Activation of a C-terminal Transcriptional Activation Domain of ERK5 by Autophosphorylation*

ERK5 plays a crucial role in many biological processes by regulating transcription. ERK5 has a large C-terminal-half that contains a transcriptional activation domain. However, it has remained unclear how its transcriptional activation activity is regulated. Here, we show that the activated kinase activity of ERK5 is required for the C-terminal-half to enhance the AP-1 activity, and that the activated ERK5 undergoes autophosphorylation on its most C-terminal region. Changing these phosphorylatable threonine and serine residues to unphosphorylatable alanines significantly reduces the transcriptional activation activity of ERK5. Moreover, phosphomimetic mutants of the C-terminal-half of ERK5 without an N-terminal kinase domain are shown to be able to enhance the AP-1 activity in fibroblastic cells. These results reveal the role of the stimulus-induced ERK5 autophosphorylation in regulation of gene expression.

ERK5 is activated by oxidative stress, hyperosmolarity, and several growth factors including epidermal growth factor and nerve growth factor (14 -18). MEK5 is the upstream MAPKK that specifically phosphorylates and activates ERK5 (16,19). It has been shown that ERK5 directly phosphorylates and activates several transcription factors including c-Myc, Sap1a, c-Fos, Fra-1, and MEF2 family members (15, 17, 20 -22). ERK5 knock-out mice die in embryonic stages with angiogenic failure and cardiovascular defects, and in the adult stages, ERK5 is required for endothelial cell survival and maintenance of blood vascular integrity (23)(24)(25)(26). The targeted deletion of MEK5 also causes early embryonic death with cardiovascular defects (27). Moreover, knockdown studies in Xenopus have shown that ERK5 and MEK5 play an essential role in neural differentiation (28).
ERK5 has a large, unique C-terminal-half not found in other MAPK family members (19,29). Because of this unique 400amino acid extension, ERK5 is also called big MAP kinase 1 (BMK1). The C-terminal-half of ERK5 has two proline-rich domains and a bipartite NLS, and shares no high homology with other proteins (19,30,31). Kasler et al. (32) demonstrated that the C-terminal-half of ERK5 has a potent transcriptional activation domain. It was subsequently shown that the ERK5 C-terminal-half is essential for transactivation of MEF2C (30). Moreover, a recent study has shown that ERK5 has two transactivation domains in its C-terminal region (33). However, the molecular mechanisms regulating the transcriptional activation activity of the ERK5 C-terminal region have remained elucidated.
Here we show that the activated kinase activity of ERK5 is required for the C-terminal-half to enhance the transactivation activity of AP-1, and that the activated ERK5 undergoes autophosphorylation on its C-terminal-half. Furthermore, our results show that phosphorylation of the C-terminal-half of ERK5 is necessary for the enhancement of the transactivation activity of AP-1. These results reveal the importance of the ERK5 autophosphorylation in ERK5 signaling, and thus identify a novel role for the C-terminal-half of ERK5 in MAPK signal transduction mechanisms.
Luciferase Assay-Cells split onto 12-well dishes were harvested for assay 24 h after transfection. The luciferase activity in cell lysates was measured by the luciferase assay system (Promega). To determine transfection efficiency, co-expressed ␤-galactosidase activity was measured, and the data were normalized for ␤-galactosidase activities. The data represent means and standard deviations of at least two independent experiments.
Immunoprecipitation and Kinase Assay-Cells were lysed in lysis buffer A. Tagged proteins were immunoprecipitated from cell lysates by incubation with anti-Myc antibody and protein A-Sepharose beads (Amersham Biosciences) for 2 h at 4°C. The precipitates were washed twice with TBS (20 mM Tris-Cl, 150 mM NaCl (pH 7.5)) and then incubated for 10 min at 30°C with myelin basic protein (final concentration, 2 mg/ml) or purified FLAG-ERK5N protein in reaction buffer (20 mM HEPES-Cl (pH 7.5), 10 mM MgCl 2 , 100 M ATP, 2 Ci of [␥-32 P]ATP).

Both the N-terminal-and the C-terminal-halves of ERK5 Are
Required for the Activation of AP-1 Activity-To investigate a role of the unique C-terminal-half of ERK5 in its signaling mechanism, we made two deletion constructs of ERK5: ERK5N (residues 1-407) and ERK5C (residues 401-806) (Fig. 1A). ERK5N has a kinase catalytic domain and a Thr-Glu-Tyr (TEY) activation motif, whereas ERK5C has an NLS. We previously reported that activation of the ERK5 pathway is able to activate the transactivation activity of AP-1 and that the N-terminalhalf of ERK5 alone is not enough for maximal activation of the transactivation activity of AP-1 (22). Thus we examined whether the C-terminal-half of ERK5 is required for maximal activation of AP-1 activity. We expressed full-length ERK5 or ERK5N or ERK5C with or without MEK5D, a constitutively active form of MEK5 (17), in NIH 3T3 cells, and performed a reporter assay measuring the transcriptional activity of AP-1. Expression of ERK5 or MEK5D alone had negligible effects, but co-expression of ERK5 and MEK5D enhanced the AP-1 activity by 21-fold over the control (Fig. 1B). Co-expression of ERK5N and MEK5D enhanced it by 6-fold over the control, and coexpression of ERK5C and MEK5D by 2-fold (Fig. 1B). However, co-expression of ERK5N and ERK5C with MEK5D increased the AP-1 activity by 20-fold, the extent being nearly the same as that in the case of co-expression of full-length ERK5 and MEK5D (Fig. 1B). These results suggest that both the N-terminal-and C-terminal-halves of ERK5 are required for maximal activation of AP-1 activity and that co-expression of ERK5N and ERK5C is able to mimic full-length ERK5 expression in this transcriptional activation.
It has been shown that the kinase activity of ERK5 is necessary for the activation of AP-1 activity induced by full-length ERK5 (22). We then examined whether the kinase activity of ERK5N was required for the enhancement of AP-1 activity by co-expression of ERK5N and ERK5C. We made two mutants of ERK5N, ERK5N AEF, and ERK5N KM. ERK5N AEF is an unphosphorylatable mutant, in which threonine and tyrosine residues at the activating phosphorylation sites (TEY) are replaced by alanine and phenylalanine, respectively. ERK5 KM is a kinase-deficient mutant, in which Lys 84 is replaced by methionine. The reporter assay showed that ERK5N AEF or ERK5N KM did not stimulate the AP-1 activity when co-expressed with ERK5C and MEK5D (Fig. 1C). These results indicate clearly that the kinase activity of ERK5N is required for the enhancement of AP-1 activity.
The N-terminal-half of ERK5 Phosphorylates the C-terminal-half of ERK5-We found that several mobility-shifted bands (slowly migrating bands) of ERK5C in SDSpolyacrylamide gel electrophoresis appeared when co-expressed with ERK5N and MEK5D ( Fig. 2A). All the shifted bands disappeared when incubated with -protein phosphatase in vitro (Fig. 2B), indicating that the shifted bands of ERK5C are caused by phosphorylation of ERK5C. It is likely that there are multiple phosphorylation states in ERK5C. It has previously been shown that ERK5 is able to undergo autophosphorylation on its own C-terminal domain at several sites (35,36). It is therefore likely that activated ERK5N phosphorylates ERK5C in our assay system. In fact, the shifted bands of ERK5C did not appear when co-expressed with ERK5N AEF or ERK5N KM in the presence of MEK5D or co-expressed with ERK5N WT in the absence of MEK5D ( Fig. 2A). Then, we examined the substrate specificity of ERK5N against ERK5C in comparison with that of ERK5N against c-Fos. Under conditions of low amounts of ERK5N and MEK5D (15-0.5 ng), ERK5C, but not c-Fos, was phosphorylated by ERK5N (Fig. 2C). This result suggests that ERK5N phosphorylates ERK5C more efficiently than c-Fos. These results together with the results shown in Fig. 1 suggest the possibility that activated ERK5N-mediated phosphorylation of ERK5C is required for the maximal activity of ERK5C to enhance AP-1 activity.
Phosphorylatable Sites in the C-terminal-half of ERK5-To elucidate a role of phosphorylation of the  DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 JOURNAL OF BIOLOGICAL CHEMISTRY 35451 C-terminal-half of ERK5, we investigated the phosphorylation sites in ERK5C, which are responsible for the mobility shifts. There are five putative MAPK phosphorylation sites (S/TP sites) in the C-terminal region; Ser 421 , Ser 433 , Ser 697 , Thr 723 , and Ser 793 . Moreover, activated human recombinant ERK5 is shown to undergo autophosphorylation on several residues in its C-terminal region in vitro (35) erated seven mutant forms of ERK5C, in which each of the seven putative phosphorylation sites was replaced by alanine, and expressed them with ERK5N and MEK5D (Fig.  3A). Only in the Thr 723 mutant ERK5C, the most slowly migrating band disappeared although other mobility-shifted bands remained (Fig. 3A), suggesting that Thr 723 is one of several phosphorylation sites in ERK5C.

Role of ERK5 Autophosphorylation in Transcriptional Activation
To identify other sites, we generated three additional deletion constructs of ERK5C, ERK5C ⌬C (residues 401-740), ERK5C ⌬N (residues 514 -806), and ERK5C⌬N⌬C (residues 514 -740) (Fig. 3B, left). ERK5C⌬N alone exhibited large mobility shift when co-expressed with ERK5N and MEK5D (Fig. 3B, right), suggesting that there may be one or several cooperative phosphorylation sites in the C-terminal region, residues 740 -806. In this region, there are five serine residues that are not proline-directed but are conserved among Xenopus, mouse and human ERK5 (Fig. 3C, upper). Therefore, we made several mutant forms of ERK5C in which, in addition to Thr 723 , one or several of these conserved serine residues were replaced by alanines. As shown in Fig. 3C, when consecutive three serines (Ser 760 , Ser 764 , and Ser 766 ) and Thr 723 were mutated, the mobility shift did not take place, suggesting that Thr 723 , Ser 760 , Ser 764 , and Ser 766 are the sites responsible for the ERK5C mobility shifts induced by ERK5N-mediated phosphorylation.
Next, we performed in vitro kinase assay using separately purified ERK5N and ERK5C proteins.  The obtained result showed that ERK5C protein was phosphorylated in the presence of ERK5N protein, but not in the absence of ERK5N protein (Fig. 3D). Moreover, the mutant ERK5C, ERK5C 4A (T723, S760, S764, S766A) was much less phosphorylated than wild-type ERK5C (Fig. 3D). These results suggest that ERK5N directly phosphorylates ERK5C.
We made an additional mutant, ERK5C S760, S764, S766A, and termed it ERK5C 3A. ERK5C T723A and ERK5C T723, S760, S764, S766A were also termed ERK5C 1A and ERK5C 4A, respectively (Fig. 4A). We also made phosphomimetic mutants of ERK5C, ERK5C T723D and ERK5C T723, S760, S764, S766D and termed them ERK5C 1D and ERK5C 4D, respectively (Fig. 4A). We then expressed each of them with ERK5N in the absence or presence of MEK5D. As already shown in Fig. 3C, ERK5C 4A did not show mobility shift, whereas ERK5C 1A (T723A) and 3A (S760, S764, S766A) did show mobilityshifted bands except for the most slowly migrating band when co-expressed with ERK5N and MEK5D (Fig. 4B, left). In contrast, both ERK5C 1D (T723D) and ERK5C 4D (T723, S760, S764, S766D) migrated more slowly than ERK5C WT in the absence of MEK5D (Fig. 4B, right). In the case of ERK5C 4D, co-expression of MEK5D did not further shift its migrating position (Fig. 4B,  right). These results also support our idea that phosphorylation of these four sites is responsible for the ERK5C mobility shifts. Next, we made four mutant forms of full length ERK5, in which one (T723) or all of the four sites were mutated to alanines or aspartic acids (ERK5 1A, ERK5 4A, ERK5 1D and ERK5 4D), and expressed each of them with or without MEK5D. Their behaviors in electrophoresis with or without MEK5D were almost identical to those of the corresponding ERK5C mutants (Fig. 4C, see Fig.  4B), except that ERK5 4A and ERK5 4D still showed a small mobility shift when co-expressed with MEK5D. As the mobility-shifted band of ERK5 4A is similar to that of ERK5 KM, which does not undergo autophosphorylation, this mobility shift could be caused by MEK5D-dependent phosphorylation of ERK5 on the TEY site in the N-terminal-half of ERK5. All these results are consistent with our idea that the four sites are ERK5 autophosphorylation sites. Next, we examined when and how the C-terminal region of endogenous ERK5 is phosphorylated upon biological stimulation. The obtained results showed that endogenous ERK5 exhibited several mobility-shifted bands upon stimulation with EGF or H 2 O 2 and the most slowly migrating band appeared at 10 min after EGF treatment (30 M) or at 2 min after H 2 O 2 treatment (100 mM) (Fig. 4D, upper and lower). An exogenously expressed ERK5 WT with MEK5D exhibited the mobilityshifted band that corresponded to the most slowly migrating band of endogenous ERK5, whereas an exogenously expressed ERK5 KM with MEK5D did not exhibit this band (Fig. 4C). Therefore, the most slowly migrating band of endogenous ERK5 results from the phosphorylation of the C-terminal region of ERK5. These results taken together suggest that the C-terminal region of ERK5 is phosphorylated in response to biological stimulation.
We then examined the kinase activity of each full-length mutant form of ERK5. ERK5 1A, 4A, 1D, and 4D were all phosphorylated on the TEY activation sites by MEK5D, and their activated form exhibited nearly the same kinase activity toward myelin basic protein as compared with wild-type ERK5 (Fig. 4E).
Phosphorylation of the C-terminal-half of ERK5 Is Necessary for the Maximal Enhancement of Its Transcriptional Activation Activity-To investigate the role of the phosphorylation of the C-terminal-half of ERK5 in its transcriptional activation activity, we examined the effect of expression of several ERK5 or ERK5C mutants on the transactivation activity of AP-1. In the reporter assay, expression of ERK5 1A or ERK5 4A with MEK5D enhanced the AP-1 activity less than wild-type ERK5; ERK5 4A was less potent than ERK5 1A (Fig. 5A, left). As no significant differences were found in the kinase activity among ERK5 WT, ERK5 1A, and ERK5 4A (Fig. 4D), the lower enhancement of AP-1 activity with ERK5 1A or ERK5 4A would result from the partial or complete lack of the autophosphorylation on the C-terminal region. In the reporter assay in which each ERK5C mutant was co-expressed with ERK5N and MEK5D, ERK5C 1A, ERK5C 3A, or ERK5C 4A enhanced the AP-1 activity less than wild-type ERK5C; ERK5C 4A was much less potent than others (Fig. 5A, right), suggesting also the importance of the phosphorylation of these sites. In addition, expression of ERK5C 4D alone without ERK5N and MEK5D enhanced the AP-1 activity by 4-fold over the control (Fig. 5B). These unexpected findings suggest that the phosphorylated C-terminal-half of ERK5 by itself without kinase activity is able to perform part of ERK5 function. However, our observation that ERK5 4A or ERK5C 4A still had the activity to enhance the AP-1 activity also suggests that the C-terminal region of ERK5 should have a basal transcriptional activation activity, which could be enhanced by its phosphorylation.
ERK5 is also known to enhance the transcriptional activity of MEF2. We then investigated whether autophosphorylation of ERK5 is involved in the enhancement of MEF2 activity. Coexpression of wild-type ERK5 and MEK5D enhanced the MEF2 activity by 3-fold over the control, whereas co-expression of ERK5 4A and MEK5D enhanced the MEF2 activity to lesser extent (Fig. 5C, left). Moreover, expression of ERK5C 4D enhanced the MEF2 activity by 1.7-fold over the control (Fig.   5C, right). Therefore, these results suggest that autophosphorylation of the C-terminal-half of ERK5 is also important for the activation of MEF2 activity.
Our results show firstly that the activated kinase activity of ERK5 is required for the C-terminal-half to maximally enhance the AP-1 activity and that the activated ERK5 undergoes autophosphorylation on its most C-terminal region. Subsequent analysis demonstrates that the C-terminal region phosphorylation is required for the maximal activation of the transcriptional activation activity of the ERK5 C-terminal region. Based on these results, we propose a hypothetical model, in which the activated kinase activity of ERK5 has two roles in gene expression; one role is to activate transcription factors by direct phosphorylation, and another is to activate the transcriptional activation activity of the C-terminal-half by autophosphorylation, presumably, intramolecular autophosphorylation (Fig. 5D). Two regions in the C-terminal region of ERK5 are autophosphorylated; T723 and a region containing S760, S764, and S766. These two important phosphorylation regions may additively enhance transcription and seem equally important for the transcriptional activation activity (see Fig. 5A). How autophosphorylation enhances the transcriptional activation activity of the C-terminal-half should be elucidated in future studies.