(Received for publication, January 24, 1997, and in revised form, April 3, 1997)
From the Departments of Pharmacology and
¶ Medicine, § University of North Carolina Lineberger
Comprehensive Cancer Center, University of North Carolina,
Chapel Hill, North Carolina 27599
In rat liver epithelial cells (GN4), angiotensin
II (Ang II) and thapsigargin stimulate a novel
calcium-dependent tyrosine kinase (CADTK) also known as
PYK2, CAK, or RAFTK. Activation of CADTK by a
thapsigargin-dependent increase in intracellular calcium
failed to stimulate the extracellular signal-regulated protein kinase
pathway but was well correlated with a 30-50-fold activation of c-Jun
N-terminal kinase (JNK). In contrast, Ang II, which increased both
protein kinase C (PKC) activity and intracellular calcium, stimulated
extracellular signal-regulated protein kinase but produced a smaller,
less sustained, JNK activation than thapsigargin. 12-O-Tetradecanoylphorbol 13-acetate (TPA), which slowly
activated CADTK, did not stimulate JNK. These findings suggest either
that CADTK is not involved in JNK activation or PKC activation inhibits the CADTK to JNK pathway. A 1-min TPA pretreatment of GN4 cells inhibited thapsigargin-dependent JNK activation by
80-90%. In contrast, TPA did not inhibit the >50-fold JNK activation
effected by anisomycin or UV. The consequence of
PKC-dependent JNK inhibition was reflected in c-Jun and
c-Fos mRNA induction following treatment with thapsigargin and Ang
II. Thapsigargin, which only minimally induced c-Fos, produced a much
greater and more prolonged c-Jun response than Ang II. Elevation of
another intracellular second messenger, cAMP, for 5-15 min also
inhibited calcium-dependent JNK activation by ~80-90%
but likewise had no effect on the stress-dependent JNK
pathway. In summary, two pathways stimulate JNK in cells expressing CADTK, a calcium-dependent pathway modifiable by PKC and
cAMP-dependent protein kinase and a stress-activated pathway
independent of CADTK, PKC, and cAMP-dependent protein
kinase; the inhibition by PKC can ultimately alter gene expression
initiated by a calcium signal.
Mitogen-activated protein kinases (MAPKs)1 are important intermediates in signaling pathways that transduce extracellular signals into intracellular responses and have been implicated in a wide array of physiological processes including cell growth, differentiation, and apoptosis. There are at least three and perhaps four subfamilies of MAPKs: (i) p44 and p42 MAPKs (MAPK1 and -2) also referred to as extracellular signal-regulated protein kinase 1 and 2 (ERK1 and -2) (1); (ii) p54 and p46 stress-activated protein kinases 1 and 2, also referred to as the c-Jun N-terminal kinase 1 and 2 (JNK1 and -2) (2); (iii) p38 MAPK (3), the closest mammalian homologue of the yeast osmosensing ERK HOG1; and (iv) a potential subfamily member, the as yet unsequenced, 88-kDa c-Fos-regulating protein kinase (4). The known subfamilies are related by three common characteristics. First, they share high sequence homology and presumably similar structure and conformation (5). Second, all appear to be activated by protein kinase cascades initiated by small guanine nucleotide-binding proteins in the Ras superfamily with subsequent mitogen-activated protein kinase kinase kinases (MEKKs) activation (6, 7). Third, the last step in activation requires a dual specificity kinase, e.g. the mitogen-activated protein kinase kinases (8) and stress-activated protein kinase kinases (2, 7), which phosphorylate both Thr and Tyr residues. Dual site phosphorylation of the sequence TEY, TPY, or TGY at the lips of the catalytic clefts of these protein kinases is the defining property of ERK, JNK, and p38 MAPK, respectively (9).
On the other hand, each subfamily also has distinct characteristics (7). First, they are activated by different stimuli, ERKs by growth factor receptor tyrosine kinases (6) or PKC (5) and JNK by stress signals (2), such as inflammatory cytokines or UV irradiation. Second, each has substrate preferences (10), e.g. Elk-1, SAP1, and phospholipase A2 for ERKs, c-Jun and ATF2 for JNK, and ATF2 and Max for p38 MAPK. Third, activation regulates distinct cellular responses; ERKs lead to cell growth and differentiation (6), while in some cells JNKs and p38 MAPK inhibit cell growth or may promote either necrotic or apoptotic cell death (2).
The present study examines the calcium-dependent pathway to
JNK activation (11) and defines two mechanisms (PKC- and
PKA-dependent) that modify its output. We had previously
shown that the calcium-dependent JNK pathway was
PKC-independent as demonstrated by the fact that Ang II stimulates JNK
more effectively in PKC-depleted cells than in control cells (11). This
phenomenon was initially thought to be secondary to abrogation of the
PKC-dependent negative feedback on the Ang II inositol
1,4,5-trisphosphate/calcium signal (12). The stimulation of JNK by Ang
II or thapsigargin appears to be initiated by a nonreceptor
calcium-dependent tyrosine kinase (CADTK) that we have
purified and cloned (13, 14). CADTK is highly related to the focal
adhesion tyrosine kinase, p125FAK, and is the same tyrosine
kinase identified recently by three other groups (PYK2, CAK, and
RAFTK) (15-17). Thapsigargin-dependent activation of JNK is
well correlated with CADTK expression and activation (14), and Tokiwa
et al. have provided direct evidence for the CADTK/PYK2 to
JNK pathway (18). However, the role of CADTK in JNK activation was
thrown into doubt when we established that tetradecanoylphorbol
13-acetate (TPA)-dependent activation of PKC stimulated
CADTK but not JNK. We therefore investigated whether PKC activation had
a second role, the inhibition of the CADTK signal to JNK. Our results
showed that prior stimulation of PKC decreased
calcium-dependent JNK activation by ~80-90%; inhibition
was observed with TPA treatment either before or shortly after the
addition of thapsigargin to cultured cells. Further studies showed that
this inhibitory TPA action was mediated by PKC and not by the
TPA-dependent secondary activation of ERK. Another second
messenger pathway, PKA, also inhibited calcium-dependent JNK activation. However, neither PKC nor PKA blunted
anisomycin-dependent JNK activation, demonstrating the
existence of at least two independent pathways to JNK (calcium and
stress) in cells expressing CADTK. The PKC-dependent
attenuation of JNK activation has biological consequences,
e.g. thapsigargin, which produces a calcium signal, is a
much stronger inducer of c-Jun mRNA than Ang II, which generates both calcium and PKC signals. In summary, cells expressing CADTK may
use PKC or PKA to balance ERK- or JNK-dependent alternation in gene expression in response to G-protein-coupled receptors.
Human recombinant EGF was purchased from Life
Technologies, Inc. Angiotensin II (Ang II) was purchased from Sigma and
prepared in 50 mM acetic acid as stock solution. TPA,
thapsigargin, and bis-(o-aminophenoxy)ethane-N,N,N,N
-tetraacetic
acid, tetra(acetoxymethyl) ester were purchased from Sigma and Biomol,
respectively, and prepared as stock solutions in dimethyl sulfoxide
(Me2SO). 8-CPT-cAMP was purchased from Biomol. Anti-Tyr(P)
monoclonal antibodies RC20H and pT66 were purchased from Transduction
Laboratories and Sigma, respectively. Anti-ERK polyclonal antibody k-23
was purchased from Santa Cruz Biotechnology, and anti-CADTK polyclonal
antibody was produced by using GST-CADTK fusion protein as
described previously (14).
Rat liver epithelial cells, GN4, were grown in Richter's improved minimal essential medium with 0.1 µM insulin supplemented with 10% fetal bovine serum as described previously (13). Cell lysate preparation was performed essentially as described previously (13). Briefly, cells treated with agonists were scraped into ice-cold cell lysis buffer (150 mM NaCl, 20 mM Tris (pH 7.5), 1% Triton X-100, 5 mM EDTA, 50 mM NaF, and 10% (v/v) glycerol) with freshly added 1 mM Na3VO4, 20 µg/ml phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, and 100 kallikrein inhibitor units of aprotinin/ml. Cell lysates were clarified by centrifugation at 14,000 × g for 10 min at 4 °C prior to use.
In Vitro JNK AssayIn vitro kinase assays were
carried out as described previously with GST-c-Jun-(1-79) linked to
Sepharose beads as the substrate (11). Briefly, 50 µg of cell lysates
were incubated with 12 µl of GST-c-Jun beads in lysis buffer. The
reaction mixture was then rotated for 2 h. After the incubation,
the reaction mixture was pelleted by centrifugation and washed three
times with lysis buffer and once with kinase buffer (20 mM
HEPES (pH 7.6), 20 mM MgCl2, 20 mM
-glycerophosphate, 20 mM p-nitrophenyl
phosphate, 0.1 mM Na3VO4, 2 mM dithiothreitol). The pellets were resuspended in 35 µl
of kinase buffer containing 0.5 µCi of [
-32P]ATP.
Samples were incubated for 10 min at 30 °C and chilled to stop the
kinase reaction. After removing excess kinase buffer, samples were
boiled with 10 µl of 3 × SDS-PAGE sample buffer and subjected
to 12% SDS-PAGE, followed by Coomassie Blue staining, autoradiography,
and PhosphorImager analysis.
Activation of ERKs was determined as described previously (14). Briefly, 15 µg of cell lysates were resolved by 15% low bisacrylamide SDS-PAGE. Proteins were then transferred to nitrocellulose membranes for analysis by immunoblotting with the anti-ERK antibody, K-23. The immunoblot was incubated with goat anti-rabbit horseradish peroxidase-conjugated antibodies and developed according to the manufacturer's procedure (Amersham Corp.).
Anti-Tyr(P), CADTK Immunoprecipitation, and ImmunoblottingIn a typical experiment, 500 µg of cell lysates were immunoprecipitated by incubation with the antibody for 2 h at 4 °C, and then 20 µl of protein A or protein A/G-agarose beads (Santa Cruz) were added for an additional 1 h. Immune complexes were collected by centrifugation, washed three times with lysis buffer, and then resuspended in SDS-PAGE sample buffer. Samples were subjected to SDS-PAGE and transferred to Immobilon (Millipore Corp.). Proteins were detected by incubating the blots with the appropriate antibody and visualized by using ECL reagents as described before (14).
Northern BlottingTotal RNA of GN4 cells, treated with Ang
II (1 µM), EGF (100 ng/ml), and thapsigargin (2 µM) for the indicated times, were prepared with Trizol®
solution according to the manufacturer's instructions (Life
Technologies, Inc.). 30 µg of total RNAs were separated on 1.2%
formaldehyde agarose gel and transferred to Zeta-Probe® GT blotting
membrane according to the manufacturer's instructions (Bio-Rad). The
membrane was probed with cDNA probes labeled with
[-32P]dCTP by the random primed DNA labeling method
(Boehringer Mannheim), washed with washing buffer according to the
manufacturer's instructions, and then subjected to autoradiography. 28 and 18 S ribosomal RNA were stained by 1% methylene blue to show even
loading.
Previous attempts to dissect the role of calcium and PKC in GN4 cell signaling revealed that both Ang II and thapsigargin increased tyrosine phosphorylation as well as JNK and AP-1 activity in GN4 cells depleted of >95% of their PKC (11, 19, 20). Moreover, Ang II-dependent JNK activation was at least 2-fold greater in PKC-depleted cells, a fact that we initially ascribed to the negative feedback of PKC on Ang II signaling (12). When we identified CADTK as the major calcium-dependent tyrosine kinase in GN4 cells (13) and correlated its stimulation and expression with JNK activation (14), a linkage between CADTK and JNK activation was apparent with one exception. TPA treatment of GN4 cells activated CADTK but not JNK. Either CADTK was not causally related to JNK activation or TPA inhibited JNK activation at a site distal to CADTK.
To elucidate an inhibitory role for PKC, we first examined in detail
the time course of calcium-dependent JNK activation by thapsigargin and Ang II. We had previously demonstrated that Ang II was
a stronger activator of JNK than thapsigargin in cells depleted of
>95% of their PKC by an overnight TPA pretreatment (11). In the
present studies, we used unpretreated GN4 cells and showed the opposite
result; thapsigargin stimulated JNK activity to a greater extent and
for a longer duration than Ang II (Fig. 1). The initial
burst of JNK activity was similar for both agonists, but by 20-30 min
the magnitude of the thapsigargin response was consistently 2-2.5-fold
greater in repeated experiments.
To directly test the hypothesis that the PKC signal inhibited the
calcium-dependent JNK pathway, we pretreated GN4 cells with 100 nM TPA for 1-10 min. Pretreatment for as little as 1 min decreased thapsigargin-dependent JNK activation by
80-90% (Fig. 2A), and a 15-s pretreatment
showed a significant inhibition (Fig. 3A). In
contrast, TPA pretreatment did not alter the stress pathway typified by
anisomycin-dependent JNK activation (Fig. 2A) or
exposure to 100 J/m2 of UV (data not shown). These results
delineated separate JNK activation pathways, one inhibited by TPA, the
other not. Interestingly, TPA added for 1 or 2.5 min after thapsigargin
also decreased calcium-dependent JNK activation by ~50%
at 30 min, mimicking the difference observed between treatment with
thapsigargin (calcium) and Ang II (calcium and PKC) at 30 min (Fig. 1).
TPA treatment 5 min after thapsigargin had little or no inhibitory
effect, suggesting that there is a limited time span in which the PKC
signal can influence the CADTK to JNK pathway (Fig. 2B). A
dose response performed using a 10-min TPA preincubation showed that as
little as 10 nM TPA blocked
thapsigargin-dependent JNK activation (Fig. 3B).
This effect of TPA was mediated by its ability to activate PKC; the TPA
inhibitory effect was abolished by depleting cells of >95% PKC using
an overnight TPA (5 µM, 18 h) pretreatment (data not
shown). Last, because the PKC signal produced by TPA activates ERK, we
tested whether it was the PKC or the ERKs activation attenuating the
CADTK to JNK pathway. EGF strongly activates ERKs in GN4 cells, but
pretreatment with EGF for 1-5 min failed to inhibit
thapsigargin-dependent JNK activation (Fig. 3C).
This suggests that inhibition of the CADTK to JNK pathway is mediated
directly by a TPA-sensitive PKC isoform.
To test whether TPA simply blocked the putative first step of Ang II or
thapsigargin action, the activation of CADTK, GN4 cells were pretreated
with or without TPA, and CADTK tyrosine phosphorylation was examined
(Fig. 4). TPA pretreatment did not inhibit, and in fact
enhanced, thapsigargin-dependent CADTK tyrosine phosphorylation. In addition, TPA also increases CADTK tyrosine kinase
activity assessed by immune complex kinase activity (14) using
poly(Glu4Tyr) as a substrate (data not shown). Therefore, PKC must interfere with elements downstream of CADTK in the
calcium-dependent JNK activation pathway.
Thapsigargin Stimulates Prolonged JNK Activation and c-Jun Expression
To test the biological consequences of the calcium/PKC
signaling cross-talk, we investigated whether Ang II, EGF, and
thapsigargin treatment differentially affected Jun and Fos family gene
expression. In GN4 cells, the calcium and PKC signals produced by Ang
II activate both JNK and ERK and EGF predominately activates ERK and
has little effect on JNK, whereas thapsigargin strongly activates JNK
(Fig. 1) and has little effect on ERK activation (11, 14). GN4 cells were stimulated with Ang II, EGF, and thapsigargin for the indicated times, and total RNA was isolated. Northern blot analysis (Fig. 5) revealed that thapsigargin strongly stimulated c-Jun
expression with expression increasing for up to 2 h; thapsigargin
had little effect on c-Fos expression. Ang II also stimulated c-Jun
expression, but the induction was weaker and markedly truncated
compared with thapsigargin. Ang II significantly stimulated c-Fos
expression, consistent with the effect on the ERK pathway. These data
support the hypothesis that the negative effect of PKC on Ang
II-dependent JNK activation has important consequences for
c-Jun expression.
As expected, EGF strongly activated c-Fos expression. Interestingly,
EGF, which had little stimulatory effect on JNK in GN4 cells (Figs.
3C and 7) (11), rapidly and potently stimulated c-Jun
expression. In contrast to thapsigargin, EGF-dependent
c-Jun expression reached its peak after about 30 min and then gradually decreased to its basal level.
cAMP Inhibits the Calcium-dependent but Not Stress-dependent JNK Pathway
In some cell types,
increased intracellular cAMP and PKA activity inhibits ERK (21-23) or
JNK activation (24-26). We investigated whether increasing cAMP would
inhibit calcium-dependent JNK activation. GN4 cells were
briefly pretreated with forskolin/IBMX or 8-CPT-cAMP. As shown in Fig.
6, these compounds have little or no effect on JNK
activation. However, pretreatment with forskolin/IBMX or cAMP for even
1 min inhibited the subsequent thapsigargin-dependent JNK
activation (Fig. 6A). Longer pretreatment (5-15 min) inhibited thapsigargin-dependent JNK activation by ~80-90%.
Treatment with forskolin/IBMX or 8-CPT-cAMP after the addition of
thapsigargin had little or no effect on JNK activation. These data
suggest that forskolin/IBMX and 8-CPT-cAMP are effective if added prior to the calcium signal and are most effective when provided 5-15 min
prior to the calcium signal. This may indicate that cAMP has a
different site of inhibition than TPA, which inhibits the CADTK to JNK
signal maximally with a 30-60-s preincubation and can inhibit even
when added 1-2 min after administering thapsigargin. However, it is
possible that PKA and PKC may share the same site of inhibition with
the PKA signal taking slightly longer to develop.
We also tested the effect of cAMP on ERK activation. Elevation of cAMP in GN4 cells had no effect on Ang II-dependent ERK activation (Fig. 6B). This is different from the action of cAMP in other cell types (21-23). Since the effect of Ang II on ERK in GN4 cells is primarily a Ras/Raf-independent process,2 the lack of cAMP inhibition may not be comparable with other Ras-dependent ERK activation systems. To study the selectivity of cAMP inhibition for the calcium-dependent JNK activation, we pretreated cells with forskolin/IBMX followed by Ang II, thapsigargin, or anisomycin treatment. Like TPA, cAMP inhibited thapsigargin and Ang II-dependent JNK activation but failed to affect JNK activation by anisomycin (Fig. 7), providing further evidence of distinct pathways to JNK stimulated by calcium and stress.
We have previously shown that a calcium signal in GN4 cells stimulates the p70S6K pathway but have not generated direct evidence for CADTK involvement (27). The calcium-dependent p70S6K pathway is blocked by wortmannin and rapamycin (27), known inhibitors of phosphatidylinositol-3 kinase and the rapamycin-sensitive kinase, respectively. Both of these enzymes have been implicated in the activation of p70S6K (28, 29). To determine whether phosphatidylinositol-3 kinase or rapamycin-sensitive kinase were involved in calcium-dependent JNK activation, GN4 cells were pretreated with rapamycin (10 mM, 10 min) and wortmannin (50 mM, 10 min), doses that totally inhibit thapsigargin-dependent p70S6K activation (27). Neither wortmannin (data not shown) nor rapamycin (Fig. 7) had any effect on calcium-dependent JNK activation. Thus, the calcium-dependent JNK pathway diverges from the calcium-dependent p70S6K pathway; rapamycin and wortmannin only inhibit the latter.
To formally rule out an upstream effect of forskolin/IBMX, rapamycin,
or wortmannin on CADTK activation, these compounds were preincubated
with GN4 cells before the addition of Ang II. As shown in Fig.
8, neither wortmannin, rapamycin, nor forskolin/IBMX altered Ang II-stimulated CADTK tyrosine phosphorylation. Thus, in GN4
cells, the phosphatidylinositol-3 kinase kinase and/or rapamycin-sensitive kinase may be downstream of a CADTK signal linked
to p70S6K activation, but these enzymes are not involved in
the CADTK to JNK pathway or the Ang II to CADTK pathway.
Recent studies using cells other than fibroblasts, e.g. cells of epithelial and hematopoietic origin, increasingly demonstrate a role for G-protein-coupled receptors and/or rises in intracellular calcium as mediators of growth control pathways (30-32) and gene expression (33, 34). Although the role of CADTK in proliferation has not yet been defined, CADTK is expressed in tissue and cell types that respond to G-protein-coupled receptors and calcium-dependent growth signals (14-17). When activated, CADTK appears to stimulate tyrosine phosphorylation of cytoskeleton proteins and activate intracellular signaling. For example, our data and the data of others suggest that CADTK tyrosine-phosphorylates paxillin tensin and p130CAS and forms complexes with paxillin and p130CAS (35, 36, 57). In GN4 cells, CADTK also appears to stimulate two protein kinase cascades leading to JNK and p70S6K activation (14, 27) and Tokiwa et al. (18) have directly demonstrated a CADTK/PYK2 to JNK activation pathway. In PC12 cells calcium-dependent CADTK/PYK2 activation regulates ERKs (15, 37), but in rat liver epithelial cells, the pure calcium signal produced by thapsigargin does not result in Raf, ERK, or p90RSK activation (14).2 Interestingly, in T lymphocytes, a calcium signal alone is not sufficient to activate JNK (38), although CADTK is expressed in the Jurkat cell line (17). We are currently working to uncover the bases for these cell type specificities, but in summary, CADTK is a potential mediator of G-protein-coupled receptor and calcium signal-regulated alterations in cell shape and gene expression.
The present study begins to define the calcium/CADTK pathway to JNK and to clearly differentiate this JNK activation pathway from the better studied stress-dependent pathway initiated by anisomycin, UV, etc. Our results show that calcium-dependent JNK activation is inhibited by TPA-sensitive PKC, while the stress pathway is not (Fig. 3). Since TPA does not inhibit thapsigargin-dependent CADTK tyrosine phosphorylation (Fig. 4), the site of PKC action is distal to CADTK. The inhibition by PKC, even when initiated 1-2 min after a calcium signal (Figs. 2 and 3), demonstrates why thapsigargin, which does not stimulate CADTK activity as well as Ang II (Fig. 4) (14), activates JNK more potently (Fig. 1). The better CADTK stimulus, Ang II, produces both positive (CADTK) and negative (PKC) signals to the JNK pathway, while thapsigargin produces only a positive signal.
The biological relevance of this PKC attenuation was uncovered by examining agonist-dependent c-Jun and c-Fos expression (Fig. 5). c-Jun has a relatively simple promoter, and most inducers activate its expression through one major cis-element, the c-Jun AP-1 site (39). This AP-1 site differs from the consensus AP-1 sequence due to a 1-base pair insertion, resulting in preferential recognition by c-Jun-ATF2 heterodimers rather than the conventional AP-1 complex (Jun-Fos) (40). Furthermore, ATF2 is preferentially activated by JNK to mediate c-Jun induction in response to some stimulators (41). c-Jun AP-1 is constitutively occupied in vivo, and the occupying c-Jun and ATF2 are phosphorylated, stimulating their ability to transactivate and increase c-Jun expression (42). The phosphorylation-induced c-Jun expression provides newly synthesized c-Jun, continuing the induction. In GN4 cells, Ang II-induced c-Jun expression lasted for about 60 min and was weaker than that of thapsigargin. In contrast, the expression induced by thapsigargin increased steadily for 2 h (Fig. 5). These data are most readily explained by the negative regulation of JNK activation by PKC in Ang II-treated cells. We have not formally ruled out cessation of Ang II c-Jun induction by ERK phosphorylation of an inhibitory site within the c-Jun carboxyl-terminal DNA binding domain (42). However, induction of c-Jun by EGF, a maximal activator of ERK, was greater and more sustained than that of Ang II but not as prolonged as that of thapsigargin. This suggests that ERKs do not inhibit c-Jun expression in GN4 cells and that the inhibitory effect of PKC on the CADTK to JNK pathway is not due to its secondary activation of ERKs (Fig. 3C). The EGF-dependent pathway probably involves a mechanism in addition to JNK, since EGF is a poor activator of JNK in these cells (Fig. 3C) (11).
The 40-50-fold thapsigargin-dependent JNK activation begins to decrease after 30 min, while c-Jun expression continues upward for at least 2 h. The explanation may lie in the continued, unopposed 20-fold JNK activation and the two phases of c-Jun induction (JNK phosphorylation of pre-existing c-Jun and ATF2 and the effect of newly synthesized transcriptional factors). It has been reported that JNK can also mediate Elk-1 phosphorylation inducing c-Fos expression (10). However, this phenomenon must be minimal in GN4 cells, because substantial thapsigargin-dependent JNK activation leads only to barely detectable c-Fos expression (Fig. 5).
The physiological consequence of constitutively increasing c-Jun
expression may vary in different cell types. In differentiated PC12
cells, the withdrawal of NGF results in apoptosis. NGF withdrawal correlated with the activation of the JNK and p38 MAPK and decreased ERK activity. Thus, in these cells persistent activation of the JNK
and/or p38 MAPK pathways promotes apoptosis particularly in these
stress activated enzymes and are not accompanied by an increase in ERK
(43). Persistent JNK activation has also been implicated in
ceramide-, Fas-, and -radiation-induced apoptosis (44-47). However, activation of JNK mediated by CD40 ligation in B cells protects against, rather than stimulates, apoptosis (48, 49). In
addition, JNK activation is not involved in induction of apoptosis by
tumor necrosis factor-
(50), and competitive inhibitory mutant JNK
does not attenuate MEKK1-stimulated cell apoptosis in Swiss 3T3 cells
(51). In GN4 cells, Ang II activates both ERKs and JNK balancing c-Fos
and c-Jun expression and leading to cell proliferation (31).
Thapsigargin, which strongly activates JNK but not ERK, disrupts the
balance between c-Fos and c-Jun expression, but it remains to be
determined whether thapsigargin promotes apoptosis and whether TPA
could block this putative thapsigargin action. There are examples of
diglyceride attenuation of ceramide-induced apoptosis in human
myeloid leukemia cells (52, 53) and sphingosine-1-phosphate suppression
of ceramide-mediated cell death (45) which could involve
PKC-dependent attenuation of a CADTK to JNK pathway.
In GN4 cells, increases in cAMP also blocks calcium-dependent JNK activation. These findings are consistent with recent reports that cAMP inhibits thrombin-induced JNK activity in vascular smooth muscle cells (25, 26) and in T lymphocytes (24), since both T cells (17) and rat aortic smooth muscle cells3 express CADTK/RAFTK. CADTK is activated by platelet-derived growth factor and Gq-coupled receptors, e.g. Ang II,4 and this activation is correlated with JNK activation in rat aortic smooth muscle cells.2 The target for cAMP-dependent inhibition may be downstream from the component inhibited by PKC, or it may simply take longer to develop.
Based on our data, the model in Fig. 9 is proposed.
Thapsigargin and Ang II produce a calcium signal and by an unknown
mechanism activate CADTK. This presumably sends a positive signal to
both the JNK (11, 14) and p70S6K (27) pathways. Ang II also
stimulates PKC, sending an inhibitory signal to the JNK pathway. The
calcium-dependent pathway to p70S6K passes
through wortmannin and rapamycin-sensitive steps, which are well
characterized in many cell types, but neither wortmannin nor rapamycin
inhibited CADTK tyrosine phosphorylation or the calcium-dependent JNK pathway. It is interesting to note
that TPA, which does stimulate JNK (54) in some cell types, does not
activate these enzymes in GN4 cells; rather, PKC (or specific isoforms)
are utilized in an inhibitory network regulating the extent and
duration of JNK activation. Last, while evidence of a calcium/CADTK to
JNK pathway is substantial, we must be somewhat circumspect about the
CADTK to p70S6K pathway until more direct evidence is
obtained.
A more complete understanding of the effect of PKC on inhibition of the
CADTK to JNK pathway (as opposed to the stress-dependent pathway) will only come by defining pathway components.
Stress-dependent JNK activation appears to be mediated by
small GTPase proteins, such as Rac and CDC42H and PAK-like protein
kinases that may in turn activate the MEKK/stress-activated protein
kinase kinase/JNK pathway (2, 7). Recently, PAK1 and -2 were shown to
be activated by chemoattractants (55). These PAKs may mediate a JNK
activation signal by heterotrimeric G-protein-coupled receptors and
their G-protein components, e.g. subunits (56). Since
thapsigargin mimics the effect of G-protein-coupled receptors in GN4
cells, it is unlikely that
are involved. Whether the CADTK to
JNK pathway uses PAK-like enzyme is unknown, but our best clue to the
mechanism at this point is that the calcium/CADTK pathway uses some
elements distinct from the stress-dependent pathway.
We thank Ruth Dy and Debra Hunter for excellent technical assistance. We also thank Drs. Channing Der and John Westwick for providing GST-c-Jun, c-Jun, and c-Fos DNA probes. We particularly thank Drs. Adrienne Cox and Al Baldwin for critical reading of this manuscript. We also thank Darla Nichols for manuscript preparation.