Identification and characterization of a novel MAP kinase kinase kinase, MLTK.

The MAPK cascades regulate a wide variety of cellular functions, including cell proliferation, differentiation, and stress responses. Here we have identified a novel MAP kinase kinase kinase (MAPKKK), termed MLTK (for MLK-like mitogen-activated protein triple kinase), whose expression is increased by activation of the ERK/MAPK pathway. There are two alternatively spliced forms of MLTK, MLTKalpha and MLTKbeta. When overexpressed in cells, both MLTKalpha and MLTKbeta are able to activate the ERK, JNK/SAPK, p38, and ERK5 pathways. Moreover, both MLTKalpha and MLTKbeta are activated in response to osmotic shock with hyperosmolar media through autophosphorylation. Remarkably, expression of MLTKalpha, but not MLTKbeta, in Swiss 3T3 cells results in the disruption of actin stress fibers and dramatic morphological changes. A kinase-dead form of MLTKalpha does not cause these phenomena. Inhibition of the p38 pathway significantly blocks MLTKalpha-induced stress fiber disruption and morphological changes. These results suggest that MLTK is a stress-activated MAPKKK that may be involved in the regulation of actin organization.

MAPKK kinases (MAPKKKs) are a family of serine/threonine protein kinases that can activate one or several of the MAPKK/MAPK cascades (8). MAPKKKs phosphorylate two serine/threonine residues in the activation phosphorylation sites of MAPKKs. The identified MAPKKKs include the Raf family, the MEK kinase (MEKK) family, and the mixed-lineage kinase (MLK) family. The Raf family kinases selectively activate the ERK pathway. On the other hand, MEKK1, MEKK2, and MEKK3 activate both the ERK pathway and the JNK/ SAPK pathway (12). MLK3, MUK, TAK1, and ASK1 activate the JNK/SAPK and p38 pathways (8). Recently, Cot/Tpl-2 was reported to activate all known MAPK pathways: the ERK, JNK/SAPK, p38 (p38␥ isoform), and ERK5/BMK1 pathways (13). Some of the MAPKKKs were shown to be activated by particular extracellular stimuli. For example, Raf-1 is activated by mitogens and phorbol esters, TAK1 by transforming growth factor-␤ (TGF-␤) and cytokines (14,15), and ASK1 by tumor necrosis factor-␣ (TNF␣) (16). Thus, various MAPKKKs appear to respond to a variety of extracellular stimuli and activate a limited number of the MAPKK/MAPK cascades.
Actin reorganization underlies changes in cell morphology and is essential for cell migration and motility. The Rho family GTPases are known to regulate particular types of actin reorganization (17,18). Rho and Rac promote stress fiber formation and membrane ruffling, respectively (19,20). Their effects on the actin cytoskeleton are at least partially mediated by protein kinases such as p160 Rho kinase and p21-activated protein kinase (17), which are reported to affect actin dynamics through regulation of the phosphorylation state of myosin light chain. However, there is no MAPKKK that has been shown to participate in actin reorganization.
In this study, we have identified a novel MAPKKK, designated MLTK for MLK-like mitogen-activated protein triple kinase, whose expression is up-regulated by activation of the ERK pathway in Swiss 3T3 cells. We show that MLTK can be activated by osmotic shock with hyperosmolar media and induce the disruption of actin stress fibers and dramatic changes in cell morphology.

EXPERIMENTAL PROCEDURES
cDNA Subtraction-Swiss 3T3 cells were stably transfected, first with p3ЈSS (Stratagene) and then with HA-LASDSE MEK1 (21) subcloned into pOPI3 (Stratagene). Clones were selected using hygromycin B (200 g/ml; Life Technologies, Inc.) and Geneticin (400 g/ml; Life Technologies, Inc.) as the selection drugs. A clone (3L-9) that showed a high inducibility of MEK1 was obtained and used for the experiment. 5 mM IPTG was added to the medium to induce MEK1 expression (Lac-Swich system, Stratagene).
cDNA subtraction was performed using a PCR-Selected cDNA Subtraction kit (CLONTECH). The mRNA from untreated 3L-9 was used as the "driver." The mRNA from 3L-9 cultured for 6 h after the addition of IPTG and that from 3L-9 cultured for 18 h were mixed, and the mixture was used as the "tester." The subtractive screening was carried out according to the manufacturer's instructions. A 1022-base pair fragment of MLTK was obtained as a candidate positive cDNA clone.
Cloning of MLTK and DNA Construction-5Ј-and 3Ј-RACE analyses for the cDNA fragment obtained were performed using mouse heart Marathon Ready cDNA (CLONTECH) as a template. The internal primers used were 5Ј-GTCCTGTGATATCCATTTGGCTC-3Ј (antisense for 5Ј-RACE) and 5Ј-CGGGAGAGACGTCTCAAGATGTGGG-3Ј (sense for 3Ј-RACE). Each RACE product was cloned into pCR2.1-TOPO (Invitrogen) and sequenced with an ABI 377 sequencer. Full-length mouse MLTK␣ and MLTK␤ were obtained by PCR using mouse heart Marathon Ready cDNA (CLONTECH) as a template. Full-length mouse MLTK␣ and MLTK␤ with BamHI sites at both ends of the cDNA were cloned into pSR␣-HA. For human MLTKs, a human expressed sequence tag data base was searched using the mouse MLTK sequence obtained in the subtraction. Based on the corresponding human sequence, RACE analyses and full-length cloning were done using a Superscript human fetal brain cDNA library (Life Technologies, Inc.) as a template for PCR.
Cell Culture and Transfection-COS-7 and Swiss 3T3 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics (100 units/ml penicillin and 0.2 mg/ml kanamycin). Transfection of the expression vectors was performed using LipofectAMINE and LipofectAMINE PLUS (Life Technologies, Inc.) according to the manufacturer's protocol.
Immunoprecipitation and Kinase Assay-COS-7 cells transfected with HA-MLTK␣ or HA-MLTK␤ were exposed to the various stimuli for 10 min. Cells were lysed, and HA-MLTKs were immunoprecipitated by anti-HA antibody 12CA5 (3 g/300 l of lysate) at 4°C for 2 h. The precipitates were washed three times with phosphate-buffered saline (pH 7.4) containing 0.05% Tween 20 and added to 15 l of kinase reaction buffer (20 mM Tris (pH 7.5), 10 mM MgCl 2 , 100 M ATP, and 2 Ci [␥-32 P]ATP) containing 5 g of His-tagged MKK6 generated in Escherichia coli. The mixtures were incubated at 25°C for 5 min and stopped by adding SDS sample buffer. Samples were subjected to SDS-PAGE, and 32 P incorporated into His-MKK6 was detected by autoradiography. A portion of each precipitate was subjected to immunoblotting using anti-HA antibody.
For the coupled kinase assay, the immunoprecipitates were first incubated with His-MEK1, His-SEK1, or His-MKK6 recombinant protein (0.2 g each/reaction) (26) at 30°C for 15 min in kinase reaction buffer without the radiolabeled ATP. The reactions were then added with the kinase-negative form of GST-ERK, GST-SAPK, or GST-p38 (27) together with 2 Ci of [␥-32 P] ATP and further incubated at 30°C for 7 min. The reactions were stopped by adding SDS sample buffer, and 32 P incorporated into each MAPK protein was detected by autoradiography. Coimmunoprecipitation assay for detecting dimerization of MLTK was performed essentially as described (28) using lysis buffer containing 50 mM Hepes (pH 7.5), 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 10% glycerol, 2 mM dithiothreitol, 10 mM ␤-glycerophosphate, 1 mM NaF, 1 mM Na 2 VO 3 , 1 mM phenylmethylsulfonyl fluoride, and 1% aprotinin.
Cell Staining-Cells were fixed and permeabilized as described previously (22). HA-tagged proteins were detected with anti-HA antibody 12CA5. Actin filaments were visualized using tetramethylrhodamine B isothiocyanate (TRITC)-or fluorescein isothiocyanate (FITC)-labeled phalloidin. Fluorescence images were observed using a Zeiss Axiophoto or a confocal microscope (Bio-Rad).

Identification of a Novel Protein
Kinase, MLTK-To isolate genes whose expression is increased after activation of the ERK/MAPK pathway, we established a stable Swiss 3T3 cell line in which expression of an active MEK1 mutant, LASDSE MEK1 (21), was induced by the addition of IPTG to the culture medium (Fig. 1A, upper). In one such cell line (clone 3L-9), expression of LASDSE MEK1 induced by IPTG treatment FIG. 1. Identification of MLTK as a molecule whose expression is increased by ERK activation. A, cloning of a stable Swiss 3T3 cell line (3L-9) in which expression of HA-tagged LASDSE MEK1 was induced by the addition of IPTG. The cloned cell line (3L-9) was treated with 5 mM IPTG for the indicated times, and the cell lysates were subjected to immunoblotting using anti-HA antibody (upper). After 24 h, morphological transformation of the cells was observed (lower). B, the MLTK transcript is increased by activation of the ERK pathway. The 3L-9 clone was treated with 5 mM IPTG for 0, 6, or 18 h. The total RNA was collected and then subjected to reverse transcription reaction using an oligo(dT) primer. Part of the cDNA obtained was used for RT-PCR analysis using a sense primer (5Ј-CTGCTGACGGAGTGCT-GAAG-3Ј) and an antisense primer (5Ј-GTCGCTCAAGGGTTGCCTCA-3Ј). Note that the amplified PCR fragment corresponds to nucleotides 464 -916 of MLTK, which is common to MLTK␣ and MLTK␤. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was adopted as a control for RT-PCR. The fragment obtained in the subtraction corresponds to nucleotides 2-1023. Our 5Ј-RACE analysis revealed two different upstream nucleotide sequences, both of which contain two and three in-frame stop codons, respectively. Therefore, we conclude that translation starts at nucleotides 38 -40 (ATG). B, schematics of MLTK␣ and MLTK␤. S/T kinase, serine/threonine kinase domain (residues 16 -277); L-zip, leucine zipper motif (residues 287-322); SAM, sterile ␣-motif (residues 337-408). C, comparison of the kinase domain sequences of MLTK, MLK2, and TAK1. Identical residues are shown in dark gray, and homologous residues are shown in light gray. The homology of the kinase domain between MLTK and MLK2 is 64.6%, and that between MLTK and TAK1 is 51.8%. D, Northern blot analysis. The 1022-base pair fragment of mouse MLTK cDNA (obtained in the subtraction) was labeled with digoxigenin using a DIG High Prime kit (Roche Molecular Biochemicals). The probe (25 ng/ml) was then hybridized to human multiple-tissue Northern membranes (CLONTECH). The bands were detected by alkaline phosphatase-conjugated anti-digoxigenin antibody (Roche Molecular Biochemicals) according to the manufacturer's protocol. Arrowheads indicate the detected bands. E, nuclear export of MLTK. Upper, the candidate NES sequences in MLTK are aligned with the NES sequence in MEK1 (44). Both MLTK␣ and MLTK␤ have two NES-like sequences. The presumably important hydrophobic residues are shown in boldface. Lower, leptomycin B (LMB) treatment resulted in the nuclear accumulation of MLTK. HA-MLTK␣ and HA-MLTK␤ were expressed in COS-7 cells, and the cells were either left untreated (ϪLMB) or treated with leptomycin B (2 ng/ml) for 2 h (ϩLMB). The subcellular localization of MLTKs was determined by immunostaining using anti-HA antibody. Experiments were repeated twice and gave similar results.
caused morphological transformation of the cells (Fig. 1A, lower). We performed a subtractive cDNA screening using mRNAs from untreated and IPTG-treated 3L-9 cells. Among the iso-lated cDNA clones, we found a cDNA fragment that encodes a novel protein kinase that appears to be a member of the MAP-KKK family. RT-PCR analysis revealed that the corresponding FIG.2-continued mRNA existed in unstimulated cells and was being up-regulated at least 18 h after the induction of active MEK1 (Fig. 1B). Northern blot analysis of the cDNA fragment obtained showed that the transcript (ϳ7.7 kilobases) was expressed ubiquitously, with higher expression levels in heart and skeletal muscle (Fig. 2D). There were two minor bands of ϳ3.3 and 1.6 kilobases in human heart and skeletal muscle (Fig. 2D).
The 5Ј-and 3Ј-RACE analyses showed that at least two mRNA variants with different 3Ј-sequences are expressed in mouse heart. cDNA cloning of the two variants and the open reading frame prediction revealed that the two proteins have identical amino acid sequences in the N-terminal region (residues 1-311), which contains a kinase domain and a leucine zipper motif (Fig. 2, A and B). The kinase domain shows the highest homology to that of MLK2 with 44.7% identity and is 32.7% identical to that of TAK1 (Fig. 2C). Thus, we termed this kinase MLTK for MLK-like mitogen-activated protein triple kinase. The two isoforms, designated MLTK␣ and MLTK␤, respectively, may be alternatively spliced forms. The C-terminal region of MLTK␤ (residues 268 -395) shows homology to the C-terminal region of TAK1 (42.5% identity). The C-terminal region of MLTK␣ has a sterile ␣-motif that has been shown to be involved in the protein-protein interactions and dimer formation of several signaling molecules and transcription regulators (29). The calculated molecular masses of MLTK␣ and MLTK␤ are 91.7 and 51.3 kDa, respectively.
Each MLTK contains two putative nuclear export signal (NES)-like sequences, which are characterized by four hydrophobic residues with appropriate spacing (Fig. 2E, upper). We expressed HA-tagged MLTK␣ and MLTK␤ in COS-7 cells and determined their subcellular localization. Both HA-MLTK␣ and HA-MLTK␤ were present mainly in the cytoplasm (Fig.  2E, lower). When the cells were treated with leptomycin B, which inhibits the function of CRM1, an NES receptor, HA-MLTK␤ accumulated in the nuclei (Fig. 2E, lower). HA-MLTK␣ also tended to appear in the nucleus after leptomycin B treatment, but nuclear accumulation of HA-MLTK␣ was observed in ϳ10% of the expressing cells (Fig. 2E, lower). These results suggest that the cytoplasmic localization of both MLTK␣ and MLTK␤ may be maintained by their NESs and that both molecules may be shuttling between the nucleus and the cytoplasm. However, we could not detect significant changes in the subcellular distribution of MLTK␣ or MLTK␤ after activation of these kinases by stimulation with osmotic shock (data not shown; see below).
We then examined whether the activity of MLTK was upregulated by extracellular stimuli. COS-7 cells were transfected with HA-MLTK␣ and exposed to various stimuli for 10 min. HA-MLTK␣ was then immunoprecipitated and subjected to the in vitro kinase assay using His-MKK6 as an exogenous substrate. MLTK␣ had a relatively high activity when overexpressed in the cells (Fig. 4A). Treatment with fetal calf serum (15%), insulin (5 g/ml), TGF-␤ (20 ng/ml), 12-O-tetradecanoylphorbol-13-acetate (100 ng/ml), lysophosphatidic acid (100 ng/ml), UV (80 J/m 2 ), and anisomycin (100 g/ml) did not activate MLTK␣ (Fig. 4A) or MLTK␤ (data not shown). On the other hand, treatment with NaCl (0.5 M) significantly activated MLTK␣ (Fig. 4, A and B) and MLTK␤ (data not shown). The NaCl treatment caused a slight mobility retardation of MLTK␣ in SDS-PAGE (Fig. 4A). Treatment with sorbitol (0.5 M) or KCl (0.5 M) as well as NaCl (0.5 M) also caused a mobility shift of MLTK␣ and MLTK␤ (Fig. 4B, lower) and induced an increase in their kinase activity (the activity of MLTK␣ for MKK6 and the autophosphorylation activity of MLTK␤) (Fig. 4B, upper). The in vitro treatment of immunoprecipitated HA-MLTK␣ with calf intestine alkaline phosphatase increased its mobility in SDS-PAGE and decreased its kinase activity (Fig. 4C). Thus, the mobility retardation correlates with the kinase activity of MLTK.
We then produced and expressed kinase-negative mutants of MLTK, MLTK␣-KM and MLTK␤-KM, in which Lys 45 was replaced by Met. COS-7 cells were transfected with each HAtagged form of the MLTKs and either left untreated or stimulated by osmotic shock (0.5 M NaCl). No kinase activity for MKK6 was detected in MLTK␣-KM and MLTK␤-KM. Moreover, although a clear mobility shift in SDS-PAGE was seen in wild-type MLTK␣ and wild-type MLTK␤ in response to osmotic shock, no mobility shift took place in the case of MLTK␣-KM and MLTK␤-KM (Fig. 4B, left), indicating that the stimulusdependent mobility shift of MLTK is caused by autophosphorylation. It was confirmed that MLTK␣-KM and MLTK␤-KM cannot activate any of the coexpressed MAPKs (Fig. 4D, right; and data not shown). Taken together, these results suggest that both MLTK␣ and MLTK␤ are activated through autophosphorylation in response to osmotic stress with hyperosmolar media.
It has been reported that autophosphorylation of MEKK1, TAK1, or Ssk2 is achieved through an intramolecular reaction (30 -32). To examine whether the autophosphorylation of MLTK occurs through an inter-or intramolecular reaction, we tested if the autophosphorylation occurs between two isoforms of MLTK. COS-7 cells were transfected with HA-tagged MLTKs and either left untreated or stimulated by osmotic shock (0.5 M NaCl). Both wild-type MLTK␣ and wild-type MLTK␤ showed a mobility shift in SDS-PAGE in response to osmotic shock (Fig. 5, first through sixth lanes), whereas MLTK␣-KM and MLTK␤-KM did not (eleventh and twelfth lanes). When MLTK␣-KM was coexpressed with wild-type MLTK␤, however, MLTK␣-KM showed a relatively weak but significant mobility shift in response to osmotic shock (Fig. 5,  ninth and tenth lanes). In contrast, MLTK␤-KM did not show a mobility shift in response to osmotic shock when coexpressed with wild-type MLTK␣ (Fig. 5, seventh and eighth lanes). Therefore, part of the autophosphorylation of MLTK␣, but not MLTK␤, may be achieved through an intermolecular reaction. The stimulus-dependent mobility shift of kinase-negative MLTK␣ seems not to be complete compared with wild-type MLTK␣, suggesting that both inter-and intramolecular reactions contribute to the full autophosphorylation of MLTK␣ in response to osmotic shock. The autophosphorylation of MLTK␤ may occur through an intramolecular reaction.
MLTK␣, but Not MLTK␤, Causes Disruption of Actin Stress Fibers and Changes in Cell Morphology-We examined the effect of MLTK expression on cell morphology and the cytoskeleton. HA-MLTK␣ or HA-MLTK␤ was expressed in Swiss 3T3 cells, and actin filaments were stained with FITC-labeled phal-loidin. Expression of MLTK␣, but not MLTK␤, caused disruption of actin stress fibers and cell shrinkage (Fig. 6). The number of stress fibers or actin bundles decreased, and each fiber became thin 3 or 5 h after transfection of MLTK␣ (Fig. 6A,  3h and 5h panels). At 7-9 h, the stress fibers disappeared almost completely, and the dot-like staining of actin, probably representing amorphous aggregates of actin filaments, appeared in the cytoplasm (Fig. 6A, 7h and 9h panels). At the same time, the cells gradually shrunk, and several protrusions were observed and became larger. At 18 h, cell shrinkage proceeded dramatically, and the cells appeared to be on the point of being detached from the culture dish (Fig. 6A, 18h  panel). Despite the dramatic change in cell shape, the nuclear morphology was normal throughout (Fig. 6A, MLTK␣ panels), which was also seen in the 4Ј,6-diamidino-2-phenylindole (DAPI) staining of the nuclear DNA (data not shown). After further incubation (30ϳ40 h), the cells expressing MLTK␣ often showed apoptotic nuclei and finally were detached from the culture dish (data not shown). It can be speculated that the loss of adequate attachment to the extracellular matrix might trigger apoptosis. Expression of MLTK␤ also caused the diminution of thickly bundled stress fibers to some extent. The effect was, however, much weaker than that caused by MLTK␣; expression of MLTK␤ did not cause the complete disruption of stress fibers (Fig. 6, B and C) or a dramatic change in cell morphology (B and D).
To examine whether or not other MAPKKKs also cause these changes in stress fibers and cell morphology, we tested ⌬N-MEKK1 (an activated form of MEKK1) (33), ASK1 (16), and ⌬N-TAK1 (an activated form of TAK1) (14). Expression of any of these MAPKKKs did not cause apparent changes in stress fibers or cell morphology within 4 h of transfection (data not shown). After 20 h, in the cells expressing any of the MAP-KKKs, the stress fibers became significantly decreased and thinner, and cell shrinkage took place (Fig. 6E). Semiquantitative analysis showed that these MAPKKKs caused almost complete loss of stress fibers in 20 -30% of the cells and dramatic cell shrinkage in 20 -35% of the cells. Thus, all the MAPKKKs tested can induce cellular changes similar to those caused by MLTK␣. However, MLTK␣ was stronger than other MAPKKKs in causing these changes in the cell. Especially, extraordinarily severe cell shrinkage was seen only in the cells expressing MLTK␣. It is worth mentioning that the number of cells expressing ⌬N-MEKK1 was drastically reduced 20 h after trans- fection, presumably due to cell death (data not shown).

MLTK␣-induced Stress Fiber Disruption and Cell Shrinkage
Require the p38 Pathway-We expressed the kinase-negative form of MLTK␣ (MLTK␣-KM) to determine whether the kinase activity of MLTK␣ is required for causing these cellular changes. As shown in Fig. 7 (A and B), expression of MLTK␣-KM did not cause stress fiber disruption or cell shrinkage. Thus, the kinase activity of MLTK␣ is indispensable. As MLTK␣ can activate four MAPK pathways (ERK, JNK/SAPK, p38, and ERK5/BMK1), we then examined which pathway is important for causing these changes. To do this, we expressed dominant-negative forms of each MAPKK (MEK1-SASA, MKK6-SATA, and MEK5-SATV, in which two activation phosphorylation sites were replaced by alanine and/or valine; and MKK7-KL, in which a lysine residue involved in ATP binding was replaced by leucine) together with MLTK␣. It has previously been demonstrated that each dominant-negative MAPKK specifically inhibits activation of its own downstream MAPK. Expression of MKK6-SATA, but not that of MEK1-SASA, MEK5-SATV, or MKK7-KL, inhibited MLTK␣-induced stress fiber disruption (Fig. 7A). Moreover, expression of MKP-5, a dual specificity phosphatase that strongly inactivates p38 (34), inhibited MLTK␣-induced stress fiber disruption as well (data not shown). Thus, MLTK␣-induced activation of the p38 pathway is required for MLTK␣-induced stress fiber disruption. Similarly, expression of the dominant-negative form of MKK6 (Fig. 7B) or MKP-5 (data not shown) most severely impaired MLTK␣-induced cell shrinkage. Although it is possible that MAPK pathways other than the p38 pathway are also involved in the MLTK␣-induced cellular changes by cooperating with the p38 pathway, coexpression of other dominantnegative forms of MAPKKs with dominant-negative MKK6 did not further enhance the inhibitory effect of dominant-negative MKK6 (data not shown).
To test if activation of the p38 pathway is sufficient for the MLTK␣-induced cellular changes, we expressed MKK6, which is able to specifically activate the p38 pathway (35). Expression of MKK6 did not cause changes in stress fibers and cell morphology (Fig. 8). Coexpression of p38 and MKK6 also did not cause these cellular changes (data not shown). Therefore, although activation of the p38 pathway is necessary, it is not sufficient for the MLTK␣-induced cellular changes. It is possible that kinase activity-independent actions of MLTK␣ might COS-7 cells transfected with the HA-tagged wild-type or kinase-negative mutant of MLTK␣ or MLTK␤ were either left untreated or stimulated with 0.5 M NaCl for 15 min. HA-MLTKs were then immunoprecipitated and subjected to in vitro kinase assays using His-MKK6 as an exogenous substrate. We could not detect marked activation of wildtype MLTKs by osmotic shock in this experiment, in which the kinase assay was performed at a higher temperature (30°C) and for longer times (15 min) than in A or B to detect the possible low kinase activity of MLTK␣-KM and MLTK␤-KM. Right, lysates of COS-7 cells cotransfected with HA-MLTKs and HA-ERK were subjected to immunoblotting with anti-phosphorylated (P) ERK antibody as in the legend to Fig. 3A.

FIG. 4. MLTK is activated by osmotic shock.
A, COS-7 cells transfected with HA-MLTK␣ were either left untreated (NT, no treatment) or exposed to various stimuli (15% fetal calf serum (FCS), 5 g/ml insulin, 100 ng/ml 12-O-tetradecanoylphorbol-13-acetate (TPA), 20 ng/ml TGF-␤, 100 ng/ml lysophosphatidic acid (LPA), 0.5 M NaCl, 80 J/m 2 UV, and 100 g/ml anisomycin) for 10 min. HA-MLTK␣ was immunoprecipitated (IP) and subjected to the in vitro kinase assay at 25°C for 5 min using His-MKK6 as the exogenous substrate. The results from autoradiography are shown (upper). A portion of the immune complex was analyzed by immunoblotting with anti-HA antibody (lower). Experiments were repeated twice with similar results. The mobility retardation of HA-MLTK␣, which correlated with its activation, was observed. B, shown is the activation of MLTK␣ by osmotic shock. COS-7 cells transfected with either MLTK␣ (left) or MLTK␤ (right) were exposed to the indicated osmotic stimuli (0.5 M each). The phosphorylation activity of MLTK␣ for His-MKK6 in vitro was measured (left), as was the autophosphorylation activity of MLTK␤ (right). Immunoprecipitated MLTKs were also subjected to immunoblotting with anti-HA antibody (lower). In the upper left panel, -fold increases in phosphate incorporation into MKK6 are shown at the bottom of each lane. C, overexpressed MLTK␣ undergoes autophosphorylation in the cells. HA-MLTK␣ was immunoprecipitated in the presence of phosphatase inhibitors (12 mM ␤-glycerophosphate, 5 mM NaF, 1 mM sodium PP i , and 1 mM Na 2 VO 3 ). Immunoprecipitated HA-MLTK␣ was washed and either left untreated or treated with calf intestine alkaline phosphatase (ϩPPase) or a control buffer (ϩmock). Then, the kinase activity for MKK6 was measured (upper), and the mobility in SDS-PAGE was analyzed as described for A (lower). Immunoprecipitation in the absence of phosphatase inhibitors abrogated both the kinase activity and mobility retardation of MLTK␣ (data not shown). Note that in this autoradiography (upper), the exposure time was longer than that in A and B. Experiments were repeated twice with similar results. D, shown are the effects of the kinase-negative mutation of MLTK. Left, be involved in causing these cellular responses. Then, to test this idea, we coexpressed the kinase-negative form of MLTK␣ (MLTK␣-KM) with MKK6, but coexpression of MLTK␣-KM and MKK6 did not cause these cellular changes (Fig. 8). Expression of MKK7, an activator of JNK/SAPK, with or without MLTK␣-KM did not cause these cellular changes at all (Fig. 8). Expression of an active form of MEK5 (MEK5(D)) or coexpression of MKK7 and MKK6 did not cause these responses, either. Thus, an MLTK␣ substrate(s) other than the known MAPKKs may cooperate with the MLTK␣-activated p38 pathway to cause such cellular responses. It should be noted that the strong activation of the ERK pathway by expression of a strongly active form of MEK1 (LASDSE), but not the moderate activation of ERK by an active MEK (SDSE), caused stress fiber disruption and cell shrinkage ( Fig. 8; data not shown), as previously reported (36). DISCUSSION In this study, we have identified and characterized a novel member of the MAPKKK family (designated MLTK) that can be activated by osmotic shock with hyperosmolar media and that can regulate actin organization and cell morphology. MLTK contains a leucine zipper motif just after the N-terminal kinase domain in the primary structure (Fig. 2B). All of the members of the MLK subfamily of MAPKKKs have leucine zipper motifs (usually two). Moreover, the kinase domain of MLTK is most similar to that of MLK2. Thus, MLTK may be a close relative of the MLK subfamily of MAPKKKs and was therefore termed MLTK for MLK-like mitogen-activated protein triple kinase, although MLTK seems to have only one leucine zipper motif. MLTK␣, but not MLTK␤, possesses a sterile ␣-motif just after the leucine zipper motif in the primary sequence. Through these motifs, MLTKs might interact with other proteins that are involved in the regulation of MLTK or with downstream targets of MLTK. Interestingly, both MLTK␣ and MLTK␤ have NES sequences (Fig. 2E). Consistent with this, they are present mainly in the cytoplasm; and after treatment with the NES inhibitor leptomycin B, they come to localize also in the nucleus (Fig. 2E). The physiological significance of the NES-dependent cytoplasmic localization of MLTK should be examined in future studies.
Activation of MLTK-MLTKs are specifically activated by osmotic shock with hyperosmolar media. In budding yeast, Ssk2, a MEKK family MAPKKK, is shown to be activated by Ssk1, a regulatory protein in the two-component osmosensor whose binding to Ssk2 induces the autophosphorylation of Ssk2 (32). MLTKs might also lie downstream of a mammalian osmosensor system whose molecular identity is unknown.
As for some of the MAPKKKs, their possible activation mechanisms have been analyzed (28,31,37,38). For example, MLK3 has a CRIB (Cdc42 and Rac interactive binding) domain, and binding of activated Rac/Cdc42 to MLK3 enhances dimeriza- tion of MLK3. Dimerization of MLK3 is achieved through its leucine zippers and is required for both its autophosphorylation and kinase activity for downstream targets (28). TAK1 undergoes autophosphorylation in response to upstream stimuli by an unknown mechanism (which may involve its association with adapter proteins TAB1 and TAB2). This autophosphorylation of TAK1 is required for activation of TAK1 activity for downstream targets (31).
We showed that MLTK forms a dimer; both MLTK␣ and MLTK␤ were able to form either a homodimer or a heterodimer, at least when they were overexpressed in cells. However, dimerization of MLTK per se might not be directly involved in the regulation of MLTK since kinase-negative forms of MLTK (MLTK␣-KM and MLTK␤-KM) were able to form dimers. 2 On the other hand, because phosphatase treatment of MLTK markedly decreased the kinase activity of MLTK for substrates, phosphorylation of MLTK induces its activation. Importantly, this stimulus-dependent phosphorylation of MLTK seems to be caused by autophosphorylation since the mobility shift of MLTK in SDS-PAGE, which results from phosphorylation, was not observed at all for a kinasenegative form of MLTK. Thus, MLTK may be activated by enhancement of the autophosphorylation activity of MLTK, the molecular mechanism of which is not known at present.
As overexpressed MLTK has kinase activity for substrates under unstimulated conditions, MLTK seems to be able to 2 M. Adachi, I. Gotoh, and E. Nishida, unpublished data. undergo autophosphorylation when overexpressed. Our data show that overexpressed MLTK␣ migrates more slowly than a presumably non-phosphorylated form of MLTK␣ that was obtained by phosphatase treatment and that wild-type MLTK␣ migrates always more slowly than kinase-negative MLTK␣. Unlike MLTK␣, however, a difference in mobility was not detected between wild-type and kinase-negative MLTK␤. It is possible that the difference in their mobility was below the limit of our observation, and/or there might be an additional autophosphorylation site(s) in MLTK␣ that is not present in MLTK␤.
MLTK-induced Activation of the MAPK Pathways-MLTKs are capable of activating the four major MAPK pathways: ERK, JNK/SAPK, p38, and ERK5/BMK1. The kinase domain of MLTK␣ is identical to that of MLTK␤, and the activation profiles for various MAPKs are similar between the two isoforms. So far, only two MAPKKKs, Cot/Tpl-2 and MEKK3, have been reported to activate ERK5/BMK1 (13,39). MLTKs have been shown here to strongly activate the ERK5/BMK1 pathway, suggesting that MLTK might be one of the important activators of the ERK5/BMK1 pathway in vivo.
MLTK␣-induced Stress Fiber Disruption and Cell Shrinkage-Expression of MLTK␣ in Swiss 3T3 cells causes the disruption of actin stress fibers and cell shrinkage. A kinasenegative mutant of MLTK␣ did not cause these cellular changes, suggesting that the kinase activity of MLTK␣ is essential. Our results also show that the kinase activity of MLTK␣ is indispensable for MLTK␣-induced activation of the MAPK pathways and that, among the MAPK pathways activated by MLTK␣, the p38 pathway is specifically required for the MLTK␣-induced cellular changes. This is in line with previous studies demonstrating the requirement of the p38 pathway for stimulus-dependent rearrangement of the actin cytoskeleton (40 -42). Our results further show that activation of the p38 pathway alone is not sufficient to cause these changes and that MLTK␣ may have a substrate(s) other than the known MAPKK family molecules that cooperates with the p38 pathway to cause these cellular changes. In this regard, it is interesting to note that MLTK␤, which has almost the same capacity to activate the MAPK pathways as MLTK␣, had little effect on actin organization and cell morphology. As MLTK␣ has a long C-terminal region that is not present in MLTK␤, it is possible that this C-terminal region of MLTK␣ is important for the MLTK␣-induced cellular responses. We speculate that the C-terminal region is involved in docking to some substrate(s) whose phosphorylation is required for the MLTK␣-induced cellular changes. Moreover, to understand the molecular mechanism of the cellular responses, it is important to determine whether these cellular responses require de novo protein synthesis. It should also be examined in future studies if endogenous MLTK is significantly functional in regulating organization of the actin cytoskeleton and cell morphology.
The present observation that several MAPKKKs other than MLTK␣ can also induce cellular changes similar to those caused by MLTK␣ is surprising. As activation of the MAPK pathways alone appears not to be sufficient to induce these effects, it is possible that these MAPKKKs, like MLTK␣, can have some substrates or activate some pathways (other than the MAPK pathways) that affect the actin cytoskeleton and/or cell morphology. Although activation of these different MAPKKKs would induce apparently similar phenotypes (i.e. rearrangement of the actin cytoskeleton), this would not necessarily imply that these MAPKKKs have a common direct target.
It has been reported that the low molecular weight GTPase Rho regulates the formation of actin stress fibers and the assembly of focal adhesions and that the inhibition of Rho function by C3 toxin blocks stress fiber formation (19,43). It is possible that MLTK␣ modulates the activity of Rho or other Rho family proteins. However, our preliminary analysis with GST-PBD (p21 Rac/Cdc42 binding domain) did not detect any marked change in the activation state of Rac when coexpressed with MLTKs in Swiss 3T3 or COS-7 cells. 2 Conversely, it is also possible that MLTK␣ is an effector of these GTPases. In our preliminary experiments, however, expression of an active form of Ras, Cdc42, Rac, or Rho did not activate coexpressed MLTK␣ in COS-7 cells. Thus, MLTK␣ might not be an effector of these GTPases. In agreement with this, we could not detect binding between MLTK␣ and these GTPases (an activated form or a dominant-negative form of Cdc42, Rac, or Rho) when they were coexpressed in COS-7 cells. 2 Further detailed experiments are needed to clarify the relationship between MLTK␣ and the Rho family GTPases.