• extracellular signal-regulated kinase (ERK);
  • metal-dependent protein phosphatase (PPM);
  • phosphatase;
  • phosphorylation;
  • substrate specificity


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Protein phosphorylation is an important post-translational modification that regulates almost every aspect of signal transduction in cells. Activation of the mitogen-activated protein kinase (MAPK) family kinase extracellular signal-regulated kinase (ERK) is a point of convergence for many cellular activities in response to external stimulation. With stimuli, ERK activity is significantly increased by the phosphorylation of Thr202 and Tyr204 at its activation loop. Downregulation of ERK phosphorylation at these two sites by several phosphatases, such as protein phosphatase 2A, HePTP and MAPK phosphatase 3, is essential for maintaining appropriate ERK function in different cellular processes. However, it is unknown whether metal-dependent protein phosphatase (PPM) family phosphatases directly dephosphorylate ERK. In this study, we found that PPM1A negatively regulated ERK by directly dephosphorylating its pThr202 position early in EGF stimulation. Additional kinetic studies revealed that key residues participated in phospho-ERK recognition by PPM1A. Importantly, PPM1A preferred the phospho-ERK peptide sequence over a panel of other phosphopeptides through the interactions of basic residues in the active site of PPM1A with the pThr-Glu-pTyr motif of ERK. Whereas Lys165 and Arg33 were required for efficient catalysis or phosphosubstrate binding of PPM1A, Gln185 and Arg186 were determinants of PPM1A substrate specificity. The interaction between Arg186 of PPM1A and Glu203 and pTyr204 of phospho-ERK was identified as a hot-spot for phospho-ERK–PPM1A interaction.

Structured digital abstract


epidermal growth factor


extracellular signal-regulated kinase






human embryonic kidney 293


c-Jun N-terminal kinase


mitogen-activated protein kinase


2-amino-6-mercapto-7-methylpurine ribonucleoside


mitogen-activated protein kinase phosphatase 3


Protein Data Bank


p-nitrophenyl phosphate


protein phosphatase 2A


metal-dependent protein phosphatase


protein tyrosine phosphatase


small interfering RNA


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Protein phosphorylation is one of the most common post-translational modifications that modulate mammalian signaling pathways, and has been extensively studied since its discovery 57 years ago. To date, more than 100 000 reports have described phosphorylation events, and this number is still increasing [1]. Two classes of enzyme – protein kinases and protein phosphatases – coordinate the regulation of protein phosphorylation levels in cells [2-5]. The human genome encodes > 500 kinases and > 140 phosphatases. Many of these enzymes form clusters in one or several pathways, and are major transducers of signaling cascades.

The activation of mitogen-activated protein kinases (MAPKs) occurs downstream of many extracellular stimuli, such as hormonal actions or immune responses, through G-protein-coupled receptors, receptor tyrosine kinases, and ion channels. In mammals, there are three major MAPK families: the extracellular signal-regulated kinase (ERK) family (ERK1–ERK8), the c-Jun N-terminal kinase (JNK) family (JNK1–JNK3), and the p38 kinase family (p38α, p38β, p38γ, and p38δ) [6, 7]. Among these kinases, ERK was the first identified MAPK member, and serves as the pivotal component in transmitting the extracellular signal to cause diverse downstream outcomes, such as cell cycle progression, apoptosis, differentiation, and migration [8, 9]. Given the plethora of ERK functions, it is not surprising that > 50 cytoplasmic proteins, as well as many transcriptional factors in the nucleus, are known ERK substrates [10]. To precisely transmit extracellular signals to distinct cellular mechanisms, ERK activation is delicately regulated both temporally and spatially, and has been reported with a duration ranging from 20 min to 3 h [7]. ERK is activated by the action of upstream kinases and adaptor proteins, and it is often marked by dual phosphorylation of Thr202 and Tyr204 at its activation loop. However, inactivation of ERK2 is mediated mainly by members of protein phosphatase families.

Human protein phosphatases consist of 140 members, and can be divided into two major classes, Ser/Thr phosphatases and protein tyrosine phosphatases (PTPs). Protein Ser/Thr phosphatases can be further divided into two structurally unrelated families, phosphoprotein phosphatases and metal-dependent protein phosphatases (PPMs). Unlike Ser/Thr phosphatases, which show pSer/pThr substrate specificity, PTPs have a broad variety of substrates [2]. In particular, MAKP phosphatase 3 (MKP3), which is a dual-specificity phosphatase, can hydrolyse both pTyr204 and pThr202 of phospho-ERK. Whereas the action of MKP3 on ERK2 dephosphorylation reduces ERK activity by 10 000-fold, the expression of MKP3 is only induced 30 min after ERK activation. Therefore, the regulation of ERK activation at an early time point is most likely attributable to the effects of other phosphatase members.

The dephosphorylation of doubly phosphorylated ERK by tyrosine phosphatases, such as HePTP, PTP-SL, and STEP, has been extensively investigated. With regard to Ser/Thr phosphatases, only protein phosphatase 2A (PP2A) has been shown to be an ERK phosphatase in cells [11-13]. The Km of PP2A for doubly phosphorylated ERK2-pT202/pY204 protein in vitro is > 20 μm, suggesting that ERK2-pT202/pY204 may not be a preferred substrate of PP2A alone [14].

The PPM family phosphatase PPM1A also shows reasonable phosphatase activity on purified doubly phosphorylated ERK in vitro [14]. PPM1A is the prototype PPM family member, and has been reported to be a major negative regulator of the MAPK signaling pathway through genetic screens in budding yeast [15, 16]. In mammals, phospho-p38, phospho-JNK and phospho-SMAD have all been demonstrated to be PPM1A substrates, but direct evidence of PPM1A as an ERK phosphatase has not been reported [17-19]. Therefore, in this study, we determined whether phospho-ERK is a direct substrate of PPM1A. Furthermore, we explored the underlying mechanism of phospho-ERK recognition by PPM1A through combined peptide Ala scanning and kinetic analysis. We identified PPM1A as a negative regulator of phospho-ERK at early time points through specific interactions of its unique basic residues located in the active site of the phospho-ERK protein.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

PPM1A negatively regulates epidermal growth factor (EGF)-induced ERK phosphorylation

Previous studies have demonstrated that the expression of the dual-specificity ERK phosphatase MKP3 is only induced 30 min after ERK activation [20]. Therefore, other phosphatases, such as PPM1A, may be responsible for controlling the ERK phosphorylation state at earlier time points. To investigate PPM1A regulation of ERK phosphorylation, we knocked down PPM1A expression in human embryonic kidney 293 (HEK293) cells, and followed this with treatment with 5 ng·mL−1 EGF to induce ERK activation. PPM1A small interfering RNA (siRNA) reduced PPM1A expression by nine-fold, which resulted in a three- to four-fold increase in ERK phosphorylation after 5 min of EGF stimulation (Fig. 1A,B). We then monitored the time-course of ERK phosphorylation after PPM1A knockdown. Cells with or without PPM1A knockdown were stimulated with 5 ng·mL−1 EGF for 0, 0.5, 2, 5, 30, 60 and 120 min. After PPM1A knockdown, the level of ERK phosphorylation was significantly increased in the early phase from 30 s to 30 min, but it was reduced in the later phase after 30 min (Fig. 1C,D). These results demonstrated that PPM1A is a negative regulator of ERK phosphorylation at early time points.


Figure 1. PPM1A negatively regulates EGF-induced ERK phosphorylation at early time points. (A) Knockdown of PPM1A increased EGF-induced ERK phosphorylation. PPM1A expression was specifically knocked down by transfecting HEK293 cells with PPM1A siRNA. Cells were stimulated with 5 ng·mL−1 EGF for 5 min after 12 h of starvation, and phospho-ERK was detected with a specific phospho-ERK1/2-pT202pY204 antibody. Actin levels were used as the loading control, and the expression of PPM1A was monitored with a specific PPM1A antibody. (B) Statistical analysis of ERK phosphorylation for the western blot presented in (A). Data were from at least three independent experiments. *P < 0.05, cells stimulated with EGF as compared with nonstimulated HEK293 cells. #P < 0.05, knockdown of PPM1A in HEK293 cells as compared with control. (C) Time course of EGF-induced ERK phosphorylation with or without PPM1A knockdown. After starvation, cells were stimulated with 5 ng·mL−1 EGF for various times, as indicated. Levels of phospho-ERK1/2 were monitored with phospho-ERK1/2-pT202pY204 antibody. (D) Statistical analysis of ERK phosphorylation for the immunoblot shown in (C). *P < 0.05, cells with PPM1A knockdown as compared with control cells. IB, immunoblot; RNAi, RNA interference.

Download figure to PowerPoint

Direct interaction between phospho-ERK and PPM1A

We next used the glutathione-S-transferase (GST) pull-down method to determine whether PPM1A negatively regulates ERK phosphorylation through direct interaction with phospho-ERK. PPM1A is a metal-dependent phosphatase, and its activity requires binding of a divalent metal at its active site to develop the negative charge for a bridging water [21, 22]. Whereas Mn2+, Mg2+ and Fe2+ efficiently activate PPM1A, some divalent metals, such as Zn2+ and Ca2+, inhibit its phosphatase activity [22]. Therefore, we incubated equal amounts of GST–PPM1A with EGF-treated cell lysates in the presence of either 2 mm EDTA, Mn2+, or Zn2+, with GST protein as a negative control. Neither the interaction between phospho-ERK and GST nor the interaction between ERK and PPM1A in the presence of EDTA or its activator Mn2+ were detectable by western blot (Fig. 2A). However, the interaction between phospho-ERK and PPM1A in the presence of its inhibitor, Zn2+, was detected (Fig. 2A). As an inhibitor of PPM1A, Zn2+ stabilized the active site of PPM1A, but EDTA did not. Thus, the phospho-ERK–PPM1A interaction was dependent on the conformation of the PPM1A active site. Phosphate was also found to be important for the phospho-ERK–PPM1A interaction, which was demonstrated by the reduction in ERK–PPM1A interaction in the presence of the PPM1A activator Mn2+.


Figure 2. PPM1A directly interacts with phospho-ERK2. (A) GST–PPM1A specifically interacted with phospho-ERK2 in cell lysates. HEK293 cells were stimulated with 5 ng·mL−1 EGF for 5 min at 37 °C. Equal amounts of EGF-stimulated HEK293 cell lysates were incubated with GST–PPM1A or GST protein under the following three conditions: 2 mm EDTA, 1 mm Mn2+, or 2 mm Zn2+. The PPM1A-interacting proteins were pulled down by GST beads and subjected to western blot analysis. (B) PPM1A interaction with phospho-ERK2 protein in vitro. GST–PPM1A (100 nm) or GST (100 nm) protein was incubated with 100 nm purified bisphosphorylated ERK protein in vitro under different conditions. Phospho-ERK binding was examined by western blot analysis.

Download figure to PowerPoint

We further assessed the phospho-ERK–PPM1A interaction in vitro with purified PPM1A and phospho-ERK, to eliminate the potential effects of other proteins in cells. The phospho-ERK–PPM1A interaction was detected in the presence of Zn2+ and, to a lesser extent, in the presence of EDTA, but the interaction was not detectable in the presence of Mn2+. These results demonstrated the direct interaction between PPM1A and phospho-ERK (Fig. 2B). Detection of the ERK–PPM1A interaction in the presence of EDTA may be attributable to the increased level of phospho-ERK.

Preference of PPM1A for the phosphopeptide sequence

Previous studies have shown that PPM1A has phosphatase activity on the doubly phosphorylated ERK protein in vitro, with a Km of 0.2 μm at 30 °C [14]. The substrate specificity of a phosphatase can be defined by both the phosphopeptide sequence and other interacting domain/motifs [4, 5]. In the case of PPM1A, its activity on phosphopeptide substrates is 90-fold higher than that on the small artificial substrate p-nitrophenyl phosphate (pNPP), indicating that the peptide sequence significantly contributes to the substrate recognition by PPM1A [22].

Therefore, we next measured the phosphatase activity of PPM1A on phosphopeptides derived from ERK, p38, histone proteins, and glutamate receptors. p38 is a known substrate of PPM1A, and the histone proteins and glutamate receptors are proposed to be substrates for another PPM family phosphatase, PPM1G [23, 24]. PPM1A efficiently dephosphorylated the phospho-ERK peptide, with a four-fold higher activity than that on the phosphopeptide derived from p38, which is a known PPM1A substrate (Table 1). Whereas no difference was seen in the kcat/Km of PPM1A for the monophosphorylated or biphosphorylated peptide derived from p38, PPM1A had a two-fold preference for the biphosphorylated ERK peptide over the monophosphorylated-Thr ERK peptide. This result is consistent with the previous finding that PPM1A prefers the biphosphorylated over the monophosphorylated ERK protein in vitro [14].

Table 1. Kinetic parameters of the activity of PPM1A towards phosphopeptides. The reactions were performed in Tris/Bis-Tris/acetate buffer at pH 7.0 and 37 °C. All data were fitted to the Michaelis–Menten equation with nonlinear least-squares regression (graphpad prism). The data are the average of at least three independent measurements
PhosphopeptidespSer or pThr peptide sequenceKmm)kcat (s−1)kcat/Km (104, m−1·s−1)
ERK-pT202pY204Ac-TGFLpTEpYVAT R-NH27.30 ± 0.260.43 ± 0.055.90 ± 0.51
ERK-pT202Ac-TGFLpTEYVAT R-NH218.47 ± 2.010.36 ± 0.042.57 ± 0.31
p38pT180pY182Ac-TDDEMpTGpYVAT-NH228.1 ± 8.50.22 ± 0.050.78 ± 0.092
p38pT180Ac-TDDEMpTGYVAT-NH233.5 ± 6.30.19 ± 0.060.57 ± 0.063
Histone H2ApS140Ac-KKATQApSQEYNH2 > 100
Histone H2BpS11Ac-APKKGpSKKAVT K-NH2> 100
Histone H2BpS13Ac-KGATIpSKKGFK K-NH2> 100
Glutamate R3-pS419Ac-HLNRFpSVSGT-NH2 > 100

The two-fold preference for the doubly phosphorylated ERK over the singly phosphorylated ERK indicated that the pThr+2 position residue, pTyr204, may mediate some interactions between ERK and PPM1A. However, this result contradicts the finding of no difference between the singly and doubly phosphorylated p38. One explanation could be that the binding modes of the p38 phosphopeptide and ERK phosphopeptide for PPM1A are different, which may be partially attributable to the residue preceding the pTyr residue. Whereas phospho-ERK has a Glu203 preceding pTyr204, p38 has a Gly181 at the corresponding position. Consistent with this hypothesis, when the solved phospho-ERK and phospho-p38 structures were superimposed, the activation loop and pTyr of these kinases assumed different conformations (Fig. S1).

In addition, PPM1A showed poor activity on phosphopeptides derived from histone proteins or glutamate receptors, which are the predicted substrates of PPM1G. Together, these results demonstrated that the phospho-ERK peptide sequence is one of the determinants of of phospho-ERK recognition by PPM1A.

Ala scanning of PPM1A-catalyzed phospho-ERK peptide dephosphorylation

To reveal the determinants of PPM1A-catalyzed ERK dephosphorylation localized in the phospho-ERK peptide sequence, we mutated each amino acid in the phospho-ERK peptide to Ala, and measured the kinetics of phosphopeptide dephosphorylation by PPM1A (Fig. 3A,B). Mutations of the two C-terminal acidic residues after pThr (E203A and pY204A) significantly decreased the kcat/Km (Fig. 3A,B). Both of the mutations had pronounced effects only on Km, with a 6.5-fold increase for the E203A mutant and a 2.5-fold increase for the pY204A mutant, suggesting that these two residues are involved only in substrate binding. In conclusion, PPM1A recognizes Glu203 and pTyr204 in the phospho-ERK peptide sequence.


Figure 3. Kinetic parameters of PPM1A phosphatase activity on an Ala scanning library derived from a phospho-ERK peptide. (A) Graph of relative kcat/Km values of PPM1A activity on phosphopeptide. Data are from at least three independent experiments. *P < 0.05. (B) kcat, Km and kcat/Km values of PPM1A activity on variants of phospho-ERK2 peptide in which every residue from Thr198 to Arg208 was replaced by Ala. Data are the averages of three independent measurements.

Download figure to PowerPoint

Effect of Glu203 mutation in the phospho-ERK peptide on PPM1A catalysis

The E203A mutation had a pronounced effect on Km, which prompted us to investigate whether the pTyr+1 residue of Glu203 plays an important role in the phospho-ERK–PPM1A interaction. Therefore, we mutated Glu203 to the corresponding residue in the same position of histone-H2A, glutamate receptor, histone-H2B, and p38. The Glu[RIGHTWARDS ARROW]Gln mutation increased the Km by two-fold, the Glu[RIGHTWARDS ARROW]Lys or Glu[RIGHTWARDS ARROW]Val mutation increased the Km by three-fold, and the Glu[RIGHTWARDS ARROW]Gly mutation increased the Km by six-fold (Table 2). These results suggested that a longer side chain at the pTyr+1 position in the phospho-ERK peptide aids its binding to PPM1A.

Table 2. Kinetic parameters of the activity of PPM1A toward mutants of ERK phosphopeptides. Reactions were performed in Tris/Bis-Tris/acetate buffer at pH 7.0 and 37 °C. All data were fitted to the Michaelis–Menten equation with nonlinear least-squares regression (graphpad prism). The data are the average of at least three independent measurements
PhosphopeptidesPeptide sequenceKmm)kcat (s−1)kcat/Km (104, m−1·s−1)kcat/Km
ERK-pT202EpY204Ac-TGFLpTEpYVAT R-NH27.30 ± 0.260.43 ± 0.055.90 ± 0.511
ERK-pT202QpY204Ac-TGFLpTQpYVAT R-NH216.25 ± 0.370.84 ± 0.035.20 ± 0.141.13
ERK-pT202VpY204Ac-TGFLpTVpYVAT R-NH223.98 ± 2.490.94 ± 0.0163.90 ± 0.081.51
ERK-pT202KpY204Ac-TGFLpTKpYVAT R-NH224.86 ± 3.220.91 ± 0.123.64 ± 0.461.62
ERK-pT202GpY204Ac-TGFLpTGpYVAT R-NH242.10 ± 7.030.70 ± 0.041.66 ± 0.533.55

Requirement of Arg33 for phospho-ERK recognition

We next investigated which PPM1A residues are responsible for recognizing Glu203 and pTyr204 of phospho-ERK. In a previous study on several phosphatases, the residues localized in the active site were found to be determinants for peptide sequence recognition [4]. Therefore, we searched the PPM1A active site for potential determinants of ERK–PPM1A interactions.

Four basic charged residues are localized in the PPM1A active site: Arg33, His62, Lys165, and Arg186. Crystallographic studies have identified a direct interaction of Arg33 with a phosphate, suggesting that Arg33 is required for the phosphosubstrate interaction. However, five PPM family members have a Lys and two members have an Asn at the positions corresponding to PPM1A Arg33, indicating that an Arg is not required for all phosphosubstrate recognition by PPM family members (Fig. 4C). Thus, we generated the PPM1A mutations R33A, R33K, and R33N, and investigated the effects of the mutations on PPM1A activity on pNPP, a phosphopeptide derived from ERK and phospho-ERK protein (Fig. 4B; Table 3).

Table 3. Kinetic parameters of the activity of PPM1A and its mutants towards pNPP and dual phospho-ERK peptide
 pNPPERK-pT202pY204 peptide
Km (mm)kcat (s−1)kcat/Km (m−1·s−1)kcat/KmKmm)kcat (s−1)kcat/Km (104, m−1·s−1)kcat/Km
PPM1A-WT3.52 ± 0.402.13 ± 0.51605.11 ± 29.617.30 ± 0.640.43 ± 0.265.90 ± 0.511
PPM1A-R33A29.37 ± 6.080.55 ± 0.03218.72 ± 5.0832.32143.21 ± 8.650.032 ± 0.00230.022 ± 0.067268.2
PPM1A-R33K8.44 ± 1.541.53 ± 0.10181.27 ± 4.583.3441.62 ± 2.360.148 ± 0.0320.36 ± 0.2316.39
PPM1A-R33N22.48 ± 0.591.41 ± 0.1662.72 ± 5.279.6454.16 ± 1.260.146 ± 0.0470.27 ± 0.05421.85
PPM1A-K165A6.81 ± 1.461.24 ± 0.12182.08 ± 8.313.3228.34 ± 4.680.169 ± 0.00390.60 ± 0.169.83
PPM1A-Q185A3.68 ± 0.790.97 ± 0.19263.58 ± 8.072.3016.15 ± 3.690.179 ± 0.00861.10 ± 0.105.36
PPM1A-R186A7.34 ± 0.844.29 ± 1.05584.49 ± 9.221.03134.63 ± 16.370.235 ± 0.00780.17 ± 0.07934.70

Figure 4. Kinetic parameters of phospho-ERK protein dephosphorylation by PPM1A and its mutants. (A) Dependence of the initial velocity (A360 nm/min) on ERK2/pTpY concentration for the reaction catalyzed by PPM1A and its mutants. Data were fitted to the Michaelis–Menten equation. (B) Kinetic parameters of the activity of wild-type or mutated PPM1A phosphatase on phospho-ERK protein. (C) Sequence alignment of PPM family members around mutated sites. Numbers represent amino acid positions in PPM1A. (D) Stereo view of the participation of PPM1A Lys165 in the hydrogen network in the active site (PDB codes: 2ERK and 3FXK). Yellow dots indicate Mn2+, and blue asterisks indicate water. Residues from PPM1A are in cyan, and residues from ERK are in green. The blue dashed lines depict hydrogen bonds. Positively charged residues are labeled (+), and negatively charged residues are labeled (−). The figure was prepared with pymol.

Download figure to PowerPoint

At pH 7.0, the R33A mutation increased the Km for pNPP by 10-fold and the the Km for phospho-ERK peptide by 20-fold (Table 3). We next measured phospho-ERK protein dephosphorylation with the established enzyme-coupled continuous assay, and recapitulated the reaction for wild-type PPM1A at 37 °C [25]. As compared with previous results, wild-type PPM1A had a similar Km value for doubly phosphorylated ERK at 37 °C, but had a slight increase in kcat (Fig. 4A,B). We then tested the effect of the R33A mutation on phospho-ERK protein catalysis. The initial rate analysis indicated that the Km of R33A-catalyzed phospho-ERK dephosphorylation was > 4 μm, and that the kcat/Km of PPM1A-R33A was 100-fold less than that of the wild type. These results are in agreement with previous kinetic studies that reported a 10-fold increase in the substrate inhibition constant, Kis, for the R33A mutant [26]. Therefore, the R33A mutation impairs PPM1A activity on the phosphosubstrate mainly through the perturbation of substrate phosphate binding (Fig. 5B).


Figure 5. Interaction of PPM1A with phospho-ERK. (A) Overall mode of binding of PPM1A (tangerine) to phospho-ERK (green). Phosphorylated Thr202 is specifically labeled. (B) Structural model of the interaction of PPM1A-R33 with the phosphate of ERK-pT202. The blue dashed lines depict hydrogen bonds. Positively charged residues are labeled (+), and negatively charged residues are labeled (−). (C) Structural representation of the interaction between PPM1A-R185 and ERK-pY204, as well as the interaction between PPM1A-Q186 and ERK-K220. Residues from ERK are in green, and residues from PPM1A are in cyan. Also shown are kinetic parameters of the activity of PPM1A-R186A towards phospho-ERK peptide mutants in which pTyr204 was replaced by Tyr or Ala, or Glu203 was replaced by Ala.

Download figure to PowerPoint

Five PPM phosphatases – PPM1J, PPM1M, PHLLPP1, PPM1H, and PTPC7 – have a Lys instead of an Arg at the corresponding PPM1A Arg33 position (Fig. 4C). Thus, we made the PPM1A R33K mutant to investigate its effect on the phosphatase activity. The R33K mutation decreased the activity on pNPP and phosphopeptide hydrolysis by three-fold and 16-fold, respectively, whereas its Km increased by 2.6-fold and six-fold, respectively, as compared with the wild type (Table 3). This mutation also increased the Km for the phospho-ERK protein by three-fold, suggesting that the R33K mutation reduces the binding of PPM1A to all phosphosubstrates to a similar extent (Fig. 4A,B).

We also designed the R33N mutation to mimic two other phosphatases, PPM2C and PDP2 (Fig. 4D). This mutation increased the Km for pNPP, phospho-ERK peptide and phospho-ERK protein by four-fold to seven-fold, which was consistent with the basic charge requirement of Arg33 for phosphosubstrate recognition (Table 1; Fig. 4A,B). Taken together, the above results demonstrated that Arg33 is required for PPM1A recognition of phospho-ERK, mainly through binding to the phosphate of the substrate.

Important role of Lys165 in PPM1A phosphatase activity and its recognition of phospho-ERK protein

We next investigated, with kinetic studies, the effect of the basic residue Lys165 on PPM1A substrate catalysis and its recognition of phospho-ERK. Lys165 is a nonconserved residue in the PPM family, with Ser, Thr, Asn, Ile, Arg and Gln at the corresponding positions in the other family members (Fig. 4C). Interestingly, the K165A mutation increased the Km and decreased the kcat by approximately two-fold for the small artificial substrate pNPP (Table 3). The K165A mutation decreased the activity on phospho-ERK protein and pNPP to a similar extent (Fig. 4A,B). A detailed examination of the PPM1A crystal structure revealed that Lys165 participates in both the charge–charge interaction and a hydrogen bond network connecting to the important metal ion-coordinated residue Asp239 (Fig. 4D). The correct conformation of Asp239 is required for efficient catalysis, and removal of the Asp239 side chain significantly impairs PPM1A catalysis [23]. Therefore, the K165A mutation reduces the PPM1A activity on phosphosubstrates through perturbation of the active site.

Gln185 and Arg186 are determinants of phospho-ERK recognition

We also investigated the contributions of Arg186 and Gln185 to phospho-ERK recognition. The complex structural model predicted that they would participate in the phospho-ERK protein interaction (Fig. 5A–C). The Q185A mutation had no effect on the Km of pNPP, but it decreased the kcat by two-fold. For the phosphopeptide, the Q185A mutation increased the Km by two-fold, which was consistent with its closing position in the active site (Table 3). Interestingly, the Km for the phospho-ERK protein increased by 10-fold, suggesting that the Q185A mutant interacts with ERK through sites outside of the phospho-ERK peptide. In the model, Gln185 was predicted to interact with Lys220 of ERK (Fig. 5C). Lys220 has been shown to undergo substantial conformational change after ERK activation [27, 28]. Therefore, Gln185 is one of the determinants of phospho-ERK recognition by PPM1A.

Arg186 is a positively charged residue located in the PPM1A active site (Fig. 5C). The R186A mutation had no effect on kcat/Km for pNPP, but it decreased the activity on phospho-ERK peptide by 34-fold and that on phospho-ERK protein by 17-fold (Table 3; Fig. 4A,B). The impaired activity on the phosphopeptide and phospho-ERK protein was attributable to an increased Km, indicating that the Arg186 mutation mainly affected peptide and protein binding. Protruding towards the phosphopeptide, Arg186 can undergo direct interactions with the acidic residues Glu203 and pTyr204. In the model, Arg186 not only formed a salt bridge with pTyr204, but could also engage in polar and hydrophobic interactions with Glu203 of the phospho-ERK peptide (Fig. 5C). Therefore, we examined the catalytic action of the R186A mutant on several phosphopeptides, in which either pTyr204 was replaced by Tyr or Ala, or Glu203 was replaced by Ala. The R186A mutant had similar catalytic constants for these mutated phosphopeptides as the wild-type phospho-ERK peptide (Fig. 5D). In contrast, the Km of the wild-type PPM1A was increased by two-fold for the pTyr204 mutated phosphopeptides and by more than six-fold for the E203A peptide (Fig. 3B). These results may explain the large Km increases for the PPM1A Arg186 mutant, which underwent a combined interaction with Glu203 and pTyr204. Thus, we speculate that R186 is the determinant residue for both phospho-ERK peptide and phospho-ERK protein recognition through interaction with the pTyr at the pTyr+2 position (pTyr204) and the Glu pTyr+1.

Four PPM family members have either a Lys or an Asp at the corresponding Arg186 position of PPM1A, and 15 PPM family members have different residues at the Gln185 position (Fig. 4C). With the currently limited number of known PPM family member substrates, no clear correlation of the substrate sequence was found with the variety of Arg186 or Gln185 residues in corresponding PPM family members (Table S1). More kinetic experiments and more known substrates of PPM family members would help to evaluate the importance of Arg186 and Gln185 in the recognition of PPM1A substrates.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information


All of the peptides were synthesized by China Peptides Co. Ltd (Shanghai, China). The Biomol Gree Reagent for phosphate detection (BML-AK111) was from Enzo Life Sciences (Lausen, Switzerland). pNPP (4264-83-9) was from Sangon Biotech Co. Ltd (Shanghai, China). Glutathione (GSH)–Sepharose 4B and Ni2+–nitrilotriacetic acid agarose were from Amersham Pharmacia Biotech (Piscataway, NJ, USA). The ERK-pT202/pY204 antibody and the His-probe (H-3) horseradish peroxidase antibody were from Cell Signaling Technology (Danvers, MA, USA). The ERK2 antibody (H-9), GST antibody (1-109) and β-actin antibody (C4) were from Santa Cruz Biotechnology (Dallas, TX, USA). 2-Amino-6-mercapto-7-methylpurine ribonucleoside (MESG) was prepared as previously described [29]. The purine nucleoside phosphorylase was from Sigma (St. Louis, MO, USA), as were most of the other chemicals.

Constructs and mutagenesis

Human PPM1A was a kind gift from P. T. W. Cohen (MRC Protein Phosphorylation Unit). The cDNA of full-length PPM1A was subcloned into PET-30a and pGEX-6P2 expression vectors with an N-terminal His-tag and an N-terminal GST-tag, respectively. The PPM1A mutants (R33A, R33N, R33K, K165A, Q185A, and R186A) were generated with the QuikChange site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA). The OPC-purified oligonucleotide primers were from Beijing Genomics Institute, and all mutants were verified by DNA sequencing. The His-tagged ERK2 construct was a kind gift from R. J. Lefkowitz (Duke University).

Cell culture, PPM1A RNA interference, and western blotting

HEK293 cells were cultured in DMEM containing 10% fetal bovine serum, 25 mm glucose and 1% penicillin/streptomycin in a humidified incubator with 5% CO2 at 37 °C. To deplete PPM1A, the cells were transfected with siRNA against human PPM1A with Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) for 48 h. The sequence of the siRNA against PPM1A was 5′-UACAAAGUAAACCUCUUGAGUCUCC-3′. After 12 h of starvation, the cells were stimulated with 5 ng·mL−1 EGF for 0, 0.5, 2, 5, 30, 60 and 120 min at 37 °C. Subsequently, the cells were washed twice with ice-cold NaCl/Pi, and then harvested with lysis buffer [150 mm NaCl, 50 mm Tris/HCl pH 7.5, 1% Triton X-100, 1 mm Na3VO4, 1 mm EDTA, 50 mm NaF, 10% (v/v) glycerol, 0.25% (w/v) sodium deoxycholate, protease inhibitor cocktail tablet, and 5 mm iodoacetamide]. The cell lysate was incubated on ice for 30 min, and centrifuged at 17 000 g for 30 min at 4 °C. The sample protein concentrations were measured with a bicinchoninic acid protein quantitation kit (Beyotime, Shanghai, China). Equal amounts of cell lysates were denatured in 2× SDS loading buffer, and boiled for 10 min. Protein samples were then subjected to western blot analysis.

Protein expression and purification

The GST–PPM1A fusion protein and GST alone were generated as previously described [4]. In general, 3 L of Escherichia coli cells, which were transformed with GST–PPM1A or GST plasmids, were cultured at 37 °C, induced with 0.4 mm isopropyl thio-β-d-galactoside at 25 °C for 12 h, and then centrifuged at 2 000 g. The pellets were washed with GST buffer (50 mm Tris/HCl, pH 8.0, 150 mm NaCl, 0.1% Triton X-100, 5% glycerol, 1 mm EDTA, 2 mm dithiothreitol), and resuspended in 30 mL of ice-cold GST buffer. The pellets were then broken three times in a Processor cell press at 4 °C, and the lysates were centrifuged for 40 min at 17 000 g and 4 °C. The supernatant was incubated with 1 mL of GST beads for 1.5 h, and then pelleted at 100 g for 10 min at 4 °C. The supernatant was discarded, and GST beads were washed with 100 mL of ice-cold GST buffer at 4 °C. The bound GST–PPM1A protein was eluted with GSH buffer (10 mm GSH, 50 mm Tris/HCl, pH 8.0, 5% glycerol). Finally, the concentrated protein was diluted in Bis/Tris-Tris/acetate buffer (0.05 m Tris, 0.05 m Bis-Tris, 0.1 m acetate, pH 8.0). The protein was 99% pure as assessed by Coomassie staining after SDS/PAGE electrophoresis. The protein concentration was measured with the bicinchoninic acid protein quantitation kit (Beyotime), and the protein was then stored at −80 °C. Expression and purification of the His-tagged PPM1A and its mutants were performed as previously described [3].

Preparation of bisphosphorylated ERK2

The in vitro phosphorylation of His-tagged ERK2 was performed as previously described [24, 30], by adding 1 mg·mL−1 purified His-tagged ERK2 and 0.1 mg·mL−1 His-tagged MEK1/G7B to a 500-μL reaction system, which contained 10 mm Hepes (pH 7.4), 100 mm NaCl, 20 mm Mg(COOH)2, 2 mm dithiothreitol, and 0.5 mm ATP. The reaction mixture was incubated at 30 °C for 90 min. After the reaction, phospho-ERK2 was separated from free ATP by passing the mixture through a Superdex-200 column (GE Healthcare, Little Chalfont, UK). Phospho-ERK2 was further purified on a Mono-Q column, and the protein was then concentrated and stored at −80 °C.

GST pull-down assays

HEK293 cells were treated with 5 ng·mL−1 EGF for 5 min at 37 °C after overnight starvation. Cells were harvested, and lysed with lysis buffer [150 mm NaCl, 50 mm Tris/HCl pH 7.5, 1% Triton X-100, 1 mm Na3VO4, 1 mm EDTA, 50 mm NaF, 10% (v/v) glycerol, 0.25% (w/v) sodium deoxycholate, protease inhibitor cocktail tablet, 5 mm iodoacetamide] with or without 2 mm MnCl2 or 2 mm ZnCl2. Cell lysates were centrifuged at 17 000 g for 20 min. The supernatant was then incubated with an equal molar amount of purified GST or GST–PPM1A fusion protein together with 15 μL of GST beads under different conditions. The mixture was incubated in the binding buffer for 2 h at 4 °C with end-to-end mixing, and then centrifuged at 100 g for 20 min. The precipitated beads were washed four times in 1 mL of ice-cold binding buffer. Binding of phospho-ERK was detected by western blot analysis.

For purified protein, 100 nm purified GST protein or GST–PPM1G fusion protein was preincubated with 15 μL of GST beads in binding buffer (20 mm Hepes, pH 7.5, 150 mm NaCl, 1 mm dithiothreitol) with 1 mm EDTA or metal ions (2 mm MnCl2 or ZnCl2). Bisphosphorylated ERK2 (1 μg) was added and subjected to further end-to-end rotation at 4 °C. The precipitated beads were washed seven times in 1 mL of ice-cold binding buffer, and subjected to western blot analysis.

Assay for pNPP activity

The activity of PPM1A-catalyzed pNPP hydrolysis was determined as previously described [22]. All assays were performed at 37 °C in Tris/Bis-Tris/acetate buffer (0.05 m Tris, 0.05 m Bis-Tris, 0.1 m acetate) containing 1 mm dithiothreitol and 10 mm Mn2+ (pH 8.0). The reactions were initiated by adding PPM1A protein and halted by adding 0.5 m EDTA (pH 10.0). The activity was assessed by monitoring the absorbance of pNPP at 405 nm. Kinetic parameters were determined with nonlinear least-squares regression fitted to the Michaelis–Menten kinetic model [31].

Determination of kinetic parameters of phosphopeptide substrates

The hydrolysis of pSer/pThr-containing phosphopeptide catalyzed by PPM1A was determined with an inorganic phosphate assay, as previously described [3, 4]. Phosphopeptides were incubated with PPM1A in Tris/Bis-Tris/acetate buffer (0.05 m Tris, 0.05 m Bis-Tris, pH 7.0, 0.1 m acetate) containing 10 mm Mn2+ at 37 °C. The reactions were halted by the addition of BIOMOL GREEN (Enzo Life Sciences), and the absorbance at 620 nm was measured to determine phosphate release. The Km, kcat and kcat/Km values were fitted to the Michaelis–Menten equation with graphpad prism as follows:

  • display math(1)

Enzyme-coupled continuous assay for determination of PPM1A activity towards bisphosphorylated ERK2 protein

Kinetic parameters for the dephosphorylation of bisphosphorylated ERK by PPM1A and its mutants were determined with a continuous spectrophotometric assay [14]. All experiments were performed in the MESG-coupled system, which included dual-phosphorylated ERK, 0.1 mg·mL−1 purine nucleoside phosphorylase and 50 μm MESG in reaction buffer (50 mm Mops, pH 7.0, 100 mm NaCl, 1 mm dithiothreitol, 10 mm MnCl2) at room temperature. The reactions were initiated by adding PPM1A protein. The progression of the reaction was measured at A360 nm. For the condition with a similar molecular ratio of PPM1A and phospho-ERK protein, the kinetic parameters were fitted to the following equation with graphpad prism [14]:

  • display math(2)

For the phospho-ERK substrate concentration, which was lower than the Km, kinetic parameters were fitted to the following equation with graphpad prism [14]:

  • display math(3)

Generation of the docking model

The crystal structures of ERK [Protein Data Bank (PDB) ID: 2ERK] and PPM1A (PDB ID: 3FXK) were combined with hex 6.3. The complex was soaked in an orthorhombic box of simple point charge water molecules and 0.15 m NaCl with a margin of 10 Å [32]. Then, molecular minimization was performed into this system with 2000 maximum iterations under default parameters with desmond3.1.

Data analysis

Data analysis was conducted with graphpad prism5 and image j (National Institutes of Health, Bethesda, MD, USA). All data are presented as the mean ± standard error. Statistical comparisons were made by ANOVA, with graphpad prism5.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

This work was supported by grants from the National Key Basic Research Program of China (2012CB910402 to J.-P. Sun; 2013CB967700 to X. Yu), the National Natural Science Foundation of China (31100580 and 31271505 to J.-P. Sun; 31000362 and 31270857 to X. Yu; and 81171062 to Q. Pang), the Foundation for Excellent Young and Middle-Aged Scientists of Shandong Province, China (BS2011SW020 to J.-P. Sun), the Foundation of Program for New Century Excellent Talents in University, China (NCET-09-0531 to X.Yu) and the Independence Innovation Foundation of Shandong University (2012TS114 to J.-P. Sun). We also thank I. Bruce for language editing.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information
  • 1
    Seo J & Lee KJ (2004) Post-translational modifications and their biological functions: proteomic analysis and systematic approaches. J Biochem Mol Biol 37, 3544.
  • 2
    Chen M, Sun JP, Liu J & Yu X (2010) Research progress of several protein tyrosine phosphatases in diabetes. Sheng Li Xue Bao 62, 179189.
  • 3
    Liu J, Chen M, Li R, Yang F, Shi X, Zhu L, Wang HM, Yao W, Liu Q, Meng FG et al. (2012) Biochemical and functional studies of lymphoid-specific tyrosine phosphatase (Lyp) variants S201F and R266W. PLoS One 7, e43631.
  • 4
    Yu X, Chen M, Zhang S, Yu ZH, Sun JP, Wang L, Liu S, Imasaki T, Takagi Y & Zhang ZY (2011) Substrate specificity of lymphoid-specific tyrosine phosphatase (Lyp) and identification of Src kinase-associated protein of 55 kDa homolog (SKAP-HOM) as a Lyp substrate. J Biol Chem 286, 3052630534.
  • 5
    Yu X, Sun JP, He Y, Guo X, Liu S, Zhou B, Hudmon A & Zhang ZY (2007) Structure, inhibitor, and regulatory mechanism of Lyp, a lymphoid-specific tyrosine phosphatase implicated in autoimmune diseases. Proc Natl Acad Sci USA 104, 1976719772.
  • 6
    Liu S, Sun JP, Zhou B & Zhang ZY (2006) Structural basis of docking interactions between ERK2 and MAP kinase phosphatase 3. Proc Natl Acad Sci USA 103, 53265331.
  • 7
    Wortzel I & Seger R (2011) The ERK cascade: distinct functions within various subcellular organelles. Genes Cancer 2, 195209.
  • 8
    Ahn NG, Seger R, Bratlien RL, Diltz CD, Tonks NK & Krebs EG (1991) Multiple components in an epidermal growth factor-stimulated protein kinase cascade. In vitro activation of a myelin basic protein/microtubule-associated protein 2 kinase. J Biol Chem 266, 42204227.
  • 9
    Seger R, Ahn NG, Boulton TG, Yancopoulos GD, Panayotatos N, Radziejewska E, Ericsson L, Bratlien RL, Cobb MH & Krebs EG (1991) Microtubule-associated protein 2 kinases, ERK1 and ERK2, undergo autophosphorylation on both tyrosine and threonine residues: implications for their mechanism of activation. Proc Natl Acad Sci USA 88, 61426146.
  • 10
    Roskoski R Jr (2012) ERK1/2 MAP kinases: structure, function, and regulation. Pharmacol Res 66, 105143.
  • 11
    Alessi DR, Gomez N, Moorhead G, Lewis T, Keyse SM & Cohen P (1995) Inactivation of p42 MAP kinase by protein phosphatase 2A and a protein tyrosine phosphatase, but not CL100, in various cell lines. Curr Biol 5, 283295.
  • 12
    Anderson NG, Maller JL, Tonks NK & Sturgill TW (1990) Requirement for integration of signals from two distinct phosphorylation pathways for activation of MAP kinase. Nature 343, 651653.
  • 13
    Sontag E, Fedorov S, Kamibayashi C, Robbins D, Cobb M & Mumby M (1993) The interaction of SV40 small tumor antigen with protein phosphatase 2A stimulates the map kinase pathway and induces cell proliferation. Cell 75, 887897.
  • 14
    Zhou B, Wang ZX, Zhao Y, Brautigan DL & Zhang ZY (2002) The specificity of extracellular signal-regulated kinase 2 dephosphorylation by protein phosphatases. J Biol Chem 277, 3181831825.
  • 15
    Maeda T, Tsai AY & Saito H (1993) Mutations in a protein tyrosine phosphatase gene (PTP2) and a protein serine/threonine phosphatase gene (PTC1) cause a synthetic growth defect in Saccharomyces cerevisiae. Mol Cell Biol 13, 54085417.
  • 16
    Maeda T, Wurgler-Murphy SM & Saito H (1994) A two-component system that regulates an osmosensing MAP kinase cascade in yeast. Nature 369, 242245.
  • 17
    Duan X, Liang YY, Feng XH & Lin X (2006) Protein serine/threonine phosphatase PPM1A dephosphorylates Smad1 in the bone morphogenetic protein signaling pathway. J Biol Chem 281, 3652636532.
  • 18
    Lin X, Duan X, Liang YY, Su Y, Wrighton KH, Long J, Hu M, Davis CM, Wang J, Brunicardi FC et al. (2006) PPM1A functions as a Smad phosphatase to terminate TGFbeta signaling. Cell 125, 915928.
  • 19
    Takekawa M, Maeda T & Saito H (1998) Protein phosphatase 2Calpha inhibits the human stress-responsive p38 and JNK MAPK pathways. EMBO J 17, 47444752.
  • 20
    Ekerot M, Stavridis MP, Delavaine L, Mitchell MP, Staples C, Owens DM, Keenan ID, Dickinson RJ, Storey KG & Keyse SM (2008) Negative-feedback regulation of FGF signalling by DUSP6/MKP-3 is driven by ERK1/2 and mediated by Ets factor binding to a conserved site within the DUSP6/MKP-3 gene promoter. Biochem J 412, 287298.
  • 21
    Das AK, Helps NR, Cohen PT & Barford D (1996) Crystal structure of the protein serine/threonine phosphatase 2C at 2.0 A resolution. EMBO J 15, 67986809.
  • 22
    Fjeld CC & Denu JM (1999) Kinetic analysis of human serine/threonine protein phosphatase 2Calpha. J Biol Chem 274, 2033620343.
  • 23
    Flajolet M, Rakhilin S, Wang H, Starkova N, Nuangchamnong N, Nairn AC & Greengard P (2003) Protein phosphatase 2C binds selectively to and dephosphorylates metabotropic glutamate receptor 3. Proc Natl Acad Sci USA 100, 1600616011.
  • 24
    Huang Z, Zhou B & Zhang ZY (2004) Molecular determinants of substrate recognition in hematopoietic protein-tyrosine phosphatase. J Biol Chem 279, 5215052159.
  • 25
    Zhang ZY, Maclean D, Thieme-Sefler AM, Roeske RW & Dixon JE (1993) A continuous spectrophotometric and fluorimetric assay for protein tyrosine phosphatase using phosphotyrosine-containing peptides. Anal Biochem 211, 715.
  • 26
    Jackson MD, Fjeld CC & Denu JM (2003) Probing the function of conserved residues in the serine/threonine phosphatase PP2Calpha. Biochemistry 42, 85138521.
  • 27
    Canagarajah BJ, Khokhlatchev A, Cobb MH & Goldsmith EJ (1997) Activation mechanism of the MAP kinase ERK2 by dual phosphorylation. Cell 90, 859869.
  • 28
    Zhang J, Zhang F, Ebert D, Cobb MH & Goldsmith EJ (1995) Activity of the MAP kinase ERK2 is controlled by a flexible surface loop. Structure. 3, 299307.
  • 29
    Webb MR (1992) A continuous spectrophotometric assay for inorganic phosphate and for measuring phosphate release kinetics in biological systems. Proc Natl Acad Sci USA 89, 48844887.
  • 30
    Zhao Y & Zhang ZY (2001) The mechanism of dephosphorylation of extracellular signal-regulated kinase 2 by mitogen-activated protein kinase phosphatase 3. J Biol Chem 276, 3238232391.
  • 31
    Marszalek J, Kostrowicki J & Spychala J (1989) LEHM: a convenient non-linear regression microcomputer program for fitting Michaelis–Menten and Hill models to enzyme kinetic data. Comput Appl Biosci 5, 239240.
  • 32
    Ritchie DW & Venkatraman V (2010) Ultra-fast FFT protein docking on graphics processors. Bioinformatics 26, 23982405.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information
febs12275-sup-0001-FigS1_TableS1.zipZip archive298K

Fig. S1. Structural comparison of phospho-ERK and phospho-p38.

Table S1. The sequence alignment of PP2C family substrates.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.