(Received for publication, November 14, 1996, and in revised form, March 24, 1997)
From the Department of Molecular Biology, Yokohama
City University School of Medicine, 3-9, Fukuura, Kanazawa-ku, Yokohama
236, Japan, the ¶ Genetics Division, National Cancer Center
Research Institute, Tsukiji 5-chome, Chuo-ku, Tokyo 104, Japan, the
Diabetes Unit and Medical Services, Massachusetts General
Hospital East, and the Department of Medicine, Harvard Medical School,
Charlestown, Massachusetts 02129, and the ** Dana-Farber Cancer
Institute, Harvard Medical School,
Boston, Massachusetts 02115
c-Jun N-terminal kinases/stress-activated protein kinases (JNKs/SAPKs) are mitogen-activated protein kinase (MAPK)-related protein kinases that are involved in several cellular events, including growth, differentiation, and apoptosis. Mixed lineage kinases (MLKs) form a family of protein kinases sharing two leucine zipper-like motifs and a kinase domain whose primary structure is similar to both the tyrosine-specific and the serine/threonine-specific kinase classes. We have reported that a member of the MLK family, MUK/DLK/ZPK, can activate JNK/SAPK in vivo, and here we show that another member of the MLK family, MST/MLK2, activates JNK/SAPK. Both MUK/DLK/ZPK and MST/MLK2 cause a slight activation of p38/Mpk2 when overexpressed in COS-1 cells, whereas MST/MLK2, but not MUK/DLK/ZPK, activates extracellular response kinase (ERK) to a certain degree. The activity of SEK1/MKK4/JNKK, a MAPK kinase class protein kinase designated as a direct activator of JNK/SAPK, is also induced by MUK/DLK/ZPK or MST/MLK2 overexpression. Furthermore, recombinant MST/MLK2 produced in bacteria directly phosphorylates and activates SEK1/MKK4/JNKK in vitro, showing that MST/MLK2 acts like a MAPK kinase kinase. Taken together, these results suggest that MLK family members are MAPK kinase kinases preferentially acting on the JNK/SAPK pathway.
Protein kinases of the MAPK1 family
are involved in a variety of cellular responses to cytokines, hormones,
and stress-inducing reagents. MAPKs are classified into at least three
groups, including ERK, JNK/SAPK, and p38/Mpk2, and each group
phosphorylates a different set of substrates, including other protein
kinases and transcription factors (1-3). JNKs/SAPKs were first
identified as c-Jun N-terminal kinases (JNKs) or stress-activated
protein kinases (SAPKs), both of which are activated in response to UV
irradiation, osmotic shock, heat shock, and treatment with tumor
necrosis factor- or antibiotics known to be inhibitors of eukaryotic
protein synthesis (4-7). The activity of JNKs/SAPKs depends greatly on
the phosphorylation of specific threonine and tyrosine residues located
in kinase subdomain VIII catalyzed by upstream activators or MAPKK
class protein kinases (5). One such activator, known as SEK1/MKK4/JNKK, has been cloned, and its activity has been shown to be greatly dependent on the phosphorylation of specific serine and threonine residues located between kinase subdomains VII and VIII, as reported for MEKs, activators of ERKs (8-10). The activation-phosphorylation of
SEK1/MKK4/JNKK is catalyzed by upstream protein kinases or MAPKKK class
protein kinases such as MEKK1 (9, 11). Since JNK/SAPK is induced by
several kinds of cellular stimuli, diverse MAPKKKs may act upstream of
the JNK/SAPK pathway. In fact, increasing numbers of protein kinases
are now identified as MAPKKKs acting on SEK1/MKK4/JNKK, including MEKK2
and MEKK3 (12) and Tpl-2/Cot (13). However, the molecular mechanism
that regulates these MAPKKKs remains mostly unknown. In addition to
SEK1/MKK4/JNKK, at least two other stress-induced JNK/SAPK activators
have been identified in column fractions of cell extracts, suggesting
the presence of multiple pathways for JNK/SAPK activation (14, 15).
Recently, we reported that a protein kinase, MUK, activates the JNK/SAPK pathway when overexpressed in cultured cells (16). MUK is a rat homologue of mouse DLK (17) and human ZPK (18), whose kinase domain shows structural features of both the tyrosine-specific and serine/threonine-specific protein kinase classes. In addition to a unique kinase domain, MUK/DLK/ZPK contains two leucine zipper-like structures located proximal to the C-terminal end of the catalytic domain. These features are common to members of the mixed lineage kinase (MLK) family, including MLK1 (19), MST/MLK2 (20, 21), and SPRK/PTK1/MLK3 (22-24). Therefore, MUK/DLK/ZPK has been identified as the fourth member of the MLK family (16-18).
The finding that MUK/DLK/ZPK activates the JNK/SAPK pathway raises the intriguing possibility that other members of the MLK family also activate the JNK/SAPK pathway or other MAPK pathways. More recently, we have shown that SPRK/PTK1/MLK3 also activates the JNK/SAPK pathway preferentially through SEK1/MKK4/JNKK (25). However, the ability to activate SEK1/MKK4/JNKK directly has not been demonstrated for either SPRK/PTK1/MLK3 or MUK/DLK/ZPK, leaving the possibility that another direct activator of SEK1/MKK4/JNKK may be involved. In this report, we show that MST/MLK2, like MUK/DLK/ZPK, activates JNK/SAPK and SEK1/MKK4/JNKK in COS-1 cells. Furthermore, we show that recombinant MST/MLK2 directly phosphorylates and activates a recombinant SEK1/MKK4/JNKK in vitro. These results show that MST/MLK2 and probably other members of the MLK family also are MAPKKKs that act preferentially on the JNK/SAPK pathway.
The mammalian expression vectors for MUK
(16), GST-SEK1 (8), and the epitope-tagged enzymes T7-tagged MUK,
T7-tagged MUK/KN, T7-tagged JNK1, and T7-tagged ERK2 (16) and HA-tagged p38 (25) have been described elsewhere. The expression vector for MST
was constructed with the mammalian expression vector SR and human
MST cDNA (20). T7-tagged MST was constructed with SR
, a DNA
fragment encoding a His tag/T7 epitope derived from pBlueBac
(Invitrogen), and MST cDNA. The first nine amino acid residues of
the MST protein produced from this expression vector were replaced by a
His tag/T7 epitope consisting of 40 amino acid residues. A
kinase-defective mutant of MST, MST/KN, was constructed with the
T7-tagged MST expression vector by in-frame deletion of residues
140-185, corresponding to kinase subdomains III-V.
COS-1 cells were routinely grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were transfected by an electroporation method using 16 µg of DNA for 5 × 106 cells seeded in four 10-cm dishes after the transfection. Cells were further cultured for 48 h before harvest. When required, 100 ng/ml 12-O-tetradecanoylphorbol-13-acetate or 0.7 M NaCl was added 10 or 40 min before harvest. For UV irradiation, cells were exposed to 100 J of UV light/m2 40 min before harvest.
Immunoprecipitation and Solid-phase Kinase Assay of Epitope-tagged JNK1, ERK2, and p38Cells grown in 10-cm dishes
were lysed in 250 µl of lysis buffer consisting of 20 mM
Tris-HCl, pH 8.0, 2 mM EDTA, 50 mM
-glycerophosphate, 0.1 mM
Na3VO4, 1% Triton X-100, 0.5% deoxycholate,
0.1% SDS, 10% glycerol, 1 mM phenylmethylsulfonyl
fluoride, 10 µg/ml leupeptin, and 10 µg/ml aprotinin. Deoxycholate
and SDS were omitted when antibody against the hemagglutinin tag was
used for immunoprecipitation. Immunoprecipitation of epitope-tagged
enzymes was carried out at 4 °C for 1 h using 1 µg of mouse
monoclonal antibody against either the T7 epitope (Invitrogen) or the
HA tag (Boehringer Mannheim) pre-fixed on 5 µl of protein G-Sepharose
4 Fast Flow (Pharmacia Biotech Inc.). The Sepharose resin was washed
five times with lysis buffer and finally washed with detergent-free
lysis buffer, and aliquots were collected in two tubes: one was used
for the solid-phase kinase assay, and the other was extracted with SDS sample buffer for Western blot analysis. For the solid-phase kinase assay, the epitope-tagged enzymes fixed on 2 µl of protein
G-Sepharose were suspended in 20 µl of assay mixture containing 20 mM HEPES, pH 7.5, 15 mM MgCl2, 15 mM
-glycerophosphate, 0.1 mM
Na3VO4, 2 mM dithiothreitol, 25 µM ATP, 5 µCi of [
-32P]ATP, and 1 µg
of recombinant c-Jun, recombinant CREBP-1/ATF2, or bovine brain myelin
basic protein (MBP) (Sigma M1891). The kinase reaction was carried out
at 30 °C for 20 min and stopped by adding 25 µl of SDS sample
buffer. The phosphorylation of each substrate was detected by SDS-PAGE,
and the amount of 32P incorporated was quantified with a
Fuji BAS2000 image analyzer.
Expression vectors for GST-SEK1, GST-JNK1, and GST-ERK1
were constructed with the bacteria expression vector pGEX-2T or pGEX-3X (Pharmacia) and cDNA encoding SEK1, the SEK1(K/R) mutant (9), MEK1,
the MEK1(K/R) mutant (37), JNK1, or ERK1 (16). To construct the
MalE-MST expression vector, a cDNA fragment encoding MST amino acids 10-757 was cloned into the pMAL-c2 vector (New England Biolabs Inc.). The expression vector for c-Jun was described elsewhere (16).
The expression vector for GST-CREBP-1 was the kind gift of Drs. T. Maekawa and S. Ishii (38). The production of GST and MalE fusion
proteins in bacteria (Escherichia coli DH1) was started by
adding 0.5 mM
isopropyl--D-thiogalactopyranoside and continued for
18 h at 25 °C. The fusion proteins were purified according to
the standard protocol using glutathione-Sepharose (Pharmacia) or
amylose resin (New England Biolabs Inc.). Routinely, 1~5 mg of fusion
proteins were recovered from 500 ml of bacteria culture.
COS-1
cells grown in 10-cm dishes were lysed in 250 µl of G buffer
containing 20 mM Tris-HCl, pH 8.0, 2 mM EDTA,
50 mM -glycerophosphate, 1 mM
Na3VO4, 1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, and
10 µg/ml aprotinin. GST-SEK1 was recovered using 5 µl of
glutathione-Sepharose. The resin was washed five times with G buffer
and then washed one time with detergent-free G buffer. GST-SEK1 was
eluted from the resin with 100 µl of 5 mM glutathione in
detergent-free G buffer.
For the in vitro kinase assay, 5 µl of the GST-SEK1
solution were first incubated with 50 ng of GST-JNK1 in 20 µl of the
assay mixture described above containing 50 µM ATP at
30 °C for 15 min. Then, 2 µl of substrate mixture containing 1 µg of c-Jun, 2.5 mM ATP, and 5 µCi of
[-32P]ATP were added, and the mixture was further
incubated for 15 min. The reaction was stopped by adding 25 µl of SDS
sample buffer, and the phosphorylation of c-Jun was quantified as
described above.
The in vitro phosphorylation of GST-SEK1(K/R) or GST-MEK1(K/R) by MalE-MST-(10-757) was carried out at 30 °C for 45 min in 20 µl of assay mixture containing 1 µg of GST-SEK1(K/R) or 2 µg of GST-MEK1(K/R) and 1 or 4 µg of MalE-MST-(10-757), adjusting the total amount of protein to 6 µg by adding MalE protein. The reaction was stopped by adding 25 µl of SDS sample buffer, and the proteins were analyzed by SDS-PAGE with Coomassie Brilliant Blue staining and autoradiography.
The activation of GST-SEK1 or GST-MEK1 by MalE-MST-(10-757) was tested
by a two-step kinase reaction. The first step was carried out at
30 °C for 45 min in 20 µl of assay mixture containing 25 µM unlabeled ATP, 50 ng of GST-JNK1 or GST-ERK1, 200 ng
of GST-SEK1 or GST-MEK1, and 1 or 4 µg of MalE-MST-(10-757),
adjusting the total amount of protein to 5.25 µg by adding MalE
protein. The second reaction was started by adding 2 µl of substrate
mixture containing 1 µg of c-Jun or myelin basic protein, 2.5 mM ATP, and 5 µCi of [-32P]ATP and
incubating the reaction mixture for a further 15 min. The
phosphorylation of c-Jun or myelin basic protein was analyzed by
SDS-PAGE as described above.
To test the ATP dependence of the activation of GST-SEK1 by MalE-MST-(10-757), 10 µg of MalE-MST-(10-757) and 2 µg of GST-SEK1 were incubated in 200 µl of assay mixture in the presence or absence of 50 µM ATP at 30 °C for 30 min. GST-SEK1 was then affinity-purified and assayed for JNK1 activation activity as described for the in vitro assay of GST-SEK1 expressed in COS-1 cells.
The ability
of MST to activate three different MAPK-related protein kinase pathways
(JNK/SAPK, p38/Mpk2, and ERK/MAPK) was tested in cotransfection
experiments using COS-1 cells with immunoprecipitation of the
epitope-tagged enzymes and solid-phase kinase assay using appropriate
substrates as indicated in Fig. 1 (A-C). The
overexpression of MUK and MST, like UV irradiation, caused a >10-fold
induction of tagged JNK1 activity (Fig. 1A). The activity of
tagged p38 was induced modestly (1.2~1.5-fold) by MUK or MST under
conditions where osmotic shock with 0.7 M NaCl caused a
>5-fold induction (Fig. 1B). The activity of tagged ERK2
was not induced by MUK at all (Ref. 16 and Fig. 1C), whereas
it was induced modestly (2~3-fold) by MST, and
12-O-tetradecanoylphorbol-13-acetate stimulation caused a
>10-fold induction (Fig. 1C). The immunoprecipitated enzymes were also analyzed by SDS-PAGE followed by Western blotting using anti-tag antibodies (Fig. 1D). Note that the amounts
of enzymes used for the solid-phase kinase assay were roughly the same
for each assay and that the activation of tagged JNK1 and tagged ERK2
could also be detected as a mobility delay by SDS-PAGE. Taken together,
these data show that MST, like MUK, preferentially activates the
JNK/SAPK pathway with a somewhat loose downstream specificity.
MST Activates SEK1 in COS-1 Cells
The activity of JNK/SAPK
depends greatly on the phosphorylation of specific threonine and
tyrosine residues catalyzed by upstream MAPKK class protein kinases
such as SEK1 (7). Therefore, the ability of MUK and MST to activate
SEK1 was tested in cotransfection experiments using COS-1 cells.
GST-SEK1 was coexpressed with tagged MUK or tagged MST in COS-1 cells,
and the activity of GST-SEK1 was measured in vitro after its
recovery from the cells with a GST affinity resin. Two recombinant
proteins, GST-JNK1 (a substrate for SEK1) and c-Jun (a substrate for
JNK1), were produced in bacteria and used for the in vitro
assay of GST-SEK1 activity; activity was finally detected as the
incorporation of radiolabeled phosphate into c-Jun (Fig.
2A). The overexpression of tagged MUK or
tagged MST in COS-1 cells caused a 10~30-fold activation of
coexpressed GST-SEK1. On the other hand, no such activation of GST-SEK1
was observed when the tagged MST or tagged MUK mutant lacking a part of
the kinase domain (tag-MST/KN and tag-MUK/KN) was
overexpressed, although the expression of these mutants was confirmed
by Western blot analysis using anti-tag antibody (Fig. 2B,
upper panel). The incorporation of radiolabeled phosphate
into c-Jun depended greatly on the presence of GST-JNK1, and very
little activity was recovered from cells not transfected with the
expression vector for GST-SEK1. Furthermore, roughly the same amount of
GST-SEK1 was recovered from different sets of cells coexpressing
different proteins (Fig. 2B, lower panel). Taken
together, these results demonstrate the ability of MUK and MST to
induce SEK1 activation in COS-1 cells.
Recombinant MST Directly Phosphorylates and Activates SEK1 in Vitro
The kinase domains of MLKs show considerable amino acid
sequence homology to those of MAPKKK class protein kinases such as c-Raf and TAK. The MST kinase domain shares 35 and 40% amino acid identity with c-Raf and TAK, respectively. Furthermore, the activation of SEK1 in vitro by overexpressed SPRK immunoprecipitated
from COS cells has been reported (25). These observations suggest that
members of the MLK family are direct activators of SEK1, but do not
rule out the possibility that they act farther upstream as activators
of certain MAPKKKs. We then produced recombinant MST in bacteria as a
fusion protein with maltose-binding protein (MalE) to test its ability
to activate SEK1 in vitro. The MalE-MST-(10-757) fusion
protein affinity-purified on amylose resin appeared as several bands
ranging between 80 and ~140 kDa on SDS-PAGE (Fig. 3A). The uppermost band corresponds to the
expected molecular size of MalE-MST-(10-757), whereas the other bands
may represent partially degraded protein since these as well as the
140-kDa band react with anti-MalE antibody (data not shown). The
ability of MalE-MST-(10-757) to phosphorylate recombinant GST-SEK1 or GST-MEK1 was first tested. To limit the autophosphorylation of GST-SEK1
or GST-MEK1, an invariant lysine residue located in kinase subdomain II
of these kinases was mutated to arginine. MalE-MST-(10-757) could
phosphorylate GST-SEK1(K/R) as well as MalE-MST-(10-757) itself (Fig.
3B). However, the phosphorylation of GST-MEK1(K/R) was
hardly detected. In addition, neither MalE nor GST was
phosphorylated. Therefore, SEK1 is a rather specific substrate
for MST in vitro.
It has been reported that the activity of SEK1 is greatly dependent on the phosphorylation of serine and threonine residues at positions 219 and 223 of SEK1 (11). To test whether the threonine residue is phosphorylated by MalE-MST-(10-757), GST-SEK1(K/R) phosphorylated by MalE-MST-(10-757) was analyzed by Western blotting using a Thr(P)-223-specific SEK1 antibody (New England Biolabs Inc.). As shown in Fig. 3C, the reactivity of the antibody toward GST-SEK1(K/R) increased significantly after phosphorylation by MalE-MST-(10-757). Therefore, MalE-MST-(10-757) induces the phosphorylation of the threonine residue responsible for the activation of GST-SEK1.
We next tested whether MalE-MST-(10-757) activates GST-SEK1 in
vitro. The activity of GST-SEK1 was estimated by its ability to
activate GST-JNK1, whose activity in turn was estimated by its ability
to phosphorylate c-Jun. When GST-JNK1 was preincubated with GST-SEK1
and MalE-MST-(10-757) in the presence of ATP, it phosphorylated c-Jun
quite efficiently, whereas only limited phosphorylation was observed
when MalE was used instead of MalE-MST-(10-757) (Fig. 4A, first three lanes). Notably,
the enhancement of c-Jun phosphorylation induced by MalE-MST-(10-757)
was totally dependent on the presence of GST-SEK1, and neither the
direct activation of GST-JNK1 nor the direct phosphorylation of c-Jun
by MalE-MST-(10-757) was observed (Fig. 4A, last
three lanes). The basal incorporation of phosphate into c-Jun
observed in the absence of GST-SEK1 depends on the intrinsic activity
of GST-JNK1 since very little phosphorylation of c-Jun was observed in
the absence of GST-JNK1 (data not shown). We also tested whether
MalE-MST-(10-757) activates GST-MEK1 in vitro. The activity
of GST-MEK1 was estimated by its ability to activate GST-ERK1, whose
activity in turn was estimated by its ability to phosphorylate MBP.
When GST-ERK1 was preincubated with GST-MEK1 and MalE-MST-(10-757) in
the presence of ATP, the phosphorylation of MBP was slightly enhanced
(Fig. 4B, first three lanes). However, a slight
enhancement of MBP phosphorylation was also observed in the absence of
GST-MEK1 (Fig. 4B, last three lanes), and
MalE-MST-(10-757) itself can weakly phosphorylate MBP (data not
shown). Therefore, the activation of GST-MEK1 by MalE-MST-(10-757)
must be little if anything, as suggested by the very weak activity of
MalE-MST-(10-757) in phosphorylating GST-MEK1(K/R) (Fig.
3B).
To test whether the activation of GST-SEK1 depends on phosphorylation by MST, GST-SEK1 was preincubated with MalE-MST-(10-757) in the presence or absence of ATP and assayed for JNK1 activation activity. As shown in Fig. 4C, incubation with MalE-MST-(10-757) in the absence of ATP did not cause the activation of GST-SEK1, whereas a significant activation of GST-SEK1 was caused by incubation in the presence of ATP. Therefore, MalE-MST-(10-757) acts like a MAPKKK and directly phosphorylates and activates GST-SEK1 in vitro.
We show here that a member of the MLK family, MST/MLK2, activates MAPK pathways. We have already reported that other members of the MLK family, MUK/DLK/ZPK and SPRK/PTK1/MLK3, also activate MAPK pathways (16, 25). All these kinases activate the JNK/SAPK and p38/Mpk2 pathways when overexpressed in COS cells, while the activation of p38/Mpk2 is rather modest. However, only MST/MLK2 slightly activates the ERK pathway. We also show that recombinant MST/MLK2 produced in bacteria directly phosphorylates and activates SEK1 in vitro. Therefore, MST/MLK2 is a novel MAPKKK, and together with MST/MLK2, all members of the MLK family may be classified into the MAPKKK class of protein kinases that act preferentially on the JNK/SAPK pathway.
All members of the MLK family share a well conserved kinase domain, especially MLK1, SPRK/PTK1/MLK3, and MST/MLK2, showing 75-77% amino acid identity. An additional member of the MLK family, MUK/DLK/ZPK, has a somewhat divergent kinase domain, sharing 46~48% amino acid identity with the others. Among protein kinases outside the MLK family, all show the highest amino acid identity (34~40%) to the kinase domains of TAK and c-Raf, which are known to act like MAPKKKs (26, 27). This agrees with our conclusion that MLKs act like MAPKKKs. However, the kinase domains of MLK family members share rather limited numbers of amino acids (29~34%) with MEKKs, MEKK1, MEKK2, and MEKK3, which also act like MAPKKKs (12, 28). In addition, the amino acid sequence identity with MEKKs is comparable to that with protein kinases situated farther upstream such as PAK1 (29) and GCK (30) (28-33%). The amino acid sequence identity is also low between the kinase domains of MEKKs and c-Raf or TAK, ranging from 26 to ~31%. Therefore, MEKKs are rather distantly related to other MAPKKKs, including Raf, TAK, and MLKs. Instead, a relatively high number of amino acid residues (31~34%) are shared between the kinase domains of MEKKs and Tpl-2/Cot, a recently identified MAPKKK that activates the ERK and JNK/SAPK pathways (13), suggesting the presence of at least two groups of MAPKKKs carrying structurally distinguishable kinase domains. These two groups are also characterized by their sequence similarity to tyrosine kinases such as c-Abl and c-Met; the kinase domains of Raf, TAK, and MLKs share 29~37% amino acid sequence identity with these tyrosine kinases, whereas the kinase domains of MEKKs and Tpl-2/Cot share rather limited amino acid sequence identity, ranging from 24 to ~27%. In contrast to the considerably diverged features of MAPKKK class protein kinases, MAPKK class protein kinases carry highly conserved kinase domains, showing at least 40% amino acid sequence identity among them (9). Therefore, MAPKKKs may have evolved from two or more different ancestral protein kinases, whereas all MAPKKs may have originated from a single ancestor.
The selective activation of target MAPK pathways by MAPKKK depends on the substrate specificity of each MAPKKK. The substrate specificity might be determined not only by the direct interaction of a MAPKKK and its substrate(s), MAPKK(s), but also by an additional factor, such as STE5 protein of budding yeast, that acts as an adapter molecule for a certain set of protein kinases (31). The rather weak activity of MST/MLK2 in phosphorylating SEK1/MKK4/JNKK in vitro could be explained by the lack of such an additional factor as well as by the structural abnormality of MST/MLK2 produced in bacteria. In any case, the ability of MST/MLK2 to activate SEK1/MKK4/JNKK indicates that the activation of JNK/SAPK by MST/MLK2 in vivo is mediated by SEK1/MKK4/JNKK. In addition, some other MAPKKs that have not been cloned yet (14, 15) could also be involved in the activation of JNK/SAPK by MST/MLK2 in vivo.
The specific response of each MAPK pathway to several different extracellular stimuli may depend on which MAPKKK(s) is activated by the stimulation. Furthermore, it may also depend on which MAPKKKs and MAPKKs are expressed in the cell. Both MST/MLK2 and SEK1/MKK4/JNKK are highly expressed in skeletal muscle and brain among the human tissues tested, implying that they play an important role in the activation of JNK/SAPK in these tissues (20, 21). In addition, several tumor cell lines show high levels of MST expression, implying some function of MST/MLK2 in cellular transformation (20). Notably, MST/MLK2 can partly activate the ERK pathway, which plays an important role in cellular transformation (32-34). However, the question of whether the constitutive activation of MST causes cellular transformation remains to be answered.
The mechanisms by which MST/MLK2 or other members of the MLK family are activated remain unclear. Overexpressed MST is active by itself, and stimulation of cells by serum, UV light, or osmotic shock does not produce any synergistic effect with overexpressed MST.2 Furthermore, the overexpression of other members of the MLK family, MUK/DLK/ZPK and SPRK/PTK1/MLK3, as well as other MAPKKKs such as MEKKs and Tpl-2/Cot also causes a potent activation of JNK/SAPK without any extracellular stimuli (12, 13, 16, 25). Therefore, the activity of these protein kinases may be regulated by a mechanism that does not function well with overexpressed enzymes. The control of mRNA or protein synthesis is a candidate for such a mechanism. In addition, the activity could be regulated by a limited amount of enzymatic or nonenzymatic inhibitor.
MST/MLK2, as well as other members of the MLK family, contains two leucine zipper-like motifs proximal to the C-terminal end of the kinase domain and a Rac/Cdc42-binding motif (35) proximal to the C-terminal end of the leucine zipper-like motifs. In addition, MST/MLK2 and SPRK/PTK1/MLK3 have SH3 domains proximal to the N-terminal end of the kinase domain. These motifs may contribute to the interaction, if any, with other proteins. Recently, it was reported that an upstream protein kinase, HPK1, interacts with SPRK/PTK1/MLK3 through the SH3 domain (36). However, the functional significance of these motifs remains to be clarified.
We thank Drs. Toshio Maekawa and Shunsuke Ishii for the gift of the GST-CREBP-1 expression vector. We also thank Kumi Noda for excellent technical support.