Fig. 1. Structures of p38 and MAPKAPK-3/3pk. (A) Schematic representation of the primary structures of 3pk and p38. (B) Three-dimensional structure of p38α. Yellow circles indicate acidic amino acids of the CD domain (Asp313, Asp315 and Asp316 for human p38α), and white circles the ED site (Glu160 and Asp161 for human p38α). An asterisk indicates the position of the active center. An arrowhead indicates Glu163. See text for details. This figure was made using RasMol, based on the crystallographic data (Wilson et al., 1996; Wang et al., 1997).

Fig. 2. Glu160 and Asp161 of p38α are necessary for the docking interaction with 3pk. (A) Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Myc antibody. Co-immunoprecipitated wild-type (wt) and mutant forms of p38 were detected [upper panel, αHA (IP)]. The expression levels of p38 in each sample were similar [middle panel, αHA (whole)]. Comparable amounts of 3pk were immunoprecipitated in each lane [lower panel, αMyc (IP)]. Similar results were obtained in three different experiments. (B) An equal amount (7 µg) of GST or GST–3pk was incubated with each lysate from NIH 3T3 cells expressing HA-wild-type (wt) or mutant forms of p38 (upper two panels). The proteins were precipitated with glutathione–Sepharose beads and analyzed by immunoblotting with anti-HA antibody (αHA). Comparable amounts of p38 were expressed [αHA (whole)]. Similar results were obtained in three different experiments. The direct binding between p38 and 3pk in vitro was examined using GST–p38 and His-3pk (lower two panels). GST fusions of p38 and His-3pk were expressed in bacteria and purified by the method described previously (Tanoue et al., 2000). Purified His-3pk (1 µg) was mixed with purified GST–p38 wt (∼1 µg) or GST–p38 all (∼1 µg) in a solution containing 50 mM HEPES pH 7.4 and 150 mM NaCl, and incubated for 1 h at 4°C. His-3pk was then precipitated with Probond resin (Invitrogen) and co-precipitated GST–p38 was detected by immunoblotting with anti-GST antibody (bound). (C) The binding between wild-type p38 and wild-type or mutant 3pk was examined by GST pull-down assay. An equal amount (10 µg) of GST or GST–p38 was incubated with each lysate from NIH 3T3 cells expressing Myc-wild-type (wt) 3pk or a mutant (mut) 3pk. Co-precipitated 3pk was detected (upper panel, αMyc). Comparable amounts of 3pk were expressed [lower panel, αMyc (whole)]. Similar results were obtained in three different experiments. (D) The binding between the mutant forms of p38 and the C-terminal portion of 3pk (residues 326–382) (GST–3pk Ct) was examined by GST pull-down assay. An equal amount (30 µg) of GST or GST–3pk Ct was incubated with each lysate from NIH 3T3 cells expressing HA-wild-type (wt) or mutant forms of p38. Co-precipitated p38 was detected (upper panel, αHA). Comparable amounts of p38 were expressed [lower panel, αHA (whole)]. Similar results were obtained in three different experiments. (E) NIH 3T3 cells transfected with either SRα-Myc-wild-type 3pk (wt) or SRα-Myc-mutant 3pk (mut) were exposed to 0.5 M NaCl for 15 min. Immunoprecipitated 3pk activity was measured using MBP as substrate (Activity). Comparable amounts of endogenous p38 were activated in both cell types after the stimulation (α-p-p38). Comparable amounts of 3pk were immunoprecipitated (αMyc). Similar results were obtained in three different experiments. (F) The ability of p38 to phosphorylate the docking-deficient 3pk (3pk mut) in vitro. Myc-3pk (wt) or Myc-3pk (mut) was prepared by immunoprecipitation from cells transfected with each construct. Activated p38 was prepared by incubating with His-tagged MKK6. The fold phosphorylation rates are shown (fold). Similar results were obtained in three different experiments. (G) The ability of the mutant forms of p38 to phosphorylate GST–3pk in vitro. Various forms of HA-tagged p38 were prepared by immunoprecipitation from COS7 cells transfected with each construct. The expression levels of the mutant forms of p38 are shown (IP). The p38 in the immunoprecipitates was then activated in vitro by incubating with His-tagged MKK6 (for 1 h at 37°C). Comparable phosphorylation levels of various forms of p38 were comfirmed by immunoblotting with anti-phospho-p38 (α-p-p38). The phosphorylation of MBP by these phosphorylated (thus activated) forms of p38 is shown (MBP). The rate of GST–3pk phosphorylation relative to that of MBP was quantified and the value relative to the wild-type p38 is shown (3pk/MBP). (H) Quantification of the data from four independent experiments of (G).

Fig. 3. The possible role of the docking interaction in the subcellular localization of 3pk. (A) Subcellular localization of a GFP-tagged C-terminal portion of 3pk (residues 326–382) (GFP–3pk Ct) in cultured cells. NIH 3T3 cells were transfected with GFP–3pk Ct. The cells were treated with leptomycin B (LMB, 8 ng/ml) for 40 min (right panel). (B) The subcellular localization of GFP-tagged wild-type 3pk (3pk wt) or a GFP-tagged docking-deficient 3pk (3pk mut) in cultured cells. The cells were stimulated with osmotic stress (0.4 M sorbitol) for 20 min (right panels). (C) Subcellular localization of GFP-tagged wild-type 3pk in the absence or presence of overexpressed wild-type p38 (p38 wt), kinase-negative p38 (p38 KD) or p38 all (p38 all). Lys53 was replaced by methionine in the kinase-negative form of p38. p38 KD, p38 all and p38 wt showed the same pan-cellular localization (data not shown). (D) Quantification of the data from three independent experiments in (C). The three experiments gave essentially the same results. The cells in which the concentration of GFP–3pk in the nucleus was much higher than that in the cytoplasm (see C, 3pk alone) were regarded as the cells showing ‘nuclear localization of GFP–3pk’, and are shown as percentages of total cells examined. In each case, 300 cells were examined in total. Comparable amounts of HA-p38 were expressed (lower panel, representative of three experiments).

Fig. 4. Glu160 and Asp161 in p38 and Thr157 and Thr158 in ERK2 are the determinants for the docking specificity towards 3pk. (A) Three-dimensional structure of ERK2. A yellow circle indicates the CD domain (Asp316 and Asp319 for rat ERK2, and Asp321 and Asp324 for Xenopus ERK2). A white circle indicates the site corresponding to the ED site of p38 (Thr157 and Thr158 for rat ERK2, and Thr162 and Thr163 for Xenopus ERK2). This figure was made using RasMol, based on the crystallographic data (Zhang et al., 1994). (B) The primary sequences of the ED site (or the TT site) and the CD domain of p38 and ERK2. The numbers shown are the sequences of Xenopus ERK2 and human p38. (C) The binding of p38-like ERK2 (p38L-ERK2) to 3pk. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Myc antibody. Co-immunoprecipitated wild-type (wt) p38, wild-type (wt) ERK2 or p38-like ERK2 (p38L-ERK2) were detected [upper panel, αHA (IP)]. The expression levels of these HA-tagged constructs were similar [middle panel, αHA (whole)]. Comparable amounts of 3pk were immunoprecipitated in each lane [lower panel, αMyc (IP)]. Similar results were obtained in three different experiments. (D) The binding of the mutant forms of ERK2 to 3pk. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Myc antibody. Co-immunoprecipitated wild-type or the mutant forms of ERK2 were detected [upper panel, αHA (IP)]. The expression levels of these HA-tagged constructs were similar [middle panel, αHA (whole)]. Comparable amounts of 3pk were immunoprecipitated in each lane [lower panel, αMyc (IP)]. Similar results were obtained in three different experiments. (E) The ability of wild-type ERK2 (w) or p38-like ERK2 (m) to phosphorylate 3pk in vitro (upper panel). GST–3pk was produced in bacteria. The GST portion was cut off by precision protease treatment, and the remaining 3pk was used as substrate. The amount of 3pk used was 6, 3, 2, 1 or 0.5 µg in 20 µl of reaction buffer, as indicated. The HA-tagged wild-type ERK2 or the HA-tagged p38-like ERK2 was expressed in ΔB-Raf:ER cells (Pritchard et al., 1995) and activated by estrogen treatment. The activated ERK2 constructs were then immunoprecipitated and used. Comparable amounts were immunoprecipitated (data not shown), and they showed comparable kinase activity towards MBP (lower panel; left, wild-type ERK2; right, p38-like ERK2). Similar results were obtained in three different experiments. (F) Activation of 3pk in cells by wild-type ERK2 or p38-like ERK2. NIH 3T3 cells co-transfected with SRα-Myc 3pk and SRα-HA-tagged wild-type (wt) ERK2 or HA-tagged p38-like ERK2 (p38L-ERK2) were serum deprived and then stimulated by 10% FCS for 15 min. Myc-3pk was then immunoprecipitated and assayed for kinase activity using MBP as substrate (upper panel, MBP). To remove co-precipitated ERK2, the immunoprecipitated 3pk was washed extensively with a high-salt buffer before kinase assay. Comparable amounts of 3pk were immunoprecipitated in each lane [middle panel, αMyc (IP)]. Comparable amounts of ERK2 were expressed in each lane [lower panel, αHA (whole)].

Fig. 5. Involvement of the ED site in the docking interaction of MAPKs with MSK2 and PRAK. (A) The binding of the mutant forms of p38 to MSK2. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Flag antibody for Flag-MSK2. Co-immunoprecipitated wild-type p38 (wt) and mutant forms of p38 (HA-tagged) were detected [upper panel, αHA (IP)]. The expression levels of p38 in each sample were similar [middle panel, αHA (whole)]. Comparable amounts of Flag-MSK2 were immunoprecipitated in each lane [lower panel, αFlag (IP)]. pMT-Flag-MSK2 was used. Similar results were obtained in three different experiments. (B) The binding of p38-like ERK2 or ERK2-like p38 to MSK2 was examined as in (A). Similar results were obtained in three different experiments. (C) The binding of the mutant forms of ERK2 to MSK2 was examined as in (A). SRα-HA-wild-type ERK2, HA-ERK2 SD, HA-ERK2 TETD and HA-p38-like ERK2 were used. Similar results were obtained in three different experiments. (D) The binding of the mutant forms of p38 and ERK2 to PRAK was examined as in (A). SRα-Myc-PRAK was used. A longer exposure of the most upper panel is shown in the second upper panel [αHA (IP) long exp]. Similar results were obtained in three different experiments.

Fig. 6. Docking interactions of p38 and ERK2 with RSK2. (A) The binding of the mutant forms of p38 and ERK2 to RSK2. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Myc antibody for Myc-RSK2. Co-immunoprecipitated HA-ERK2 wt, HA-p38L-ERK2, HA-p38 wt and HA-ERK2L-p38 were detected [upper panel, αHA (IP)]. The same membrane was exposed for a longer time (long exp). The expression levels of these HA-tagged constructs were similar [middle panel, αHA (whole)]. Comparable amounts of Myc-RSK2 were immunoprecipitated in each lane [lower panel, αMyc (IP)]. Similar results were obtained in three different experiments. (B) The binding of ERK2 CDm to RSK2 was examined as in (A). Similar results were obtained in three different experiments.

Fig. 7. Docking interactions of p38 and ERK2 with MSK1 or MNK1. (A) The binding of the mutant forms of p38 to MSK1. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Flag antibody for Flag-MSK1. Co-immunoprecipitated HA-p38 wt, HA-p38 ETDT, HA-p38 CDm and HA-p38 all were detected [upper panel, αHA (IP)]. The expression levels of these HA-tagged constructs were similar [middle panel, αHA (whole)]. Comparable amounts of Flag-MSK1 were immunoprecipitated in each lane [lower panel, αFlag (IP)]. pMT-Flag-MSK1 was used. Similar results were obtained in three different experiments. (B) The binding of p38-like ERK2 or ERK2-like p38 to MSK1 was examined as in (A). SRα-HA-ERKL-p38 and SRα-HA-p38L-ERK2 were used. Similar results were obtained in three different experiments. (C) The binding of ERK2 CDm to MSK1 was examined as in (A). Similar results were obtained in three different experiments. (D) The binding of the mutant forms of ERK2 to MSK1 was examined as in (A). SRα-HA-ERK2 wt, SRα-HA-ERK2 SD, SRα-HA-ERK2 TETD and SRα-HA-p38L-ERK2 were used. Similar results were obtained in three different experiments. (E) The binding of the mutant forms of p38 and ERK2 to MNK1. Myc-tagged MNK1 was expressed. Similar results were obtained in three different experiments.

Fig. 8. Involvement of the ED site in the docking interaction of MAPKs with MKPs. (A) The binding of the mutant forms of p38 and ERK2 to MKP-5. Lysates of NIH 3T3 cells co-transfected with the indicated combinations of constructs were immunoprecipitated with anti-Myc antibody for Myc-MKP-5. Co-immunoprecipitated HA-p38 wt, HA-p38 ETDT, HA-ERK2 wt and HA-p38L-ERK2 were detected [upper panel, αHA (IP)]. The same membrane was exposed for a longer time (long exp). The expression levels of these HA-tagged constructs were similar [middle panel, αHA (whole)]. Comparable amounts of Myc-MKP-5 were immunoprecipitated in each lane [lower panel, αMyc (IP)]. SRα-Myc-MKP-5 was used. Similar results were obtained in three different experiments. (B) The binding of the mutant forms of p38 and ERK2 to MKP-3. SRα-Myc-MKP-3 was used. Similar results were obtained in three different experiments.

Fig. 9. Schematic representation of the docking groove. The three-dimensional structures of the docking pocket of p38 and ERK2 are shown (upper panels). The CD domain and the ED (TT) site are indicated. The arrows indicate Glu163 of p38 or Asp160 of ERK2. See text for details.