Stress-induced Inhibition of ERK1 and ERK2 by Direct Interaction with p38 MAP Kinase*

We have identified a direct physical interaction between the stress signaling p38α MAP kinase and the mitogen-activated protein kinases ERK1 and ERK2 by affinity chromatography and coimmunoprecipitation studies. Phosphorylation and activation of p38α enhanced its interaction with ERK1/2, and this correlated with inhibition of ERK1/2 phosphotransferase activity. The loss of epidermal growth factor-induced activation and phosphorylation of ERK1/2 but not of their direct activator MEK1 in HeLa cells transfected with the p38α activator MKK6(E) indicated that activated p38α may sequester ERK1/2 and sterically block their phosphorylation by MEK1.

. Whereas the selective activation of distinct MAP kinase pathways in response to different extracellular stimuli has been extensively documented, there is increasing evidence for cross-talk between distinct MAP kinase pathways. A p38-dependent ERK1/2 activation was observed in several mammalian cell lines including the human embryonic kidney cell line HEK293 upon arsenite treatment (5). It was also found that inactivation of p38 by SB202190 treatment resulted in a delayed and prolonged activation of ERK1/2 in the human hepatoma cell line, HepG2 (6). In both cases, MEK1 was implicated in the activation of ERK1/2. Here we report that in HeLa and HEK293 cells, stress stimuli lead to an inhibition of ERK1/2 via p38␣. Phosphorylated p38␣ is capable of forming a complex with ERK1/2, and it prevents their phosphorylation by MEK1/2.
Expression of GST Fusion Proteins in Bacteria-GST fusion protein plasmids were transformed into DH5␣ bacteria. Expression of GST fusion proteins was induced by 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside at 37°C for 3 h. GST fusion proteins were purified as described previously on glutathione-agarose (Sigma) and eluted from beads with reduced glutathione; GST tag was cleaved with thrombin when necessary (8).
GST Fusion Protein Pull-down-Two mg of rat brain lysate or 1 mg of HeLa cell lysate was mixed with 20 g of immobilized GST-ERK1, GST-p38␣, or GST alone on glutathione-agarose beads at 4°C with rotation. After a 2-h incubation, beads were washed three times with 50 mM Tris (pH 8.0), 150 mM NaCl, and 1% Nonidet P-40 followed by separation of bound proteins by SDS-polyacrylamide gel electrophoresis (PAGE). Proteins were transferred to nitrocellulose membrane, and immunoblotting was performed with either anti-p38␣ or anti-ERK1-CT antibody (StressGen, Victoria, British Columbia, Canada).
In Vivo Association of ERK1 and p38 -HEK293 or HeLa cells cultured in Dulbecco's minimum essential medium (DMEM, Life Technologies, Inc.) containing 10% fetal bovine serum were grown to 50 -60% confluence and transfected with pcDNA3-p38␣ using SuperFect reagent (Qiagen, Mississauga, Ontario, Canada) as per the manufacturer's instructions. For each 100-mm dish, 5 g of plasmid DNA were introduced into cells using 30 l of SuperFect reagent in serum-free medium. After a 3-h incubation, cells were starved in fresh serum-free DMEM. 24 h after transfection, cells were stimulated with anisomycin (Sigma) or arsenite (Sigma) alone or in combination with SB203580 (Calbiochem) in serum-free DMEM as indicated in the figure legends. Cells were lysed in 500 l of lysis buffer (150 mM NaCl, 20 mM Tris pH 8.0, 0.5% (w/v) Nonidet P-40, 1 mM dithiothreitol (DTT), 20 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin) and sonicated for 30 s. Cell debris was removed by centrifugation at 13,000 rpm for 15 min at 4°C. Protein concentration was determined by the Bradford assay (9). Cell lysates were precleared with 40 l of protein A-Sepharose (Amersham Pharmacia Biotech) beads and then incubated with anti-ERK1-CT antibody (1 g/ml of lysate) at 4°C overnight. 40 l of protein A-Sepharose beads were then added to precipitate immunocomplexes and washed with lysis buffer three times; immunoprecipitates were analyzed by immunoblotting with anti-p38␣ antibody.
MBP Kinase Assay-Anti-ERK1 immunoprecipitates bound to protein A beads were further washed twice in assay dilution buffer (25 mM ␤-glycerophosphate, 20 mM MOPS, pH 7.2, 5 mM EGTA, 2 mM EDTA, 20 mM MgCl 2 , 1 mM Na 3 VO 4 , 0.25 mM DTT) and then incubated with 20 g of MBP (Sigma) in the same buffer supplemented with 50 M ATP and 10 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech) in a volume of 20 l at 30°C for 15 min. Reactions were stopped by spotting onto P81 paper. After three washes in 1% H 3 PO 4 , phosphotransferase activity was quantitated by liquid scintillation counting.
Interaction of Bacterially Expressed p38␣ and ERK1 in Vitro-Eluted GST-p38␣ wild-type or GST-p38␣ (AF) mutant fusion protein was first incubated with bacterially expressed MKK6(E), a constitutively active form of MKK6, at 30°C for 2 h in 1ϫ kinase buffer (20 mM Hepes, pH 7.4, 10 mM MgCl 2 , 1 mM DTT, 150 M phenylmethylsulfonyl fluoride, 0.4 mM ATP). GST-p38␣ proteins were then absorbed onto to glutathione-agarose beads and incubated with bacterially expressed ERK1 at 4°C for 2 h. Following three washes in lysis buffer, the beads were boiled in 2ϫ Laemmli sample buffer (10), and the bound proteins were separated on SDS-PAGE. Conversely, p38␣ was first incubated with GST-MKK6(E) beads for activation under the conditions described above. GST-MKK6(E) beads were pelleted and discarded, and GST-ERK1 beads were added to the supernatants. After a 2-h incubation, the bound proteins were resolved by SDS-PAGE.

RESULTS AND DISCUSSION
Evidence for the Inhibition of ERK1/2 by p38 -By using antibody specific for the phosphorylated (activated) form of ERK1/2 in Western blotting studies, we observed that anisomycin treatment resulted in decreased ERK1/2 phosphorylation. By contrast, SB203580, a specific p38 MAP kinase inhibitor, induced increased ERK1/2 phosphorylation in HeLa cells similar to that observed with epidermal growth factor (EGF) exposure (Fig. 1A). These findings are consistent with the results in HepG2 cells (6) and indicate that p38 MAP kinase somehow exerts an inhibitory effect on ERK1/2 activation. Surprisingly, an increase of myelin basic protein (MBP) phosphotransferase activity associated with anti-ERK1 immunopre-cipitates was detected upon anisomycin treatment, which could be inhibited by including SB203580 in kinase assay reactions (Fig. 1B). This observation indicated that the increase of MBP phosphotransferase activity precipitated by anti-ERK1 antibody upon anisomycin treatment was at least partially due to p38 MAP kinases. The apparent contradiction between the ERK1/2 phosphorylation state in Western blotting and the MBP phosphotransferase activity associated with anti-ERK1 immunoprecipitates could be potentially reconciled if p38 is coprecipitated with ERK1/2 in response to anisomycin treatment. Previously, we reported that ERK1 was found in immunoprecipitates of a p38 homologue in immature sea star oocytes (11). Based on these observations, we postulated that the direct interaction between p38 and ERK1/2 MAP kinases might play a role in coordinating the regulation of these two distinct MAP kinase pathways.
Interaction of ERK1/2 and p38 in Vitro-To examine whether a physical interaction occurs between p38 and ERK1, we expressed glutathione S-transferase (GST) fusion proteins of the full-length human ERK1 and p38␣ in bacteria and used them to affinity purify proteins from a rat brain lysate. Immunoblotting with anti-p38␣ antibody revealed the presence of p38␣ protein on the GST-ERK1 beads at a level well above that bound to GST alone ( Fig. 2A). Under similar conditions, GST-p38␣ was able to pull down both ERK1 and ERK2 proteins (Fig.  2B). These results indicated a specific, direct interaction between ERK1/2 and p38␣.
Interaction of ERK1/2 and p38 Is p38 Activitydependent-We next examined the interaction between ERK1 and p38␣ in HeLa cells treated with stimuli that specifically activate either the ERK1/2 or p38 MAP kinase pathways. The treatment of anisomycin resulted in a significant increase of p38␣ that was bound by GST-ERK1 (data not shown). No apparent difference of the amount of p38␣ protein purified by GST-ERK1 fusion protein between EGF-treated and untreated control HeLa cells was observed. Treating HeLa cells with SB203580 prior to anisomycin stimulation diminished the association of ERK1 with p38␣. The correlation of the enhancement of the binding of ERK1 and p38␣ with p38 kinase activation indicated that the interaction between these two MAP kinases was dependent upon the p38 but not the ERK1 activity status.
We further monitored the physical interaction of these two MAP kinases in mammalian cells by coimmunoprecipitation. HeLa cells were transfected with human p38␣ full-length DNA and then treated with anisomycin 24 h later. Cell lysate was prepared and precipitated with anti-ERK1-CT antibody. Anti-ERK1 immunoprecipitates were probed with anti-p38␣ antibody in Western blot analysis for coprecipitated p38␣. As shown in Fig. 3A, a much higher level of p38␣ protein was detected in the anti-ERK1 precipitates from anisomycintreated HeLa cells than in those from untreated control cells. This finding was consistent with the GST fusion protein pulldown results and further confirmed the requirement of p38 kinase activity for the interaction between ERK1 and p38␣.

FIG. 2. Interaction of ERK1 and p38␣ in rat brain lysate revealed by GST fusion protein pull-down experiments.
A, after incubation with rat brain lysate, GST-ERK1 and GST beads were washed and immunoblotted with anti-p38␣ antibody. B, ERK1/2 were found on GST-p38␣ beads after incubation with rat brain lysate.
Similarly, interaction between ERK1 and p38␣ in HEK293 cells transfected with p38␣ was observed in a p38 activity-dependent manner (Fig. 3B). Compared with untreated control cells, an increasing level of p38␣ protein precipitated by anti-ERK1-CT antibody was found in arsenite-treated HEK293 cells. Pretreating HEK293 cells with SB203580 prior to arsenite reduced the level of p38␣ precipitated by anti-ERK1-CT to that of untreated control. These results confirmed a p38 activitydependent association of p38␣ with endogenous ERK1 in mammalian cells.
Direct Interaction Occurs between Purified ERK1 and p38 in Vitro-We further examined whether a direct interaction occurs between p38␣ and ERK1 using recombinant forms of these MAP kinases. Bacterially expressed ERK1 protein was mixed with GST-p38␣ wild-type or dominant negative mutant p38␣ (AF) that was preincubated with MKK6(E), a p38-specific upstream activating kinase. p38␣ (AF) is a kinase-inactive mutant with the substitution of two activating phosphorylation sites by alanine and phenylalanine (5), whereas MKK6 is constitutively activated by replacement of two activating phosphorylation sites with glutamic acid (12). The ERK1 protein was affinity-purified by p38␣ wild-type protein after its activation by MKK6(E) (Fig. 4A). No MKK6 protein was detected in the p38␣/ERK1 complexes (data not shown), indicating that p38␣ and ERK1 can interact with each other without involvement of other proteins. Consistent with the requirement of p38 kinase activity in the formation of ERK1/p38 complexes, ERK1 precipitated by p38␣ wild-type protein was present at a higher level than that obtained with the p38␣ (AF) mutant. Similar observations were made in the converse experiment, where GST-ERK1 bound active wild-type but not inactive wild-type and dominant negative forms of p38␣ (Fig. 4B). Furthermore, Western blotting analysis with anti-phospho-p38␣ antibody revealed the p38␣ protein coprecipitated with ERK1 was in its phosphorylated form (Fig. 4C).
p38 Does Not Suppress MEK1 Phosphorylation to Inhibit ERK1/2-To assess the roles of p38-ERK1/2 interaction in regulating the activation of ERK1/2, we activated endogenous p38 kinases by transfecting HeLa cells with MKK6(E) DNA and then treated the cells with EGF and monitored the activation of ERK1/2 and their upstream activating kinase, MEK1. In response to EGF treatment, the phosphorylated forms of ERK1/2 were present at much lower levels in MKK6(E)-transfected cells than in nontransfected HeLa cells (Fig. 5A), indicating an inhibitory effect of active p38 on ERK1/2. Moreover, no apparent difference in the phosphorylation of MEK1 at its activation sites was observed under these two circumstances (Fig. 5B). These results indicate that p38 suppresses ERK1/2 by a mechanism independent of MEK1 phosphotransferase activity.
In summary, our study provides the first experimental evidence for the direct interaction between two MAP kinases lying in two distinct signaling cascades, raising the possibility that the direct interaction between p38␣ and ERK1 may play a role in coordinating activation of these two distinct MAP kinase pathways. These findings differ from those revealed by previous studies in that the cross-talk between the p38 and ERK1/2 signaling pathways was believed to mediate through upstream activating kinases of the ERK1/2 cascade. Moreover, the necessity of p38, but not ERK1, phosphotransferase activity for the interaction between ERK1 and p38␣ indicates the cross-talk between these two MAP kinase pathways is a one-way process.
Combined with previous studies, our results indicate that the communication between p38 and the ERK1/2 pathways may act through two distinct modes. Active p38 may suppress ERK1/2 phosphotransferase activity either through inhibition of upstream activating kinases of ERK1/2 or through direct interaction between p38 and ERK1/2. Based on our observation of the direct association between p38␣ and ERK1 and the inhibitory effect of p38 on ERK1/2 phosphotransferase activities independent of MEK1 phosphorylation, we hypothesize that activated p38 may sequester ERK1/2 and sterically block phosphorylation of these MAP kinases by MEK1/2. The ability of activated p38 to regulate another protein kinase allosterically is not restricted to ERK1/2. We have recently reported that activated p38 can also form a complex with casein kinase CK2 in HeLa cells, but in that instance p38 activates CK2 (13).  3. Coimmunoprecipitation of ERK1 and p38␣ in human cell lines in a p38 kinase activity-dependent manner. A, after transfection with pcDNA3-p38␣ and treatment with anisomycin (10 g/ml, 30 min), HeLa cells were lysed, and the lysate was precipitated by anti-ERK1-CT antibody. Anti-ERK1 precipitates were immunoblotted with anti-p38␣ antibody. B, HEK293 cells were transfected with pcDNA3-p38␣ 24 h prior to treatment with arsenite (0.5 mM, 4 h) or a combination of SB203580 (5 M, 60 min prior to arsenite) and arsenite (SBϩArsenite). Cell lysates were prepared and immunoprecipitation was performed with anti-ERK1-CT antibody. Immunoprecipitates were probed with anti-p38␣ antibody. p38␣ was present in higher levels in arsenite-treated cells than found in SB203580-treated and untreated control cells.

FIG. 4. Direct interaction between bacterially expressed ERK1 and p38␣ in vitro.
A, following incubating with 20 g of MKK6(E), 50 g of eluted GST-p38␣ wild-type (WT) or GST-p38␣ (AF) was precipitated with glutathione-agarose beads. 50 g of thrombin-digested GST-ERK1 was added to GST-p38␣ beads, and ERK1 that bound to p38 was detected by Western blotting with anti-ERK1-CT antibody. B, 50 g of thrombin-digested p38␣ wild-type or p38␣ (AF) was first incubated with MKK6(E) and then incubated with 50 g of GST-ERK1 immobilized on glutathione-agarose beads. GST-ERK1-bound p38␣ was detected by Western blotting with anti-p38␣ antibody. C, a similar blot to that shown in B was probed with anti-active p38␣ antibody (New England Biolabs, Beverly, MA).