Insulin Signal Transduction Pathways and Insulin-induced Gene Expression*

Insulin regulates metabolic activity, gene transcription, and cell growth by modulating the activity of several intracellular signaling pathways. Insulin activation of one mitogen-activated protein kinase cascade, the MEK/ERK kinase cascade, is well described. However, the effect of insulin on the parallel p38 pathway is less well understood. The present work examines the effect of inhibiting the p38 signaling pathway by use of specific inhibitors, either alone or in combination with insulin, on the activation of ERK1/2 and on the regulation of gene transcription in rat hepatoma cells. Activation of ERK1/2 was induced by insulin and was dependent on the activation of MEK1, the kinase upstream of ERK in this pathway. Treatment of cells with p38 inhibitors also induced ERK1/2 activation/phosphorylation. The addition of p38 inhibitors followed by insulin addition resulted in a greater than additive activation of ERK1/2. The two genes studied, c-Fos and Pip92, are immediate-early genes that are dependent on the ERK1/2 pathway for insulin-regulated induction because the insulin effect was inhibited by pretreatment with a MEK1 inhibitor. The addition of p38 inhibitors induced transcription of both genes in a dose-dependent manner, and insulin stimulation of both genes was enhanced by prior treatment with p38 inhibitors. The ability of the p38 inhibitors to induce ERK1/2 and gene transcription, both alone and in combination with insulin, was abolished by prior inhibition of MEK1. These data suggest possible cross-talk between the p38 and ERK1/2 signaling pathways and a potential role of p38 in insulin signaling.

Insulin regulates metabolic activity, gene transcription, and cell growth by modulating the activity of several intracellular signaling pathways. Insulin activation of one mitogen-activated protein kinase cascade, the MEK/ERK kinase cascade, is well described. However, the effect of insulin on the parallel p38 pathway is less well understood. The present work examines the effect of inhibiting the p38 signaling pathway by use of specific inhibitors, either alone or in combination with insulin, on the activation of ERK1/2 and on the regulation of gene transcription in rat hepatoma cells. Activation of ERK1/2 was induced by insulin and was dependent on the activation of MEK1, the kinase upstream of ERK in this pathway. Treatment of cells with p38 inhibitors also induced ERK1/2 activation/phosphorylation. The addition of p38 inhibitors followed by insulin addition resulted in a greater than additive activation of ERK1/2. The two genes studied, c-Fos and Pip92, are immediateearly genes that are dependent on the ERK1/2 pathway for insulin-regulated induction because the insulin effect was inhibited by pretreatment with a MEK1 inhibitor. The addition of p38 inhibitors induced transcription of both genes in a dose-dependent manner, and insulin stimulation of both genes was enhanced by prior treatment with p38 inhibitors. The ability of the p38 inhibitors to induce ERK1/2 and gene transcription, both alone and in combination with insulin, was abolished by prior inhibition of MEK1. These data suggest possible cross-talk between the p38 and ERK1/2 signaling pathways and a potential role of p38 in insulin signaling.
Insulin is a primary regulator of energy metabolism, is essential for normal growth and development, and can directly or indirectly stimulate DNA synthesis (1). Upon binding of insulin, the insulin receptor becomes activated, resulting in its phosphorylation and the phosphorylation of insulin receptor substrate (IRS) proteins. The growth factor binding protein 2 (Grb2)-son of sevenless complex (SOS) can then be recruited directly along with Shc or via insulin receptor substrate 1 or 2 (2). Usually, this leads to activation of Ras, Raf, and the MEK 1 kinase cascade, resulting in phosphorylation and activation of ERK1/2 (3,4). A parallel MAPK signaling pathway is the mixed lineage kinase/apoptosis signaling kinase, MAPK kinase 3/6, p38 kinase cascade. The effects of insulin on the p38 pathway are inconsistent. In fetal neurons, insulin can inhibit the p38 pathway as well as apoptosis (5). Conversely, in cultured myoblasts, insulin can induce p38 activity and is required for cytoskeletal rearrangement and myoblast differentiation (6,7). Recent studies have indicated that insulin stimulates p38 activity in L6 myotubes and that this activation is important for insulin-induced glucose transport (8 -11).
Following fasting, the liver is the major source of plasma glucose, and hepatic glucose output is tightly regulated by insulin (12). Recently, the significance of hepatic insulin signaling was elegantly demonstrated in liver-specific insulin receptor knockout mice. These animals demonstrated severe hyperglycemia in the presence of dramatic hyperinsulinemia (13). The minimal deviation hepatoma cell line H35 (and its derivatives Fao as well as the H4IIE cells used in the present studies) has long been used as a model for the effect of insulin on intracellular signaling and hepatic gene expression (14,15). These cells express receptors that bind insulin with a similar affinity to isolated rat hepatocytes and adipocytes, and insulin induces glycogen synthesis at physiologic concentrations (16). Further, as in the liver, in these cell lines insulin regulates the transcription, mRNA levels, and activity of gluconeogenic enzymes such as phosphoenolpyruvate carboxykinase, glucose-6phosphatase, and ornithine decarboxylase (17)(18)(19)(20)(21)(22)(23)(24).
Aside from its metabolic role, another important action of insulin is promoting growth. The induction of gene expression, including several immediate-early genes, may be important in cell cycle progression (25,26). The proto-oncogene c-Fos is an early response gene and a member of a family of genes that, in combination with c-Jun and its family members, binds to AP-1 sites. In addition to insulin, c-Fos is induced by a variety of stimuli including a number of mitogens (26 -28).
Pip92 is a proline-rich cytoplasmic protein that shares little homology to any known proteins but is another regulated immediate-early gene. Its mRNA is induced by serum in fibroblasts and pheochromocytoma cells (29,30) and is increased in activated T-cells, and differentiated HL-60 cells (31,32). We have previously shown that insulin regulates Pip92 mRNA abundance in rat hepatoma cells (33). Studies examining the mechanism of regulation of the Pip92 and c-Fos genes found that their induction in response to stimuli is dependent upon a serum-responsive element, in addition to other elements, within each gene (34,35).
Little is known about interactions between the p38 and ERK1/2 MAPK signaling pathways. A few recent reports have suggested that some cross-talk may occur but with variable results. The data published to date indicate that any interplay between these two MAPK pathways may be specific to the cell type and the particular stimulus. Even less is known about the effects of the p38 pathway on insulin activation of ERK1/2. The present work describes cross-talk between the insulin-activated MEK/ERK pathway and the p38 pathway and the effects of the cross-talk on expression of two insulin-regulated immediate-early genes, c-Fos and Pip92. We identify the MEK/ERK pathway as a necessary component of insulin induction of these genes and demonstrate the ability of p38 inhibitors to modulate MEK/ERK signaling and insulin-and MEK/ERK-dependent gene expression.
Gene Expression-The transcription rates were assayed by the nuclear run-on method (36). Briefly, transcriptionally active nuclei were labeled with [ 32 P]UTP (ICN, Costa Mesa, CA), RNA-isolated, hybridized with cDNA spotted on nitrocellulose, and autoradiographed. Densitometric data from autoradiographs were analyzed using ZeroD Scan from Scanalytics. The rat Pip92 cDNA was cloned from a library of insulin-induced genes as previously described (33), and the complete murine c-Fos cDNA in pBR322 was obtained from the ATCC (pc-fos3).
Total and Active ERK1⁄2 Measurements-Serum-deprived cultures were washed with 37°C phosphate-buffered saline. Whole cell (SDS) lysates were prepared as previously reported (37) and were assayed for protein content using the Bio-Rad DC assay. The lysates (50 g) were resolved by discontinuous polyacrylamide gel electrophoresis with a 10% resolving gel (38) using the Bio-Rad MiniProtean II system and transferred to S & S Protran BA85 membranes using the Mini Transblot apparatus (39,40). Total proteins or phosphoproteins were visualized by ECL Plus (Amersham Biosciences) reagents and autoradiography or digital image recording (41). Using the AlphaInnotech Fluorchem FC imager, sequential images of the chemiluminescent membranes were recorded. The range of emission intensities between faint bands and bright bands detected provides a linear data relationship with antigen concentration (42)(43)(44)(45)(46). Only bands in the linear optical density (O.D.) range were analyzed, and those beyond this range in each exposure were identified and rejected by the onboard software. Comparison of experimental bands to control bands of intermediate intensity allowed a broad range of signal intensities to be analyzed. Densitometric data from three to six experiments were averaged.
In Vitro Kinase Assays-Nonradioactive kinase assay kits were obtained from Cell Signaling Technology and used according to the supplied protocols. Briefly, p38 and ERK kinases were purified by immunoprecipitation with specific immobilized antiserum from 0.5-1.0 mg of soluble cell lysate by standard methods. The immune complex was washed two times in lysis buffer (1% Triton X-100, 10% glycerol, 20 mM Tris, pH 7.5, 150 mM NaCl, 7.5 g/ml aprotonin, 5 mM benzamidine, 5 mM phenylmethylsulfonyl fluoride, 50 mM NaF, and 0.125 mM NaVaO 4 ) and then twice in kinase buffer (25 mM Tris, pH 7.5, 10 mM MgCl 2 , 5 mM ␤-glycerophosphate, 0.1 mM Na 3 VaO 4 , and 2 mM dithiothreitol). The immune complexes were then incubated at 30°C for 30 min with 200 M ATP and 2 g of the purified recombinant substrate, GST-ATF-2 or GST-Elk, respectively (47)(48)(49). The kinase reaction was terminated by the addition of 1% SDS loading buffer and boiling for 5 min. Phosphorylated substrate was detected by Western blotting with antisera specific for the phosphorylated form of ATF-2 or Elk1.
Statistical Analysis-Analysis of variance and Student's t tests were performed using the Instat program (Graphpad Software Inc., San Diego, CA). The values from the experimental treatments were expressed as fold change compared with the vehicle-treated controls within each experiment. The relative strength of correlations was calculated using the Spearman rank correlation coefficient (r s ).

Insulin, MEK⁄ERK, and p38
Signaling Cross-talk-To directly characterize the intracellular signaling pathways activated by insulin in rat hepatoma cells, Western blots were performed using antisera to total and phosphorylated ERK1/2 (P-ERK1/2). As expected, treatment of H4IIE cells with insulin caused a large and rapid increase in phosphorylation of both ERK1 and ERK2 ( Studies were performed to investigate the cross-talk between the MEK/ERK and p38 signaling pathways. To determine whether regulation of p38 was involved in acute insulin activation of ERK1/2, serum deprived cells were pretreated with the p38 inhibitor SB202190 (SB) at concentrations of 2 or 10 M over a period of 5-270 min and compared with control (vehicletreated) cells. Western blots revealed that, by itself, 2 M SB, and to a greater extent 10 M SB, induced ERK1/2 activation in a time-dependent manner (P-ERK1/2; Fig. 1, B and C). The levels of total ERK1/2 immunoreactive protein were not significantly or consistently altered by any of the treatments (Fig. 1, A-C, lower panels), indicating that changes in P-ERK1/2 bands reflected changes in the fraction of activated kinase. The results of three or more experiments with each concentration and at each time of SB treatment were compiled and expressed as the average fold increases of band intensity compared with untreated control cell lysate within each experiment (Fig. 1D). Treatment with 2 M SB resulted in a maximum of 15.6-fold induction of P-ERK1/2 over control by 30 min compared with 33-fold above control by 60 min in cells treated with 10 M SB. For both concentrations of SB, P-ERK1/2 remained elevated above control levels even at 150 and 270 min, the longest time points tested.
Studies were performed to ensure that our measurements of P-ERK1/2 by Western blot were linear within the range of proteins used and the amount of P-ERK1/2 generated. Using a highly sensitive digital video camera, chemiluminescent Western blots were imaged directly. Over a 1-30-fold range, P-ERK1/2 band intensity in Western blots correlated with whole cell lysate concentration and, thus, P-ERK1/2 concentration in a linear fashion (Fig. 1E).
a representative autoradiograph of a Western blot of whole cell lysates with the treatments as indicated above the lanes. Ins, insulin. B and C, serum-deprived H4IIE cells were treated for the indicated times with 2 or 10 M SB. The upper panels were probed with anti-P-ERK1/2, and the lower panels were probed with the total ERK1/2 antiserum. D, averages of three to five separate experiments at each time point with error bars (S.E.) indicated. The band densitometric intensity for the controls was arbitrarily set to 1 in each experiment, and other treatment values were expressed as fold increases versus control. #, greater than control (p Ͻ 0.05); *, greater than 2 M (p Ͻ 0.05). E, serial dilutions of identically treated whole cell lysates were probed with the anti-P-ERK1/2 antiserum (left panel) and visualized using direct digital imaging. In the right panel, the band intensity corresponding to each lysate concentration is plotted on the y axis, and the resultant curve fit is represented as a solid line connecting each point. The extrapolated correlation is shown as a dashed line. F, in vitro ERK kinase assays were performed on cells treated as indicated. Following incubation with insulin, the immunoprecipitated ERK1/2 was incubated with GST-Elk, and P-GST-Elk was detected by Western blot with P-Elk antiserum followed by direct digital image analysis of the chemiluminescent blot.
To examine whether the p38 inhibitor alters the ability of insulin to activate ERK1/2, H4IIE cells were pretreated with either 2 or 10 M SB for 30 min. The cells were then either harvested immediately or treated with insulin for 5-240 min. The upper panels were probed with anti-P-ERK1/2, and the lower panels were probed with the total ERK1/2 antiserum. C, averages of three to five separate experiments at each time point with error bars (S.E.). In the left panel, the fold increase in ERK1/2 phosphorylation by insulin treatment alone (from Fig. 1) is compared with the time course of the combination of SB plus insulin treatments. In the right panel, ERK1/2 phosphorylation by SB alone (from Fig. 1) is compared with the time course of the combination of SB plus insulin treatments. The portion of the graph to the left of the zero time point indicates the activation of ERK1/2 that occurs during the pretreatment with SB; insulin treatments begin at time 0. #, greater than SB alone (p Ͻ 0.02); *, greater than insulin alone (p Ͻ 0.03). D, after incubation with SB alone or followed by insulin, kinase activity of immunopurified ERK was determined by Western blot with P-Elk antiserum followed by direct digital image analysis of the chemiluminescent blot. Pretreatment with SB alone caused the expected, concentrationdependent large increase in P-ERK1/2. In addition, the induction of P-ERK1/2 was greatly augmented by insulin treatment (Fig. 2, A and B). Insulin treatment alone caused a maximum increase in ERK1/2 phosphorylation of ϳ28-fold by 5 min followed by a decrease to a plateau of around 2-3-fold above control levels between 120 and 240 min (Fig. 2C, left panel).
Pretreatment with the lower concentration of SB (2 M) caused an approximate doubling of the maximum level of insulin-induced P-ERK1/2 (28 versus 59-fold), and the higher concentration of SB (10 M) caused an ϳ4-fold increase in maximum ERK1/2 phosphorylation compared with insulin alone (110-fold above serum free control). These increases in response to insulin are in addition to the elevated P-ERK1/2 at the beginning of insulin treatment that results from the 30-min incubation with SB. For comparison, in the right panel of Fig.  2C is the effect of SB alone (replicated from Fig. 1C, but on the expanded scale necessary for these experiments) or the combination of SB and insulin as presented in the left panel. Regulation of ERK activity in in vitro kinase assays parallels the induction of P-ERK1/2 in H4IIE cells treated with SB alone or prior to insulin addition (Fig. 2D).
This indicates a greater than additive effect of 2 and 10 M SB in combination with insulin to induce P-ERK1/2 as compared with the effects of the agents added separately (Fig. 2C). For example, if the effects are compared (treatment with SB 2 M plus insulin versus the addition of the effects of SB 2 M alone plus the effects of insulin alone), at 5, 15, and 120 min, there were 42, 122, and 266% increases above additivity, respectively, for induction of P-ERK1/2, suggestive of a synergistic effect of the combination of low dose SB and insulin. The effect of pretreatment with the higher dose of SB (10 M), followed by insulin addition resulted in 81, 161, and 150% increases above additivity for induction of P-ERK1/2 at 5, 15, and 120 min, respectively, above the simple addition of the effects of 10 M SB alone and insulin alone. Throughout all of these experiments, there were no significant changes in the amount of total ERK1/2 protein, indicating that changes in the phosphorylation of ERK1/2 reflected changes in the fraction of activated kinase (Fig. 2, A and B, lower panels).
To ensure that the effects of p38 inhibition by SB are not a unique feature of this compound, a separate but chemically similar inhibitor was also tested. The distinct inhibitor, SB203580, was also found to induce ERK1/2 phosphorylation when added alone (Fig. 3A, lane 1). SB203580 also acted synergistically with insulin to increase ERK1/2 activation (Fig. 3A,  lanes 2-4). As a further control, a structurally related but inactive compound, SB202474, which does not inhibit p38 kinase activity, was also tested on H4IIE cells and was unable to increase ERK1/2 phosphorylation on its own or in combination with insulin treatment (Fig. 3B).
Insulin, MEK⁄ERK, and p38 Regulation of Gene Expression-To examine the effects of activation of the ERK1/2 pathway and the possible cross-talk of the ERK and p38 pathways in insulin regulation of gene expression, nuclear run-on assays were performed. H4IIE cells were either not treated or treated with 10 nM insulin, and the transcription rate of c-Fos and Pip92 was measured. In addition, other cells were pretreated either with vehicle (control) or with the MEK1 inhibitor (PD) to inhibit signaling through the MEK1/ERK1/2 pathway. As we have previously reported (27,33), following 30 min of insulin treatment, there were significant increases in transcription of both genes (Fig. 4). There was no significant effect on the transcription of either gene when cells were treated with PD alone (50 M). In cells pretreated with PD, the effects of insulin on c-Fos and Pip92 were reduced ϳ70 -80%, to levels not significantly different from basal (control) levels of transcription. These results indicate that MEK1, and most likely ERK1/2 activation, is necessary for the full effect of insulin-mediated induction of both c-Fos and Pip92 genes. A control gene, ␤-tubulin, was not significantly altered by insulin or PD alone or by the combination of the two.
We then determined whether regulation of p38 could effect insulin-mediated regulation of c-Fos and Pip92 transcription. Serum-deprived cells were pretreated with the p38 inhibitor (SB) at concentrations of 2 or 10 M and compared with vehicle control. Following a pretreatment period, the cells were stimulated acutely with 10 nM insulin for 30 min or left untreated. Low dose (2 M) SB alone increased c-Fos transcription 5.5-fold while having a slight but nonsignificant effect on Pip92 ( Fig.  5A; insulin data from Fig. 4B added to this graph for comparison). Pretreatment with 2 M SB followed by stimulation with insulin for 30 min resulted in a 26.5-fold increase in c-Fos and a 12.2-fold induction of Pip92 transcription (Fig. 5A). For the two induced genes, the effect of 2 M SB alone was less than that of insulin (48% for c-Fos and 21% for Pip92), whereas the combination of insulin and SB on transcription is greater than additive. If the effects are compared (treatment with SB 2 M plus insulin versus the addition of the effects of SB 2 M alone plus the effects of insulin alone), there was a 55% increase above additivity for c-Fos and a 61% increase above additivity for Pip92, suggestive of a synergistic effect (Fig. 5A), similar to that found in P-ERK1/2 induction. There was no significant effect of any of these treatments on transcription of the control gene, ␤-tubulin.
The higher dose of SB (10 M) by itself caused a dramatic 36-fold increase in c-Fos transcription, much larger than the 11.5-fold effect of insulin alone (the insulin data is again shown in Fig. 5B for comparison). Unlike the lower dose of SB, which was ineffective, the 10 M concentration of SB induced Pip92 transcription 5.5-fold, similar to the 6.3-fold effect of insulin alone. When H4IIE cells were pretreated with the higher dose of SB and were then exposed to insulin for 30 min, there was no further increase in c-Fos transcription compared with 10 M SB alone. However, this regimen resulted in an induction of Pip92 transcription that is 83% larger (21.5-fold) than the sum of the effects of 10 M SB alone (5.5-fold) and insulin alone (6.3-fold), again suggesting a synergistic interaction of insulin and SB. The synergistic effects of SB and insulin on gene transcription coincided well with the synergistic effects of the p38 inhibitor and insulin on the induction of P-ERK1/2.
Previous studies from our laboratory (27)   It is clear that the addition of SB can increase phosphorylation of ERK1/2 and result in the increased transcription of insulin-regulated, ERK-dependent genes. Additionally, there is a synergistic induction of P-ERK1/2 and the ERK-dependent genes, c-Fos and Pip92, when SB and insulin are combined. However, whether a common pathway was used by both insulin and p38 inhibitors to activate ERK1/2 and subsequently induce c-Fos and Pip92 expression was tested by again using the MEK1 inhibitor, PD98059. After 30 min of pretreatment with PD, the cells were treated or not for a further 30 min with SB. Greatly overexposed Western blots (Fig. 5C) indicated that the MEK1 inhibitor completely blocked SB-induced increases in ERK1/2 phosphorylation, as well as slightly reducing basal phosphorylation, suggesting that MEK1 activity is required for the activation of ERK1/2 by the p38 inhibitor.
When the effects of MEK1 inhibition were studied on the stimulation of c-Fos and Pip92 expression, there was an inhibition of the SB (10 M) induction of both genes by PD pretreatment (Fig. 5B). In addition, when cells were pretreated with PD prior to the combination of SB and insulin, the large induction of both c-Fos and Pip92 transcription were completely inhibited (Fig. 5B). This suggests that SB induces transcription of these ERK1/2-dependent genes, acting at the level of MEK1 or at a point in the insulin signaling pathway upstream of MEK1.
The similarities in the induction of ERK1/2 activation/phosphorylation and transcriptional regulation of both c-Fos and Pip92 was striking. To further examine the relationship between these two variables, the correlation was quantitated (52). Linear regression lines were obtained for induction of each gene (represented as the y variable) and the corresponding level of P-ERK induction (represented as the x variable) for each treatment described. The slopes of the lines obtained for c-Fos and Pip92 were 0.32 and 0.21, respectively (Fig. 6). The strengths of the correlation described by these regression lines (r s ϭ 0.9084 and 0.8772, respectively) and their significance (p Ͻ 0.0001) (53) further support the idea that insulin alone, p38 inhibitor alone, or the combination of treatments all regulate c-Fos and Pip92 via ERK1/2. We then measured whether insulin had a direct effect on p38 activity in H4IIE cells. Consistent with our cross-talk hypothesis, basal activity of p38 was observed in quiescent cells. A modest, but not statistically significant increase in p38 activity was sustained over a time course of insulin treatment at all times greater than 5 min (Fig. 7). Small changes in p38 activity in response to insulin are consistent with recent reports of studies in a cultured skeletal muscle cell line (10).

DISCUSSION
To study the regulation of insulin signaling, the activation (phosphorylation) state of ERK1/2 was examined in rat H4IIE hepatoma cells. Insulin treatment alone resulted in a rapid increase in ERK1/2 phosphorylation and activity, whereas pretreatment with the MEK1 inhibitor, PD98059, abolished the effect of insulin on P-ERK1/2 as well as insulin induction of the c-Fos and Pip92 genes, implying that insulin control of these genes proceeded primarily through the MEK1-ERK1/2 pathway.
Our findings agree with work in rat hippocampal neurons Following incubation with insulin (Ins), the immunoprecipitated p38 was incubated with GST-ATF-2 and P-GST-ATF-2 was detected by Western blot with P-ATF-2 antiserum followed by direct digital image analysis of the chemiluminescent blot. A, a representative blot is shown. B, the results graphed are the averages of six separate experiments at each time point with error bars (S.E.) indicated. The band densitometric intensity for the controls was arbitrarily set to 1 in each experiment, and other treatment values were expressed as fold change.
indicating that Raf-or fibroblast growth factor-mediated activation of ERK1/2 via MEK1 is required for the rapid activation of Pip92 transcription (34). In 3T3-L1 adipocytes, insulin induction of a c-Fos reporter construct was prevented by MEK1 inhibition (54). Thus, MEK/ERK activation is important for the activation of these genes and, as presented here, is of specific importance for insulin induction of c-Fos and Pip92 in rat H4IIE hepatoma cells.
When H4IIE cells were treated with SB alone, there was a dose-dependent induction of P-ERK1/2 and of c-Fos and Pip92 transcription. Further, when cells pretreated with SB were given a secondary insulin stimulus, there was a greater-thanadditive activation of ERK1/2 and of c-Fos and Pip92 transcription. It is not entirely clear why the lower concentration of SB alone is more potent with respect to transcription of c-Fos than Pip92, but presumably there are other factors involved in c-Fos and Pip92 transcription because they are not uniformly induced in these studies. One possibility is that Pip92 has a higher "threshold" of ERK1/2 activity necessary for induction than does c-Fos.
The activation of ERK1/2 observed by treatment of cells with the p38 inhibitor SB202190 was also observed when cells were treated with a different p38 inhibitor, SB203580. The chemically related compound SB202474, which does not inhibit p38, has no effect on ERK1/2 phosphorylation either alone or in combination with insulin. This implies that it is the inhibition of p38 itself and not a nonspecific action of one of the p38 inhibitors that results in induction of P-ERK1/2.
One could hypothesize that p38 activity inhibits transcription through a mechanism separate from the ERK1/2 pathway used by insulin in the regulation of these two genes and that the synergistic actions of the p38 inhibitors and insulin were due to SB having effects downstream of ERK1/2 activity. Thus, when p38 was inhibited by SB, this separate inhibitory pathway would have been neutralized, and the c-Fos and Pip92 genes would be induced to a greater degree. Clearly, this hypothesis was incorrect because the p38 inhibitors resulted in activation of ERK1/2 and acted to synergistically induce P-ERK1/2 in combination with insulin.
Another hypothesis initially considered was that basal p38 activity was responsible for a separate and distinct "off" signal for transcription. In this model, transcription would be turned on by insulin, but the off signal would be deficient in the presence of p38 inhibition, and transcription would continue unchecked. This also was not supported by the evidence, because the p38 inhibitors alone could induce P-ERK1/2 and transcription of the c-Fos and Pip92 genes, even in the absence of insulin.
Alternatively, the p38 pathway may have a tonic, inhibitory effect on MEK/ERK signaling, and inhibition of p38 by SB increased ERK1/2 pathway activity resulting in induction of ERK1/2-dependent genes. When a stimulator of the MEK/ERK pathway was added in combination with the p38 inhibitors, a greater fold induction of P-ERK1/2 and the c-Fos and Pip92 genes was obtained. This is a potential explanation of the data obtained in the present work, because the addition of an active, but not the inactive p38 inhibitor by itself resulted in activation of ERK1/2 signaling and subsequent transcription of c-Fos and Pip92. Our data also suggested that the activation occurs at or above the level of MEK1, because inhibition of MEK1 activity blocked the induction of transcription and P-ERK1/2 by insulin alone, SB alone, and the combination of SB and insulin. Conversely we found the basal level of p38 activity is modestly induced by insulin. The fact that this occurs with a slightly delayed time course, corresponding to the downward slope of the peak of insulin-induced ERK1/2 activity, fits well with the idea that p38 may play a negative regulatory role.
Data regarding the activation of other, parallel, MAPK pathways by insulin has been somewhat contradictory. However, recent studies have indicated a major role of the p38 signaling pathway in insulin-inducible glucose transport. In these studies, insulin induced p38 activity in L6 myotubes. The p38 inhibitors, SB202190 and SB203580, reduced insulin-stimulated glucose transport by 40 -60% without altering insulininduced Akt phosphorylation (8 -11). In the present studies, the same inhibitors activated ERK and worked synergistically with insulin to stimulate ERK-dependent genes. This suggests a potential complex role of the p38 signaling pathway in the multiple actions of insulin. This may also imply a role for p38 that varies in the different insulin target tissues.
Several recent reports suggest possible cross-talk between the MEK/ERK and p38 pathways, in ways supportive of our findings. For instance, in PANC-1, HL-60, L1210, and ECV304 cells, ERK1/2 activity or phosphorylation is increased by SB203580 or SB202190 (55)(56)(57)(58). In human skin fibroblasts, phorbol ester-and Raf-mediated induction of MEK1 and ERK1/2 are inhibited when p38 is activated by constitutively active MAPK kinase 3b, the kinase upstream of p38 (59), and in the human hepatoma cell line, HepG2, low density lipoprotein receptor mRNA and P-ERK1/2 are increased by SB treatment and reduced by constitutively active MAPK kinase 6b, another kinase upstream of p38 (60). These observations suggest a role for the p38 pathway in MEK-ERK-dependent processes.
However, several other reports obtained results different from those presented here; in baboon smooth muscle cells and Swiss 3T3 fibroblasts use of the p38 inhibitor activates Raf, the upstream kinase for MEK1, but somehow fails to activate MEK1 and ERK1/2 (61,62). Treatment with SB203580 did not alter epidermal growth factor-induced ERK activation in Swiss 3T3 cells or IGF-1-induced ERK activation in L6 myotubes (62), and in mIMCD3 cells, inhibition of p38 blocks activation of P-ERK1/2 by hypertonic NaCl treatment (63). Thus, there is extensive heterogeneity in the nature of the interactions between the MEK-ERK and p38 pathways. There appear to be cell type differences, differences in interactions with various stimulators of intracellular signaling, and differences in the effects of p38 activation or inhibition on the downstream effects being measured following activation of the MEK-ERK pathway.
The present studies have identified potential cross-talk between the MEK-ERK and p38 pathways and a role of this interaction in insulin signaling and insulin action. The modest and slightly delayed effect of insulin on p38 activity may indicate that it functions to limit ERK1/2 activation and subsequent gene transcription rather than actively decreasing ERK1/2 activity and transcription. Future studies are required to identify the target at or upstream of MEK1 in the MEK-ERK signaling pathway affected by addition of p38 inhibitors, and why these effects may be tissue-specific. Future experiments will also be needed to determine whether insulin regulation of other insulin-sensitive genes is modulated by changes in p38 activity and subsequent changes in MEK1/ERK signaling.