Impaired synergistic activation of stress-activated protein kinase SAPK/JNK in mouse embryonic stem cells lacking SEK1/MKK4: different contribution of SEK2/MKK7 isoforms to the synergistic activation.

Stress-activated protein kinase/c-Jun NH(2)-terminal kinase (SAPK/JNK), which is a member of the mitogen-activated protein kinase (MAPK) family, plays an important role in a stress-induced signaling cascade. SAPK/JNK activation requires the phosphorylation of Thr and Tyr residues in its Thr-Pro-Tyr motif, and SEK1 (MKK4) and MKK7 (SEK2) have been identified as the upstream MAPK kinases. Here we examined the activation and phosphorylation sites of SAPK/JNK and differentiated the contribution of SEK1 and MKK7alpha1, -gamma1, and -gamma2 isoforms to the MAPK activation. In SEK1-deficient mouse embryonic stem cells, stress-induced SAPK/JNK activation was markedly impaired, and this defect was accompanied with a decreased level of the Tyr phosphorylation. Analysis in HeLa cells co-transfected with the two MAPK kinases revealed that the Thr and Tyr of SAPK/JNK were independently phosphorylated in response to heat shock by MKK7gamma1 and SEK1, respectively. However, MKK7alpha1 failed to phosphorylate the Thr of SAPK/JNK unless its Tyr residue was phosphorylated by SEK1. In contrast, MKK7gamma2 had the ability to phosphorylate both Thr and Tyr residues. In all cases, the dual phosphorylation of the Thr and Tyr residues was essentially required for the full activation of SAPK/JNK. These data provide the first evidence that synergistic activation of SAPK/JNK requires both phosphorylation at the Thr and Tyr residues in living cells and that the preference for the Thr and Tyr phosphorylation was different among the members of MAPK kinases.

The SAPK/JNK 1 is a member of the family of mitogen-acti-vated protein kinases (MAPKs). This MAPK is activated not only by many types of cellular stresses including heat shock, UV irradiation, and inflammatory cytokines (IL-1␤ and tumor necrosis factor-␣) but also by heterotrimeric G-proteins, phorbol esters, and co-stimulatory activation of T lymphocytes. The activated SAPK/JNK phosphorylates c-Jun, Jun D, and ATF-2 to regulate gene expression for the stress response (1).
To examine the physiological roles of SAPK/JNK in the stress-induced signaling pathway, we previously disrupted sek1 gene in mouse embryonic stem (ES) cells and in mice (2,3). In the sek1(Ϫ/Ϫ) ES cells, heat shock-induced activation of SAPK/JNK was almost completely abolished, indicating that SEK1 certainly functions as an activator of SAPK/JNK. However, evidence has indicated that there might be SAPK/JNK activator(s) other than SEK1. Indeed, several groups including us have isolated the cDNA encoding SEK2 (also called MKK7) as another activator of SAPK/JNK (4 -10). There are six different isoforms of MKK7 (␣1, ␣2, ␤1, ␤2, ␥1, and ␥2) due to alternative splicing from the same gene (11). SEK1-deficient embryos displayed severe anemia and died between embryonic day 10.5 and embryonic day 12.5 mainly because of defective liver formation (3,12). However, biochemical roles of SEK1 in the liver formation in vivo remain to be resolved. Interestingly, SAPK/JNK activation in response to phorbol ester plus Ca 2ϩ ionophore was lost in sek1(Ϫ/Ϫ) thymocytes but not in sek1(Ϫ/Ϫ) peripheral T cells (13). These findings allow us to speculate that SEK1-induced phosphorylation of SAPK/JNK may contribute to the kinase activation in different manners dependent on cell types.
In this regard, Lawler et al. (14), using Escherichia coliexpressed recombinant enzymes, have shown that SAPK1c/ JNK1 was activated synergistically in vitro by SEK1 and MKK7 and found that SEK1 had a preference for a Tyr residue and that MKK7 had a preference for a Thr residue within the Thr-Pro-Tyr motif in the MAPK. Lisnock et al. (15) also reported similar results using recombinant JNK3␣1 expressed in E. coli. In a more detailed analysis with SAPK/JNK isoforms (JNK1␣1, JNK2␣2, and JNK3␣1) expressed in insect cells, it has been shown that both SEK1 and MKK7 were required in vitro for the maximum activation (16). These results give us a hint why two kinds of SAPK/JNK activators, SEK1 and MKK7, exist in cells and explain the differences of the biochemical phenotypes in SEK1-deficient cells to some extent.
To elucidate the role of SEK1 in the SAPK/JNK activation in living cells, we first examined the activation and phosphorylation sites of SAPK/JNK in SEK1-deficient mouse ES cells that retain MKK7 expression. The contribution of various MKK7 isoforms to the SAPK/JNK activation was further investigated in HeLa cells that had been co-transfected with each isoform of MKK7 and SEK1. Our present results clearly show that both phosphorylation at the Thr and Tyr residues was required for the synergistic activation of SAPK/JNK in living cells and that the preference for the Thr and Tyr phosphorylation was different among the two members of the MAPKKs, SEK1 and MKK7.

EXPERIMENTAL PROCEDURES
Cell Culture and Materials-The murine ES cell line E14K (wild type) and the sek1(Ϫ/Ϫ) mutant cell line made by sek1 gene targeting were maintained in Dulbecco's modified Eagle's medium supplemented with 15% fetal calf serum and leukemia inhibitory factor as described previously (2). The ES cells (ϳ1 ϫ 10 6 cells/ml) were treated with a pulse of heat shock (42°C for 15 min), UV irradiation (15 min using DNA Stratalinker from Stratagene), 1,10-phenanthroline (1 mM for 1 h), lysophosphatidic acid (10 g/ml for 15 min), or IL-1␤ (10 ng/ml for 15 min) for stress signals. The human cell line HeLa was maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum.
Construction of Plasmids and Transfection-cDNAs encoding FLAGtagged SEK1, MKK7␣1, MKK7␥1, and MKK7␥2 and HA-tagged JNK1 were cloned into mammalian expression vector pCMV5. For gene expression analysis, HeLa cells were plated to ϳ50% confluence and transfected 1 day later with expression vectors using the Lipo-fectAMINE method (Life Technologies, Inc.). The cells, after being cultured for a 0.5 day, were serum-starved for 12 h and subsequently stimulated with the heat shock at 42°C for 15 min. Cell extracts were prepared for the assays of SAPK/JNK activity and the Thr and Tyr phosphorylation as described below.
Immunoprecipitation and Immunoblotting-The HeLa cells (ϳ1 ϫ 10 6 cells) were suspended at 4°C in 0.5 ml of a lysis buffer consisting of 20 mM Na-Pipes, pH 7.0, 10 mM NaCl, 0.5% Nonidet P-40, 0.05% 2-mercaptoethanol, 5 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 100 M Na 3 VO 4 , 20 g/ml leupeptin, 50 mM NaF, and 1 mM benzamidine. The cell lysates were incubated with anti-HA-agarose at 4°C for 2 h, and the immunocomplexes were washed several times with the lysis buffer. The samples were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotting. Proteins were electrophoretically transferred to a polyvinylidene difluoride membrane (Bio-Rad) and probed with anti-FLAG, -HA, -phospho-Tyr, -phospho-Thr, and -phospho-SAPK/JNK Abs. The bands were visualized by SuperSignal West Pico Chemiluminescent Substrate for the development of immunoblots using a horseradish peroxidase-conjugated secondary Ab according to the manufacturer's instructions (Pierce).
Assay of SAPK/JNK Activity-SAPK/JNK proteins were immunoprecipitated at 4°C for 2 h using the anti-SAPK/JNK polyclonal Ab. The SAPK/JNK activity in the precipitated fractions was measured with GST-c-Jun as an in vitro substrate in the presence of 60 M [␥-32 P]ATP as described previously (2,13). The amounts of the precipitated SAPK/ JNK that had been monitored by immunoblotting with the anti-SAPK/ JNK polyclonal Ab were almost constant in a series of the present experiments.
All experiments were repeated at least three times with different batches of the cell samples, and the results were fully reproducible. Hence, most of the data shown are representative of several independent experiments.

Impaired Stress-induced SAPK/JNK Activation in SEK1deficient ES Cells-
To investigate the role of SEK1 in stressinduced SAPK/JNK activation in intact cells, we first used sek1(Ϫ/Ϫ) mouse ES cells that lacked SEK1 protein. Fig. 1 shows the time course of SAPK/JNK activity in response to a pulse of heat shock in wild-type and sek1(Ϫ/Ϫ) ES cells. The activity was measured by the ability of SAPK/JNK to phosphorylate GST-c-Jun as a substrate. The heat shock markedly stimulated SAPK/JNK activity in wild-type ES cells; the maximum activation was observed at 30 min. However, such SAPK/ JNK activation was greatly impaired in sek1(Ϫ/Ϫ) ES cells in accordance with our previous report (2). These results suggest that SEK1 is the main stimulator for the heat shock-induced SAPK/JNK activation at least in ES cells.
UV irradiation was also used for the stress signal in the ES cells, and phosphorylated SAPK/JNK together with other members of the MAPK family (ERK and p38) were analyzed by immunoblotting with mAbs specifically recognizing their phosphorylated forms. As shown in Fig. 2, the UV-induced phosphorylation of ERK (p44) and p38 was clearly observed in both wild-type and sek1(Ϫ/Ϫ) ES cells; their phosphorylated levels were not significantly different from each other (Fig. 2, panels A and C). In contrast, UV-induced phosphorylation of SAPK/ JNK (Fig. 2, panel E, p46 and p54) was greatly impaired in sek1(Ϫ/Ϫ) ES cells that lacked p45 SEK1 (Fig. 2, panel G), although the cells contained another SAPK/JNK activator, p48 MKK7, at the same level as wild-type cells (Fig. 2, panel H). It is interesting to note here that the p48 MKK7 in mouse ES cells may be the isoform of ␥1 (or ␤1), but not ␣1, ␣2, ␤2, or ␥2, based on our findings by reverse transcription-polymerase chain re-action analysis (data not shown). These results indicate that SEK1 is essential for the UV-induced full activation of SAPK/ JNK and is not involved in the activation of other members of MAPK family, such as ERK and p38, in ES cells. In other words, the ␥1 isoform of MKK7 present in ES cells is not capable of compensating for the role of SEK1 in the stressinduced SAPK/JNK activation.
Impaired Stress-induced SAPK/JNK Activation in SEK1deficient ES Cells Arising from the Loss of Its Tyrosine Phosphorylation-Besides heat shock and UV irradiation, a variety of stimuli, such as 1,10-phenanthroline, lysophosphatidic acid, and IL-1␤, also induced SAPK/JNK activation to different extents in wild-type ES cells (Fig. 3). This activation was again markedly impaired in sek1(Ϫ/Ϫ) ES cells (Fig. 3, panels A and D). It has recently been reported that the phosphorylation of two amino acid residues on the Thr-Pro-Tyr motif within SAPK/JNK was required for the full activation of the MAPK in vitro (14 -16). Therefore, we identified the phosphorylated sites of SAPK/JNK in the stimulated ES cells using anti-phospho-Tyr (PY20). The Tyr phosphorylation was almost completely abolished in the SEK1-deficient ES cells (Fig. 3, panels B and  E). These results indicate that SEK1 preferentially phosphorylates the Tyr residue of the MAPK in ES cells. Thus, it is very likely that the SAPK/JNK full activation observed in wild-type ES cells seems to be because of the cooperative actions of SEK1 and MKK7 (␥1) to phosphorylate both sites on the MAPK.
Synergistic Activation of SAPK/JNK by Dual Phosphorylation of Its Thr and Tyr Residues-In ES cells, SEK1 clearly contributed to the Tyr phosphorylation of SAPK/JNK, and the stress-induced full activation of the MAPK appeared to require the dual phosphorylation on the Thr-Pro-Tyr motif. Therefore, we further investigated the cooperative action of MKK7, which was supposed to be involved in the Thr phosphorylation of SAPK/JNK (14 -16). For the analysis, we used HeLa cells for transient transfection because of low efficiency of transfection into ES cells. FLAG-tagged MKK7 (␣1, ␥1, or ␥2) was coexpressed with FLAG-SEK1 and HA-JNK1/SAPK␥ in the cells using pCMV5 mammalian vectors. The three isoforms of MKK7 used in this study were the shortest ␣1 (347 amino acids), ␥1 which has an 89-residue extension at the amino terminus of ␣1, and the longest ␥2, which has a 33-residue extension at the carboxyl terminus of ␥1 resulting from alternative splicing of one gene locus (see Fig. 8A later). The transfected cells, after being starved for 12 h to decrease the basal activity of endogenous SAPK/JNK, were stimulated with heat shock at 42°C for 15 min. The cell lysates were immunoprecipitated with anti-HA Affinity Matrix and analyzed for activity and phosphorylation of SAPK/JNK using anti-phospho Abs. In a series of the experiments, the expression of HA-tagged JNK1 was almost constant (see Figs. 4 -6, panel C). Endogenous SEK1 and MKK7 in HeLa cells did neither phosphorylate nor activate HA-tagged JNK1 under the present conditions (Figs. 4 -6, lane 1). The different ratios of SEK1 and MKK7 expression vectors induced varied expression of SEK1 and MKK7 proteins, but the sum of the expressed proteins was almost constant in each experiment (Figs. 4 -6, panels A and B). Fig. 4 shows typical results showing the cooperative effects of SEK1 and MKK7␥1, which appeared to be present in ES cells, on the heat shock-induced activation and phosphorylation of HA-JNK1. The expression of SEK1 or MKK7␥1 alone clearly phosphorylated the Tyr or Thr of HA-JNK1, which was recognized with the anti-phospho-Tyr or -phospho-Thr Ab (Fig. 4,  panels D and E, lanes 2 and 6). In other words, the two upstream kinases, SEK1 and MKK7␥1, strongly favored either one amino acid or the other. The kinase activity measured by GST-c-Jun phosphorylation of HA-JNK1 was, however, rather low in either case (Fig. 4, panel G). When both Tyr and Thr residues of HA-JNK1 were phosphorylated by SEK1 and MKK7␥1 (Fig. 4, panels D and E, lanes 3-5), there was synergistic stimulation of the kinase activity (Fig. 4, panel G). The dual phosphorylation of HA-JNK1 (Fig. 4, panel F) was also evident in the mobility shift of the Tyr-phosphorylated form on SDS-polyacrylamide gel electrophoresis (Fig. 4, panel D).
The more quantitative analysis was performed with different batches of the transfected HeLa cells, and the results are summarized in Fig. 7. As shown in Fig. 7A, SEK1 preferentially phosphorylated the Tyr residue of SAPK/JNK, whereas MKK7␥1 mainly phosphorylated its Thr residue. These results indicate that MKK7␥1 has a preference for the Thr phosphorylation of SAPK/JNK and that the MAPK was synergistically activated by the dual phosphorylation because of the independent actions of the two upstream MAPKKs.

Different Properties of MKK7 Isoforms in the Tyr and Thr
Phosphorylation of SAPK/JNK-The effects of other MKK7 isoforms, ␣1 and ␥2, were also investigated under the same conditions. As shown in Figs. 5 (lane 6) and 7B, the expression of MKK7␣1 alone failed to phosphorylate the Thr residue of HA-JNK1. This was in sharp contrast to the action of MKK7␥1. However, MKK7␣1-induced Thr phosphorylation became apparent when HA-JNK1 was once phosphorylated at the Tyr residue by SEK1 (Fig. 5E, lanes 3-5, and Fig. 7B, middle columns). The synergistic activation of HA-JNK1 (Fig. 5G) was again accompanied by the dual phosphorylation of the MAPK at both residues (Fig. 5, D-F). Thus, MKK7␣1 preferentially phosphorylated the Thr residue of Tyr-phosphorylated SAPK/ JNK rather than that of its nonphosphorylated form.
The action of MKK7␥2 was the most striking among the three MAPKK isoforms. As shown in Figs. 6 (lane 6) and 7C, this isoform was capable of phosphorylating the Tyr in addition to the Thr residue of HA-JNK1. Therefore, activity of HA-JNK1 phosphorylated by MKK7␥2 alone (Fig. 6G) was significantly higher than that by MKK7␥1 or -␣1, although cooperative action of SEK1 was still apparent in the synergistic activation of SAPK/JNK (Fig. 6, panels F and G, lanes 3-5). These results indicate that both phosphorylation at the Thr and Tyr residues was required for the synergistic activation of SAPK/JNK in living cells and that the preference for the Thr and Tyr phosphorylation was dependent on the MAPKKs SEK1 and MKK7 (see Fig. 8).

DISCUSSION
It has been reported in in vitro experiments that synergistic SAPK/JNK activation requires both Tyr and Thr phosphorylation within the Thr-Pro-Tyr motif by the two different activators, SEK1 and MKK7 (14 -16). This idea was confirmed and further extended by our present intact cell experiments. First, SEK1 appeared to be a selective MAPKK for the Tyr phosphorylation of SAPK/JNK based on the following results. SEK1deficient ES cells had a defect in synergistic SAPK/JNK activation in response to a variety of stimuli (Figs. 1-3). The Tyr phosphorylation of SAPK/JNK observed in the wild-type cells was almost completely abolished in the SEK1-deficient ES cells (Fig. 3). Moreover, transfected SEK1 selectively phosphorylated the Tyr residue of SAPK/JNK in HeLa cells; the Thr phosphorylation was never observed under the present conditions (Figs. 4 -6, lane 2, and Fig. 7).
Second, the properties of another MAPKK, MKK7, which was initially supposed to be involved in the Thr phosphoryla- tion of SAPK/JNK, appeared to be different among the isoforms (Figs. 7 and 8B). MKK7 isoforms were produced from alternative splicing of one gene locus (11). As shown in Fig. 8A, MKK7␥1 and -␥2 but not -␣1 have an NH 2 -terminal extension that interacts with SAPK/JNK, and MKK7␥2 has further an extra COOH-terminal region whose biochemical properties remain unknown. This gene locus also produces other isoforms including MKK7␣2, -␤1, and -␤2. However, apparent differences among the isoforms were not reported in the previous studies. In the present study, we found that the MKK7␥2 isoform expressed in HeLa cells was intrinsically a "dual-specific" protein kinase for SAPK/JNK because it phosphorylated the MAPK not only at the Thr residue but also at the Tyr (Figs. 6, D and E, lane 6, and Fig. 7C). This was in sharp contrast to the action of MKK7␥1, which phosphorylated only the Thr residue of SAPK/JNK (Figs. 4, D and E, lane 6, and Fig. 7A). Moreover, MKK7␣1 failed to phosphorylate SAPK/JNK at the Thr residue unless the Tyr residue of the MAPK was phosphorylated by SEK1 (Figs. 5E and 7B). These differences in MKK7 isoforms may explain the molecular mechanisms of the lost and maintained SAPK/JNK activation observed between sek1(Ϫ/Ϫ) thymocytes and peripheral T cells (see the Introduction). In other words, the SEK1-deficient phenotype would be varied in the types of cells expressing MKK7 isoforms.
Recently, Dong et al. (17) have generated mkk7(Ϫ/Ϫ) ES cells, which lack MKK7 but retain SEK1 expression. They showed that SAPK/JNK activation in response to UV irradiation, heat shock, and other stress signals was greatly reduced in the MKK7-deficient cells as had been observed in SEK1deficient ES cells (2,18). Thus, the two MAPKKs SEK1 and MKK7 (probably its ␥1 isoform) appeared to be required for the synergistic activation of SAPK/JNK in ES cells. We have also made mkk7(Ϫ/Ϫ) bone marrow-derived mast cell lines that showed increased cell proliferation in response to the growth factors stem cell factor (c-Kit ligand) and interleukin-3. 2 The SAPK/JNK activation by a variety of stimuli including UV irradiation and Fc⑀ receptor stimulation was lost in the mkk7(Ϫ/Ϫ) mast cells. Interestingly, the expression of SEK1 protein was strongly up-regulated in the mkk7(Ϫ/Ϫ) mast cells, and SEK1 was phosphorylated upon stimulation. These data indicate that MKK7 is essential, and SEK1 is not enough for SAPK/JNK activation in mast cells. Thus, the same synergistic SAPK/JNK activation by SEK1 and MKK7 has been observed not only in ES cells but also in differentiated mast cells.
In the present study, we mainly used heat shock as a stress signal especially in HeLa cells. The molecular mechanism by which heat shock induces SAPK/JNK activation is reported to be due to the inhibition of SAPK/JNK phosphatase(s) rather than the activation of SEK1 and MKK7 (19). This allowed us to investigate the MAPK activation without specific stimulators of the MAPKKs. Indeed, we could observe the progressive SAPK/JNK activation in a manner dependent on the expressed amounts of SEK1 and MKK7 (see Figs. 4 -7). Although we have not investigated the effects of other stress signals under the same conditions, it is very likely that SAPK/JNK activation induced by those signals may be different from the present heat shock results. In relation to this, it has been reported that IL-1␤ and tumor necrosis factor-␣ mainly activate MKK7 rather than SEK1 (6,9). In contrast, SEK1 appeared to be activated by the ␤␥ subunits of G-proteins (20). These results indicate that SEK1-and MKK7-dependent signaling pathways exist independently in living cells. Our transfection systems presented in this study would be useful to distinguish such differences in intact cell levels.
In the stress-responsive SAPK/JNK cascade, full activation of the MAPK mostly requires both stimulations of the two separate MAPKKs, and the activated MAPK phosphorylates its downstream substrates, such as c-Jun and the members of the AP-1 transcription family, to regulate proper gene expression. Therefore, this signaling should strictly proceed without errors. It is thus tempting to speculate that the existence of the two separate pathways leading to SAPK/JNK activation may physiologically function as a fail-safe mechanism.