsive elements (2). The response of c-jun to mitogens is controlled by two mechanisms. First, phosphorylation of the N-terminal region of c-Jun by Jun N-terminal kinase (JNK) 1 activates its transcriptional activation function, and second, mitogens induce transcription from the c-jun promoter (5)(6)(7)(8). Transcription of the c-jun gene increases rapidly in response to serum and growth factors such as EGF and nerve growth factor, as well as to stimuli such as UV and 12-Otetradecanoylphorbol-13-acetate (6 -11).
Reporter genes containing the c-jun promoter can reproduce stimulation by many of the agents that stimulate c-jun expression. This has allowed the mapping of sequence elements required for induction. The c-jun promoter contains binding sites for SP1, CTF, MEF2, and ATF factors. There are two AP1-like elements, at Ϫ72 and Ϫ190, which have been termed the jun1 and jun2 sites, respectively (12). We have previously found that the jun1 site binds predominantly ATF1 and lesser amounts of the closely related factor CREB in nuclear extracts of HeLa cells (13). In contrast, the jun2 site binds ATF2-c-Jun heterodimers (12,14). We mapped regions of the mouse c-jun promoter responsible for EGF induction and found that the ATF1 (jun1) site at Ϫ72 and a MEF2 site at Ϫ59 are critical for induction (13,15). Whereas point mutations show that both sites are required for EGF induction, we did not detect cooperative binding to these closely spaced sites. 2 Induction by genotoxic agents such as UV light requires both the jun2 and ATF1 sites (14).
The Ϫ59 site is bound by the MEF2 family of proteins, which include MEF2A through MEF2D. We found that MEF2D is the predominant factor binding to the c-jun MEF2 site in HeLa cell nuclear extracts with minor binding by MEF2A (16) although others have found more equal binding by MEF2A and MEF2D (17).
The intracellular pathways linking the EGF receptor to the c-jun promoter are not completely known. We previously investigated the role of a number of signaling pathways using activated and dominant negative factors. These experiments showed that EGF activates the c-jun promoter through a Ras to Rac to MEKK pathway (13). Because MEKK activates JNK, these results showed that the same pathway activates both the c-Jun protein product and transcription from the c-jun promoter. It has also been shown that serum and EGF activation of the c-jun promoter requires ERK5 signaling using MEF2 factors with mutations in their ERK5 phosphorylation site (18,19). Activation of the c-jun promoter by a G protein-coupled receptor (human muscarinic acetylcholine receptor activated by carbachol) was found to utilize multiple pathways including JNK, p38, and ERK5, because all of these pathways needed to be blocked to completely reduce c-jun luciferase induction by carbachol (20). Activated RhoA was also found to activate the c-jun promoter in NIH3T3 cells (21), although we observed only very weak activation in HeLa cells (13). RhoA was found to activate c-jun in NIH3T3 cells via is activation of p38␥ and required both the ATF1 and MEF2 sites (21). Activation of MEF2 factors by MAPK family members has been shown using GAL4-MEF2 constructs (18,19,22,23). The p38 family phosphorylates and activates MEF2A and MEF2C whereas ERK5 activates MEF2A, MEF2C, and MEF2D (18,19,22,23). This provides one mechanism for activation of the c-jun promoter although these MAPKs are not induced by all inducers of c-jun expression.
We report here on our investigation of the pathway activating the ATF1 site. ATF1 is inducibly phosphorylated on serine 63 to activate its transcriptional activity (24). This site is homologous to the serine 133 site in CREB that is phosphorylated upon activation and that is involved in binding to the coactivator CREB-binding protein (24). ATF1/CREB phosphorylation can be activated by ERK1/2 and p38 MAPK pathways (25). ERK1/2 can activate members of the related RSK and MSK kinase families whereas and p38 can also activate the MSK family (26 -29). The ribosomal S6 kinase (RSK) and MSK families can then directly phosphorylate ATF1 and CREB on their activating sites (27)(28)(29)(30). Mouse knock-outs of MSK1 and MSK2 show that these kinases are the major kinases required for EGF-induced phosphorylation of CREB and ATF1 in mouse embryo fibroblasts (31). Mutations in RSK2 were found to be associated with Coffin-Lowry syndrome (CLS) (32), and fibroblasts from CLS patients were found to be defective for EGFinduced CREB phosphorylation, suggesting that RSK2 is required for this modification in human fibroblasts (30). This difference in the requirement for MSK1/2 or RSK2 may be because of cell-type specificity, because platelet-derived growth factor and insulin-like growth factor-1-induced phosphorylation of CREB was normal in RSK2-null mouse embryo fibroblasts (26).
We have tested here whether ATF1 phosphorylation is required for EGF induction of the c-jun promoter and what the pathway is for EGF-induced ATF1 phosphorylation. We find that ATF1 is rapidly phosphorylated after EGF treatment along with MSK1 kinase activity. Inhibition of MEK1 in the ERK1/2 MAPK pathway blocked EGF-induced ATF1 phosphorylation, MSK1 kinase activity, and c-jun expression. Dominant negative mutants of ATF1 and MSK1 inhibited EGF induction of c-jun mRNA suggesting that the ERK1/2 to MSK to ATF1 pathway is required for induction of c-jun expression.

EXPERIMENTAL PROCEDURES
Materials-Human recombinant EGF was obtained from BIO-SOURCE International. The inhibitors of the MAPK, U0126, SB203580, and JNKII, were from Calbiochem. Phospho-specific antibodies to ERK, JNK, and p38 were obtained from Cell Signaling whereas the phospho-CREB/ATF1 antibodies were from Upstate Biotechnology. Anti-ATF1 sera were from Tsonwin Hai, Ohio State University (13). M2 antibodies against the FLAG tag, Protein A-agarose, the expression vector p3x FLAG-CMV, and the protease inhibitor mixture (P1860) were from Sigma.
Plasmids and Expression Vectors-Plasmid pJC6GL3 contains position Ϫ225 to ϩ150 of the murine c-jun promoter and a luciferase reporter gene (13). A Renilla luciferase gene with the SV40 promoter, pRLSV40P, was used as an internal control (33). MSK1 and its mutants at residues 195 and 565 were subcloned into the expression vector p3xFLAG-CMV using the EcoRI and XbaI sites, with the FLAG tag at the N terminus. ATF1 and ATF1-S63A, with a serine to alanine mutation at position 63, were subcloned into p3xFLAG-CMV with the tag at the C terminus. For promoter transfection and RNase protection assays, IMAGE clone 3968444 (ATCC) that contains Ϫ50 to ϩ4005 of the Mus musculus c-jun cDNA was digested with BamHI and XhoI and cloned into the BamHI and SalI sites of pJC6GL3 to make a construct, pcJUN, containing Ϫ225 to ϩ4005 nucleotides of the mouse c-jun gene. Note that the genomic c-jun gene does not contain introns (34). MSK1 and the MSK1 dominant negative mutants were from Dario Alessi, University of Dundee. The JNKK2-JNK1 fusion construct with three hemagglutinin tags was obtained from Anning Lin (35). pRaf BXB was used as an activated form of c-Raf-1 (36). MEKK1 was overexpressed in vector pCMV5 (37). Activated MEK5 (MEK5DD), dominant negative MEK5 (MEK5AA), and ERK5 had three hemagglutinin tags at the N terminus and were a gift of Jack Dixon (University of Michigan), whereas the p38 expres-sion vector had a FLAG tag at the N terminus. Activated MKK6 (MKK6EE) and dominant negative MKK3b(A) were gifts from Jiahuai Han (Scripps Research Institute). Dominant negative MEK1A was as described (38). The ERK2 and Ras valine 12 expression vectors were from C. Chandra Kumar (Schering-Plough Research Institute).
Transfections and Luciferase Assays-HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% newborn calf serum (NCS). Duplicates of 6-well plates were transfected by the calcium phosphate coprecipitation method (39)). The c-jun reporter plasmid pJC6GL3 (0.25 g), the internal control pRLSV40P (0.1 g), and the indicated amounts of expression vectors were transfected keeping the total DNA at 2.5 g per well with herring sperm DNA. Medium was changed 16 -20 h after transfection; 8 to 12 h later cells were serum-starved in DMEM supplemented with 0.2% NCS for 24 to 30 h. Luciferase assays were done using the dual luciferase reporter assay system (Promega). Preparation of cell extracts was as described by the manufacturer. All firefly luciferase values were normalized to Renilla luciferase levels.
Immunoprecipitation and Kinase Assays-HeLa cells were transfected with 3ϫ FLAG tagged MSK1 (2.5 g) alone or with the activators MEKK1, JNKK2-JNK1, Raf BXB and ERK2, MKK6EE and p38, or MEK5DD and ERK5 (2.5 g of each activator) keeping the total DNA at 10 g per 10-cm plate with herring sperm DNA. After 16 h of transfection, the medium was changed with DMEM and 10% NCS, and after another 8 h the medium was changed to DMEM with 0.2% NCS. After 60 h of transfection cells were treated with drugs U0126 (10 M), SB203580 (10 M), and/or JNKII (20 M) for 40 min followed by EGF (100 ng/ml) for 30 min. Plates were washed with ice-cold phosphatebuffered saline and lysed with 500 l of lysis buffer containing 25 mM Tris-HCl, pH 8.0, 137 mM NaCl, 2.5 mM KCl, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 2 mM dithiothreitol, and proteinase inhibitor mixture. Cells in lysates were scraped from the plates, transferred to microfuge tubes, and rotated at 4°C for 30 min. Tubes were centrifuged for 10 min at 13,0000 rpm, and the supernatant was incubated with anti-FLAG antibodies (1:200 dilution) and rotated overnight at 4°C. Immune complexes were precipitated with 20 l of 50% protein A-agarose. The immunoprecipitates were washed three times with lysis buffer and resuspended in 50 l of 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA. To measure the kinase activity of FLAG-MSK1, bacterially purified His-tagged ATF1 was used as the substrate. His-tagged ATF1 was purified using nickel-nitrilotriacetic acid metal chelate affinity chromatography as described (40). The MSK1 immunoprecipitates (10 l) were mixed with 14 l of Kinase Buffer K (25 mM HEPES, pH 7.4, 5 mM ␤-mercaptoethanol, 10 mM MgCl 2 , 90 M sodium vanadate, 10% glycerol, and 10 M ATP) along with 1 l of purified His-ATF1 (1 mg/ml). The kinase reactions were carried out at 30°C for 30 min. The reaction was stopped by adding 12 l of 3ϫ SDS sample buffer (150 mM Tris-HCl, pH 6.8, 6% SDS, 0.3% Bromphenol Blue, 30% glycerol, 3% ␤-mercaptoethanol). The samples were immunoblotted with anti-phospho-CREB antibodies (1:1000 dilution). The MSK1 immunoprecipitates (10 l) were also immunoblotted with anti-FLAG antibodies (1:1000) to control for the expression levels of transfected MSK1.
Immunoblotting-Cell lysates were prepared from HeLa cells grown on 35-mm plates by resuspending the cells in 50 l of Nonidet P-40 lysis buffer (25 mM Tris-HCl, pH 8.0, 1% Nonidet-P40). Cell lysates were clarified by centrifugation at 13,000 rpm for 5 min, 3ϫ SDS sample buffer (25 l) was added, and the samples were immunoblotted with 1:1000 dilutions of the indicated antibodies. Horseradish peroxidaseconjugated goat anti-mouse or anti-rabbit antibodies (1:5000) were used as secondary antibodies, and the signal was visualized using the ECL plus chemiluminescence kit (Amersham Biosciences).
RNase Protection Assays-HeLa cells (10-cm plates) were transfected with pcJUN containing Ϫ225 to ϩ4005 of the mouse c-jun gene along with the internal control hsp70 promoter (phsp70CAT) (15). Where indicated the cells were also transfected with WT or mutant ATF1 (0.5 to 10 g) and MSK1 genes (2.5 g). After 60 h of transfection cells were treated with the indicated inhibitors and with or without EGF as described for the kinase assays. Cells were lysed with 1 ml of Trizol reagent (Invitrogen), and total RNA was isolated as described by the manufacturer and stored in formamide at Ϫ80°C. Total RNA (10 g) was hybridized with 1 ϫ 10 5 cpm of the RNase protection probe at 48°C overnight. The RNase protection probe for transfected mouse c-jun, pJG4, was linearized with HindIII and transcribed with SP6 polymerase and [␣-32 P]CTP (15). This probe contains the mouse c-jun promoter and 250 nucleotides of the CAT gene such that it recognizes the transfected c-jun gene (150 nucleotides) and the CAT portion of the hsp70CAT internal control (250 nucleotides) (15). The hybridized samples were digested with 2 g/ml of RNase T1 and 40 g/ml of ribonuclease A at 30°C for 1 h as described (15). Digests were resolved on 6% polyacrylamide, 7 M urea gels and exposed to x-ray film. The endogenous level of human c-jun mRNA in untransfected HeLa cells was measured using the RNase protection probe pAP1SP65 that was linearized by EcoRI and transcribed with SP6 polymerase and [␣-32 P]CTP (15).

Phosphorylation of ATF1 and Activation of MSK1 in EGF-
treated Cells-We found previously that ATF1 is the predominant factor in HeLa nuclear extracts that binds to the jun1, AP1-like site in the c-jun promoter and that this site is critical for EGF induction of the c-jun promoter (13). To investigate how ATF1 might contribute to regulation of the promoter, we used phospho-specific antiserum to CREB phosphorylated at serine 133, which also recognizes ATF1 phosphorylated at serine 63. We found a strong and rapid increase in phosphorylated ATF1 after EGF treatment of HeLa cells, whereas there was no significant change in the amount of total ATF1 (Fig. 1A). The kinetics of the increase precedes the increase in c-jun mRNA consistent with a role in increasing c-jun transcription (6,15). The increase was transient, returning to basal levels by 60 min.
To identify a kinase responsible for phosphorylation of ATF1 we tested MSK1. MSK1 has been shown to phosphorylate CREB and to be activated by EGF, as well as ERK1/2 and p38 MAPKs (27). HeLa cells were transfected with 3ϫ FLAGtagged MSK1, immunoprecipitated with anti-FLAG antibodies, and assayed for ATF1 serine 63 kinase activity using recombinant ATF1 and anti-phospho-CREB serum. The transfected HeLa cells were treated with EGF or anisomycin, an activator of the p38 pathway that has been shown to activate MSK1 (27). Both EGF and anisomycin caused a strong increase in MSK1 kinase activity, without affecting MSK1 levels (Fig.  1B). These results show that MSK1 can be activated to phosphorylate ATF1 by EGF treatment comparably to anisomycin and suggest that it is involved in activating the c-jun promoter upstream of ATF1.
Because we previously found that MEKK1 strongly activates the c-jun promoter (13), we tested whether MEKK1 and other MAPK pathway family members could activate MSK1 kinase activity. Overexpression of MEKK1 alone is sufficient to cause activation of downstream components; however the MAPKs require activation. For JNK1 we transfected an activated form, JNKK2-JNK1, that is fused to its upstream kinase JNKK2/ MKK7 (35). ERK2, p38, and ERK5 were transfected with activated forms of their upstream kinases Raf-1, MKK6, and MEK5, respectively. MEKK1 activated MSK1 kinase activity, but its downstream kinase JNK1 did not, suggesting that MEKK1 activates MSK1 through a different pathway (Fig. 1C). ERK2 and p38 activated MSK1 kinase activity as expected whereas ERK5 did not activate. ERK5 was highly activated as shown by induction of MEF2 reporter genes (data not shown). JNKK2-JNK1 was active as shown by its activation of a c-jun reporter gene (described below). These results suggest ERK2 and p38 as candidates for mediating MEKK1 and EGF activation of MSK1 and ATF1 phosphorylation.
Requirement of the ERK Pathway for EGF-induced Phosphorylation of ATF1-To determine the MAPK pathways responsible for EGF-induced phosphorylation of ATF1, we used chemical inhibitors of the different MAPK pathways. U0126 inhibits the ERK1/2 pathway by inhibiting the upstream kinase MEK1/2 (41). SB203580 inhibits the p38 pathway whereas the recently identified JNKII inhibits JNK1/2 (42)(43)(44)(45). HeLa cells were treated with EGF and the indicated inhibitors and immunoblotted with anti-phospho-CREB serum to determine the effects of the inhibitors on endogenous ATF1 phosphorylation. Only U0126 was able to inhibit ATF1 phosphorylation suggesting that the MEK1-ERK pathway is required ( Fig. 2A). As controls to show that the chemicals inhibited their respective MAPKs, we found that U0126 inhibited EGF induction of phosphorylated ERK1, SB203580 inhibited anisomycin-induced phosphorylation of ATF1, and JNKII inhibited anisomycininduced phosphorylation of ATF2 (Fig. 2, B-D). Anisomycin activates both p38 and JNK although it is JNK that phosphorylates ATF2 (46). SB203580 inhibition of anisomycin-induced ATF1 phosphorylation (Fig. 2C) suggests that p38 is required for this pathway as shown previously (27). In addition to showing that EGF induces ERK phosphorylation in HeLa cells (Fig.  2B), we also tested for EGF activation of JNK and p38 using phospho-specific sera. EGF induced JNK phosphorylation but had no effect on p38 (Fig. 2, E and F). As a control, anisomycin induced p38 phosphorylation (Fig. 2F). Because EGF did not activate p38, this further suggests that it is not required for EGF induction of ATF1 phosphorylation or c-jun expression.
MSK1 Kinase Activity Is Induced by EGF through the ERK Pathway-We used the MAPK pathway inhibitors to test which pathway is required for EGF activation of MSK1 kinase activity. The MEK inhibitor U0126 inhibited the EGF-induced activity whereas the other inhibitors had weak or negligible effects (Fig. 3A). We also used a dominant negative form of MEK1, MEK1A, to test for the requirement of the MEK1 pathway. MEK1A has a serine to alanine change at one of its FIG. 1. EGF induction of ATF1 phosphorylation and MSK1 kinase activity. A, HeLa cells were serum-starved for 24 h and treated with EGF (100 ng/ml) for the indicated times. Cell lysates were immunoblotted with anti-phospho-CREB (top) or anti-ATF1 antibodies (bottom). B, HeLa cells were transiently transfected with 3ϫ FLAG-MSK1, serum-starved, and treated with EGF (100 ng/ml) or anisomycin (25 g/ml) for 30 min followed by immunoprecipitation with anti-FLAG antibodies as described under "Experimental Procedures." The ATF1-kinase activity of the immune complex was measured using recombinant ATF1 as a substrate. The level of phosphorylated ATF1 was assayed by immunoblotting with anti-phospho-CREB sera (top). The amount of immunoprecipitated MSK1 was detected by immunoblotting with anti-FLAG antibodies (bottom). C, activators of MSK1 kinase activity. HeLa cells were cotransfected with 3ϫ FLAG-MSK1 and the indicated MAPK pathway kinases. ERK2, p38, and ERK5 were also cotransfected with activated forms of their upstream activators Raf BxB, MKK6, or MEK5, respectively. The cells were serum-starved, and the ATF1-kinase activity of transfected MSK1 was measured as described above.
activating phosphorylation sites (amino acid 222) (38). We found that it also inhibited EGF induction of MSK1 kinase activity but had no effect on anisomycin induction (Fig. 3B). We further tested for a role of the p38 and ERK5 pathways using dominant negative forms of their upstream kinases MKK3 and MEK5, respectively, that have alanines in place of their activating phosphorylation sites. The dominant negative MEK5 was used, because U0126 has been found to inhibit MEK5 activity, as well as MEK1 (47,48), and because the MEK5/ ERK5 pathway has been reported to be required for EGF induction of c-jun expression (19). Neither dominant negative MKK3A nor MEK5AA inhibited EGF induction of MSK1 kinase activity (Fig. 3C). In contrast, MKK3A inhibited anisomycin induction of MSK1 kinase activity consistent with the requirement for p38 activation in this pathway. These results show that EGF activation of MSK1 is by the MEK1-ERK pathway, the same as the pathway required for EGF-induced ATF1 phosphorylation (Fig. 2).
Dominant Negative ATF1 and MSK1 Block MEKK1 Induction of the c-jun Promoter-We tested whether ATF1 phosphorylation and MSK1 are required for activation of the c-jun promoter using dominant negative mutants. For ATF1 a phosphorylation site mutant was used (S63A) such that the protein can bind to ATF1 sites but cannot be activated (24,49). MSK1 contains two protein kinase domains, and we used mutations in conserved residues of either the first or second kinase domains (D195A and D565A) (27). We initially tested the effects of the mutants on MEKK1 activation of the c-jun promoter reporter gene, pJC6GL3, because we found previously that MEKK1 gives strong activation of the promoter similar to Rac1 (13). MEKK1 activated the promoter 25-fold, and this activation was strongly inhibited by ATF1-S63A (Fig. 4A). There was only a modest inhibition with WT ATF1. The two MSK1 mutants also inhibited although the MSK1-D565A mutant was more effective. WT MSK1 had only a slight effect. As a control, we tested the mutants with an activated form of MKK6 (MKK6EE) and p38. MKK6EE phosphorylates and activates p38 kinase activity (50). Although activated p38 can cause ATF1 phosphorylation, it can also activate the c-jun promoter through its phosphorylation of MEF2 factors (22,51). The c-jun promoter was strongly activated by MKK6EE and p38, but this activation was unaffected by the ATF1 and MSK1 mutants (Fig. 4A). The effect of the ATF1 phosphorylation site mutant strongly suggests that phosphorylation at serine 63 is required for MEKK1 activation of the c-jun promoter. The MSK1 kinase mutants also suggest that MSK1 or a related factor are required for the activation of the promoter.
Because ATF1 phosphorylation and MSK1 kinase activity were activated by the MEK/ERK pathway, we tested whether MEKK1 activation of the c-jun reporter gene also required this Cell lysates were immunoblotted with anti-phospho-CREB serum. B, U0126 inhibits ERK phosphorylation. HeLa cells were treated with or without U0126 followed by EGF treatment for 30 min, as indicated. Cell lysates were immunoblotted with anti-phospho-ERK serum. C, SB203580 blocks anisomycin-induced ATF1 phosphorylation. HeLa cells were treated with SB203580 for 40 min followed by anisomycin (25 g/ml) for 30 min. Lysates were immunoblotted with anti-phospho-CREB antibodies. D, JNKII inhibits JNK activation. HeLa cells were treated with JNKII for 40 min followed by anisomycin for 30 min. Lysates were immunoblotted with anti-phospho-ATF2 serum. E, EGF treatment induces the JNK pathway. HeLa cells were treated with EGF (100 ng/ml) for 30 min. Lysates were immunoblotted with anti-phospho-JNK serum. F, EGF does not induce the p38 pathway. HeLa cells were treated with EGF or anisomycin for 30 min. The lysates were immunoblotted with anti-phospho-p38 antibodies.

ATF1 Phosphorylation Regulates the c-jun Promoter
pathway. Dominant negative MEK1 strongly inhibited MEKK1 activation of the promoter, whereas it had no effect on activation by MKK6EE and p38 (Fig. 4B). Because MEKK1 also activates JNK kinase activity we tested for a requirement for JNK using the JNK binding domain of the JNK pathway scaffold JIP1 (52,53). This domain of JIP1 also strongly inhibited MEKK1 activation of the promoter, suggesting that JNK activity is required (Fig. 4B).
We found previously that Ras is required for EGF activation of the c-jun promoter and that activated Ras is also sufficient for activation (13). We tested whether Ras activation also requires both the JNK and MEK1 pathways and found that inhibitors of both of these pathways strongly inhibited Ras activation of the c-jun promoter (Fig. 4B).
Although MEKK1 predominantly activates the JNK pathway, when overexpressed it can activate ERK1/2, as well (37, 54 -56). However, we found previously that activated ERK (Raf BxB and ERK2) only weakly activated the c-jun promoter (13). Because EGF-induced phosphorylation of ATF1 is dependent upon the ERK pathway, and because ERK and JNK pathway inhibitors blocked MEKK1 and Ras activation of the c-jun promoter, these results suggested that the ERK and JNK pathways may cooperate to activate the c-jun promoter. We therefore transfected activated JNK1 (JNKK2-JNK1) with activated ERK2 to test for activation of the c-jun promoter. Although activated ERK2 and JNK1 both activated pJC6GL3 modestly (4-fold), together they gave a strong activation of 14-fold (Fig.  4C). These results suggest that these two pathways can cooperate to activate the c-jun promoter.
EGF-induced c-jun Expression Requires ERK and JNK Pathways-We found that EGF induction of the c-jun-luciferase gene, pJC6GL3, was only about 3-fold such that it was difficult to use this reporter to test the effect of inhibitors (data not shown). This low induction may be because of the stability of the luciferase mRNA and/or protein that makes it difficult to detect the transient induction over basal levels. We therefore used a full-length mouse c-jun gene (Ϫ225 to ϩ4005) that contains the 3Ј untranslated region of c-jun including mRNA destabilizing sequences (34). The mouse c-jun gene was trans-fected into human HeLa cells such that the transfected mouse gene could be distinguished from the endogenous transcript by RNase protection. An hsp70 promoter plasmid was cotransfected as an internal control for transfection efficiency. With this system, there was almost no basal expression of the transfected c-jun gene and strong induction by EGF (Fig. 5A). We then tested for the effect of ATF1 and MSK1 mutants. WT ATF1 and MSK1 had no effect on EGF induction whereas the mutants dramatically inhibited expression of the mouse c-jun gene. These results suggest that MSK1 and ATF1 are required for EGF activation of the promoter.
Because activation of MSK1 and ATF1 by EGF are ERK-dependent, we tested whether this or other MAPK pathway were required for EGF induction of the transfected c-jun gene using the MAPK inhibitors. There was no effect of the p38 inhibitor SB203580, but U0126 and JNKII completely blocked EGF induced expression (Fig. 5B).
We finally tested whether the endogenous human c-jun gene required the same pathways using an RNase protection assay for the human c-jun transcript. Similarly as for the transfected mouse c-jun gene, we found that EGF-induced expression of the human gene was blocked by U0126 and JNKII (Fig. 5C). These results show that the MEK-ERK and JNK pathways are required for EGF induction of c-jun expression.

DISCUSSION
One of the earliest and crucial events caused by mitogenic stimulation of mammalian cells is the expression of the protooncogene c-jun (1,2). We found previously that the MEF2 site at Ϫ59 and ATF1 site at Ϫ72 in the c-jun promoter are required for EGF-induced activation of expression (13,15). We also found that the Ras to Rac to MEKK pathway was required for EGF induction of the c-jun promoter (13). We have shown here that the MEK1 to ERK pathway is also required for EGF induction of the promoter because of its phosphorylation of ATF1. The MEK1 inhibitor U0126 blocked EGF-induced phosphorylation of ATF1 as serine 63, as well as EGF-induced expression of exogenous and endogenous c-jun genes. We also found that an ATF1 phosphorylation site mutant changing An activated form of MKK6 (MKK6EE) was cotransfected with p38 to cause its activation. The WT or mutant forms of ATF1 or MSK1 were included as indicated. After transfection the cells were serum-starved for 30 h and assayed for luciferase activity. All luciferase values were normalized to the internal control Renilla luciferase values and are presented as -fold induction over the reporter gene activity without activators. Each point is the mean of three determinations and shown with the mean Ϯ S.E. B, MEK and JNK pathways are required for MEKK1 and Ras activation of the c-jun promoter. Activation of pJC6GL3 was measure as in A with the indicated activators and inhibitors except that 0.5 g of MKK6EE and p38 were used. The JNK binding domain (amino acids 130 -281) of JIP1 was used, as was the dominant negative MEK1A. Activation by RasV12 reproducibly activated the SV40 promoter Renilla luciferase internal control 2.5-fold such that this increase was accounted for in the normalization of all samples transfected with RasV12. C, ERK and JNK pathways cooperate to activate the c-jun promoter. HeLa cells were transfected as above with pJC6GL3, pRLSV40, and expression vectors for activated Raf (Raf BXB) and ERK2 and/or activated JNK1 (JNKK2-JNK1). After transfection, cells were lysed and assayed for luciferase activity as described above.
ATF1 Phosphorylation Regulates the c-jun Promoter serine 63 to alanine blocked EGF-induced c-jun expression. These results strongly suggest that ATF1 phosphorylation by the ERK pathway is required for EGF-induced c-jun expression.
Because MSK1 and MSK2 are required for EGF-induced CREB phosphorylation in mouse embryo fibroblasts and are directly activated by ERK phosphorylation (27,28,31), we checked for the requirement of MSK factors in EGF induction of c-jun expression. The ATF1-kinase activity of transfected MSK1 was strongly activated by EGF. Dominant negative "kinase dead" MSK1 mutants also blocked EGF-induced c-jun expression. These results suggest that MSK1, or a closely related gene, is the protein required for EGF-induced ATF1 phosphorylation. Besides its close relative MSK2, the RSK family of kinases (RSK1, RSK2, RSK3) are also closely related to MSK1. We and others (30) (data not shown) have found that RSK2 kinase activity is also activated by EGF and that it can phosphorylate CREB and ATF1. The RSK2 gene is mutated in cells from CLS patients, a disease characterized by severe psychomotor retardation among other defects. EGF-induced phosphorylation of CREB was lost in fibroblasts from CLS patients that lack RSK2 (32). This suggests that RSK2 is required for CREB/ATF1 phosphorylation in some cells. Indeed, there was a low level of residual EGF-induced CREB phosphorylation in MSK1/MSK2 double null cells that could be because of RSK family members (31). Because we have used a dominant negative MSK1 mutant we cannot address which exact family member is required in HeLa cells. However, the loss of CREB kinase activity in MSK1/MSK2 double null cells and the maintenance of that activity in RSK2 null mouse embryo fibroblasts suggests that MSK1 and MSK2 are primarily required for EGF-induced CREB/ATF1 phosphorylation (26,31).
We found previously (13) that MEKK1 overexpression was sufficient for activation of a c-jun reporter gene. Because MEKK1 overexpression can activate ERK, as well as JNK, MEKK1 activation of the c-jun promoter is consistent with a requirement for ERK activation of MSK1 and ATF1. We confirmed this by showing that the ATF1 and MSK1 dominant negatives blocked MEKK1 activation of the c-jun promoter. We also found that a MEK1 dominant negative and a JNK inhibitor (JIP1) blocked MEKK1 activation of the promoter. We further found that activation of the JNK and ERK pathways (by JNKK2-JNK1 and activated Raf) synergized to cause strong activation of the c-jun promoter. Because ERK and JNK pathway inhibitors blocked EGF induction of c-jun expression, these results strongly suggest that these two pathways cooperate to activate the c-jun promoter.
The MEF2 site at Ϫ59 is also required for EGF-induced activation of the c-jun promoter and binds predominantly the MEF2D factor in HeLa cells (15,16). A potent activator of MEF2D is ERK5, which is also activated by EGF (19,57). Mapping of EGF-induced phosphorylation sites on MEF2D has suggested that the same site (serine 179) is phosphorylated by ERK5 and in response to EGF (19). A MEF2D serine 179 mutant also blocked EGF induction of the c-jun promoter (19). However, we and others (23) 3 have found no effect of MEF2D 3 R. Wang and R. Prywes, unpublished results. Transcripts from the mouse c-jun and hsp70CAT genes were detected by RNase protection as described under "Experimental Procedures." The positions of migration of the specific transcripts are indicated. B, U0126 and JNKII block EGF induction of the transfected mouse c-jun gene. HeLa cells were transfected with the mouse c-jun gene as described above and treated with the inhibitors U0126, SB203580, or JNKII for 40 min followed by EGF for 30 min. The mouse c-jun and hsp70CAT transcripts were detected by RNase protection. C, U0126 and JNKII block EGF induction of the endogenous human c-jun. HeLa cells were serum-starved 24 h prior to inhibitor and EGF treatment as described above. The endogenous human c-jun transcript was detected by RNase protection as described under "Experimental Procedures." serine 179 mutation for ERK5 activation of GAL4-MEF2D, such that the effect of the MEF2D serine 179 mutant may not reflect a requirement for ERK5. MEKK1 activation of a c-jun reporter gene was also not affected by a MEK5 dominant negative such that ERK5 activation does not appear to be required for this pathway (20).
MEKK1 and Ras activation of the c-jun promoter were strongly inhibited by a JIP1 mutant that inhibits JNK activation (20) (Fig. 4). We further found here that EGF activation of a transfected c-jun gene and endogenous c-jun expression were inhibited by the JNK inhibitor JNKII. Because activated JNK cooperated with activated ERK for induction of the c-jun promoter, and because activated JNK did not activate MSK1 phosphorylation of ATF1, it is intriguing to speculate that JNK may contribute by inducing MEF2D activity by an unknown pathway.
Together our results suggest a model whereby EGF induces c-jun expression by at least two MAPK pathways (Fig. 6). EGF induction of Ras is mediated by GRB2 and Sos (58,59). Ras then activates Raf-1 and the rest of the ERK1/2 pathway (5). The ATF1 site on the c-jun promoter is activated by ERK1/2 through its phosphorylation and activation of an MSK family member that phosphorylates ATF1 on serine 63. Activation of the EGF receptor also results in activation of Rac1, which activates the downstream kinase JNK (58,60). We find here that JNK is also required for c-jun expression; however the mechanism remains to be determined. It is unclear how ERK5 fits into these pathways although it is a potent activator of MEF2D (19,23). ERK5 can bind to Raf-1; however activated Raf-1 did not activate ERK5 kinase activity (61). The ERK5 kinase pathway is activated by MEK5, which is activated by MEKK2/3 (62,63), such that it will be important to link EGF activation of MEKK2/3 to the EGF receptor.