Research ArticleStructural Biology

A Distinct Interaction Mode Revealed by the Crystal Structure of the Kinase p38α with the MAPK Binding Domain of the Phosphatase MKP5

See allHide authors and affiliations

Science Signaling  20 Dec 2011:
Vol. 4, Issue 204, pp. ra88
DOI: 10.1126/scisignal.2002241

Abstract

The mitogen-activated protein kinase (MAPK) cascades play a pivotal role in a myriad of cellular functions. The specificity and efficiency of MAPK signaling are controlled by docking interactions between MAPKs and their cognate proteins. Many MAPK-interacting partners, including substrates, MAPK kinases, phosphatases, and scaffolding proteins, have linear sequence motifs that mediate the interaction with the common docking site on MAPKs. We report the crystal structure of p38α in complex with the MAPK binding domain (KBD) from MAPK phosphatase 5 (MKP5) at 2.7 Å resolution. In contrast to the well-known docking mode, the KBD binds p38α in a bipartite manner, in which two distinct helical regions of KBD engage the p38α docking site, which is situated on the back of the p38α active site. We also determined the crystal structure of the KBD of MKP7, which closely resembles the MKP5 KBD, suggesting that the mechanism of molecular recognition by the KBD of MKP5 is conserved in the cytoplasmic p38- and c-Jun N-terminal kinase–specific MKP subgroup. This previously unknown binding mode provides new insights into how MAPKs interact with their binding partners to achieve functional specificity.

Introduction

The mitogen-activated protein kinase (MAPK) signaling pathways play an important role in cellular signal transduction (13). There are 14 MAPKs in humans, which define seven distinct MAPK signaling pathways. The basic assembly of MAPK pathways is a three-tier kinase module that establishes a sequential activation cascade: a MAP kinase kinase kinase activates a MAP kinase kinase, which in turn activates a MAP kinase. Different MAPK modules respond to distinct extracellular stimuli. The three best-characterized MAPK signaling pathways are mediated by the kinases extracellular signal–regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38. The ERK pathway is activated by various mitogens and phorbol esters, whereas the JNK and p38 pathways are stimulated mainly by environmental stress and inflammatory cytokines (2, 4, 5). After activation, each MAPK phosphorylates a distinct spectrum of substrates, which include key regulatory enzymes, cytoskeletal proteins, nuclear receptors, regulators of apoptosis, and transcription factors.

The physiological outcome of MAPK signaling depends on both the magnitude and the duration of kinase activation. Modulation of MAPK activities in response to extracellular stimuli is mainly achieved by the coordinated action of protein kinases and phosphatases. Full activation of MAPKs requires dual phosphorylation of a Thr-X-Tyr sequence located on a flexible activation loop, where X is Glu in ERKs, Pro in JNKs, and Gly in p38s. Dual-specificity kinases termed MAPK kinases (MAPKKs) catalyze the dual phosphorylation of threonine and tyrosine residues in this sequence, resulting in an increase of more than 1000-fold in MAPK activity. In contrast, down-regulation of MAPK activity can be carried out by the dephosphorylation of these residues by various serine-threonine phosphatases, tyrosine phosphatases, and dual-specificity phosphatases termed MAPK phosphatases (MKPs). That MAPK activation is catalyzed by a single kinase and its inactivation involves multiple phosphatases suggests that the regulated dephosphorylation of MAPKs may play a crucial role in controlling cellular responses to external stimuli and determining the time course, threshold for activation, and hence the physiological outcome of signaling.

The specificity in the functional interactions between MAPKs and their cognate proteins appears to be generated by a common docking (CD) site in MAPKs for interaction partners that have docking motifs or D motifs [also referred to as kinase interaction motifs (KIMs)] (68). KIMs are found in all the MAPK interacting proteins, including phosphatases, substrates, and scaffolding proteins, as well as activating kinases. There are two major elements in nearly all the KIMs: a cluster of basic residues at the N terminus and a hydrophobic motif near the C terminus. Crystal structures of mammalian MAPKs bound with the KIM sequences derived from their interacting proteins have been reported (914). These structures revealed that linear docking motifs in interacting proteins bind to a CD site on MAPKs outside the kinase active site. The docking site is composed of a highly acidic patch and a hydrophobic groove, which mediate interactions with the basic and with the ΦA-X-ΦB residues, respectively, in the KIM sequences. Both variations in the number and position of hydrophobic and basic residues within KIM sequences and amino acid composition of the docking sites on MAPKs likely contribute to the pathway specificity among MAPK modules. Similarly, docking interactions were demonstrated for the complexes of yeast MAPKs, Fus3 and Kss1, with specific peptides, indicating that the mechanism of molecular recognition by the docking site is conserved evolutionarily in MAPK networks from yeast to mammals (15).

To understand the molecular basis of p38α recognition by MAPK phosphatase 5 (MKP5), we have determined the crystal structure of p38α in complex with the MAPK binding domain (KBD) of MKP5. To our surprise, the nature of the interaction between p38α and KBD is fundamentally different from those of other MAPK complexes including MAPKs and both KIM peptides and their cognate partners. Thus, this structure reveals a new mode of docking interaction between MAPK and its cognate proteins, providing a hitherto unrecognized diversity of potential interaction partners for mediating and regulating the signal transduction by the MAPK cascades.

Results

Interaction of p38α with KBD of MKP5

MKPs are crucial inhibitors of MAPK signaling because they dephosphorylate both phosphotyrosine and phosphothreonine in the activation loop of MAPKs. The MKPs constitute a structurally distinct subfamily of 10 catalytically active enzymes within the larger family of dual-specificity protein phosphatases (DUSPs) encoded in the human genome (16). On the basis of sequence similarity, protein structure, substrate specificity, and subcellular localization, the MKP family can be further divided into three groups (17). The first group is composed of four inducible nuclear MKPs: MKP1, MKP2, PAC1, and hVH3. The second group contains three closely related cytoplasmic ERK-specific phosphatases: MKP3, MKP4, and MKPX. The third group is composed of MKP5, MKP7, and hVH5, which selectively inactivate the stress-activated MAPKs JNK and p38. All MKPs share a common structure comprising a C-terminal catalytic domain and an N-terminal KBD (18). Despite sequence divergence within KBD, a conserved KIM is present in the N-terminal region of all these MKPs. The KIM is characterized by a region rich in basic amino acids, followed by a ΦA-X-ΦB motif (where Φ is Leu, Ile, or Val), and plays an important role in MAPK substrate recognition and binding (Fig. 1A).

Fig. 1

The KBD of MKP5 forms a stable heterodimer with p38α. (A) Sequence alignment of kinase interaction motifs (KIMs), which are proposed to mediate MAPK recognition by the 10 human MKPs. On the basis of sequence similarity, protein structure, substrate specificity, and subcellular localization, the MKP family can be further divided into three groups, indicated by the colored backgrounds. Identical and similar residues are colored orange and blue, respectively. The conserved basic residues (++) and hydrophobic residues (ΦA and ΦB) in the KIM sequence are proposed to be important for MAPK recognition. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) Domain organization of human MKP5. NT, N-terminal domain; KBD, MAPK binding domain; CD, catalytic domain. (C) The KBD of MKP5 binds either unphosphorylated p38α or phosphorylated p38α. Gel filtration profiles for unphosphorylated p38α or phosphorylated p38α alone, KBDMKP5 alone, and a mixture of the two. (D) Inhibition of MKP5-catalyzed phosphorylated p38α dephosphorylation by the KBD of MKP5 (KBDMKP5). As assessed by the competitive binding assay, the KBD of MKP5 binds phosphorylated p38α with a dissociation constant of 1.01 μM to form a 1:1 complex. The assay was performed for MKP5 by varying the concentrations of both phosphorylated p38α and KBDMKP5 as indicated at 25°C.

MKP5, which belongs to the MKP family, preferentially inactivates p38 and JNK but not ERK (table S1) (19, 20). MKP5 is unique among the MKPs in that it has an additional segment of unknown function at the N terminus (Fig. 1B). Deletion of the KBD in MKP5 leads to a 400-fold increase in Km (Michaelis constant) for p38α substrate, but it has little effect (1.5-fold) on kcat, suggesting that the KBD of MKP5 is critical for p38α binding and efficient p38α dephosphorylation by MKP5 (21). To test whether isolated KBD of MKP5 and p38α could form a stable binary complex in solution, we analyzed the binding of KBD with unphosphorylated and phosphorylated p38α by size exclusion chromatography (Fig. 1C). The chromatographic profiles of unphosphorylated and phosphorylated p38α showed that both proteins eluted with a similar retention time, corresponding to a molecular mass of 43.5 kD as determined from the protein calibration curve. When 5 molar equivalents of KBD were mixed with 1 molar equivalent of p38α, a small amount of KBD was shifted to earlier fractions, forming stable complexes, and the excess amount of KBD was eluted from the size exclusion column as a monomer.

We used competitive binding assays to measure the binding affinity of the KBD of MKP5 for phosphorylated p38α. The initial rates of MKP5-catalyzed reaction were measured in the presence of various concentrations of p38α substrate and four different fixed concentrations of KBD. The initial velocity of MKP5 catalysis depended on the p38α concentration, and the KBD inhibited the MKP5 activity in a concentration-dependent manner (Fig. 1D). Because the KBD of MKP5 binds p38α tightly, this inhibition probably reflects competition for p38α binding to full-length MKP5. The data sets of initial velocities were analyzed with a kinetic equation by global fitting procedure. The binding affinity of the KBD of MKP5 for phosphorylated p38α had a Ki (inhibition constant) of 1.01 ± 0.01 μM.

Structure of p38α in complex with the KBD of MKP5

We solved the structure of the binary complex of unphosphorylated p38α bound to the KBD of MKP5 by molecular replacement (Fig. 2A). The model of the heterodimer was refined at 2.7 Å resolution to an R factor of 21.6% (Rfree of 26.0%) with good stereochemistry (Table 1). When bound to p38α, the KBD of MKP5 covers a contiguous surface area situated opposite of the p38α active site (Fig. 2B). The docking site on p38α contains the previously identified (negatively charged) CD region and hydrophobic groove between helices αD and αE and the β7-β8 reverse turn (9, 22, 23). We found that the interaction of the KBD of MKP5 with p38α mainly involved two helices, α2 and α3′, from the KBD. In the p38α-KBD complex, the conformation of KBD is similar to that of the isolated KBD of MKP5 (24), except several residues local to the binding region undergo conformational changes upon binding (fig. S1). However, the structure of p38α in the present complex has some differences from the apo-p38α structure. The p38α structure of the present heterodimer superimposed on the apo-p38α (25, 26) with a root mean square deviation (RMSD) of 1.2 Å for equivalent Cα atoms. The two lobes are rotated by 5.6° toward one another in p38α of the complex (Fig. 2C). The activation loop exhibits perhaps the largest change on binding of the KBD. In the present complex structure, many of the residues of the activation loop, including Thr180 and Tyr182, are crystallographically disordered (no electron density was visible for residues 170 to 183), suggesting that flexibility in the p38α activation loop caused by docking to the KBD could release this loop for binding to the active site of MKP5 and aid in dephosphorylation of p38α. Similar observation of enhanced disorder in the activation loop was reported previously by Chang et al. (9) and Zhou et al. (12), and they suggested that the disorder may be important for the action of the docking peptide.

Fig. 2

Overall structure of p38α in complex with the KBD of MKP5. (A) Ribbon diagram of p38α (cyan and blue) in complex with KBDMKP5 (magenta). Dotted lines indicate the disordered activation loop. (B) Surface representation of p38α in complex with KBD of MKP5 (KBDMKP5, magenta), colored by electrostatic potential. (C) Superimposition of the uncomplexed p38α (PDB ID 1P38; yellow, and activation loop in orange) and the p38α (cyan and blue) from p38α-KBDMKP5 complex with corresponding Cα atoms within the C lobe. The Cα atoms from residues 112 to 172, 202 to 242, and 269 to 309 were used to superimpose the two p38α molecules with an RMSD of 0.59 Å. (D) Surface representation of ERK2 in complex with KIM peptide of MKP3 (KIMMKP3, yellow), colored by electrostatic potential (PDB ID 2FYS).

Table 1

Data collection and refinement statistics.

View this table:

The crystal structure of ERK2 in complex with the KIM peptide from the KBD of MKP3 (KIMMKP3, residues 60 to 74) has been reported (11). Like other known docking peptide complexes, such as p38α-KIMMKK3b and p38α-KIMMEF2A, the ERK2-KIMMKP3 complex reveals that this KIM peptide binds ERK2 in an extended conformation (Fig. 2D). The ΦA-X-ΦB motif Leu71′-Pro72′-Val73′ (the residues of MKP3 or MKP5 are indicated by apostrophes after their residue numbers) in KIM of MKP3 binds to the hydrophobic groove, and the basic residues Arg64′ and Arg65′ interact with Asp316 and Asp319 in the CD site in ERK2. Compared to ERK2-KIMMKP3, the docking site in the p38α-KBDMKP5 complex overlaps with the binding surface of the MAPK that interacts with the canonical docking motif. The nature of the interaction, however, is completely different. The KBD of MKP5 binds to p38α in the opposite polypeptide direction compared to how the KIM sequence of MKP3 binds to ERK2. The ΦA and ΦB residues in the proposed KIM of MKP5 do not participate in the interaction between p38α and MKP5. Instead, the KBD interacts with p38α in a bipartite manner, such that two distinct helical regions of KBD engage the p38α docking site situated on the back of the p38α active site.

The solution structure of the KBD (residues 1 to 154) from MKP3 has been determined by nuclear magnetic resonance (27). Distinct structural differences are observed between the KBD of MKP5 and that of MKP3 (24, 27). The KIM sequence that is important for ERK2-MKP3 interaction lies within different structural elements and is located at different positions in the two structures. The residues of KIM sequence in MKP3 are located in a flexible region connecting β3 and α3, whereas their sequence equivalents (residues 199 to 213) in MKP5 are located in a helix (α3′) and following loop region, on the opposite face of the central β sheet. Thus, the large structural differences may explain why the binding of MKP5 to p38α adopts an interaction mode different from that of other KIM peptides.

Mutational analysis of the p38α-MKP5 docking interface

On the basis of the crystal structure, we found that the KBD of MKP5 buries a total surface area of ~920 Å in p38α. The KBD of MKP5 interacts with p38α through two major contact areas (Fig. 3A). The first area is composed of the helix α3′ and the following α3′-α3 loop in the KBD of MKP5 and CD domain in p38α, an acidic patch involving Asp313, Asp315, Asp316, and Glu81 (Fig. 3B). The side chain of Arg203′ in KBD forms four hydrogen bonds with the carboxylates of Asp316 and Glu81 in p38α, and the side chain of Arg204′ is hydrogen-bonded to Asp316, Tyr132, and Tyr311 of p38α. These charge-charge interactions are similar to those observed in the ERK2-KIMHePTP complex (fig. S2) (12). In addition, Ile200′ contacts the residues Phe129 and Tyr132 of p38α through hydrophobic interactions, and the side chain of Lys209′ forms a salt bridge with Asp313 of p38α. The second area involves the helix α2 of the KBD and the hydrophobic groove of p38α (Fig. 3C). The side chains of Phe180′ and Met181′ on helix α2 are nestled in a hydrophobic pocket formed by the side chains of Ile116, Cys119, Leu122, Leu130, Val158, and Cys162 and the aliphatic portions of Gln120 and His126. By comparing the current structure with previous ones (911, 1315), we find that the hydrophobic pocket, which accommodates the binding of the ΦA residue in ERK2-KIMMKP3 and other MAPK-KIM complex structures, is plugged into by two intervening residues, Phe180′ and Met181′, whereas the hydrophobic patch filled by the ΦB residue is unoccupied in the p38α-KBDMKP5 complex (Fig. 3D). Although MK2 binds to the p38α hydrophobic groove in the same polypeptide direction as KBDMKP5, the docking mode of MK2 is more similar to other known KIM peptides binding in the opposite direction, such as KIMMEF2A and KIMMKK3b, than KBDMKP5 (14) (fig. S3). Thus, the docking of the KBD of MKP5 to p38α represents a previously unrecognized docking class.

Fig. 3

A new docking interaction mode between p38α and the KBD of MKP5. (A) Surface representation of p38α in complex with the KBD of MKP5 (KBDMKP5, magenta), colored by electrostatic potential. Right, close-up view of interactions between KBDMKP5 and C lobe of p38α. The side chains of residues in helix α3′ and α2 of KBDMKP5, involved in p38α binding, are shown in stick representation. (B and C) Detailed interactions between KBDMKP5 (magenta) and the C lobe of p38α (blue). Ion-pair and hydrogen-bonding interactions are presented by black dashed lines. (D) Superimposition between p38α in complex with KBDMKP5 (magenta) and ERK2 in complex with KIMMKP3 (yellow). For clarity, only residues Phe180-Lys209 of KBDMKP5 are shown. The overlaid MAPK proteins are colored gray. Right, cartoon of MAPK docking surface comparing p38α-KBDMKP5 complex with ERK2-KIMMKP3 complex. (E) Comparison of kcat/Km values of dephosphorylation reactions of phosphorylated p38α catalyzed by MKP5 [wild type (WT) and the indicated mutants].

To validate the binding mode between KBDMKP5 and p38α, we first examined the effects of KBDMKP5-p38α interaction on the p38α activity toward both peptide substrate [epidermal growth factor receptor (EGFR) peptide] and KIM-containing substrate (ATF2Δ109). The initial velocity of the p38α-catalyzed reaction depends on the concentration of KBDMKP5 when ATF2Δ109 is used as substrate (fig. S4). As expected, the enzyme activity of p38α for ATF2Δ109 decreased with increases in the concentration of KBDMKP5, whereas KBDMKP5 had no effect on the p38α activity for EGFR peptide substrate because the EGFR peptide does not have a KIM domain. Using gel filtration assays, we also performed competitive binding experiments and found that KBDMKP5 disrupted the p38α-ATF2Δ109 and p38α-MKK6 complexes (fig. S5). Together, our biochemical results suggest that the physical association of KBDMKP5 with p38α may generate a steric block that prevents binding of the KIM-containing partners.

To assess the importance of the aforementioned interactions, we generated a series of point mutations in the KBD of MKP5 and examined their effect on the MKP5-catalyzed p38α dephosphorylation (Fig. 3E). When the hydrophobic residues Phe180 and Met181 on the α2 of the KBD were individually replaced by Ala, the catalytic efficiencies (kcat/Km) of the F180A and M181A mutants for p38α dephosphorylation decreased four- and twofold, respectively. Compared to wild-type MKP5, replacement of both Phe180 and Met181 with an Ala led to an eightfold drop in the kcat/Km value. However, mutating these hydrophobic residues to charged Asp (F180D, M181D, or F180D and M181D) led to marked enzymatic inactivation (8- to 30-fold) (table S2). Similarly, when the two basic residues, Arg203 and Arg204, on the α3′ of KBD were individually replaced by an Ala, the catalytic efficiencies (kcat/Km) of the mutants were decreased more than 10-fold, and substitution of both these Arg residues with Ala resulted in a further decrease (~150-fold) in kcat/Km, comparable to that observed in a mutant lacking the whole docking domain (MKP5ΔN319) (Fig. 3E and table S2). In addition, forms of MKP5 with mutations in crucial residues on the KBD interface, I200A and K209A, had decreased catalytic efficiencies. Mutations in the KBD of MKP5 caused an increase in Km but had a modest effect on kcat (table S2), supporting the notion that the major function of the KBD in MKP5 is to increase the effective concentration of the pT-X-pY motif in the vicinity of the active site of MKP5 for dephosphorylation. We also made mutations of two hydrophobic residues corresponding to ΦA and ΦB in the KIM sequence, I210A and V212A, as negative controls, and assessed their effect on catalytic function (Fig. 3E and table S2). As expected, these two mutants did not show altered catalytic efficiencies. Collectively, these data indicated that both the hydrophobic and the charged residues in the KBD of MKP5 are required for its high-affinity binding to the MAPK p38α.

Comparison of docking interactions among MKPs

MKP7 is specific for p38 and JNK but not ERK, and its unique structural feature is that it contains an additional sequence at the C terminus that is rich in prolines, glutamates, serines, and threonines (PEST) (Fig. 4A) (28, 29). Compared with MKP5, however, deletion of the N-terminal domain had only a modest effect on the phosphatase activity of MKP7 (~14-fold in kcat/Km) (table S3). Thus, it would be of interest to know whether the KBD of MKP7 can form a tight physical complex with p38α. To address this issue, we analyzed the interaction of unphosphorylated p38α with the KBD of MKP7 (residues 5 to 138) by gel filtration analysis and competitive binding assays. Gel filtration analysis indicated that p38α formed a stable heterodimer with the KBD of MKP7 in solution (Fig. 4B). Consistent with MKP7 binding to p38α through its KBD, the enzymatic activity of MKP5 toward p38α substrate was inhibited in the presence of isolated MKP7 KBD (Fig. 4C). The binding affinity of the MKP7 KBD for phosphorylated p38α had a Ki of 0.74 ± 0.13 μM.

Fig. 4

KBD of MKP7 forms a stable heterodimer with p38α. (A) Domain organization of human MKP7. KBD, MAPK binding domain; CD, catalytic domain; PEST, C-terminal sequence rich in prolines, glutamates, serines, and threonines. (B) The KBD of MKP7 binds unphosphorylated p38α. Gel filtration profiles of unphosphorylated p38α alone, KBDMKP7 alone, and a mixture of the two. (C) Inhibition of MKP5-catalyzed phosphorylated p38α dephosphorylation by KBD of MKP7 (KBDMKP7). As assessed by the competitive binding assay, the KBD of MKP7 binds phosphorylated p38α with a dissociation constant of 0.74 μM to form a 1:1 complex. The assay was performed for MKP5 by varying the concentrations of both phosphorylated p38α and KBDMKP7 as indicated at 25°C.

To gain a structural understanding of the interaction mode between p38α and MKP7, we crystallized the MKP7 KBD (residues 5 to 138) and determined its three-dimensional structure at 2.7 Å resolution (Fig. 5A). There were four molecules of MKP7 KBD in each asymmetric unit; these four molecules exhibited identical structure and were superimposable with one another (fig. S6). The compact structure comprised a central β sheet of five parallel β strands, which was surrounded by three α helices on the other sides. In addition, there was a small, two-stranded β sheet involving residues 41 to 43 and 136 to 138 at the C terminus of the domain.

Fig. 5

Structural differences among the KBDs of MKP3, MKP5, and MKP7. (A) Ribbon diagram of the KBD of MKP7. (B) Structural overlay of KBDMKP3 (yellow; PDB ID 1HZM) and KBDMKP7 (orange). Residues of KIMMKP3 involved in contacts with ERK2 (Arg64, Arg65, Leu71, and Val73) are shown as gray sticks. The equivalent residues in MKP7 (Arg56, Arg57, Val63, and Ile65) are shown as cyan sticks. (C) Structural overlay of KBDMKP5 (magenta; PDB ID 2OUC) and KBDMKP7 (orange). Residues of KBDMKP5 involved in contacts with p38α (Phe180, Met181, Ile200, Arg203, Arg204, and Lys209) are shown as gray sticks. The equivalent residues in MKP7 (Phe34, Met35, Leu53, Arg56, Arg57, and Lys62) are shown as cyan sticks. (D) Structure-based sequences of MKP3 (Gln42-Val72 in KBD), MKP5 (Phe180-Val212 in KBD), and MKP7 (Phe34-Ile65 in KBD). Residues of MKP3 and MKP5 involved in hydrophobic and hydrophilic contacts with MAPKs are colored in cyan and blue, respectively. Potential residues of MKP7 involved in p38α binding are underlined. (E) Comparison of kcat/Km values of MKP7 (WT and the indicated mutants)–catalyzed phosphorylated p38α dephosphorylation reaction. (F) Superimposition of apo KBDMKP7 (orange) and KBDMKP5 in the p38α-KBDMKP5 complex (color-coded as in Fig. 2A). (G) Amino acid sequence alignment of KBD of cytoplasmic p38- and JNK-specific MKPs including MKP5, MKP7, and hVH5. Identical and similar residues are shaded orange and purple, respectively. Residues of KBDMKP5 involved in hydrophobic and hydrophilic contacts with p38α are marked with cyan and blue triangles, respectively.

The KBD of MKP7 shares 35% and 32% sequence identity with that of MKP3 and MKP5, respectively. Although the KBDs of MKP7 and MKP3 have greater sequence similarity, differences in KBD structures do not fall strictly on MKP7 compared to MKP3 and MKP5 lines. The KBD structure of MKP7 is more similar to MKP5 (RMSD, 1.5 Å) than to MKP3 (RMSD, 8.0 Å) (Fig. 5, B and C). There are differences in many of the surface loops and helices between the KBD structures of MKP3 and MKP7 (Fig. 5B). The positively charged residues and ΦA-X-ΦB motif that are important for MAPK binding are located in different positions in the two structures. In contrast to MKP3, the KBD structure of MKP7 (residues 5 to 138) closely resembles that of MKP5 (Fig. 5C). In particular, the key residues on the helices α2 and α3′ are well conserved between these two domains, suggesting a possible similarity in the interaction mode between MKP7 and p38α (Fig. 5D). To further test this hypothesis, we generated forms of the MKP7 KBD bearing mutations corresponding to those we made on MKP5 on the basis of sequence and structural alignment and examined their effects on MKP7 phosphatase activity (Fig. 5E and table S3). As expected, the V63A and I65A MKP7 mutants showed little or no difference in phosphatase activity, whereas the other mutants showed reduced specific activities of MKP7. Thus, MKP7 binds p38α in a docking mode similar to that of MKP5, and the interaction model can be generated by superimposition of KBDMKP7 to the corresponding domain of MKP5 in the complex (Fig. 5F). Given the relatively higher sequence identity (40%) between MKP7 and hVH5 (also known as M3 or M6) (30), and because residues involved in the protein-protein interaction are highly conserved (Fig. 5G), it is reasonable to postulate that the new binding mode is conserved in this cytoplasmic and p38- and JNK-specific MKP subgroup (MKP5, MKP7, and hVH5).

Discussion

The p38α MAPK signaling pathway plays an important role in inflammation and many other physiological processes. It has been the subject of extensive efforts in both basic research and drug discovery to make specific inhibitors of p38α for the treatment of inflammatory diseases such as psoriasis, rheumatoid arthritis, and chronic obstructive pulmonary disease (3133). Many adenosine 5′-triphosphate (ATP)–competitive small-molecule inhibitors directed against p38α have been described and tested (34). Although some of these inhibitors show anti-inflammatory effects in animal models, the side effects have limited their clinical utility. The repeated failure of the current p38α inhibitors in clinical trials raises the need for additional modes of specifically interfering with kinase activity. MAPK docking sites are emerging as possible targets for drug design, because MAPKs use docking interactions to link module components and bind substrates. Docking site–directed inhibitors could block the function of a specific MAPK pathway by inhibiting both MAPKK binding and substrate binding. This strategy has opened a new way for the development of protein kinase inhibitors targeting substrate-specific docking sites, rather than the highly conserved ATP binding sites.

Here, we report the crystal structure of the MAPK p38α in complex with the KBD of the phosphatase MKP5. Our structure reveals that the binding mode of p38α to its cognate phosphatase is distinct from that observed for the recognition of ERK2 by the KBD of MKP3. The KBD of MKP5 also binds in the p38α docking groove, but the major difference with the ERK2-KIMMKP3 complex is the reversed direction of the peptide backbones of MKP5 compared to MKP3. Accordingly, although both p38α and ERK2 use a cluster of highly conserved basic residues to bind to their respective KBDs, the reverse orientations in each case imply the involvement of a whole new set of hydrophobic residues. Indeed, although such hydrophobic residues are located C-terminal to the cluster of basic residues in MKP3, they are found N-terminal to the corresponding cluster in MKP5. Additionally, the region within the KBD of MKP3 that directly docks onto ERK2 adopts an extended conformation, whereas the regions within the KBD of MKP5 in contact with p38α adopt α-helical conformations, resulting in a different interacting mode. We also determined the crystal structure of the KBD of MKP7, another member of the cytoplasmic and p38- and JNK-specific MKP subgroup. The KBD structure of MKP7 closely resembles that of the MKP5 KBD, suggesting that the mechanism of molecular recognition by the KBD of MKP5 is conserved in this MKP subgroup. Biochemical data provide further support to this notion.

In summary, our finding sheds new light on the molecular basis of MAPK-MKP interactions, helps to better understand the regulation of MAPKs, and may provide invaluable information to exploit highly selective p38α inhibitors for innate and adaptive immune diseases.

Materials and Methods

Protein expression and purification

Expression and purification of N-terminally (His)6-tagged unphosphorylated p38α was performed as previously described (21). For the KBD of MKP5 (KBDMKP5) construct, the complementary DNA (cDNA) encoding residues 139 to 287 of the human MKP5 was subcloned into the expression vector pET21b (Novagen) with the restriction sites Nde I and Xho I. The recombinant protein carried a hexahistidine tag at the C terminus. The plasmid was transformed into Escherichia coli strain BL21, and cultures were grown in LB medium at 37°C to an OD600 (optical density at 600 nm) of 0.8 and induced with 0.2 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 20°C for 14 hours. Cells were collected by centrifugation and lysed by sonication. The protein was first isolated by means of an affinity Ni-NTA column (Qiagen) and eluted with 300 mM imidazole. The eluted sample was subsequently subjected to a cation exchange SOURCE 15S column (GE Healthcare) and eluted with a NaCl gradient up to 1 M. The p38α-KBDMKP5 complex was prepared by mixing the purified proteins with KBDMKP5 present in slightly molar excess (1:1.5 molar ratio); the p38α-KBDMKP5 complex was purified by size exclusion chromatography using a Superdex 200 (S200, GE Healthcare) column equilibrated in a buffer containing 10 mM Hepes (pH 7.5), 150 mM NaCl, and 2 mM dithiothreitol (DTT). Purified protein was flash-frozen in liquid nitrogen and stored at −80°C.

For the KBD of MKP7 (KBDMKP7) construct, the cDNA encoding residues 5 to 138 of the human MKP7 was subcloned into the expression vector pET21b (Novagen) with the restriction sites Nde I and Xho I. The C-terminally (His)6-tagged KBDMKP7 was overexpressed in E. coli BL21(DE3) with 0.2 mM IPTG induction at 20°C for 14 hours. The KBDMKP7 was first purified by an Ni-NTA column and anion exchange SOURCE 15Q column and was further purified with a Superdex 200 column equilibrated in 10 mM Hepes (pH 7.5), 150 mM NaCl, and 2 mM DTT. Purified protein was flash-frozen in liquid nitrogen and stored at −80°C.

Protein crystallization and data collection

Crystals of p38α-KBDMKP5 complex were obtained at 22°C in 5 days by means of the hanging-drop technique using equal volumes of protein solution and reservoir buffer [100 mM tris (pH 7.5 to 8.0), 8 to 12% (w/v) polyethylene glycol 3350, and 8 to 12% (w/v) sucrose]. Fresh crystals were quickly transferred to a cryoprotectant buffer containing 20% ethylene glycol before snap-freezing in liquid nitrogen before data collection. Diffraction data were collected at the Shanghai Synchrotron Radiation Facility (SSRF) BL17U beamline and processed with HKL2000 (35). The crystals belonged to space group P41212 with unit cell dimensions of a = b = 72.4 Å, c = 226.1 Å, and α = β = γ = 90°. There was one 1:1 p38α-KBDMKP5 complex in the asymmetric unit.

Crystals of KBDMKP7 were obtained at 22°C in 2 days with the hanging-drop technique using equal volumes of protein solution and reservoir buffer [100 mM bis-tris (pH 5.7 to 6.2), 24 to 28% (w/v) polyethylene glycol 3350, and 0.1 M ammonium sulfate]. Microseeding was used to produce single crystals. Diffraction data were collected on a home lab x-ray system and processed with HKL2000. The crystals belonged to space group P1 with unit cell dimensions of a = 40.5 Å, b = 47.5 Å, c = 64.5 Å, and α = 91.2°, β = 97.3°, γ = 96.8°. There were four molecules of KBDMKP7 in the asymmetric unit.

Structure determination and refinement

The crystal structure of p38α in complex with KBD of MKP5 was determined by molecular replacement using Phaser (36) with the search models of unphosphorylated p38α [Protein Data Bank (PDB) ID 1P38] and apo KBD of MKP5 (PDB ID 2OUC). The crystal structure of KBDMKP7 was solved by molecular replacement using Phaser with the search model of apo KBD of MKP5 (PDB ID 2OUC). The initial solution of either structure was subjected to multiple cycles of coordinates, simulated annealing, individual B factors, and TLS (translation-libration-screw) groups with PHENIX (37) and to manual refinement with Coot (38). The crystallographic R factor and free R factor were finally reduced to 21.6% and 26.0%, respectively, at 2.7 Å resolution for the p38α-KBDMKP5 complex and 23.9% and 28.8% for KBDMKP7 at 2.67 Å resolution. PROCHECK (39) indicated that none of the residues in either structure is in the disallowed region of the Ramachandran plot. Details of the data collection and refinement statistics for the complex are summarized in Table 1. All figures displaying the protein structures were prepared with PyMOL software from Delano Scientific (http://www.pymol.org).

Generation of MKP5 and MKP7 mutants and phosphatase activity assays

The mutants were generated by polymerase chain reaction (PCR) oligonucleotide site-directed mutagenesis, and all mutations were confirmed by DNA sequencing. The MKP5 mutants were expressed and purified as previously described for the wild-type protein (21). Because we could not obtain soluble protein of full-length MKP7 expressed in bacteria, we subcloned into the expression vector pET21b the cDNA encoding residues 5 to 303 of the human MKP7 containing the KBD and catalytic domains but without the C-terminal PEST-like sequence (Fig. 4A). Deletion of the C-terminal PEST-like sequence has no effect on the binding ability and phosphatase activity of MKP7 toward MAPKs (28). Therefore, to access the binding mode between KBD of MKP7 and p38α, we generated mutations in MKP7 (5 to 303) corresponding to those on MKP5 and examined their effects on phosphatase activity toward bisphosphorylated p38α. The C-terminally His6-tagged MKP7 (5 to 303) and its mutants were expressed in E. coli BL21(DE3) and purified using standard procedures and an Ni-NTA column, followed by a cation exchange SOURCE 15S column. All proteins prepared were examined by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) analysis, and the contents of each were judged at least 90% pure. The purified protein was combined with 20% glycerol and stored at −80°C. Protein concentrations were determined spectrophotometrically using theoretical molar extinction coefficients at 280 nm.

MKP5 and MKP7 phosphatase activity toward bisphosphorylated p38α was measured by a continuous spectrophotometric assay (21, 40). This assay incorporates a coupled enzyme system, which uses purine nucleoside phosphorylase and its chromogenic substrate 7-methyl-6-thioguanosine (MESG) for the quantification of inorganic phosphate produced in the phosphatase reaction (41). All experiments were carried out at 25°C in 1.8-ml reaction mixture containing 50 mM Mops (pH 7.0), 100 mM NaCl, 0.1 mM EDTA, 100 μM MESG, and purine nucleoside phosphorylase (0.1 mg/ml). The reactions were initiated by the addition of phosphatase. The time courses of resulting absorbance change at 360 nm were recorded on a PerkinElmer LAMBDA 45 spectrophotometer equipped with a magnetic stirrer in the cuvette holder. Initial rates were determined from the linear slope of progress curves obtained, and the experimental data were analyzed with a nonlinear regression analysis program. Quantitation of phosphate release was determined with the extinction coefficient of 11,200 M−1 cm−1 for the phosphate-dependent reaction at 360 nm at pH 7.0 (42). The concentration of MESG was determined at 331 nm, using a molar extinction coefficient of 32,000 M−1 cm−1. MKP5 and MKP7 (as a control) and mutants were assayed with the same procedure mentioned above.

p38α and KBD interactions

Size exclusion chromatography was performed to determine the apparent molecular weight of p38α alone, KBD of MKP5 or MKP7 alone, and mixture of the two (with KBD present in roughly fivefold molecular excess). Each sample (0.1 ml) was loaded onto a Superdex 200 column, preequilibrated in the buffer [50 mM Hepes (pH 7.5), 150 mM NaCl, and 2 mM DTT], and run at 4°C using a flow rate of 0.5 ml/min. The column was calibrated with the molecular weight markers.

Kinetic analysis of competitive binding assays to determine the binding affinity of KBD for phosphorylated p38α

Because KBDMKP5 binds p38α tightly, we believe that this domain may compete with full-length MKP5 for p38α binding. The postulated scheme is shown below.S+EKSESKcatE+PSIIKiwhere S represents substrate bisphosphorylated p38α, and E and I represent the enzyme (full-length MKP5) and inhibitor (the KBD domain of MKP5), respectively. KS and Ki are the dissociation constants for the phosphorylated p38α to full-length MKP5 and its KBD domain, respectively, and kcat is the turnover number.

The velocity equation is given byv=kcat[E][S]KS+[S]where [E] is the concentration of enzyme (full-length MKP5) and [S] represents the concentration of free substrate available for the interaction with enzyme. [S] can be calculated for any given total inhibitor and total substrate concentrations, [S]t and [I]t, as shown below.

The expression for Ki isKi=[S][I][SI]

The mass balance equations for S and I are[S]t=[S]+[SI]+[ES][I]t=[I]+[SI]

The mass balance equations assume that the concentration of S complexed with the enzyme is negligible.

Substituting for [I] and [SI] in the expression for Ki:Ki=[S]([I]t([S]t[S]))[S]t[S]=[S]2+[S][I]t[S][S]t[S]t[S][S]2+([I]t[S]t+KS)[S]Ki[S]t=0[S]=([I]t[S]t+Ki)([I]t[S]t+Ki)2+4Ki[S]t2

The negative solution of the square root can be ignored because [S] must be a positive number.[S]=([I]t[S]t+Ki)2+4Ki[S]t([I]t[S]t+Ki)2

The expression for [S] can then be substituted into the velocity equation:v=kcat×[E]×{([I]t[S]t+Ki)2+4Ki[S]t([I]t[S]t+Ki)}2Ks+([I]t[S]t+Ki)2+4Ki[S]t([I]t[S]t+Ki)(1)

We carried out the competitive binding assays and determined the binding affinities of KBDMKP5 and KBDMKP7 for phosphorylated p38α, respectively. The initial rates of MKP5-catalyzed reactions were measured in the presence of various concentrations of inhibitor KBD and substrate p38α. Figures 1D and 4C respectively show the dependence of the initial velocity on p38α concentration at four different fixed KBDMKP5 and KBDMKP7 concentrations, which revealed that both KBD domains inhibit the MKP5 activity in a concentration-dependent manner. The data sets of initial velocities were analyzed with Eq. 1 by a global fitting procedure. The kinetic parameters for KBDMKP5 were determined to be KS = 0.065 ± 0.009 μM, Ki = 1.01 ± 0.01 μM, and kcat = 0.585 ± 0.014 s−1, and those for KBDMKP7 were KS = 0.070 ± 0.001 μM, Ki = 0.74 ± 0.13 μM, and kcat = 0.600 ± 0.014 s−1, respectively.

p38α kinase activity assays

The kinase activity of phosphorylated p38α was measured spectrophotometrically with ATF2Δ109 or EGFR peptide as substrate (43). This assay couples the production of adenosine 5′-diphosphate (ADP) with the oxidation of NADH [reduced form of nicotinamide adenine dinucleotide (NAD+)] by pyruvate kinase (PK) and lactate dehydrogenase (LDH). The standard assay was carried out at 25°C in 1.8-ml reaction mixture containing 50 mM Mops (pH 7.0), 100 mM NaCl, 0.1 mM EDTA, 10 mM MgCl2, 0.2 mM NADH, 1.0 mM phosphoenolpyruvate, LDH (20 U/ml), and PK (15 U/ml), 1 mM ATP, 30 or 12 nM phosphorylated p38α, and 4 μM ATF2Δ109 or 100 μM EGFR peptide. The progress of the reaction was monitored continuously by following the formation of NAD+ at 340 nm on a PerkinElmer LAMBDA 45 spectrophotometer equipped with a magnetic stirrer in the cuvette holder. The concentration of ADP produced in the phosphorylated p38α–catalyzed reaction was determined using an extinction coefficient for NADH of 6220 cm−1 M−1 at 340 nm.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/4/204/ra88/DC1

Fig. S1. Comparison of the uncomplexed KBDMKP5 with KBDMKP5 from the p38α-KBDMKP5 complex.

Fig. S2. Comparison of the electrostatic interactions in the CD domains of p38α and ERK2.

Fig. S3. Comparison of the complex structures of p38α-KBDMKP5 and p38α-MK2.

Fig. S4. Effect of KBDMKP5 on the kinase activity of phosphorylated p38α toward different substrates.

Fig. S5. Effect of KBDMKP5 on the interaction of p38α with different cognate partners.

Fig. S6. Superimposition of the four molecules within the asymmetric unit of the KBDMKP7 structure.

Table S1. Kinetic parameters of full-length MKP5 and its catalytic domain toward phosphorylated ERK2 or p38α.

Table S2. Kinetic parameters of full-length MKP5 and its mutants with phosphorylated p38α as substrate.

Table S3. Kinetic parameters of MKP7 (residues 5 to 303) and its mutants with phosphorylated p38α as substrate.

References

References and Notes

Acknowledgments: We thank S. C. Lin for critical reading of the manuscript. We also thank the staff at the SSRF for helping with the collection of synchrotron x-ray data. Funding: This work was supported in part by grants 2006CB503900 and 2007CB914400 from the Ministry of Science and Technology of China and grants 30425005 and 30770476 from National Natural Science Foundation of China. Author contributions: Y.-Y.Z. performed the experiments. Y.-Y.Z., J.-W.W., and Z.-X.W. analyzed the data. Z.-X.W. prepared the manuscript. Competing interests: The authors declare that they have no conflict of interest. Accession numbers: Atomic coordinates and structure factors for p38α-KBDMKP5 complex and KBDMKP7 have been deposited with the Protein Data Bank under accession codes 3TG1 and 3TG3.
View Abstract

Navigate This Article