(Received for publication, October 7, 1996, and in revised form, December 31, 1996)
From the Department of Medicine, University of British Columbia, § Kinetek Pharmaceuticals, Inc., Vancouver, British Columbia V5Z 1A1, Canada, and the ¶ Institute of Molecular Biology and Biochemistry, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
The 78-kDa protein kinase Mekk1 plays an important role in the stress response pathway that involves the activation of downstream kinases Sek1 and stress-activated protein kinase/c-Jun NH2-terminal kinase. Conserved serine and threonine residues located between the kinase subdomains VII and VIII of many protein kinases are phosphorylated for maximal kinase activation. Two threonine residues within this region in Mekk1 at positions 560 and 572, but not the serine at 557, were shown to be essential for catalytic activity in this study. When these threonine residues were replaced with alanine, there was a significant loss in phosphotransferase activity toward the primary substrate, Sek1, and a large decrease in autophosphorylation activity. Site-directed mutagenesis demonstrated that these threonine residues cannot be replaced with either serine or glutamic acid for preservation of phosphotransferase activity. Further examination of the Mekk1 mutants isolated from 32P-labeled transfected COS cells showed that Thr-560 and Thr-572 were indeed phosphorylated after two-dimensional tryptic-chymotryptic phosphopeptide analysis. Additional determinants in the NH2-terminal domain of Mekk1 also play a role in the regulation of Mekk1 activity. Although Pak3 and PKC can activate Mekk1 in vivo, this interaction is indirect and independent, since there was no direct phosphorylation of Mekk1 by Pak3 or PKC or of Pak3 by PKC, respectively.
Guanine nucleotide-binding proteins (G
proteins)1 and protein kinases are
components within the intracellular signaling pathways in diverse
eukaryotes. Modules of sequentially activating protein kinases are
coupled to hormone receptors through both monomeric and heterotrimeric
classes of G proteins. Genetic analysis of yeast mutants that are
compromised in various signal transduction pathways has led to the
identification of several related protein kinase modules, some of which
transduce signals from G proteins (1). For example, budding yeast
mating pheromones bind to serpentine seven-transmembrane
domain-containing receptors that are coupled through the - and
-subunits of a trimeric G protein to the activation of a module of
protein kinases composed of Ste20, Ste11, Ste7, and Fus3 (2, 3).
Mammalian homologs of each of these kinases have been identified, such
as p21-activated kinase (Pak), Mek kinase (Mekk), MAP/Erk kinase (Mek),
and extracellular signal-regulated kinase (Erk), respectively (4, 5).
A MAP kinase module that is activated by environmental stresses such as
heat shock (6), hyperosmotic conditions (7, 8), UV radiation (9, 10),
protein synthesis inhibitors such as cycloheximide (11), and anisomysin
(12) as well as proinflammatory cytokines like tumor necrosis
factor- (6, 8), has been reported recently. The module is proposed
to be comprised of the stress-activated protein (SAP) kinases (6) (also
known as c-Jun NH2-terminal kinase), SAP/Erk kinase (Sek1) (12, 13) (also known as MKK4), and Mekk (14, 15). However, the
intervening steps following surface receptor binding of these agents
are still obscure. There are at least three isoforms of Mekk (16). The
original 78-kDa form of Mekk (Mekk1) was first identified as a Mek1
kinase (17) but is considered now to target Sek1 (14, 15, 18).
Recently, two additional Mekk clones, Mekk2 and Mekk3, which
corresponded to proteins of 69.7 and 71 kDa, respectively, were
isolated (19). Mekk2 was reported to preferentially activate the
downstream SAP kinase, whereas Mekk3 would activate p42-Erk2 (19).
Additionally, Lange-Carter and colleagues (20, 21) have identified in
rat pheochromocytoma (PC12) cells two immunologically related Mekks of
82 and 98 kDa, where the 98-kDa form was shown to be activated in
response to nerve growth factor, epidermal growth factor, phorbol
ester, and oncogenic Ras (20, 21). A larger form of Mekk1, with a
predicted molecular mass of 161 kDa, has also been cloned (22). Another isoform may correspond to the I-Mekk detected in preadipocyte 3T3-L1
cells by Haystead et al. (23). This 56-kDa kinase is distinct from Raf1 and is rapidly and transiently activated in rat
adipocytes treated with insulin. I-Mekk was also not activated in
response to phorbol ester and was immunologically distinct from the
Mekk1 described by Lange-Carter et al. (17). Another possible member of the Mekk family is the transforming growth factor-
-activated kinase, which was also stimulated in response to
bone morphogenetic protein (24). This kinase, however, has an extremely
short amino terminus (22 residues), which appeared to be regulatory for
enzyme function.
Enzymological studies of the above mammalian protein kinases should provide further information about their roles in various G protein-dependent signaling pathways. In the present study, we have characterized two threonine residues, Thr-560 and Thr-572, in Mekk1 that are essential for its catalytic function. We have further determined that the NH2-terminal domain of Mekk1 has an additional regulatory role in the enzyme function. In the search for kinases that target Mekk1, in vivo experiments using Mekk1-transfected COS cells and Mekk1 baculovirus-infected insect cells demonstrated that Mekk1 activity can be stimulated by phorbol ester treatment. However, no direct phosphorylation of Mekk1 by PKC could be demonstrated. This implies that while regulation by PKC may be physiologically significant, PKC activation of Mekk1 is achieved indirectly. Co-expression studies with Pak3 baculovirus- and Mekk1 baculovirus-infected insect cells showed an activation of Mekk1. There was, however, no direct phosphorylation of Mekk1 by Pak3.
Murine Mekk1 cDNA was kindly provided to us
by Drs. Gary Johnson and Carol Lange-Carter of the National Jewish
Center for Immunology and Respiratory Medicine in Colorado. The
plasmids for GST-Cdc42Hs and murine Pak3 were generously donated by
Drs. Subha Bagrodia and Richard Cerione (Cornell University, Ithaca, NY). The plasmids for murine Sek1 and Sek1(K129R/T387A/S390A) were
gifts from Dr. Leonard I. Zon (Howard Hughes Medical Institute, Boston,
MA). pGEX-Mek1 and pGEX-Mek2 plasmids encoding for the respective human
proteins were kindly given by Dr. Kun-Liang Guan (University of
Michigan, Ann Arbor, MI). Baculovirus-expressing PKC II isoform was
a gift from Dr. Daniel E. Koshland (University of California,
Berkeley). Purified rat brain protein kinase C (PKC), a mixture of
,
, and
isoforms, was generously provided by Dr. Michael Walsh
(University of Calgary, Alberta, Canada).
To express Mekk1
in COS cells, a 2205-base pair (bp) region of Mekk1 containing the
entire coding sequence was amplified by PCR using the primers Mekk5-1
(5-ATG GTT GGC AAG CTC TCT CG) and Mekk3-1 (5
-GTT GCA TAT CCT GTC
TCC) with VENT DNA polymerase according to the manufacturer's
specifications supplemented with 5% Me2SO. Conditions for
amplification of the 2.2-kilobase pair fragment were preincubation at
99 °C for 2 min; 30 cycles of 99 °C for 1 min, 53 °C for 2 min, and 72 °C for 2 min; and then a final extension at 72 °C for
10 min in a Perkin-Elmer 480 thermocycler. The 2.2-kilobase pair PCR
product was purified by agarose gel electrophoresis and recovered from
the agarose using Qiaex (Qiagen Inc.). The 2.2-kilobase pair Mekk1
fragment was cloned into the mammalian expression vector pEF1 under the
control of the human elongation factor-1
promoter (25). The pEF1
vector has a simian virus 40 origin that ensures a high copy number of
replication for the vector in COS cells. The junction between the Mekk1
PCR fragment and vector pEF1 was sequenced at both the 5
- and 3
-ends of the insert to confirm sequence and orientation. The resulting construct was designated pEF1-Mekk1. Mekk1 with a polyhistidine (His6) tag on the NH2 terminus or COOH terminus
was created by PCR mutagenesis using primer Mekk5-H6 (5
-ATG CAC CAT
CAC CAT CAC CAT ATG GTT GGC AAG CTC TCT CG) with Mekk3-1 (5
-GTT GCA
TAT CCT GTC TCC) or Mekk5-1 (5
-ATG GTT GGC AAG CTC TCT CG) with
Mekk-CY (5
-GA GGA TCC CTA GTG ATG ATG GTG GTG GTG CCA CGT GGT ACG GAA GAC CGG), respectively. The truncated kinase was generated using primers Mekk-M353 (5
-ATG GCG ATG TCA GCG TCT CAG) with Mekk3-1. Mutants K432AMekk1, S557AMekk1, T560AMekk1, and T572AMekk1 were also
generated by PCR mutagenesis. All mutants were confirmed by
dideoxynucleotide sequencing.
COS-1 (ATCC number CRL 1650) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum (Life Technologies Ltd.) at 37 °C with 5% CO2. COS cells were transfected with vector, wild type Mekk1, or the various Mekk1 mutants (pEF1, pEF1-Mekk1, pEF1-Mekk1(NT-His6), pEF1-Mekk1(CT-His6), pEF1-K432AMekk1, or pEF1-S557AMekk1 or the threonine mutants, pEF1-T560AMekk1 or pEF1-T572AMekk1) by using DEAE-dextran with a 20% (v/v) glycerol shock for 2 min as described previously (26).
Fifty-two hours after transfection, the cells were washed with DMEM and
subsequently incubated in DMEM plus 0.1% bovine serum albumin (w/v).
For phorbol 12-myristate 13-acetate (PMA) stimulation, the DMEM plus
0.1% bovine serum albumin was replaced with fresh medium containing 10 nM PMA 18 h later. Unstimulated cells were used as
controls. After incubation at 37 °C for the specified period of
time, the flask of cells was placed on ice and washed three times with
cold phosphate-buffered saline, and the cells were harvested by
scrapping in cold phosphate-buffered saline containing 2.5 mM EDTA. COS cells were pelleted by centrifugation and
resuspended in homogenization buffer A (10 mM MES (pH 6.0), 50 mM -glycerophosphate, 100 µM sodium
vanadate, 2 mM MgCl2, 1 mM EGTA,
0.5% Triton X-100, 5 µg/ml leupeptin, 2 µg/ml aprotinin, and 1 mM dithiothreitol) (17). Cells were lysed with a Dounce homogenizer followed by sonication on ice in two 15-s bursts at 80%
power output (Branson Sonic Power Company, model 130). A sample of
homogenate was removed for kinase assays, and the lysate was centrifuged at 45,000 rpm (TLA45 rotor) for 1 h in a Beckman TL100 ultracentrifuge. The supernatant or cytosolic fraction was collected, and the protein concentration was determined according to the method of
Bradford (27) using bovine serum albumin as standard.
For column fractionation, cytosolic protein (1-2 mg) was loaded onto a
Mono Q column (1-ml bed volume) previously equilibrated with buffer B
containing 10 mM MOPS, pH 7.2, 25 mM
-glycerol phosphate, 5 mM EGTA, 2 mM EDTA,
0.25 mM dithiothreitol, and 2 mM sodium
vanadate. The column was eluted at a flow rate of 0.8 ml/min with a
10-ml linear 0-0.8 M NaCl gradient in buffer B using a
Pharmacia FPLC system, and 0.25-ml fractions were collected.
Recombinant baculoviruses expressing Mekk1 or Pak3 were isolated using the BaculoGold kit from PharMingen Inc. (San Diego, CA). Virtually 100% of the recombinants were achieved, and one of the plaques was propagated as a recombinant virus stock. Trichoplusia ni cells were grown as monolayers and infected with Mekk1, Pak, and PKC, separately or in combination, as per the manufacturer's instructions at a multiplicity of infection of 10. At 50 h postinfection, cells were harvested as described above for the COS cells.
In Vitro Phosphorylation of Mekk1To determine whether
Mekk1 can be phosphorylated in vitro by PKC or Pak3,
Mono-Q-fractionated Mekk1 (wild type and kinase-inactive version,
K432AMekk1) was used. Mekk1 (1 µg) was incubated at 30 °C in the
presence or absence of PKC (16 ng), 4.5 mM
CaCl2, 60 µg/ml phosphatidylserine, 6 µg/ml diolein,
and 10 mM MgCl2 in buffer C (12.5 mM MOPS, pH 7.2, 12.5 mM -glycerol
phosphate, 5 mM EGTA, 7.5 mM MgCl2,
50 mM sodium fluoride, and 250 µM
dithiothreitol). The reaction was initiated by the addition of 50 µM [
-32P]ATP (~5000 cpm/pmol). After
60 min of incubation, the phosphorylation reaction was terminated by
the addition of 5 × SDS sample buffer (250 mM
Tris-HCl, pH 6.8, 10% SDS, 25% glycerol, 5%
-mercaptoethanol, and
0.02% bromphenol blue). Subsequent to SDS-PAGE (28), proteins were
transferred to nitrocellulose membrane (pore size, 0.45 µm) as
described by Towbin et al. (29), followed by
autoradiography.
Pak3 was isolated with glutathione-agarose-immobilized GST-Cdc42Hs (see
below) and was mixed with 0.3 µg of Mekk1 in the presence of 10 mM MgCl2 in buffer C. The reaction was
initiated by the addition of 50 µM
[-32P]ATP (~5000 cpm/pmol) and was incubated for 60 min at 30 °C. An aliquot of the supernatant of the reaction mixture
was removed after the reaction, subjected to SDS-PAGE, and then
transferred to nitrocellulose membrane followed by autoradiography.
Phosphorylated Mekk1 was excised from the membrane, and the
radioactivity was determined by liquid scintillation counting.
Uninfected insect cell extract was used as control.
Mekk1 phosphotransferase activity in COS
cells was generally determined using GST-Sek1(K129R/T387A/S390A) as
substrate. (Myelin basic protein and histone H1 were also found to be
in vitro substrates for Mekk1.) An aliquot of the
Mekk1-containing fractions eluted from the Mono Q column or cytosolic
cell lysate was added to glutathione-agarose-immobilized GST-Sek1(K129R/T387A/S390A) (1 µg) in the presence of 10 mM MgCl2. The reaction was initiated by the
addition of 50 µM [-32P]ATP (~3000
cpm/pmol). After 30 min of incubation at 30 °C, the phosphorylation
reaction was terminated by the addition of 1 ml of ice-cold buffer C
containing 0.5% Triton X-100 and 150 mM NaCl. The reaction
mixture was then centrifuged at maximum speed in a microcentrifuge for
1 min, and the supernatant was discarded. The pellet was washed twice
more with buffer C followed by the addition of an aliquot of 5 × SDS sample buffer. Following SDS-PAGE, proteins were transferred to
nitrocellulose membrane. GST-Sek1(K129R/T389A/S390A) bands were
visualized by Ponceau S staining and autoradiography and then excised
for liquid scintillation counting.
Pak activity was measured using MBP as substrate. A 100-µl aliquot of
cytosolic cell extract was incubated with 2 µg of glutathione agarose-immobilized GST-Cdc42Hs for 60 min at 30 °C. The
preincubation was terminated by the addition of 1 ml of ice-cold buffer
C. The reaction mixture was then centrifuged at maximum speed in a
microcentrifuge for 1 min, and the supernatant was discarded. The
pellet was washed once more with Buffer C followed by the addition of
MBP (final concentration: 3 mg/ml) and 10 mM
MgCl2. The reaction was initiated by the addition of 70 µM [-32P]ATP (~700 cpm/pmol). After 30 min of incubation at 30 °C, the phosphorylation reaction was
terminated by the addition of an aliquot of 5 × SDS sample
buffer. Following SDS-PAGE, proteins were transferred to nitrocellulose
membrane. Bands corresponding to MBP were visualized by Ponceau S
staining and autoradiography and then excised for liquid scintillation
counting. The presence of Pak3 in the assay was confirmed by Western
immunoblotting using the anti-Pak3-NT polyclonal antibody as described
above.
PKC assays were determined using PKC MBP (residues 4-14) substrate peptide, QKRPSQRSKYL (Upstate Biotechnology, Inc., Lake Placid, NY) as substrate.
32P Labeling of Mekk1COS cells were transfected with COOH terminus polyhistidine-tagged wild type Mekk1 and Mekk1 threonine mutants, T560AMekk1 and T572AMekk1, as described above and subsequently labeled with 0.1 mCi of 32Pi/ml of phosphate-free DMEM for 4 h at 37 °C in the presence of CO2. At the end of the labeling period, the cells were harvested as described above. Cell pellets were lysed in 2 ml of 6 M guanidine HCl, 0.1 M sodium phosphate, 0.01 M Tris-HCl, pH 8.0. This lysate was passed through a 24-gauge syringe needle four times prior to binding to 1 ml of ProBond (Invitrogen). The beads were washed, and polyhistidine-tagged Mekk1 was eluted under denaturing conditions as per the manufacturer's protocol. The eluate was concentrated, exchanged into 2% SDS using Centricon-30 (Amicon), and then boiled in SDS sample buffer. Following SDS-PAGE, proteins were transferred to polyvinylidene difluoride. Mekk1 bands were visualized by autoradiography and excised for tryptic-chymotryptic two-dimensional phosphopeptide mapping (30).
The 78-kDa protein kinase
Mekk1 was originally identified as a Mek1 kinase (17), but recent
expression studies in COS cells (14) and NIH-3T3 cells (15) have
instead pointed to Sek1 as its physiological target. Low levels of
Mekk1 lead to increased activation of SAP kinases without stimulation
of Erk1 or Erk2. We have found that recombinant Mekk1 expressed in COS
cells (Fig. 1B) catalyzes the phosphorylation
of a GST fusion protein of Sek1 only about 3 times better than GST-Mek1
and twice as well as GST-Mek2 in vitro (Fig. 1D).
Furthermore, when phosphate incorporation caused by
autophosphotransferase activity of the substrates was eliminated, their
phosphorylation by Mekk1 was quite similar. This was accomplished by
using kinase-inactive versions of GST-Sek1 (Sek1(K129R/T387A/S390A))
and GST-Mek1 (Mek1(K97A)) as substrates, resulting in Mekk1 activities
of 1.9 and 1.8 pmol/min/ml, respectively. Mekk1 phosphorylation of
GST-Sek1 has been demonstrated on Ser-220 and Thr-224 just upstream of
the kinase subdomain VIII region (15). The phosphorylation of Mek1 by
Mekk1 has also been shown to occur at residues Ser-218 and Ser-222 in a
homologous kinase subdomain to that of Sek1 (31-34). All of the
following studies were conducted using the kinase-inactive version of
Sek1, which also had its two potential MAP kinase phosphorylation sites
at Thr-387 and Ser-390 mutated to alanine. These MAP kinase sites can
be phosphorylated in vitro by Erk2, but the functional
effect of this is unclear (13).
The NH2-terminal Domain of Mekk1 Is Regulatory
The NH2-terminal domain of Mekk1 has been
suggested to play a regulatory role in its phosphotransferase activity
(17). Interestingly, when a polyhistidine sequence was tagged onto the
NH2 terminus of Mekk1, the enzyme was inactive (Fig.
2). A bacterially expressed GST-Mekk1 (GST tagged at the
NH2 terminus) fusion protein was also inactive (data not
shown). However, when the polyhistidine was tagged onto the C terminus
of Mekk1, it did not affect its phosphotransferase activity. A
catalytically active Mekk1 was also generated when the
NH2-terminal domain (residues 1-352) was deleted from the
protein (Fig. 2). As reported by Minden et al. (14), this
truncated form of Mekk1 was expressed at much higher levels than all
other forms of Mekk1.
Essential Threonine Residues in the COOH-terminal Catalytic Domain
We next assessed whether the COOH-terminal catalytic
kinase domain contains regulatory residues. Of particular interest was the loop 12 region between conserved kinase subdomains VII (DFG) and
VIII (APE) (35), where activating serine, threonine, and tyrosine
residues for many protein kinases have been reported (9, 12, 13, 15,
29, 36-46). In the homologous region of Mekk1, there are three
potential serine/threonine residues, namely Ser-557, Thr-560, and
Thr-572 (Fig. 3). In the present study, these residues
were mutated to alanine, and the autophosphotransferase activity was
determined. Fig. 4A showed that the two
threonine mutants had substantially reduced autophosphotransferase
activities: 36 and 12% of wild type for T560AMekk1 and T572AMekk1
mutants, respectively. In contrast, the S557AMekk1 mutant did not
exhibit any difference in autophosphotransferase activity. Similar
patterns were obtained when the enzyme activity was determined using
GST-Sek1(K129R/T387A/S390A) as substrate (Fig. 4B). Both
T560AMekk1 and T572AMekk1 mutants were reduced to 8% of wild type
Mekk1 activity, while S557AMekk1 was essentially unchanged.
To determine whether these two threonine sites are phosphorylated
in vivo, COOH-terminal polyhistidine-tagged Mekk1 mutants were isolated from 32P-labeled transfected COS cells and
analyzed by two-dimensional tryptic-chymotryptic phosphopeptide
mapping. The wild type Mekk1 generated at least 12 phosphopeptides
(Fig. 5A). When the transfected COS cells
were transfected with T560AMekk1 mutant (Fig. 5B) or with
T572AMekk1 mutant (Fig. 5C), there was a disappearance of several phosphopeptides. In each of the phosphopeptide maps of the
threonine mutants, a unique phosphopeptide (labeled a and b for T560AMekk1 and T572AMekk1, respectively) was missing
compared with that of the wild type. Therefore, it appears that both
Thr-560 and Thr-572 of Mekk1 were phosphorylated in vivo and
probably corresponded to phosphopeptide a and b,
respectively.
To investigate further the significance of the contribution of Thr-560
and Thr-572 to the enzyme activity of Mekk1, we mutated these residues
to serine and to glutamic acid residues separately. Fig.
6A shows that when the two threonine residues
were separately mutated to serine, the autophosphotransferase
activities were reduced to 86 and 37% of wild type Mekk1 for
T560SMekk1 and T572SMekk1 mutants, respectively. Although these
autokinase activities were higher than that of the alanine mutants
(T560AMekk1 and T572AMekk1), their catalytic activities toward Sek1
were still significantly low compared with that of wild type (Fig.
6B). The T560SMekk1 and T572SMekk1 mutants had only 26 and
15% catalytic activity, respectively, compared with that of wild type.
Fig. 6, A and B, also indicate that the two
threonine residues cannot be substituted with glutamic acid residues,
which can partially mimic the introduction of phosphorylated residues
(47).
Mekk1 Is Activated in Phorbol Ester-stimulated COS Cells
We
next sought to identify the activating kinase that targets the
threonine sites. Since the primary structure of the 78-kDa Mekk1
features several potential PKC phosphorylation sites (17), we proceeded
to explore the regulation of Mekk1 by PKC. Mekk1 was expressed in COS
cells and stimulated in vivo by phorbol PMA, a known
activator of PKC. Cytosolic extracts from the stimulated and
unstimulated cells were fractionated by Mono Q chromatography. Western
immunoblotting confirmed that Mekk1 was appreciably expressed in COS
cells transfected with the Mekk1 plasmid and eluted from the anion
exchange column at a NaCl concentration of 0.4-0.45 M
(data not shown). The kinase was also very active, since it phosphorylated GST-Sek1 (Fig. 7) and autophosphorylated
(data not shown). Fig. 7 shows that PMA further stimulated,
approximately 2.5-fold, Mekk1 phosphotransferase activity toward
GST-Sek1. In these PMA-stimulated cells, PKC activity was also
stimulated by approximately 2-fold (data not shown). Mekk1 was
stimulated by as low as 1 nM PMA and optimally with 10 nM (data not shown). Higher concentrations of PMA did not
produce detectable activation of Mekk1. Additionally, the maximal
stimulatory effect with PMA was within 10 min. This phorbol
ester-stimulated increase in 32P incorporation into Sek1 is
the first indication that PKC may be involved in the activation of
Mekk1. This activation of Mekk1 by phorbol ester could also be blocked
by incubation of the transfected COS cells with 10 µM of
the PKC inhibitor Roche compound 3 for 1 h prior to harvest of the
cells for determination of Mekk1 phosphotransferase activity toward
Sek1 (data not shown).
Mekk1 Is Activated when Co-expressed with PKC in Insect Cells
We next tested whether Mekk1 could be further activated by
co-expression with PKC in insect cells. Recombinant baculoviruses encoding Mekk1 and PKCII were used individually or in combination to
infect the T. ni insect cells. When expressed alone, Mekk1 was very active, but it could again be further stimulated about 2-fold
by PMA (Fig. 8, A and C). When
co-expressed with PKC, Mekk1 phosphotransferase activity toward
GST-Sek1 was increased to levels similar to that observed when
stimulated by PMA. The level of PKC protein and enzymatic activity was
reduced about 50% when the PKC was co-expressed with Mekk1 in the same
experiments, although the expression of Mekk1 was unaffected (Fig.
8B). The addition of PMA to insect cells expressing both PKC
and Mekk1 did not further stimulate Mekk1 activity. PKC could not be
further stimulated by the addition of PMA to these cells (Fig.
8B). Apparently, co-infection of PKC and Mekk1 resulted in
the maximal activation of Mekk1.
Mekk1 Cannot Be Directly Phosphorylated by PKC
The ability of
highly purified preparations of PKC (,
, and
) from rat brain
to phosphorylate recombinant Mekk1 in vitro was tested. No
cross-phosphorylation of Mekk1, isolated from transiently transfected
COS cells and Mekk1 baculovirus-infected insect cells, could be
demonstrated (data not shown). A kinase-inactive form of Mekk1
(K432AMekk1), which was expressed in COS cells, was also not a
substrate of PKC (data not shown). However, PKC could directly phosphorylate bacterially expressed GST-Mekk1 in vitro, and
three sites of phosphorylation (Ser-194, Ser-504, and Ser-557) were identified using coupled high pressure liquid
chromatography-electrospray ionization mass spectrometry. Nevertheless,
in vivo labeling of the mutated forms of Mekk1, in which
each of these serine residues was replaced by alanine, followed by
two-dimensional phosphopeptide analysis showed that these sites were
not phosphorylated in response to PMA stimulation of COS cells.
Recent studies (15, 48) have indicated that Pak may be the
upstream kinase that targets Mekk1. Additionally, two small GTP-binding
proteins, Rac1 and Cdc42Hs, have been shown to activate the c-Jun
NH2-terminal kinase/SAP kinase signaling pathway (48, 49).
In an attempt to ascertain whether Pak is the kinase responsible for
phosphorylating the two threonine residues in vivo, we
tested whether Mekk1 is activated when co-expressed with Pak3 or
Cdc42Hs in insect cells. In cells expressing Pak3 alone, there was a
3-fold higher Mekk1 phosphotransferase activity toward GST-Sek1
compared with that expressing Mekk1 alone (Fig. 9). When
Pak3 was co-expressed with Mekk1, the Mekk1 activity was increased
8.4-fold over that expressing Mekk1 alone (Fig. 9). This Mekk1 activity
was attenuated when these cells were co-infected with Cdc42Hs. However,
when Cdc42Hs was co-infected with Mekk1, the Mekk1 activity was
increased 4-fold.
Mekk1 Cannot Be Phosphorylated by Pak3 Alone
As with PKC, we also used active Pak3 to phosphorylate recombinant Mekk1 in vitro. Similar to the findings with PKC, Pak3 was unable to cross-phosphorylate recombinant Mekk1 expressed in baculovirus-infected insect cells or transiently transfected COS cells (data not shown). The recombinant Pak3 was nevertheless very active, since it could catalyze the phosphorylation of myelin basic protein.
Pak3 Is Not Targeted by PKCAlthough Pak3 and PKC did not
directly phosphorylate Mekk1, we were interested in determining whether
the activations of Mekk1 by these kinases were independent. We thus
measured Pak activity in PMA-stimulated Mekk1-transfected COS cell
lysates. Using MBP as substrate, Pak activity was not significantly
elevated in these cells as compared with those that were unstimulated
(data not shown). To determine whether PKC directly activates Pak,
insect cells were infected with baculoviruses encoding Pak3 and
PKCII either individually or in combination. When co-expressed with PKC, Pak3 phosphotransferase activity toward MBP was not substantially different from that which expressed Pak3 alone (Fig.
10). When stimulated with PMA, there was no difference
in the levels of Pak3 activity (Fig. 10).
In mammalian cells, many modules of protein kinases transduce signals from receptors that are coupled to both monomeric and heterotrimeric G proteins. The kinases in these modules may be assembled into complexes to prevent cross-talk with related kinase modules in parallel pathways, to increase the efficiency of signal transmission, and to accept multiple negative and positive inputs from other signaling pathways. Protein phosphorylation has been shown to be critical for the activation of MAP kinases, SAP kinases, Mek, and Sek (5), which exhibit strong homology with Mekk1 in their catalytic domains. It was anticipated that the conserved serine and threonine residues located just before the subdomain VIII region of the MAP kinase/Ste-related kinase superfamily need to be phosphorylated for maximal kinase activation. The results in this study indicate that two threonine residues at positions 560 and 572 in the sequence of Mekk1 are essential for catalytic activity. First, when these threonine residues were substituted with alanine, there was a significant loss in phosphotransferase activity toward the primary substrate, Sek1, and a near complete loss in autophosphorylation activity. Second, the phosphopeptide pattern after metabolic labeling showed that there was a disappearance of major spots in the T560AMekk1 and T572AMekk1 mutants compared with that of wild type Mekk1, indicating that these two threonine residues are indeed phosphorylated in vivo. Furthermore, results here also support the notion that the NH2-terminal domain of Mekk1 has a regulatory function.
As has been suggested by Rhodes et al. (50) for Ste11 and Lange-Carter et al. (17) for Mekk1, the NH2-terminal domain of these protein kinases participates in the regulation of their activities. When amino acid residues 1-352 were deleted from Mekk1 and expressed in COS cells, there was an increase in phosphotransferase activity toward Sek1 for the truncated mutant. These results corroborate the published observations for truncated Mekk1 (15, 32, 34). Additionally, it is intriguing that Mekk1 was completely inactivated when a polyhistidine tag was fused to the NH2 terminus of the kinase but exhibited no difference in activity when the tag was fused to the COOH terminus. It is tempting to suggest that the NH2 terminus of the kinase is embedded within the protein molecule but not the COOH terminus. Thus, the addition of a peptide sequence, especially a charged one, to the NH2 terminus may disrupt the proper folding of the molecule. Further investigation will be necessary to establish the precise residue(s) or peptide region in Mekk1 that is involved in regulation.
In addition to the regulatory NH2-terminal domain, there are additional determinants of enzyme activity present in the COOH-terminal domain of Mekk1. As previously mentioned, there are two essential threonine residues in Mekk1, Thr-560 and Thr-572, which are located in a region analogous to the phosphorylation lip of MAP kinases and cyclin-dependent kinases 2 and 4 (42, 47), i.e. between the conserved kinase subdomain VII and VIII region. In the present study, when Thr-560 and Thr-572 were individually mutated to alanine residues, Mekk1 lost a significant portion of its catalytic activity both in terms of autophosphorylation and phosphotransferase activity toward Sek1. The effect was much more pronounced with the Thr-572 mutation, which is located just before kinase subdomain VIII. Wu and associates (51) recently reported that in yeast Ste20p, Thr-777 could also autophosphorylate and was crucial for its catalytic activity. In Ste20p, Thr-777 lies in a position equivalent to that of Thr-572 in Mekk1. Furthermore, the catalytic subunit of cAMP-dependent protein kinase can also autophosphorylate at Thr-197, which is located in an equivalent kinase region (41). The autophosphorylation at that threonine residue has been shown to be sufficient to activate the cAMP-dependent protein kinase. In an attempt to distinguish whether autophosphorylation or cross-phosphorylation by another kinase is required for Mekk1 activity, both Thr-560 and Thr-572 were individually mutated to serine, which was a conservative amino acid substitution. This partially restored the autophosphorylation activity about 2- (Thr-560) to 6-fold (Thr-572) for the respective Mekk1 mutant. With a serine at position 572, the mutant Mekk1 did not recover much phosphotransferase activity toward Sek1. This implies that a fully autophosphorylated Thr-572 may be crucial for catalytic activity. Despite restoration of up to 86% of wild type Mekk1 autophosphorylation activity, the T560SMekk1 mutant only had about 26% of wild type phosphotransferase activity toward Sek1. Thus, the presence of a threonine residue at position 560 may be important for the specific interaction or recognition by a regulatory protein or an activating kinase. This is based on the recent suggestion that the phosphothreonine (Thr-197) region of the cAMP-dependent protein kinase may serve as a docking surface for the regulatory subunit of the holoenzyme complex (52). Further, it has been reported that the calcium/calmodulin-dependent protein kinase I was phosphorylated on Thr-177 and that its activity was enhanced 25-fold by calcium/calmodulin-dependent protein kinase I kinase (46). Ribosomal S6 kinase (Rsk1) was also activated by MAP kinase when it was phosphorylated at Thr-562 (53). Both activating threonine residues of calcium/calmodulin-dependent protein kinase I and Rsk1 are located at equivalent positions just before the catalytic subdomain VIII region.
Efforts were also made to change the two putative threonine phosphorylation sites to glutamic acid residues to mimic the introduction of negative charges due to phosphorylation. Such mutations have been reported to produce constitutively active Mek1 (54-56). Moreover, it has been proposed that a negative charge on this loop is important for the correct alignment of the residues participating in catalysis (44). In the present study, however, these glutamic acid replacements in Mekk1 did not translate into elevated Mekk1 activity toward Sek1 nor increased autophosphorylation activity. Similar observations have been reported for analogous mutants in two other Mek1 homologs, i.e. Mek5 and MKK4 (Sek1) (57). This further demonstrates that the environment of the activating Thr-560 and Thr-572 in Mekk1 is more involved than the requirement of a negatively charged residue. Precise determination of the molecular environment of these residues will require further structural studies.
In an attempt to identify the kinase(s) that targets Mekk1, we demonstrated that both Pak3 and PKC can activate Mekk1 in vivo. Mekk1 phosphotransferase activity was stimulated 2-fold in response to PMA when expressed either in simian COS cells or in T. ni insect cells, and this could be blocked by preincubation of the cells with the PKC inhibitor Roche compound 3. When co-expressed with PKC in insect cells, Mekk1 activity was elevated to levels similar to that observed in the presence of PMA. We also tested whether the Cdc42/Rac-activated kinase, Pak, which is suggested to be upstream of Mekk1 (15, 48), can activate Mekk1. When Mekk1 was co-expressed with Pak3, Mekk1 activity was stimulated 8.4-fold. However, direct phosphorylation by Pak3 or by PKC could not be demonstrated. Prior treatment of the Mekk1 with phosphatases followed by the addition of Pak3 or PKC also did not result in any increase in 32P incorporation into Mekk1.2 Hence, activation of Mekk1 in vivo does not appear to be a consequence of direct phosphorylation by Pak3 or by PKC. Furthermore, the activation of Mekk1 by Pak3 is independent of the activation by PKC. It is still possible that Mekk1 is phosphorylated and activated by another Ste20-like member of the family of p21 G protein-activated kinases (58-61). Likewise, Pak3 may target other mammalian Mekk homologs. In Saccharomyces cerevisiae, it has been demonstrated that Ste20p can directly phosphorylate but not activate Ste11p (51).
Although Mekk1 could phosphorylate Mek1, Mek2, Sek1, and MKK3, when it
was expressed at physiological levels, only SAP kinase was markedly
activated without Hog (13) or Erk (34) stimulation. Because of the
promiscuity of enzymes like Mekk1, the question of how the cell
maintains specificity in intracellular signaling begs further
investigation. As reported for the mating pheromone signaling cascade
in S. cerevisiae, a plausible alternative may be the
mammalian homolog of Ste5p, which is suggested to be the scaffolding
protein involved in coordinating the Ste20 Ste11
Ste7
Fus3
protein kinases (2, 3, 51, 62). The mammalian Ste5-like homolog may
interact with Mekk1 in a manner that orientates Mekk1 in a conformation
that allows only limited and more specific access to activating
kinases. A more recent report suggests that extracellular stress
stimuli and Mekk may result in the induction of MAP kinase phosphatase
1 gene expression that can inactivate Erk and thus convey specificity
to the stress-activated pathway (63). Since Mekk1 is demonstrated here
to be phosphorylated in vivo, one explanation for the
inability of Pak3 to phosphorylate Mekk1 may be that
transphosphorylation of the expressed Mekk1 may already have occurred
in vivo and thus renders it catalytically competent. The
autophosphorylation at the two threonine sites may be a consequence of
an already active Mekk1. This, nonetheless, does not diminish the
significance of the finding that there is a loss of function when
either Thr-560 or Thr-572 is replaced with another amino acid. These
threonine residues may be involved in substrate recognition, and their
phosphorylation may change the conformation of the substrate binding
cleft, thus allowing access only to a specific substrate. The results
of the present study, nevertheless, cannot differentiate the order of
phosphorylation for Thr-560 and Thr-572. It should be reiterated that
there are additional determinants present in the
NH2-terminal domain of Mekk1, which may be key to
understanding the regulation of this enzyme. Regions of the
NH2-terminal domain may fold over and interact with the
active site, thus restricting access to a substrate. Resolution of this
awaits the three-dimensional structure determination of Mekk1.
We are grateful to Drs. Subha Bagrodia, Richard Cerione, Kun-Liang Guan, Gary Johnson, Carol Lange-Carter, and Leonard Zon for plasmids and to Monika Ahluwalia, Gloria Leung, and Edward Lynn for technical assistance.