Reduced MAP Kinase Phosphatase-1 Degradation After p42/p44MAPK-Dependent Phosphorylation

See allHide authors and affiliations

Science  24 Dec 1999:
Vol. 286, Issue 5449, pp. 2514-2517
DOI: 10.1126/science.286.5449.2514


The mitogen-activated protein (MAP) kinase cascade is inactivated at the level of MAP kinase by members of the MAP kinase phosphatase (MKP) family, including MKP-1. MKP-1 was a labile protein in CCL39 hamster fibroblasts; its degradation was attenuated by inhibitors of the ubiquitin-directed proteasome complex. MKP-1 was a target in vivo and in vitro for p42MAPK or p44MAPK, which phosphorylates MKP-1 on two carboxyl-terminal serine residues, Serine 359 and Serine 364. This phosphorylation did not modify MKP-1's intrinsic ability to dephosphorylate p44MAPK but led to stabilization of the protein. These results illustrate the importance of regulated protein degradation in the control of mitogenic signaling.

The control of cell division in response to mitogens is mediated at least in part by MAP kinase (MAPK) signaling pathways (1). The p42MAPK and p44MAPK enzymes [extracellular signal–regulated kinase (ERK)–2 and ERK-1] are activated in cells stimulated with mitogens, by phosphorylation on threonine and tyrosine residues within protein kinase subdomain VIII, mediated by a class of MAP kinase (or ERK) kinases typified by MEK1 (2). Inhibition of p42MAPK and p44MAPK blocks cell-cycle reentry (3) and is principally mediated in vivo by members of a family of dual specificity phosphatases, of which MAP kinase phosphatase (MKP-1, also called 3CH134, CL100, or erp) is archetypal (4, 5). At least nine distinct MKP family members have been cloned, most, if not all, of which are the products of immediate early genes and therefore under tight transcriptional control (5, 6). MKP-1, MKP-2, and MKP-3 are transiently synthesized after activation of p42MAPK and p44MAPK, suggesting the presence of a negative feedback loop to regulate p42MAPK and p44MAPK (5–7). To determine whether expression of the MKP-1 protein is also subject to control, we determined the half-life of MKP-1 in CCL39 fibroblast cells (Fig. 1A). MKP-1 was barely detectable in quiescent CCL39 fibroblasts, but its expression level was increased in cells stimulated with mitogens before [35S]methionine labeling and immunoprecipitation (8). Its half-life was on the order of 45 min.

Figure 1

Lability of MKP-1. (A) The half-life of MKP-1 was determined in quiescent CCL39 cells after cell treatment with mitogen [thrombin (1 U/ml) + bovine fibroblast growth factor (bFGF) (10 ng/ml)] for 3 hours,35S-methionine labeling for 30 min, and addition of unlabeled methionine (1 μM). MKP-1 was immunoprecipitated at the times indicated (chase; h, hours) (8), resolved on SDS-PAGE (8.5% gel), and revealed by autoradiography. The bottom panel shows quantitation of immunoprecipitated 35S-labeled MKP-1 by band excision and counting in a scintillation counter. NS, no stimulation. (B) Expression of MKP-1 was increased in quiescent CCL39 cells by fetal bovine serum (FBS) (10%, 3 hours) before addition of cycloheximide (CHX, 10 μg/ml) and either LLnL (50 μM), E64 (10 μM), or vehicle (Con). At the times indicated, cells were lysed in a solution of 1% SDS and 4 M urea, and proteins were immunoblotted to reveal MKP-1 (7). The bottom panel shows quantitation of MKP-1 in the protein immunoblot by densitometric scanning (MKP-1 at time 0 is 100%). (C) Ubiquitination of MKP-1 in vivo. HEK 293 cells were transfected with expression plasmids encoding His-ubiquitin (His-Ub) and MKP-1 (7,12). After 30 hours, cells were rinsed with ice-cold PBS, and His-ubiquitin conjugates were purified (12,13). MKP-1 (left) and ubiquitin (right) were detected by protein immunoblotting with antibody to MKP-1 or antisera to ubiquitin, respectively (7). The position of MKP-1–ubiquitin conjugates is indicated on the right of each immunoblot.

Because many labile proteins are targeted for degradation by the ubiquitin-directed proteasome complex (9), we analyzed the effects of a proteasome inhibitorN-acteyl-leu-leu-norleucinal (LLnL) (10) and a lysosomal cysteine protease inhibitor,N-[N-(l-3-trans-carboxiraine-2-carbonyl)-l-leucyl]-agmatine (E64) (10) on MKP-1 degradation (Fig. 1B). After treatment of cells with serum to induce synthesis of MKP-1 and blockade of further protein synthesis by cycloheximide, the amount of MKP-1 decreased at a similar rate in control cells and in the presence of E64, but the rate of degradation of MKP-1 was decreased in the presence of the proteasome inhibitor. Two other proteasome inhibitors Cbz-LLLal (MG132) and Lactacystin (11) also attenuated the rate of MKP-1 degradation. MKP-1 was expressed in human embryonic kidney (HEK) 293 cells together with histidine-tagged ubiquitin. The experimental design (12, 13) allows the isolation of multiubiquitinated forms of MKP-1, which are visible as ladders of immunoreactivity on an immunoblot (Fig. 1C).

The inhibition of the proteasome (Fig. 2) was sufficient, on its own, to increase the abundance of MKP-1 and, to a lesser extent, that of MKP-2 in quiescent CCL39 cells (Fig. 2A). After treatment of cells with LLnL, the addition of serum promoted a reduction in MKP-1 mobility in SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 2B). To test whether the reduction in MKP-1 mobility was a consequence of phosphorylation, we treated cell extracts containing MKP-1 with phosphatase lambda, which led to the elimination of the more slowly migrating form of MKP-1 (Fig. 2C). Thus, MKP-1 is phosphorylated in response to serum. Examination of MKP-1 primary structure revealed several consensus phosphorylation sites for p42MAPK or p44MAPK. To assess whether MKP-1 was phosphorylated by p42MAPK or p44MAPK in vivo, we used CCL39 cells expressing a Raf:estrogen receptor (ER) chimera (14), which allows conditional activation of Raf after treatment of the cells with estradiol and concomitant activation of p45MEK1, p42MAPK, and p44MAPK. Treatment of ΔRaf-1::ER-expressing cells labeled with inorganic phosphate with LLnL increased the abundance of MKP-1. When these cells were treated with estradiol, phosphorylation of MKP-1 was increased (Fig. 2D). The inclusion of the p45MEK1inhibitor PD098059 blocked this phosphorylation, indicating that MKP-1 is phosphorylated in an p45MEK1-dependent manner.

Figure 2

Expression of MKP-1 in response to FBS in the presence of proteasome inhibitors. (A) Quiescent CCL39 cells were treated with LLnL (50 μM) for the times indicated, before cell lysis and MKP-1 immunoblotting as described in the legend to Fig. 1. (B) Quiescent CCL39 cells were treated with LLnL (50 μM) for 2 hours, before being stimulated with FBS (10%) for the times indicated; cells were lysed, and MKP-1 was detected by immunoblotting (7, 8). The presence of a more slowly migrating MKP-1 species is indicated by the letter P and is visible 0.5 hours after FBS addition. At 2 hours, MKP-2 is detectable as a poorly resolved band migrating more slowly than MKP-1–P. An additional slowly migrating band is unidentified. (C) MKP-1 is phosphorylated in vivo: ΔRaf-1::ER chimera–expressing cells (14) were treated with LLnL (50 μM) for 4 hours before FBS stimulation (10%, 10 min), cell lysis, and phosphatase λ (Phos.λ) treatment (17). MKP-1 was detected by immunoblotting (7, 8). (D) ΔRaf-1::ER chimera–expressing cells were incubated with 32Pi (500 μCi/ml) in phosphate-free medium and left untreated (Con) or treated with LLnL (L, 50 μM) for 4 hours. Cells were stimulated with estradiol (E, 10 μM) with or without PD098059 (PD, 30 μM). Cell lysates were subdivided and MKP-1 was immunoprecipitated, resolved on SDS-PAGE (11% gel), and revealed by autoradiography (top) or protein immunoblotting (bottom) (7).

Purified, active p44MAPKphosphorylated MKP-1 in vitro (15, 16) (Fig. 3A). This phosphorylation was rapid and reversible after addition of phosphatase lambda (17). Purified p45MEK1 did not phosphorylate MKP-1 under similar conditions but did phosphorylate catalytically inactive p44MAPK (15). Although in vitro–translated MKP-1 dephosphorylated p42MAPKand p44MAPK, no dephosphorylation of MKP-1 itself was detected (18). Taken together, our data indicate that MKP-1 is a substrate for p42MAPK or p44MAPK. To identify the site or sites of p44MAPK-mediated MKP-1 phosphorylation, we constructed an MKP-1 truncation mutant, MKP-1(1-340), lacking the last 27 COOH-terminal amino acids (19) (Fig. 3B). This mutant did not undergo phosphorylation by p44MAPK. The p42MAPK and p44MAPKenzymes phosphorylate substrates containing PX(S/T)P (P, proline; X, any neutral or basic amino acid; S/T, serine or threonine) motifs (20). However the first proline is not absolutely required (21). Analysis of the last 27 COOH-terminal amino acids of MKP-1 revealed two possible phosphorylation sites, Ser359 and Ser365, which are in the consensus sequence XXSP (22). Each serine was mutated to alanine (A), either individually or in combination. Mobility shift experiments and in vitro kinase assays revealed that each single MKP-1 mutant exhibited a slightly reduced mobility in SDS-PAGE (Fig. 3C). Both single mutants were also phosphorylated by p44MAPK, albeit less extensively than wild-type MKP-1 (Fig. 3C). The double mutant, MKP-1 (S359A, S364A), showed no reduced mobility in SDS-PAGE and was a poor p44MAPK substrate. MKP-1 (S359A, S364A) expressed in HEK293 cells was not phosphorylated after activation of the p42MAPKor p44MAPK cascade in vivo (Fig. 3D). Hence, the major sites of p44MAPK-mediated MKP-1 phosphorylation are serines 359 and 364.

Figure 3

Phosphorylation of MKP-1 by p44MAPK. (A) MKP-1 was translated in vitro with 35S-methionine according to the manufacturer's instructions (Promega) and incubated in an in vitro kinase assay (15) with p44MAPK (16). Samples were removed at the times indicated and resolved by SDS-PAGE (8.5% gel), and MKP-1 was revealed by autoradiography. (B) MKP-1 and MKP-1(1-340) (19) were treated with p44MAPK for 10 min as described (A). (C) (Top) MKP-1 was mutated to produce the single mutants S359A and S364A and the double mutant S359-364A (19) and treated with p44MAPKfor 10 min as described (A). (Bottom) Wild-type (WT) MKP-1 and the above-mentioned mutants were translated in vitro with unlabeled methionine, subjected to an in vitro kinase assay with p44MAPK and [γ32P]ATP (50 μM, 2 μCi per sample), immunoprecipitated, and detected by protein immunoblotting and autoradiography. (D) HEK 293 cells were transiently transfected with expression plasmids encoding Myc-epitope–tagged wild-type MKP-1 or the double mutant S359-364A (SA/SA), together with plasmid encoding the ΔRaf-1::ER chimera (14). Twenty-four hours after transfection, the cells were labeled with32Pi (500 μCi/ml) in phosphate-free medium for 4 hours before stimulation with estradiol (E, 1 μM) for 15 min (Con, no stimulation). MKP-1 was immunoprecipitated, resolved on SDS-PAGE (8.5% gel), and revealed by autoradiography (top) or protein immunoblotted (bottom) as described (7). In all cases, phosphorylated MKP-1 is indicated as MKP-1–P.

Because phosphatases undergo a phosphorylation-dependent modulation of activity (23), we sought to determine whether p42MAPK- or p44MAPK-mediated phosphorylation of MKP-1 modified its activity (Fig. 4). We increased expression of MKP-1 by treating ΔRaf-1::ER-expressing cells with LLnL. After addition of estradiol, the ability of phosphorylated MKP-1 to hydrolyze para-nitrophenylphosphate (p-NPP) (Fig. 4A) and to inactivate p44MAPK was determined (Fig. 4B). Phosphorylated and nonphosphorylated MKP-1 had equal activity to hydrolyze p-NPP and dephosphorylate p44MAPK. Thus, it appears that the phosphorylation of MKP-1 by p42MAPK or p44MAPK does not appreciably modify the phosphatase activity of MKP-1.

Figure 4

Phosphorylation of MKP-1 by p42MAPK and p44MAPK does not modify MKP-1 activity. ΔRaf-1::ER chimera–expressing cells were treated with LLnL (50 μM) for 4 hours. Cells were then left untreated or treated with estradiol (1μM, 15 min), and MKP-1 was immunoprecipitated. (A) (Top) The ability of nonphosphorylated (MKP-1) or p42MAPK- or p44MAPK-phosphorylated MKP-1 (MKP-1–P) to hydrolyze p-NPP was determined at the times indicated (18). (Bottom) Immunoprecipitated MKP-1 and MKP-1–P were subjected to protein immunoblotting. (B) p44MAPK was in vitro dephosphorylated by immunoprecipitated MKP-1 or MKP-1–P [produced as in (A)] (7, 26), and remaining p44MAPK activity was determined by an in vitro kinase assay with PHAS-I as substrate (15) (top). The bottom panel shows the quantification of the data by densitometric scanning.

The transcription factors c-Fos and c-Jun, which are both subject to ubiquitin-directed proteasome degradation, are stabilized as a direct result of MAP kinase–mediated phosphorylation (24). To investigate whether p44MAPK-mediated phosphorylation of MKP-1 modified its stability, we increased expression of MKP-1 by treating ΔRaf-1::ER-expressing cells with a submaximal concentration of LLnL, which can be removed by repeated washing. After addition of cycloheximide to block further protein synthesis, cells were left untreated or were treated with estradiol to activate p42MAPK or p44MAPK and phosphorylate MKP-1 (Fig. 5). The degradation rate of MKP-1 after activation of p42MAPK or p44MAPK was less than half that in the absence of estradiol. Under the same conditions, the degradation rate of inhibitor of κBα (IκBα) in response to tumor necrosis factor–α was unaltered in ΔRaf-1::ER-expressing cells after addition of estradiol (18). Hence, phosphorylation of MKP-1 by p42MAPK or p44MAPK serves to reduce ubiquitin-dependent degradation of the phosphatase. Activation of p42MAPK or p44MAPK therefore regulates MKP-1 protein expression through both an up-regulation of the rate of transcription (7) and a reduction in the rate of proteasome-mediated degradation.

Figure 5

MKP-1 is stabilized by activation of the p42MAPK and p44MAPK cascade. ΔRaf-1::ER chimera–expressing cells were treated with LLnL (10 μM) for 2 hours, before extensive cell washing to remove LLnL. Cells were treated with cycloheximide (CHX, 10 μg/ml) or CHX with estradiol (1 μM). Cells were lysed at the times indicated, and MKP-1 was detected by protein immunoblotting (top) followed by densitometric quantitation (bottom).

The p42MAPK and p44MAPK enzymes have a central role in the capacity of cells to divide in response to growth factors. Activation of p42MAPK and p44MAPK is a prerequisite for cell-cycle reentry (3). However, inappropriate or constitutive activation of the p42MAPK or p44MAPK cascade may provoke cellular senescence (25). Taken together, these findings illustrate a complex control mechanism designed to limit undesirable long-term activation of p42MAPK and p44MAPK and further demonstrate the importance of regulated protein degradation to the control of cell division processes.

  • * Present address: The Scripps Research Institute, Department of Molecular Biology, MB-3, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA.

  • To whom correspondence should be addressed. E-mail: mckenzie{at}


Stay Connected to Science

Navigate This Article