Biochemical Journal

Research article

The Ser186 phospho-acceptor site within ERK4 is essential for its ability to interact with and activate PRAK/MK5

Maria Perander, Espen Åberg, Bjarne Johansen, Bo Dreyer, Ingrid J. Guldvik, Heidi Outzen, Stephen M. Keyse, Ole-Morten Seternes

Abstract

ERK (extracellular-signal-regulated kinase) 4 [MAPK (mitogen-activated protein kinase) 4] and ERK3 (MAPK6) are atypical MAPKs. One major difference between these proteins and the classical MAPKs is substitution of the conserved T-X-Y motif within the activation loop by a single phospho-acceptor site within an S-E-G motif. In the present study we report that Ser186 of the S-E-G motif in ERK4 is phosphorylated in vivo. Kinase-dead ERK4 is also phosphorylated on Ser186, indicating that an ERK4 kinase, rather than autophosphorylation, is responsible. Co-expression of MK5 [MAPK-activated protein kinase 5; also known as PRAK (p38-regulated/activated kinase)], a physiological target of ERK4, increases phosphorylation of Ser186. This is not dependent on MK5 activity, but does require interaction between ERK4 and MK5 suggesting that MK5 binding either prevents ERK4 dephosphorylation or facilitates ERK4 kinase activity. ERK4 mutants in which Ser186 is replaced with either an alanine residue or a phospho-mimetic residue (glutamate) are unable to activate MK5 and Ser186 is also required for cytoplasmic anchoring of MK5. Both defects seem to reflect an impaired ability of the ERK4 mutants to interact with MK5. We find that there are at least two endogenous pools of wild-type ERK4. One form exhibits reduced mobility when analysed using SDS/PAGE. This is due to MK5-dependent phosphorylation and only this retarded ERK4 species is both phosphorylated on Ser186 and co-immunoprecipitates with wild-type MK5. We conclude that binding between ERK4 and MK5 facilitates phosphorylation of Ser186 and stabilization of the ERK4–MK5 complex. This results in phosphorylation and activation of MK5, which in turn phosphorylates ERK4 on sites other than Ser186 resulting in the observed mobility shift.

  • extracellular-signal-regulated kinase (ERK)
  • mitogenactivated protein kinase (MAPK)
  • MAPK-activated protein kinase 5 (MK5)
  • phosphorylation
  • protein–protein interaction

INTRODUCTION

The MAPKs (mitogen-activated protein kinases) are critical components of signal transduction pathways involved in the regulation of important cellular processes, such as proliferation, differentiation, migration and apoptosis. They belong to the large family of proline-directed kinases that phosphorylate substrates on a serine/threonine residue immediately preceding a proline residue. MAPKs are activated through a dual phosphorylation of both a threonine (T) and a tyrosine (Y) in a conserved T-X-Y motif found in the activation loop of kinase subdomain VIII. The phosphorylation of the T-X-Y motif is catalysed by a dual-specificity MAP2K (MAPK kinase), which is phosphorylated and activated by a MAP3K (MAPK kinase kinase). The module of a MAPK, MAP2K and MAP3K is referred to as a MAPK cascade or a MAPK module [1]. Specific protein–protein interactions both within components of a MAPK module and with scaffold proteins are important in order to obtain specificity and efficiency in signalling through a MAPK cascade [2]. Activated MAPKs phosphorylate numerous cytosolic and nuclear substrates including downstream kinases and transcription factors [3]. Substrate specificity is determined by specific protein–protein interaction between the MAPK and specific MAPK-docking sites found in both substrates and MAPK regulators [4]. To date up to six distinct groups of MAPKs have been identified in mammals [5]. These are the ERKs (extracellular-signal-regulated kinases), ERK1 and ERK2, the JNKs (c-Jun N-terminal kinases), JNK1, JNK2 and JNK3, the p38 isoforms α, β, γ and δ, the BMK (BIG MAPK)/ERK5, the atypical MAPKs, ERK3 (MAPK6) and ERK4 (MAPK4), and the recently identified ERK7 and ERK8.

ERK4 (MAPK4) was originally cloned as p63 MAPK and belongs, together with its closest homologue ERK3 (MAPK6), to the atypical group of MAPKs [6]. These kinases possess approx. 50% homology with ERK1/2 within their kinase domain, but have distinct structural features which are atypical among the MAPKs. Most obvious is the lack of the canonical T-X-Y motif in the activation-loop, which in ERK3 and ERK4 is replaced by a single phospho-acceptor in an S-E-G motif. In addition, an S-P-R motif replaces the conserved A-P-E motif in kinase domain VIII and they also contain a unique C-terminal extension [7]. Recently, MK5 (MAPK-activated protein kinase 5) was shown to be a specific target for ERK3 and ERK4 [811]. Interaction with ERK3 or ERK4 resulted in both subcellular redistribution and activation of MK5. So far no MAP2K or MAP3K have been identified for ERK3 or ERK4. ERK3 is phosphorylated on Ser189 in intact cells and an unidentified kinase capable of phosphorylating ERK3 on this residue has been partially purified from rabbit muscle [12]. However, there is still no information regarding the regulation of the phosphorylation of the activation loop of ERK3 or its physiological role. Although ERK3 is a highly unstable protein, the activity of which seems to be regulated at the level of cellular abundance, ERK4 is a relatively stable protein suggesting a different mode of regulation [10,11,13].

In the present study we investigate the phosphorylation of Ser186 within the ERK4 activation loop and the role that this residue plays in the ability of ERK4 to interact with and activate its physiological substrate MK5. Our results indicate a mechanism in which complex formation between ERK4 and MK5 results in enhanced phosphorylation of Ser186 and that this both stabilizes the complex between ERK4 and MK5 and facilitates the activation of MK5.

EXPERIMENTAL

Reagents

Sodium arsenite, PMA, BSA, PDGF (platelet-derived growth factor; BB homodimer) and LMB (leptomycin B) were purchased from Sigma–Aldrich. Redivue [γ-32P]ATP was purchased from Amersham Biosciences. PRAKtide (12–363) was purchased from Upstate Biotechnologies. PP2A (protein phosphatase 2A) was a gift from Carol Mackintosh (MRC Protein Phosphorylation Unit, University of Dundee, Dundee, Scotland, U.K.) and SAP (shrimp alkaline phosphatase) was a gift from Biotech-Pharmacon (Tromsø, Norway).

Antibodies

The phospho-specific antibody recognizing ERK4 phosphorylated on Ser186 (P-S186-ERK4) was raised in sheep at the Scottish Antibody Production Unit (SAPU, Lanarkshire, U.K.) against peptide HYSHKGYLSEGLVTK corresponding to residues 178–192 of human ERK4 in which the underlined residue is phosphoserine. The antibody was affinity-purified on CH-Sepharose covalently coupled to the phosphorylated peptide and then passed through a column of CH-Sepharose coupled to the non-phosphorylated peptide. Antibodies that did not bind to the latter column were selected. The sheep polyclonal anti-ERK4 antibody has been described previously [11]. The monoclonal PRAK antibody (A-7) (sc-46667) and polyclonal ERK2 antibody (C-14, sc-154) were purchased from Santa Cruz Biotechnology. Anti-FLAG(M2)-conjugated agarose was purchased from Sigma–Aldrich. The polyclonal GFP (green fluorescent protein) antibody (ab290) used for immunoprecipitation of EGFP (enhanced GFP)–MK5 was bought from Abcam. The polyclonal rabbit anti-(phospho-JNK) (#9251), anti-(phospho-ERK1/2) (#9101), anti-(phospho-p38) (#9211), anti-[SAPK (stress-activated protein kinase)/JNK] (#9252) and monoclonal mouse anti-p38 (#9212) were bought from Cell Signaling Technology. Alexa Fluor® 594 goat anti-mouse IgG, Alexa Fluor® 680 goat anti-rabbit IgG (A21076), Alexa Fluor® 680 goat anti-mouse IgG (A-21057) and Alexa Fluor® 680 donkey anti-sheep IgG (A-21102) were purchased from Molecular Probes. IRDye 800 conjugated anti-GST (glutathione transferase) (600-132-200) and anti-GFP antibodies (600-132-215), IRDye 800CW conjugated affinity-purified anti-mouse IgG (610-131-121) and IRDye 800CW conjugated affinity-purified anti-rabbit IgG (611-131-122) were purchased from Rockland.

DNA constructs

Construction of the following plasmids has been described previously: pEGFP–MK5, pEGFP–MK5K51E, pEGFP–MK5T182A, pEGFPmutNES and pGEX4-T3 MK5 [14], pEGFP–MK5 1–423 [8], and pENTR-ERK4, pGBKT7-MK5, pEXPGAD-ERK4wt and kinase-dead, pEXPmyc-ERK4wt and kinase-dead, pDESTGAD and pDESTmyc [11]. The gateway compatible expression vector pDEST-10 for expression of proteins with N-terminal His6 tag using the baculovirus system was purchased from Invitrogen. Mutants of ERK4 where Ser186 was changed to either an alanine or glutamate residue was produced by site-directed mutagenesis using the QuikChange® method (Stratagene) and primers (only forward primer shown) 5′-CCACAAGGGTTATCTGGCAGAAGGGTTGGTAACA-3′ and 5′-TCCCACAAGGGTTATCTGGAAGAAGGGTTGGTAACAAAGTG-3′ respectively using pENTR-ERK4 as a template. The resulting Gateway entry vectors pENTR-D-ERK4 S186A, pENTR-D-ERK4 S186E, were used in recombination reactions with Gateway destination vectors pDESTGAD, pDESTmyc and pDEST-10 to construct the expression vectors pEXPGAD-ERK4, pEXPmyc-ERK4 and pEXP-10-ERK4 respectively. To construct pEXP/FRP-3xFLAG-ERK4, full length ERK4 was first amplified using the primers 5′-CGGAATTCGGCTGAGAAGGGTGACTGCATCGCCAGTGTCTATGG-3′ and 5′-CGGGATCCTCGAGTTATTGTCCGGCAGGTCCTCGGGCTTGG-3′. The PCR-product was cloned into p3xFLAG-CMV7.1 (Sigma) using the EcoRI-XhoI sites. The 3xFLAG tagged-ERK4 was then amplified from the p3xFLAG-CMV7.1 vector using the primers 5′-CACCTCGTTTAGTGAACCGTCAGA-3′and 5′-TCACCACCTTTCTTTGGAGAAGG-3′ and cloned into pENTR-D-TOPO (Invitrogen). The resulting pENTR-3xFLAG-ERK4 clone was further recombined with the Gateway destination vector pEF5/FRT/V5-DEST (Invitrogen) generating pEXP/FRP-3xFLAG-ERK4. All the plasmid constructs described were verified by DNA sequencing using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems). All PCR was performed using Pfx Platinum polymerase (Invitrogen) according to the manufacturer's protocol.

Yeast two-hybrid assay

Expression plasmids with GAL4 DB (DNA-binding domain) and AD (activation domain) fusion were transformed into the Saccharomyces cerevisiae strains PJ69-2A(MATa) and Y187(MATα) respectively, using the Frozen-EZ Yeast Transformation II (Zymo Research), and plated out on to suitable drop-out medium, according to the manufacturer's protocol. The yeast-mating procedure and semi-quantitative β-galactosidase assay were performed as described previously [8].

Cell culture and transfection

HeLa cells were maintained in Eagle's MEM (minimum essential medium) supplemented with 1× non-essential amino acids (Invitrogen), 10% fetal bovine serum (Invitrogen), 2 mM L-glutamine, penicillin (100 units/ml) and streptomycin (100 μg/ml). Flp-In-3T3 cells were maintained in DMEM (Dulbecco's Modified Eagle Medium) supplemented with 10% newborn calf serum (Invitrogen), penicillin (100 units/ml), streptomycin (100 μg/ml) and zeocin (100 μg/ml). A cell line for stable expression of 3xFLAG-tagged ERK4 (FLAG–ERK4) was generated from Flp-In-3T3 cells by Flp recombinase-mediated integration using the Flp-In system (Invitrogen). Briefly, pEXP/FRT-3xFLAG-ERK4 was transfected together with pOG44 for transient expression of Flp recombinase. At 24 h after transfection the cells were cultured in the presence of 200 μg/ml hygromycin B for selection of cells with integrated FLAG–ERK4. Hygromycin-B-resistant cells were expanded further and stable expression of FLAG-tagged ERK4 was verified with Western blotting using both anti-FLAG and anti-ERK4 antibodies. Lipofectamine™ in conjunction with PLUS reagent (Invitrogen) was used to transfect the HeLa cells and Flp-In-3T3 cells according to the manufacturer's protocol.

siRNA (small interfering RNA)

All siRNAs were purchased from Ambion and have been described previously [11]. The sequences of the siRNAs used were as follows (sense and antisense): MK5 siRNA 5′-GGAUAUGCGAAGAAAGAUCTT-3 and 5′-GAUCUUUCUUCGCAUAUCCTT-3; and ERK4 siRNA 5′-GGGUGAGCUGUUCAAGUUCTT-3′ and 5′-GAACUUGAACAGCUCACCCTG-3′. As a control siRNA the Silencer negative control siRNA (Ambion; catalogue number 4611) was used. Lipofectamine™ 2000 (Invitrogen) was used to transfect the siRNA into HeLa cells according to the manufacturer's protocol.

Immunoblotting

Proteins were resolved on SDS/PAGE (4–12% NUPAGE; Invitrogen), transferred on to a nitrocellulose membrane (Amersham Biosciences) and probed with indicated antibodies. The blots were developed using either Alexa Fluor® 680- or IRDye 800-conjugated secondary antibodies and the Odyssey IR Imaging System (Li-Cor Biosciences). Molecular masses were estimated using the MagicMark Western protein standard (Invitrogen).

Expression of His6-tagged proteins in Sf9 cells

The pEXP-10 vectors were used to generate recombinant baculovirus using the Bac-to-Bac system (Life Technologies). The resulting viruses, encoding the ERK4 wild-type and mutants with an N-terminal hexahistidine sequence, were used to infect Sf9 insect cells. The infected cells were harvested 72 h post-infection, and the histidine-tagged ERK4 proteins were purified by Ni2+-nitrilotriacetate agarose chromatography. The purified proteins were then dialysed against 50 mM Tris/HCl (pH 7.5), 50% (v/v) glycerol, 150 mM NaCl, 0.1 mM EGTA, 0.1% (v/v) 2-mercaptoethanol, 0.03% (w/v) Brij-35, 1 mM benzamidine and 0.2 mM PMSF, and stored at −20 °C.

Expression of GST fusion proteins in Escherichia coli

GST-fusion proteins were expressed in E. coli (BL21) and purified as previously described [8]. SDS/PAGE and Coomassie Blue staining were used to analyse both the expression and yield of the fusion proteins.

GST pulldown assays

Recombinant ERK4 was precleared prior to the pulldown using 4 μg of GST-protein and 30 μl of glutathione–Sepharose (Amersham Biosciences) [50% slurry equilibrated in a buffer containing 50 mM Tris/HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 1% (w/v) Triton X-100, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate and 0.27 M sucrose] for 1 h at 4 °C. Direct interaction between MK5 and ERK4 wild-type and ERK mutants was detected by mixing 2 μg of recombinant ERK4 with 4 μg of purified GST–MK5 or GST in 300 μl of the above buffer for 2 h at 4 °C. After addition of 30 μl of glutathione–Sepharose (50% slurry equilibrated in the above buffer) the samples were incubated for a further 1 h. The beads were washed five times with the above buffer and once in 50 mM Tris/HCl (pH 7.5) and then resuspended in 40 μl 2× SDS-sample buffer. GST–MK5 and co-precipitated ERK4 were detected by SDS/PAGE and Western blotting using GST antibody and ERK4 antibodies respectively.

Kinase assays

GFP-tagged MK5 was immunoprecipitated from HeLa cells using 1 μl of the polyclonal anti-GFP antibody ab290 (Abcam), and the kinase activity of MK5 was measured as previously described [14]. Activity is expressed as cpm (counts per min) incorporated. Experiments were performed three times, and the results are presented as mean values±S.E.M.

Cell staining and microscopy

Determination of the subcellular localization of GFP fusion proteins and myc-tagged ERK4 was performed as described previously [8]. Images were collected using a Zeiss LSM510 confocal laser-scanning microscope and processed using Adobe Photoshop. For cell counting experiments, several fields of cells from each transfection or co-transfection were examined and at least 100 cells were scored for the subcellular localization of the protein(s) of interest.

RESULTS

ERK4 is phosphorylated on Ser186

Instead of the conserved T-X-Y dual phosphorylation sites found within the classical MAPKs, the atypical MAPKs ERK3 and ERK4 contain an S-E-G motif with a single potential phospho-acceptor (Ser189 and Ser186 respectively) within their activation loop in kinase subdomain VIII. To investigate the phosphorylation status of ERK4 on Ser186 we first generated an antibody using a phospho-Ser186-containing peptide from the ERK4 kinase as antigen. This antibody readily detects ERK4 protein expressed and purified from baculovirus-infected Sf9 cells. Furthermore, this antibody recognizes phospho-ERK4 as prior incubation of the recombinant protein with the serine/theonine phosphatase PP2A almost completely abolished this signal (Figure 1A). To determine whether this site is phosphorylated in vivo, HeLa cells were transfected with expression plasmids containing myc-tagged reading frames encoding either wild-type ERK4, a kinase-deficient mutant of ERK4 (ERK4D168A; from here on referred to as ERK4kd), or a mutant where Ser186 is substituted by a non-phosphorylable alanine residue (ERK4S186A). Following immunoprecipitation, ERK4 was analysed by immunoblotting using the P-S186-ERK4 antibody. Both wild-type and kinase-dead ERK4 were phosphorylated on Ser186 in these cells (Figure 1B). In contrast, ERK4 protein in which Ser186 was substituted by an alanine residue was not recognized by this antiserum. We conclude that ERK4 is phosphorylated when expressed in either insect (Sf9) or mammalian cells and our observation that a kinase-dead mutant of ERK4 is also modified on this site indicates that this results from the activity of an ERK4 kinase rather than autophosphorylation. However, since ERK4 has been reported to form homodimers [10] in mammalian cells, we cannot exclude the possibility that the phosphorylation of ERK4kd in HeLa cells is a result of transphosphorylation by endogenous wild-type ERK4.

Figure 1 ERK4 is phosphorylated on Ser186

(A) Recombinant ERK4 (0.5 μg) purified from baculovirus-infected Sf9 cells was either left untreated or treated with 0.5 m-units PP2A for 1 h at 37 °C. The samples were analysed by immunoblotting with antibodies that were either raised against ERK4 phosphorylated on Ser186 (P-S186-ERK4) (upper panel) or against total ERK4 (lower panel). (B) Wild-type and kinase-dead ERK4 are phosphorylated on Ser186 in HeLa cells. HeLa cells were transfected with expression vectors encoding ERK4, kinase-dead ERK4D168A (ERK4kd) or the phosphosite mutant ERK4S186A, all containing an N-terminal myc tag. At 24 h after transfection, whole cell extracts were prepared and ERK4 proteins were immunoprecipitated using an anti-myc antibody. Immunoprecipitated ERK4 phosphorylated on Ser186 was visualized by Western blot analyses using the P-S186-ERK4 antibody and total ERK4 was detected using an anti-ERK4 antibody. IP, immunoprecipitate; WB, Western blot.

As the classical MAPKs demonstrate rapid increases in T-X-Y activation loop phosphorylation in response to extracellular stimuli we wanted to investigate whether the phosphorylation of Ser186 in the ERK4 activation loop also could be modified in response to stimuli that activate the classical MAPKs. Unfortunately, our P-S186-ERK4 antibody was not sensitive enough to detect endogenous ERK4 phosphorylated on Ser186 in HeLa, HEK (human embryonic kidney)-293 or NIH 3T3 cells. Therefore 3T3 Flp-In cells (Invitrogen) were used to generate an isogenic cell line, 3T3 Flp-In-FLAG-ERK4 that stably expresses FLAG-tagged ERK4 at low levels following insertion of a single copy of the ERK4 cDNA into the genome. 3T3 Flp-In-FLAG-ERK4 cells were treated with a wide range of extracellular stimuli and, following immunoprecipitation, FLAG–ERK4 was analysed by immunoblotting using the P-S186-ERK4 antibody. As shown in Figure 2, none of the treatments, which activated either the classical ERK1 and 2 MAPKs (serum, PDGF or PMA) or the stress-activated MAPKs p38 and JNK (sodium arsenite, H2O2 or UV-light) caused any increase (or decrease) in the level of Ser186 phosphorylation within ERK4. In contrast, we could readily observe the appropriate changes in phosphorylation of ERK1/2, p38 and JNK.

Figure 2 Ser186 phosphorylation is not induced by known MAPK activators

NIH 3T3 Flp-In ERK4 cells that have integrated one copy of an ERK4 gene construct encoding FLAG-tagged ERK4 were serum-starved for 24 h and then subjected to various cellular stimuli that activate classical MAPKs. Cells were treated with 10% newborn calf serum (NCS) for 15 min, 10 ng/ml PDGF for 15 min, 50 ng/ml PMA for 30 min, 250 μM sodium arsenite (SA) for 30 min, 5 mM H2O2 for 15 min and UV-C for 15 min before lysis. FLAG–ERK4 was then immunoprecipitated (IP) using anti-FLAG M2 antibody conjugated to agarose and the level of phosphorylation on Ser186 was determined by Western blotting (WB) using the phospho-specific P-S186-ERK4 antibody. The presence of equal amounts of immunoprecipitated ERK4 in each sample was verified by Western blotting using an anti-ERK4 antibody. Activation of ERK1/2, p38 and JNK1/2 in whole cell extracts (WCE) was analysed by Western blotting using phospho-specific antibodies specifically recognizing activation-loop phosphorylation of the indicated MAPKs.

Co-expression of MK5 leads to increased Ser186 phosphorylation of ERK4

Thus far, the only serine/threonine protein kinase known to specifically interact with ERK4 is the MAPK-activated protein kinase MK5 [10,11]. In order to determine whether MK5 could be responsible for mediating Ser186-phosphorylation of ERK4, we co-transfected kinase-dead ERK4 together with either wild-type MK5, a kinase-deficient mutant of MK5 (MK5-K51E) or a mutant MK5 lacking the conserved phospho-acceptor site necessary for its activation (MK5-T182A). ERK4kd was then immunoprecipitated from the transfected cells and analysed by Western blotting using the P-S186-ERK4 antibody. When co-expressed with wild-type MK5 the levels of Ser186 phosphorylation on ERK4kd increased several-fold. However, both a kinase-dead mutant of MK5 and also inactive MK5 containing a mutated phospho-acceptor site (Thr182) also increased levels of Ser186 phosphorylation on ERK4. In contrast, co-expression of an MK5 deletion mutant (MK51-423) that is unable to interact with ERK4 [11] was unable to increase Ser186 phosphorylation (Figure 3). The fact that both wild-type and kinase-dead mutants of MK5 could induce Ser186 phosphorylation of ERK4kd indicates that MK5 is not the ERK4 Ser186 kinase. It is possible that MK5 could act in two ways to cause the increase in Ser186 phosphorylation. First, it might protect the site from dephosphorylation by a protein phosphatase. Alternatively, its binding to ERK4 might facilitate the recruitment or activity of a specific ERK4 Ser186 kinase to the ERK4–MK5 complex.

Figure 3 Both wild-type and kinase-deficient mutants of MK5 increase Ser186 phosphorylation of kinase-dead ERK4

HeLa cells were co-transfected with vectors encoding myc–ERK4kd and GFP–MK5, GFP–MK5K51E, GFP–MK5T182A or GFP–MK5(1-423). At 24 h post-transfection, whole cell extracts were prepared and ERK4kd was immunoprecipitated (IP) with an anti-myc antibody. Ser186 phosphorylation of immunoprecipitated ERK4 was detected by Western blot (WB) analysis using the phospho-specific P-S186-ERK4 antibody. Levels of mycERK4kd in immunoprecipitates and GFP–MK5 in whole cell extracts was verified by Western blotting using antibodies against ERK4 and GFP respectively.

In a recent study, Kant et al. [10] showed that, when co-expressed with MK5, ERK4 was resolved as a doublet on SDS/PAGE and that the faster migrating form corresponds to the single band observed when ERK4 was expressed alone. Furthermore, treatment of cell extracts with calf intestinal phosphatase resulted in a loss of the retarded band, indicating that MK5 causes a phosphorylation-induced ERK4 mobility shift [10]. In order to determine the phosphorylation status on Ser186 of the differently migrating ERK4 molecules, Myc-tagged ERK4 was expressed either alone or co-expressed with either wild-type or kinase-dead MK5. ERK4 was then immunoprecipitated and analysed by Western blotting using either an anti-ERK4 antiserum or the phospho-Ser186 antibody. As reported previously, two ERK4 bands appeared in SDS/PAGE upon co-expression of MK5 (Figure 4A). Interestingly, even though the amount of protein in the two bands is approximately equal, only the retarded ERK4 molecules were phosphorylated on Ser186. In contrast, the expression of a kinase-dead mutant of MK5 did not cause any phosphoshift in ERK4 despite the fact that, like the wild-type, MK5 expression of this protein led to a significant increase in the level of phosphorylation on Ser186 (Figure 4A). First, this demonstrates that only ERK4 phosphorylated on Ser186 is able to act as a substrate for MK5 and undergo a phosphorylation-induced mobility shift. Secondly it shows that phosphorylation of Ser186 is not responsible for this altered migration. To further explore the nature of the interaction between these two pools of ERK4 and MK5 we performed reciprocal immunoprecipitations from cells co-expressing myc-tagged ERK4 and GFP-tagged MK5. Western blotting of anti-myc immunoprecipitates using an anti-ERK4 antibody revealed the expected doublet. However, when the anti-GFP immunoprecipitate was similarly analysed the only form of ERK4 detected was the retarded (phosphorylated) band (Figure 4B). Overall, our results are compatible with a model in which MK5 is only capable of forming complexes with a subset of the ERK4 molecules and this complex formation facilitates increased phosphorylation of Ser186 in ERK4. Complex formation is also accompanied by activation of MK5, which then phosphorylates ERK4 on residues other than Ser186 leading to the observed ERK4 phospho-shift. Finally, to rule out any artifactual changes in the behaviour of either ERK4 or MK5 due to overexpression of these proteins in our experiments, we have examined the electrophoretic mobility of endogenous ERK4 immunoprecipitated from HeLa cells transfected with either a control (scrambled) siRNA or siRNAs specific for either ERK4 or MK5. In control cells, a doublet of ERK4 is clearly seen and both bands are lost when cells are transfected with siRNA against ERK4. In contrast, transfection of cells with siRNA against MK5 causes loss of only the upper band of ERK4 (Figure 4C). The upper (retarded) band is due to phosphorylation of endogenous ERK4 as treatment of ERK4 immunoprecipitates with SAP leads to selective loss of the more slowly migrating species (Figure 4D). This strongly suggests that endogenous ERK4 is found in at least two different forms in mammalian cells and that the phospho-shifted form of ERK4 is a result of MK5-dependent phosphorylation.

Figure 4 MK5 induces a mobility shift of a subset of ERK4 molecules

(A) In the presence of wild-type MK5 only the retarded form of ERK4 is phosphorylated on Ser186. HeLa cells were co-transfected with expression vectors encoding myc–ERK4 and either GFP–MK5 or GFP–MK5K51E. At 24 h later, cells were lysed and myc–ERK4 was immunoprecipitated from the whole cell extracts using an anti-myc antibody. Total ERK4 and the level of Ser186 phosphorylation of ERK4 were determined by immunoblotting with an anti-ERK4 antibody and the anti-P-S186-ERK4 antibody respectively. Retarded ERK4 is marked as ERK4*. GFP–MK5 expression was verified by Western blot analyses of whole cell extracts. (B) ERK4 molecules in complex with MK5 are shifted in SDS/PAGE. Myc–ERK4 was expressed in HeLa cells either alone or together with GFP–MK5. Cell lysates were split in two and either subjected to immunoprecipitation with an anti-myc or an anti-GFP antibody. Immunoprecipitated and co-immunoprecipitated proteins were detected by Western blot analysis using an anti-ERK4 or an anti-GFP antibody. (C) Endogenous ERK4 exists in two forms that migrate differently in SDS/PAGE dependent on MK5 expression. HeLa cells were transfected with 20 nM of either control siRNA, ERK4 siRNA or MK5 siRNA. Endogenous ERK4 protein was visualized by immunoprecipitation and Western blot using anti-ERK4 antibody. Knockdown of MK5 was verified by immunoblotting using an anti-MK5 antibody. (D) The retarded endogenous ERK4 band is not detected after phosphatase treatment. Endogenous ERK4 was immunoprecipitated from HeLa cells and either left untreated or treated with 5 units of SAP for 1 h at 37° and then analysed by Western blot using an anti-ERK4 antibody. IP, immunoprecipitate; WB, Western blot; WCE, whole cell extract.

ERK4 in which Ser186 is replaced with either alanine or glutamate is unable to activate MK5

Phosphorylation of the T-X-Y motif in the activation loop is essential for the activation of the classical MAPKs. To assess the importance of Ser186 phosphorylation of ERK4 for its activity, we co-transfected HeLa cells with plasmids encoding GFP–MK5, the only known substrate of ERK4, together with plasmids encoding wild-type ERK4 or mutants of ERK4 where Ser186 is replaced with either an alanine (ERK4S186A) or glutamate (ERK4S186E) residue. GFP–MK5 was immunoprecipitated from the lysates of the transfected cells and MK5 activity towards the model MK5 peptide substrate PRAKtide (KKLRRTLSVA, derived from glycogen synthase) was assayed. While co-transfection with wild-type ERK4 led to 6–7-fold activation of MK5, no increase in MK5 activity was observed in cells co-transfected with ERK4S186A (Figure 5). Replacement of Ser186 with the phospho-mimetic glutamate caused only a slight recovery in the ability of ERK4 to activate MK5 when compared with the alanine mutant. We conclude that the integrity of Ser186 is essential for the ability of ERK4 to activate MK5.

Figure 5 Mutation of Ser186 in the activation loop of ERK4 abrogates ERK4-mediated activation of MK5 in vivo

HeLa cells were co-transfected with plasmids encoding the indicated proteins. After 24 h, cells were lysed and MK5 was immunoprecipitated using an anti-GFP antibody. The GFP–MK5 activity was assayed in an immune complex kinase assay using PRAKtide as substrate. Activity is expressed as cpm incorporated. Kinase assays were performed in quadruplicate and values are mean incorporations (cpm)±S.E.M. Expression of wild-type and mutant forms of MK5 and ERK4 was verified by Western blotting of cell lysates using appropriate antibodies. WB, Western blot.

Ser186 is important for ERK4-mediated cytoplasmic anchoring of MK5

MK5 contains both a NLS (nuclear localization signal) and a NES (nuclear export signal). However, although MK5 is capable of shuttling between the nucleus and cytoplasm it is predominantly nuclear in resting cells indicating that nuclear import of MK5 predominates [14,15]. In contrast ERK4 is mainly localized to the cytoplasm by a CRM1 (chromosome region maintenance 1)-dependent nuclear export process [11]. We and others [10,11] have previously shown that co-expression of ERK4 and MK5 results in cytoplasmic relocalization of MK5 and that this occurs independently of the kinase activity of either ERK4 or MK5. The fact that both ERK4 and MK5 shuttle between nucleus and cytoplasm through active transport mechanisms, prompted us to ask if an active nuclear export pathway is essential for the cytoplasmic anchoring of MK5 by ERK4. Both proteins were co-expressed in HeLa cells either in the absence or presence of LMB, an inhibitor of CRM1-dependent nuclear export [16,17]. LMB had no effect on the ability of ERK4 to cause cytoplasmic anchoring of MK5, indicating that this occurs independently of an active nuclear export process (Figures 6A and 6E). This observation was further strengthened by the fact that a mutant of MK5 in which the functional NES sequence has been deleted (GFP–MK5mutNES), was also exclusively localized to the cytoplasm when co-expressed with wild-type ERK4 (Figure 6A). Previous work studying the cytoplasmic anchoring of the classical ERK2 MAPK by its upstream activating kinase MEK1 (MAPK/ERK kinase 1) has shown that deletion of the phosphorylation sites (Thr183 and Tyr185) within the activation loop of ERK2 stabilizes the interaction with MEK1. This leads to the cytoplasmic retention of ERK2, even under conditions which would normally result in nuclear translocation of the kinase [18]. To investigate the role of Ser186 in the activation loop of ERK4 in the cytoplasmic anchoring of MK5, HeLa cells were co-transfected with expression vectors encoding either wild-type MK5 or the NES mutant of MK5 (GFP–MK5mutNES) together with expression vectors encoding either myc-tagged wild-type or Ser186 mutants of ERK4 (ERK4S186A or ERK4S186E). Like wild-type ERK4, both ERK4S186A and ERK4S186E were almost exclusively cytoplasmic proteins when expressed in HeLa cells (results not shown). Co-expression of these mutant forms of ERK4 caused either partial (ERK4S186A) or complete (ERK4S186E) relocalization of MK5 from nucleus to cytoplasm (Figures 6B and 6C). However, whereas exposure of cells co-expressing wild-type ERK4 and MK5 to LMB or mutation of the NES within MK5 did not interfere with the ability of ERK4 to retain MK5 in the cytoplasm, mutation of Ser186 to either an alanine or glutamate residue caused the localization of MK5 to become sensitive to LMB with the drug causing MK5 to relocalize to the nucleus in both cases. Mutation of the NES within MK5 also caused the protein to relocalize to the nucleus in the presence of the two mutant forms of ERK4 (Figures 6B and 6C). Our results clearly demonstrate that the mutation of Ser186 causes profound changes in the ability of ERK4 to bind to and relocalize MK5. Wild-type ERK4 is able to cause translocation of MK5 from nucleus to cytoplasm in a mechanism which does not depend on active CRM1-dependent nuclear export. In contrast, any relocalization of MK5 seen on co-expression of ERK4 lacking Ser186 is now dependent on CRM1 and requires the NES within MK5. The simplest explanation is that although mutation of Ser186 does not interfere with nuclear export of MK5, it does compromise the ability of ERK4 to anchor MK5 within the cytoplasm.

Figure 6 ERK4-mediated anchoring of MK5 is compromised by mutation of Ser186 in the activation loop

HeLa cells were transfected with expression vectors encoding either myc-tagged ERK4 (A), ERK4S186A (B) or ERK4S186E (C) either alone or together with expression vectors encoding either GFP–MK5 or GFP–MK5mutNES. After 24 h, cells were either left untreated or treated with 2 ng/ml LMB for 4 h and then fixed. GFP–MK5 was visualized directly (green channel) and myc–ERK4 was detected by immunostaining using an anti-myc antibody and an Alexa Fluor® 594-coupled secondary anti-mouse antibody (red channel). Several fields of cells were examined, and representative images are shown. The cells were counted and results are presented as the percentage of cells in which GFP–MK5 was nuclear or expressed all over the cell (N≥C) or predominantely cytoplasmic (C>N) (D and E). Several cells from different experiments were counted and results from one representative experiment are shown.

Inefficient complex formation with MK5 by ERK4 lacking Ser186in vivo

The results so far indicate that activation-loop phosphorylation of ERK4 is induced by MK5 expression and that the integrity of Ser186 is required for both MK5 activation and cytoplasmic anchoring of MK5. One possible explanation for the defect in cytoplasmic anchoring of MK5 observed could be that complex formation between ERK4 and MK5 is compromised by mutation of the phospho-acceptor site in the activation loop. To address this, HeLa cells were co-transfected with expression vectors for GFP–MK5 and mycERK4, mycERK4kd, mycERK4S186A or mycERK4S186E. Myc-tagged or GFP-tagged proteins were then immunoprecipitated from extracts with appropriate antibodies and the presence of GFP–MK5 or myc–ERK4 in the immunoprecipitates was determined by Western blotting. Whereas GFP–MK5 forms a stable complex with both wild-type ERK4 and ERK4kd, no GFP–MK5 was co-immunoprecipitated with ERK4S186A or ERK4S186E (Figure 7A). Likewise, neither ERK4S186A nor ERK4S186E were co-immunoprecipitated with GFP–MK5. Since expression of MK5 increases the levels of phosphorylation on Ser186, it is possible that this phosphorylation stabilizes the ERK4–MK5 complex after its formation, rather than regulating the initial recognition and interaction of the two proteins. To evaluate this, we performed GST-pulldown analyses of recombinant proteins in vitro. We were not able to see any differences in the efficiency of complex formation in vitro between bacterially produced GST–MK5 and ERK4, ERK4kd or ERK4S186A produced and purified from Sf9 cells (Figure 7B). In addition, both ERK4S186A and ERK4S186E were able to interact with MK5 as assessed using the yeast two-hybrid assay with at least the same affinities as wild-type ERK4 (Figure 7C). These observations suggest that the ability of ERK4 to initially recognize and bind to MK5 is not severely impaired by mutation of Ser186. Therefore the poor complex formation observed in vivo between MK5 and ERK4S186A or ERK4S186E might suggest that the subsequent phosphorylation on Ser186 mediated by MK5 binding acts to stabilize the complex between these two proteins in vivo.

Figure 7 Different requirement for Ser186 for complex formation in vivo and in vitro

(A) Activation-loop mutants of ERK4 do not form a stable complex with MK5 in vivo. HeLa cells were co-transfected with expression vectors encoding GFP–MK5 and either myc–ERK4, myc–ERK4kd, myc–ERK4S186A or myc–ERK4S186E. After 24 h, cells were lysed and cell extracts were split into two samples. One lysate was then subjected to MK5 immunoprecipitation using an anti-GFP antibody while ERK4 was immunoprecipitated from the other using an anti-myc antibody. GFP–MK5 and myc–ERK4 were detected by Western blotting using anti-MK5 or anti-ERK4 respectively. (B) ERK4S186A interacts with MK5 in vitro. A 2 μg quantity of ERK4, ERK4kd or ERK4S186A were incubated with 4 μg of GST or GST–MK5 and glutathione–agarose. Bound ERK4 was detected by Western blotting using an anti-ERK4 antibody and GST–MK5 was visualized using an anti-GST antibody. (C) The phospho-acceptor site mutants of ERK4 bind to MK5 in yeast. Expression vectors encoding GAL4 AD-fusion proteins of either ERK4, ERK4kd, ERK4S186A or ERK4S186E were transformed into the yeast strain Y187. Selected transformants were then mated with transformants of yeast strain PJ69-2A expressing MK5 fused to the GAL4 DB (GBK–MK5). Yeast diploids expressing both DB and AD fusions were selected on synthetic drop out medium deficient for leucine and tryptophan. Diploid yeast clones were grown overnight and yeast two-hybrid interactions were assayed by measuring the induction of the β-galactosidase gene.

DISCUSSION

In the present study we have investigated the functional role of Ser186 within the S-E-G motif in the activation loop of ERK4. By employing a phospho-specific antibody raised against this site, we demonstrate that ERK4 is phosphorylated on Ser186 when overexpressed in Sf9 cells or in mammalian cells. The fact that equal levels of phosphorylation are observed on both wild-type and a kinase-dead mutant of ERK4 strongly suggests that this modification is not due to intra-molecular autophosphorylation. A kinase activity phosphorylating the corresponding Ser189 site in ERK3 was partially purified based on its strong interaction with ERK3 [12]. However, the identity of this kinase is not known. No increase in Ser186 phosphorylation of ERK4 is observed after treatment of a NIH 3T3 cell line with a variety of growth factors and cellular stresses known to activate classical MAPKs. Therefore it is unlikely that Ser186 is phosphorylated by any of the dual-specificity MAP2Ks which are responsible for modification of both the threonine and the tyrosine residues in the T-X-Y motif of classical MAPKs. In support of this we can find no evidence of interaction between ERK4 and any mammalian MAP2Ks when examined using the yeast two-hybrid assay (results not shown).

The only kinase reported to bind to ERK3 and ERK4 is the MAPK-activated protein kinase MK5. In the present study, we report that co-expression of ERK4 and MK5 increases the level of ERK4 phosphorylation on Ser186. Interestingly, an increase is also observed when ERK4 is co-expressed with kinase-deficient mutants of MK5, but not upon co-expression of a deletion mutant of MK5 that is unable to interact with ERK4. This strongly suggests that MK5 does not increase the phosphoryation of Ser186 by acting as a Ser186 ERK4 kinase. Instead, interaction with MK5 could protect ERK4 from dephosphorylation, possibly by making the ERK4 molecules that are in complex with MK5 inaccessible to a protein phosphatase. Alternatively, MK5 binding might facilitate either the recruitment and/or activation of the putative ERK4 Ser186 kinase. Interestingly, we observed that ERK4 expressed in mammalian cells was more highly phosphorylated than ERK4 expressed in insect cells (results not shown). Since MK5 is not conserved in insect cells this may indicate a function for MK5 in regulating the basal level of ERK4 Ser186 phosphorylation when ERK4 is overexpressed in mammalian cells. Importantly, the ERK4–MK5 complex displays a different dynamic with respect to the ability of both proteins to shuttle between the nucleus and cytoplasm. This could influence the accessibility of ERK4 as a substrate for a kinase and/or for a phosphatase. We have interpreted our results showing Ser186 phosphorylation of a kinase-dead mutant of ERK4 as evidence for the existence of a distinct ERK4 kinase. However, a recent study by Kant et al. [10] has demonstrated that ERK4 is able to form homodimers and also to heterodimerize with ERK3 when overexpressed in HEK-293 cells. Therefore it cannot be ruled out that MK5 expression facilitates the dimerization of a kinase-dead mutant with wild-type ERK4, which might then transphosphorylate its inactive partner on Ser186.

Although the atypical MAPKs ERK3 and ERK4 were among the first members of the MAPK family to be described [6,19], there is still little information available on the regulation of their activity. Phosphorylation of the activation loop of a wide variety of protein kinases is instrumental in their activation. We find that Ser186 in ERK4 has a fundamental role in ERK4 activity since replacement of this residue by alanine or glutamate residues makes ERK4 unable to activate its only known downstream target MK5. The interpretation of this observation is further complicated by the fact that the ability of ERK4 to form a stable complex with MK5 is severely impaired by substituting Ser186 with either alanine or glutamate. However, in vitro kinase assays demonstrate that the ability of the ERK4S186A mutant to activate MK5 is severely compromised even though this mutant binds to MK5 in vitro (results not shown).

In the present study, we demonstrate that an intact Ser186 site in ERK4 is a prerequisite for stable complex formation with MK5 in mammalian cells since neither ERK4S186A nor ERK4S186E mutants co-immunoprecipitate with MK5 from mammalian cellular lysates. However, the same mutants interact with MK5 in both cell-free in vitro assays and in yeast with much the same affinity as the wild-type kinase. Therefore we do not think that initial recognition and complex formation between ERK4 and MK5 are dependent on phosphorylation of Ser186. This conclusion is further strengthened by the observations that the ERK4S186E and ERK4S186A mutants have either retained or partially retained their ability to redistribute MK5 from the nucleus to the cytoplasm. In contrast with what is observed when MK5 is expressed with wild-type ERK4, the cytoplasmic redistribution of MK5 expressed with ERK4S186A or ERK4S186E is reversed upon treatment of cells with LMB. Thus we propose that phosphorylation of Ser186 is important for a stabilization and cytoplasmic anchoring of the MK5–ERK4 complex after its formation. Future studies using more sophisticated cellular imaging techniques should be done to study the dynamics of the MK5–ERK4 complexes in the cytoplasm which could possibly reveal whether a third partner is involved in the cytoplasmic anchoring of the complexes.

We find that there are at least two cellular pools of endogenous ERK4 molecules that migrate differently when resolved on SDS/PAGE gels. The retardation of a subset of the ERK4 molecules is due to phosphorylation by MK5 as both phosphatase treatment and siRNA-mediated knockdown of endogenous MK5 lead to a loss of this mobility shift. This supports the results from Kant et al. [10] that demonstrates that ERK4 is a substrate for MK5. Our phospho-Ser186 antibody was not good enough to analyse the Ser186 phosphorylation state of the endogenous ERK4 molecules. However, when ERK4 is ectopically expressed with MK5, only the shifted ERK4 molecules are phosphorylated on Ser186. This supports the notion that Ser186 is required for stable complex formation with ERK4 and clearly indicates that only molecules phosphorylated on Ser186 are substrates for MK5. The reason why only a subset of ERK4 molecules is phosphorylated by MK5 in vivo is unclear. An implication of this is that only a subset of cellular ERK4 is able to activate MK5 in asynchronous cells. It will be interesting to further study the dynamics between ERK4 and MK5 by determining the migration of ERK4 under different cellular conditions, e.g. at different stages of the cell cycle.

The possible involvement of both a putative S186-ERK4 kinase and a putative P-S186-ERK4 phosphatase is illustrated in Figure 8. The presence of MK5 favours the phosphorylation of ERK4 on Ser186 and this is not dependent on the kinase activity of MK5. The results of our present study indicate that a stronger interaction between ERK4 and MK5 which is now dependent on phosphorylation of Ser186, is then a prerequisite for MK5 activation and the subsequent phosphorylation by MK5 of ERK4 on sites other than Ser186, which can be visualized by a marked mobility shift when analysed by SDS/PAGE. This also leads to cytoplasmic anchoring of the active MK5–ERK4 complex. We propose that the subsequent phosphorylation of ERK4 occurs rapidly upon increased Ser186 phosphorylation since only the retarded ERK4 band generated in the presence of MK5 is phosphorylated on Ser186. Furthermore, we suggest that the accessibility of ERK4 for a putative P-S186-ERK4 phosphatase is low when ERK4 is in complex with MK5. However, upon dissociation from MK5, the dephosphorylation of ERK4 may be rapid.

Figure 8 Phosphorylation of ERK4 on Ser186 regulates its interaction with MK5

Schematic representation of the mechanism for ERK4 and MK5 interaction and subsequent MK5 activation. Unphosphorylated ERK4 is able to bind to MK5, but the interaction is weak and transient. However, interaction with MK5 promotes ERK4 Ser186 phosphorylation either by facilitating recruitment of an S186-ERK4 kinase or by blocking the access of a P-Ser-186-ERK4 phosphatase. Once phosphorylated on Ser186, ERK4 is able to form a more stable complex with MK5 and this strong interaction is required for phosphorylation and activation of MK5 and also the cytoplasmic anchoring of this protein complex. Once activated the MK5 phosphorylates its ERK4-binding partner at sites other than Ser186 and this can be visualized as a distinct mobility shift in ERK4 when analysed by SDS/PAGE.

In conclusion, we have provided the first evidence that the single phospho-acceptor site within the activation loop of the atypical ERK4 MAPK plays a crucial role in determining its ability to form a stable complex with MK5. Future studies will address the mechanism of Ser186 modification and the identity of the kinases and/or phosphatases responsible for modulating phosphorylation of this critical regulatory site.

Acknowledgments

We acknowledge Professor Sir Philip Cohen for reagents and the protein production and antibody purification teams (Division of Signal Transduction Therapy, University of Dundee, Dundee, Scotland, U.K.) co-ordinated by Hilary McLauchlan, and James Hastie (Division of Signal Transduction Therapy, University of Dundee, Dundee, Scotland, U.K.) for expression and purification of enzymes and affinity purification of antibodies. The present study was supported by the National Programme for Research in Functional Genomics in Norway (FUGE) of the Research Council of Norway, the Norwegian Cancer Society and the Erna and Olav Aakres Foundation. S. M. K. is supported by Cancer Research UK and O. M. S. is a fellow of the Norwegian Cancer Society.

Abbreviations: AD, actvation domain; CRM1, chromosome region maintenance 1; DB, DNA-binding domain; ERK, extracellular-signal-regulated kinase; (E)GFP, (enhanced) green fluorescent protein; GST, glutathione transferase; HEK, human embryonic kidney; JNK, c-Jun N-terminal kinase; LMB, leptomycin B; MAPK, mitogen-activated protein kinase; MAP2K, MAPK kinase; MAP3K, MAPK kinase kinase; MEK1, MAPK/ERK kinase 1; MK5, MAPK-activated protein kinase 5; NES, nuclear export signal; PDGF, platelet-derived growth factor; PP2A, protein phosphatase 2A; PRAK, p38-regulated/activated kinase; SAP, shrimp alkaline phosphatase; siRNA, small interfering RNA

References

View Abstract