Structure-function studies of p38 mitogen-activated protein kinase. Loop 12 influences substrate specificity and autophosphorylation, but not upstream kinase selection.

Several mitogen-activated protein kinase (MAPK) cascades have been identified in eukaryotic cells. The activation of MAPKs is carried out by distinct MAPK kinases (MEKs or MKKs), and individual MAPKs have different substrate preferences. Here we have examined how amino acid sequences encompassing the dual phosphorylation motif located in the loop 12 linker (L12) between kinase subdomains VII and VIII and the length and amino acid sequence of L12 influence autophosphorylation, substrate specificity, and upstream kinase selectivity for the MAPK p38. Conversion of L12 of p38 to an “ERK-like” structure was accomplished in several ways: (i) by replacing glycine with glutamate in the dual phosphorylation site, (ii) by placing a six-amino acid sequence present in L12 of ERK (but absent in p38) into p38, and (iii) by mutations of amino acid residues in loop 12. Two predominant effects were noted: (i) the Xaa residue in the dual phosphorylation motif Thr-Xaa-Tyr as well as the length of L12 influence p38 substrate specificity, and (ii) the length of L12 plays a major role in controlling autophosphorylation. In contrast, these modifications do not result in any change in the selection of p38 by individual MAPK kinases.

Protein kinase cascades control a variety of functions in eukaryotic cells (1). Key components of these protein kinase cascades are members of the superfamily of mitogen-activated protein kinases (MAPKs). 1 These serine/threonine kinases act by modulating the activity of cellular proteins such as transcription factors, enzymes including other kinases, cell-surface receptors, and structural proteins (2). In yeast, distinct signal-ing pathways leading to MAPK activation have been defined using biochemical and genetic approaches (3)(4)(5).
Several individual MAPK signal transduction pathways have been characterized in mammalian cells. One pathway leads to activation of the extracellular signal-regulated protein kinase (ERK) group of MAPKs, and these enzymes have been shown to play an important role in regulating cellular responses to growth factors (2,6). A second MAPK pathway is regulated by changes in the extracellular milieu induced by physical-chemical changes or by proinflammatory cytokines leading to phosphorylation of transcription factors such as c-Jun (7,8). A third MAPK pathway leads to activation of the p38 pathway by stimuli such as UV light, increased extracellular osmolarity, proinflammatory cytokines, or products of microbial pathogens (9 -12). An additional MAPK family member has been recently described termed ERK5 (13) or BMK1 (14). MAPK family members have marked sequence homologies, with an overall sequence identity of Ͼ40%. Nonetheless, there are specificities found at the level of the upstream activators, the MAPK kinases (15)(16)(17)(18)(19)(20) and the substrates phosphorylated by the active MAPKs (7,8,10,21,22).
From the x-ray crystallographic structures of ERK2 (23) and p38 (24,25) as well as from primary sequence comparisons of the members of the MAPK family (9,14), two structural domains of MAPK have been suggested to be important in regulating function insofar as recognition by upstream activators and substrate specificity are concerned. Specifically, these are the Thr-Xaa-Tyr dual phosphorylation motif and the loop 12 linker (L12) as defined by data from the three-dimensional structure of ERK2 (23). Herein we have examined the influence of the Xaa residue in the Thy-Xaa-Tyr dual phosphorylation motif and the role of sequences that are part of L12 in several aspects of p38 function. We provide evidence that the Xaa residue in the Thy-Xaa-Tyr motif and the length of the loop are important for the substrate specificity of p38. The length of L12 also plays an important role in controlling the extent of autophosphorylation. In contrast, modification of the L12 structure alone does not change the selectivity of the upstream MAPK kinases.

MATERIALS AND METHODS
Mutagenesis of Murine p38 -A construct designed to express murine p38 with His 6 at its amino terminus was prepared by ligating the p38 coding region, amplified from p38 cDNA in BlueScript using the polymerase chain reaction (primers gcagccatATGTCGCAGGAGGCC and ccg-gatccTCAGGACTCCATTTCTTC), with pET14b vector (Novagen, Madison, WI), incorporating NdeI and BamHI sites. A construct designed to express p38 with a flag tag at the amino terminus in a mammalian expression vector (pcDNA3) was prepared as described (10,26). The p38 mutants shown in Fig. 1 were generated using a polymerase chain * This work was supported by National Institutes of Health Grants GM51417 (to J. H.) and GM37696 and AI15136 (to R. J. U.) and American Heart Association Grant-in-aid 95007690 (to J. H.). This is Publication 9797-IMM from the Department of Immunology, Scripps Research Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** Established Investigator of the American Heart Association. To whom correspondence should be addressed: Dept. of Immunology, Scripps Research Inst., 10666 N. Torrey Pines Rd., La Jolla, CA 92037. Tel.: 619-784-8704; Fax: 619-784-8239. 1 The abbreviations used are: MAPKs, mitogen-activated protein kinases; ERK, extracellular signal-regulated protein kinase; L12, linker loop 12 between kinase subdomains VII and VIII; PAGE, polyacrylamide gel electrophoresis; HA, hemagglutinin; MKK or MEK, MAPK/ ERK kinase; GST, glutathione S-transferase; ATF2, activating transcription factor 2; JNK, c-Jun N-terminal kinase; MBP, myelin basic protein.
reaction-based approach as described (27). 2 The cDNA encoding each of these mutants was ligated into pET14b vector using NdeI and BamHI sites and pcDNA3 vector with HindIII and XbaI sites as described above for wild-type p38. Each construct was sequenced to ensure the fidelity of the mutations.
Expression and Purification of p38 and Its Mutants-The pET14b vectors containing DNA encoding p38 or the p38 mutants were transformed into the BL21(DE3) strain of Escherichia coli. E. coli cells were grown at 37°C in Luria broth until A 600 ϭ 1.0, at which time isopropyl-␤-D-thiogalactopyranoside (1 mM final concentration) was added for an additional 2 h. The cells were then collected by centrifugation at 8000 ϫ g for 10 min. For each 100 ml of original bacterial culture, the bacterial pellet was resuspended in 5 ml of binding buffer (5 mM imidazole, 500 mM NaCl, and 20 mM Tris-HCl, pH 7.9). The bacteria were lysed by sonication using a cell disrupter (W-375, Ultrasonics, Inc.). Insoluble materials were removed by centrifugation at 10,000 ϫ g for 30 min. The supernatant was applied to a Ni 2ϩ -nitrilotriacetic acid-agarose column (0.5 ml; Novagen); the column was washed with 10 ml of washing buffer (60 mM imidazole, 500 mM NaCl, and 20 mM Tris-HCl, pH 7.9); and the bound protein was eluted with elution buffer (1 M imidazole, 500 mM NaCl, and 20 mM Tis-HCl, pH 7.9). The expressed p38 and altered proteins were soluble and were recovered with a yield of ϳ5 g/1 ml of bacteria culture. When analyzed by SDS-PAGE, the proteins appeared to be Ͼ90% pure after purification on an affinity column selective for His-tagged recombinant proteins (data not shown).
Transient Expression or Coexpression of Various cDNAs-COS-7 (4 ϫ 10 5 ) or CHO-K1 (6 ϫ 10 5 ) cells (in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum) on 35-mm plates were transfected with 1 g (total) of one or more plasmid DNAs using Lipofectamine (Life Technologies, Inc.). After 48 h, the cells were treated with or without stimuli as described below. The tagged proteins were immunoprecipitated with the appropriate monoclonal antibody and quantitated by Western blot analysis as described below.
Preparation of MKK3b, MKK6b, and MEK1-HA-tagged MKK3b or MKK6b in pcDNA3 vector and HA-tagged MEK1 cDNA in pSR␣ vector (1 g) were transfected into COS-7 cells as described above. 48 h after transfection, the cells were lysed by incubation at 4°C for 15 min in lysis buffer (20 mM Tris-HCl, 120 mM NaCl, 10% glycerol, 1 mM Na 3 VO 4 , 2 mM EDTA, 1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride, pH 7.5), and the HA-tagged protein was immunoprecipitated with anti-HA monoclonal antibody 12CA5 as described (15). Quantification of MKKs in the immunoprecipitates was accomplished by Western blotting using monoclonal antibody 12CA5; this quantitation provided us with a basis for normalization in subsequent in vitro kinase assays.
Western Blotting-Western blotting was performed as described (33). Anti-phosphotyrosine antibodies 4G10 (Upstate Biotechnology, Inc., Lake Placid, NY) and FB2 (obtained from American Type Culture Collection) was used to detect proteins with phosphotyrosine; antibodies M2 and 12CA5 were used in Western blotting for analysis of tagged proteins expressed in COS-7 and CHO-K1 cells. Goat anti-mouse IgG conjugated to horseradish peroxidase (Cappel, Durham, NC) was used as secondary antibody. Enhanced chemiluminescence detection reagents (ECL, Amersham Corp.) were used according to the manufacturer's instruction. Quantitation was done by densitometer analysis.
Kinase Assays-Kinase assays using 0.2-0.4 g of purified p38, p38 mutants, or immunoprecipitates containing activated kinases were performed at 37°C for 20 min using 5 g of substrate, 250 M ATP, and 10 Ci of [␥-32 P]ATP in 20 l of kinase buffer (20 mM Hepes, pH 7.6, 20 mM MgCl 2 , 25 mM ␤-glycerophosphate, 0.1 mM Na 3 VO 4 , and 2 mM dithiothreitol). The reactions were terminated with Laemmli sample buffer, and the products were resolved by SDS-PAGE and analyzed by autoradiography. The extent of protein phosphorylation was quantified by phosphoimaging.

RESULTS
p38 Mutants with Modifications in Loop 12-Sequence comparisons of the known MAPKs revealed that each group has a distinct phosphorylation lip (L12) structure (9,14). To investigate how the L12 structure is involved in the function of p38, we have created a series of p38 mutants with modifications in this region (shown schematically in Fig. 1). Specifically, we introduced mutations to make p38 more "ERK-like" in the L12 region. Thus, p38(E) contains an ERK-like Thr-Glu-Tyr dual phosphorylation site; p38(6ϩ) has an additional six amino acids from ERK inserted after Glu 178 ; p38(6ϩE) has both the additional six-amino acid sequence of p38(6ϩ) and the Thr-Glu-Tyr dual phosphorylation site; p38(VAP) has 174 HTDD 177 substituted with the ERK2 sequence VADP; p38(DL) has 178 EM 179 replaced with DL; and p38(VAPD6ϩLE) has a loop 12 structure identical to that found in ERK2.
Activation of p38 and p38 Mutants Catalyzed by MKKs-Previous studies have demonstrated that there are distinct MAPK kinases (MEKs or MKKs) responsible for activating ERK, p38, or JNK (or stress-activated protein kinase) (15). It has been suggested by others that the nature of the second amino acid in the Thr-Xaa-Tyr dual phosphorylation motif may play a role in determining the specificity of interactions between MKK and its MAPK substrate (34,35). To test this hypothesis, we examined the activation of wild-type p38 and 2 The sequences of the primers used for generating the mutations will be provided upon request. the panel of p38 mutants; MKK3b (36), 3 MKK6b (16), 3 or MEK1 (37) was used for this experiment. Using an in vitro coupled kinase assay, we observed that the enzymatic activity of p38 or p38 mutants was increased when MKK3b or MKK6b was included in the kinase reaction, but not when we used MEK1 (Fig. 2). The extent of activation was nearly the same when each of the mutants was compared with wild-type p38. As expected, MEK1 dramatically increased the activity of ERK2 (Fig. 2, lower panel). In a second approach, we cotransfected MKKs and individual epitope-tagged p38 forms in COS-7 cells to further examine the effects of modifying L12 on the ability of MKKs to activate p38 (Fig. 3A). Consistent with the in vitro experiments, the activity of p38 or p38 mutants was enhanced similarly when cotransfected with MKK3b or MKK6b, but not with MEK1. ERK2 activity was only enhanced when cotransfected with MEK1. Equal amounts of wild-type or modified p38 protein were detected in the immunoprecipitates used for the kinase assay (Fig. 3B).
We next asked whether this same pattern is observed in a setting where activation is initiated by an extracellular stimuli. To investigate this, CHO-K1 cells were transiently transfected with flag-tagged p38, p38 mutants, or ERK2, and cells were treated with 0.4 M sorbitol, UV light, epidermal growth factor ϩ insulin, or 20% fetal calf serum. The transiently expressed epitope-tagged MAPKs were immunoprecipitated with the M2 antibody, and the resultant immunoprecipitates were analyzed for kinase activity. High osmolarity or UV light enhanced the kinase activity of p38 and all p38 mutants (Fig. 4A). Addition of an ERK-like loop structure did not result in conversion of p38 to respond to growth factor signals that activate ERK. Thus, the totality of these data suggest that neither the Thr-Xaa-Tyr motif nor the L12 structure is a critical structural feature directing the selectivity of the upstream kinase.
Autophosphorylation of p38 and p38 Mutants-Histidinetagged recombinant p38 expressed in E. coli appears to undergo an autophosphorylation reaction since it is phosphorylated on tyrosine (Fig. 5). Although at present this event cannot be linked to any physiologically relevant changes, the parameter of autophosphorylation provides us with an additional way to evaluate structure-function relationships of p38. We used anti-phosphotyrosine antibody in Western blots to analyze the extent of tyrosine phosphorylation of recombinant wild-type p38 or p38 mutants with the modification in the L12 region. We observed that the proteins with added amino acids in L12 have substantially more phosphotyrosine when compared with equal amounts of wild-type p38 (Fig. 5, A and B). The same results were obtained using two different anti-phosphotyrosine monoclonal antibodies, 4G10 and FB2. The level of phosphotyrosine in p38(6ϩ), p38(6ϩE), or p38(VAPD6ϩLE) is similar to that found in recombinant ERK2 expressed and isolated identically to the means used to obtain p38 forms.
Autophosphorylation of p38 or its modified forms can also be detected by incubation of recombinant protein with [␥-32 P]ATP under conditions of an in vitro kinase assay; p38(6ϩ), p38(6ϩE), and p38(VAPD6ϩLE) display enhanced autophosphorylation when compared with wild-type p38 or the other p38 mutants (Fig. 5C). Similar results were obtained when recombinant proteins expressed in mammalian cells were used (data not shown). These data suggest that the length of L12 controls the structure of p38 in ways that regulate the autophosphorylation reaction, while the nature of the Xaa residue in the Thr-Xaa-Tyr dual phosphorylation site has little or no influence on this activity of p38.
Effect of Sequence Modifications of L12 on p38-Substrate Interaction-We next asked whether changes in L12 influence the substrate specificity of p38. To investigate this, we performed in vitro kinase assays using the following proteins as substrates: ATF2 (38), ELK1 (30), c-Myc (31), PHAS-1 (39), and myelin basic protein (MBP). Each of these proteins can be phosphorylated by p38 (Fig. 6), with the exception of GST used as a control for the GST fusion proteins (data not shown). Equal quantities of kinase (see Fig. 5A) and substrate protein (Fig.  6B) were used in each kinase assay. The level of substrate phosphorylation was observed to differ substantially when wild-type p38 and its modified forms were compared (Fig. 6A). MBP, Myc, and ELK1 contain the consensus Pro-Xaa-Ser/Thr-Pro sequence, which has been defined as the consensus recognition site for ERK1/2 (40 -42). The kinase activity of p38 toward these three proteins was increased by the insertion of six residues or by a Gly to Glu mutation. However, p38(6ϩE) and p38(VAPD6ϩLE) do not show this enhanced activity, but have reduced substrate phosphorylation. ATF2 and PHAS-1, which have different primary structure at the phosphorylation site when compared with MBP, Myc, or ELK1, behaved differently insofar as p38-catalyzed phosphorylation is concerned. p38(E) and p38(6ϩ) each demonstrated markedly reduced phosphorylation of ATF2 when compared with that produced by wild-type p38. The p38(6ϩE) and p38(VAPD6ϩLE) mutants also showed an even greater reduction in ATF2 phosphorylation, suggesting that the two modifications may synergize in terms of their effect on substrate recognition and/or catalysis; p38(6ϩ) and p38(6ϩE) had enhanced activity toward PHAS-1. The mutation of HTDD (p38(VAP)) or EM (p38(DL)) does not lead to any change in the substrate specificity. The differential phosphorylation of these substrates by p38 and some p38 mutants cannot be due to a general change in enzymatic activity because differing effects on phosphorylation by p38 and its mutants were substrate-dependent.
Autophosphorylation and Enzymatic Activity-We next investigated whether the increased autophosphorylation observed with the p38 mutants p38(6ϩ), p38(6ϩE), and p38(VAPD6ϩLE) is reflected in a similar change in enzymatic activity toward specific protein substrates. We found that a higher level autophosphorylation of p38(6ϩ), p38(6ϩE), and p38(VAPD6ϩLE) did not correlate with phosphorylation of MBP and PHAS-1 (Fig. 7). Similar results were obtained with the other substrates shown in Fig. 6 (data not shown). These data show that there are different mechanisms in controlling the autophosphorylation and substrate recognition/phosphorylation.

FIG. 3. In vivo activation of p38 and p38 mutants by various
MKKs. COS-7 cells were cotransfected with flag-tagged p38, p38 mutants, or ERK2 with MKK3b, MKK6b, MEK1, or control empty vector. The kinase activity of p38, p38 mutants, or ERK2 isolated by immunoprecipitation was measured using MBP as substrate. Quantitation of MBP phosphorylation was done by PhosphorImager. The quantity of flag-tagged protein in the immunoprecipitate was analyzed by Western blotting using the M2 antibody, as shown in B.

FIG. 4. Activation of p38 and p38 mutants by various extracellular stimuli.
CHO-K1 cells were transiently transfected with flagtagged p38 or its mutants or ERK2. 36 h after transfection, cells were serum-starved for 8 h and then treated with sorbitol (0.4 M), epidermal growth factor (EGF; 10 nM) plus insulin (0.25 units/ml), or 20% fetal calf serum (FCS) for 30 min or with UV light (50 J/m 2 ), followed by a 30-min incubation at 37°C. The kinase activity of p38, p38 mutants, or ERK2 isolated by immunoprecipitation was measured using MBP as substrate. The flag-tagged protein in the immunoprecipitate was quantitated by Western blotting using the M2 antibody, as shown in B. Modification of L12 in p38 to an ERK-like sequence does not result in p38 activation by signals that activate ERK.

DISCUSSION
Although all MAPKs are regulated by dual phosphorylation on adjacent Thr and Tyr residues, each has its distinct set of upstream activators. Whereas the members of this enzyme family have Ͼ40% overall identity, each displays very distinct substrate specificities. A structural basis for these upstream and downstream specificities has not been currently determined. Here we have utilized p38 and ERK2 to investigate structural features that may control recognition by upstream activators or may determine substrate specificity. Using recombinant DNA techniques, we have modified p38 in the phosphorylation lip L12, the locus of major differences between p38 and ERK2, to an ERK-like structure. This permitted us to evaluate the influence of such modifications on the activation and function of p38 as well as on its ability to undergo autophosphorylation.
Since this structural domain contains the dual phosphorylation sites, the target of the upstream MAPK kinase, this region of MAPK must interact with the activating kinase. Differences in the L12 structure of the known MAPKs suggested that the Xaa residue in the dual phosphorylation motif may play a role in determining the specificity of interactions with MKK (34,35). Our data indicate that neither the Xaa residue in the Thr-Xaa-Tyr motif in p38 nor the other residues present in L12 are critical in controlling p38 activation regardless of whether this is tested in fully in vitro experiments or in cells.
After submission of this paper, two other studies provided data supporting this contention. Brunet and Pouyssegur (31) expressed a p38-ERK chimeric protein that demonstrated that the amino terminus of p38 plays a predominant role in determining interactions with the upstream activator and that L12 has no role in this event. Robinson et al. (43) showed that Xaa in the dual phosphorylation motif of ERK does not have any role in the upstream kinase selectivity and that the length of the lip of ERK2 has only a small effect on the ability of MEK1 to phosphorylate ERK2. Therefore, although amino acid residues in the phosphorylation lip interact with MKK, the MKK specificity is determined by a p38-MKK interaction involving other domain(s). These may include portions of the amino terminus of p38 (31).
Information from the three-dimensional structure of p38 suggests that a surface groove of p38 is likely to be in contact with protein substrates (24,25). This groove is composed of several helices and the carboxyl-terminal domain. Based on a cAMP-activated protein kinase-peptide substrate model, the sequence amino-terminal to the phosphorylation site of the substrate binds in the surface groove of the kinase (44). Although L12 is not a part of this surface groove, residues 170 -178 in L12 were found to occupy the peptide-binding groove in unphosphorylated p38 (24). Our data have demonstrated that the changes in the L12 structure result in a change in substrate specificity. Each MAPK is specific for proline in the Pϩ1 site. The Pϩ1 pocket (Val 183 -Arg 189 ) is conserved in all MAPKs and is adjacent to L12. It is possible that the phosphorylation lip may be a part of the substrate-binding pocket in the phosphorylated form of p38 or may contribute to the overall topology of the substrate-binding groove. Analysis of the primary sequence around the phosphorylation site of the substrates we used in the experiments reveals that the changes in the substrate specificity of the p38 mutants are dependent on the primary amino acid sequence around the phosphorylation site. The substrates that contain the Pro-Xaa-Ser/Thr-Pro sequence (ELK1, c-Myc, and MBP) are phosphorylated better by the p38(6ϩ) and p38(E) mutants, while ATF2 and PHAS-1, with a different sequence adjacent to the phosphorylation site, are recognized differently by the p38 mutants. These data suggest that the L12 structure can influence the recognition of primary amino acid sequences by p38. Introducing six amino acids from the ERK sequence or incorporating a Thr-Glu-Tyr dual phosphorylation site into p38 produced an enhanced activity toward the consensus ERK phosphorylation site Pro-Xaa-Ser/Thr-Pro. But completely converting L12 of p38 to that contained in ERK2 does not change the substrate selection of p38. These data suggest that interaction of the lip with other domains may occur and that substrate recognition determinants may act in concert to control substrate specificities. Nevertheless, our data provide evidence for the first time supporting the contention that the structure of the phosphorylation lip is a determinant of substrate specificity.
We suggest that a physiological substrate of p38 should be a preferred substrate for unmodified p38, but not for the L12 mutants. Therefore, such p38 mutants may be useful tools to test the specificity of suspected p38 substrates. For example, the stronger phosphorylation of ATF2 by wild-type p38 than by p38 mutants suggests that there is a specific interaction between this protein and p38.
In vitro autophosphorylation has been studied in the ERK group of MAPKs, where it was shown to be a unimolecular reaction occurring primarily on tyrosine (45)(46)(47)(48)(49)(50)(51)(52). From the results of crystallographic studies, it is clear that the Thr residue in the Thr-Glu-Tyr phosphorylation site of ERK is on the surface, while the Tyr residue is buried in a large hydrophobic pocket. The Tyr residue is located near a putative catalytic base, Asp 147 (Asp 150 in p38), and ␥-phosphate of bound ATP. This may account for the ability of ERK to catalyze its efficient autophosphorylation. The shortened length of L12 in p38 results in the Thr-Gly-Tyr dual phosphorylation site being present on the surface of the protein. This may cause a decrease in the flexibility of the loop that prevents the conformation change required to allow the Tyr residue to be accessed.
The insufficient activation of p38 by autophosphorylation suggests that the phosphorylated amino acids may not reflect phosphorylations at all critical regulatory sites; this is the case for ERK (46). Although this study does not provide a physiological role for autophosphorylation, we can speculate that the autophosphorylation may contribute to the activation of MAPK in cells. Although it is well known that dual specificity kinases are responsible for the activation of MAPKs (34,53), it is still possible that a serine/threonine kinase can partially activate MAPK when there is a prior autophosphorylation on tyrosine.
In summary, our data demonstrate that although the sequences in loop 12 account for major differences between p38 and ERK2, this phosphorylation lip does not appear to direct MKK specificity. A change in the length of L12 affects the structure of p38 in a way that alters the level of autophosphorylation. In addition, modifications of the lip sequence lead to specific changes in substrate selectivity. These data suggest that the phosphorylation lip may be involved in p38-substrate interaction. This finding would not have been predicted from what is known about the cyclic AMP-dependent protein kinasesubstrate interaction (44). How the phosphorylation lip is involved in substrate recognition awaits further investigation.