INTRODUCTION

Insulin action is initiated by ligand activation of the insulin receptor tyrosine kinase (IRK)¹ from which ensues phosphorylation of critical intracellular intermediates, which propagate signals that ultimately determine the biological effects of the hormone (1, 2). A major substrate of the activated IRK is IRS-1 (3, 4), which in its tyrosine-phosphorylated state associates with the Src homology 2 domain of various molecules, leading to the transduction of downstream signals (1). The p85 regulatory subunit of PI3-kinase represents an important Src homology 2 domain-containing protein, the binding of which to tyrosine-phosphorylated IRS-1 activates the catalytic function of the 110-kDa subunit of PI3-kinase (5, 6, 7). Wortmannin (8, 9) and LY294002 (10) are two potent and specific chemical inhibitors of PI3-kinase, the use of which has established an important role for PI3-kinase in transducing numerous metabolic effects of insulin, including stimulation of glucose transport (11, 12, 13, 14), antilipolysis (12), protein (15) and glycogen (16, 17) synthesis, and recently, insulin regulation of gene transcription (18, 19). These inhibitors also prevent insulin-induced activation of p70^s6k, (13, 20), demonstrating that the latter lies downstream of and requires PI3-kinase for activation. Studies with rapamycin, an inhibitor of p70^s6k activation (21, 22) that does not inhibit PI3-kinase, have suggested that p70^s6k is an effector for some but not all insulin actions mediated by PI3-kinase (23).

Another major signaling cascade is activated by the association of growth factor receptor binding protein 2-Son of Sevenless (Grb2-Sos) with tyrosine-phosphorylated IRS-1 (24, 25) and Shc (25, 26), resulting in the activation of p21^ras, and the sequential activation, by serine/threonine phosphorylation, of Raf and MAP kinase kinase (MEK). MEK is immediately upstream of and highly specific for MAP kinases (ERK-1 and ERK-2), which it activates by phosphorylation on specific threonine and tyrosine residues (27, 28). The activity of MAP kinases can be negatively regulated by selective dephosphorylation by specific tyrosine/threonine phosphatases (29, 30). MAP kinases activate a number of transcription factors (31), and indirect evidence suggests that they are involved in mediating insulin-stimulated DNA synthesis, c-fos expression, and mitogenesis (32, 33, 34). The role of MAP kinases in other aspects of insulin action is under intense investigation but is presently unclear (35).

Insulin-like growth factor binding proteins consist of a family of six proteins (IGFBP-1 to -6), which bind insulin-like growth factor (IGF) peptides with high affinity (36). They are collectively viewed as modulators of the growth-promoting and metabolic actions of IGFs operating on their cognate receptors in target tissues (37). The liver is the predominant site of production of IGF-1 and IGFBPs, which are released into the circulation. Whether IGFBPs have a function within the liver is unclear because no IGF-1 receptors are expressed in this tissue (38). Because IGF-1 can bind and activate insulin receptors (IRs), it is possible that liver-derived IGFBPs can modulate the impact of IGF-1 acting through hepatic IRs.

We hypothesized that the catabolic effects of glucagon (Gluc) and the glucocorticoid dexamethasone (Dex) in liver involve the induction of IGFBPs, thus sequestering IGF-1 away from the IR. Furthermore, the anabolic effects of insulin could involve its ability to down-regulate IGFBP production, thus increasing free local IGF-1 available for IRK activation. This is consistent with the demonstration, in rat primary hepatocytes, that IGFBP-1 mRNA levels are stimulated by cAMP and by the glucose counterregulatory hormones Gluc and Dex (the effects of which are additive) and are powerfully inhibited by insulin (>90%), an effect which is not overcome by either Gluc or Dex (39). These findings are in line with previous studies in hepatoma cells of rat and human origin describing the activation of the IGFBP-1 gene promoter by cAMP and Dex (40, 41) and the rapid transcriptional inhibition, by insulin, of the IGFBP-1 gene (41, 42, 43), which is dominant over the stimulatory effects of glucocorticoids and cAMP (44). An understanding of the hormonal regulation of other hepatic IGFBP species could provide insight into their physiologic function.

Because of its abundance in liver, we have compared, in rat primary hepatocytes, IGFBP-4 and IGFBP-1 mRNA regulation by Gluc and Dex, insulin, and peroxovanadium (bpV(phen)), an insulin mimetic agent that activates the IRK (45). We have also studied the signaling pathways triggered in response to IRK activation, which mediate the effect of insulin on IGFBP-1 mRNA. We report that insulin fails to suppress IGFBP-4 mRNA levels, and that PI3-kinase and p70^s6k, but not MAP kinases, are required for the effect of insulin on IGFBP-1 gene transcription. However, bpV(phen), which activates PI3-kinase and p70^s6k, mediates inhibition of IGFBP-1 and -4 mRNA in a wortmannin- and rapamycin-independent manner. Thus, in addition to PI3-kinase/p70^s6k, an alternate signaling pathway can mediate IGFBP-1 mRNA inhibition. We provide data consistent with a role for MAP kinases in mediating the inhibitory effect of bpV(phen) on IGFBP mRNA. Furthermore, we demonstrate that MAP kinase activation by bpV(phen), unlike activation by insulin and growth factors, can be effected in a MEK-independent manner.

EXPERIMENTAL PROCEDURES

Porcine insulin was a gift from Lilly Research Laboratories (Indianapolis, IN). Gluc, Dex, MBP, protein kinase inhibitor (P0300), and wortmannin were purchased from Sigma; rapamycin was obtained from Calbiochem; and the MEK inhibitor PD98059 (46) was kindly provided by Dr. Alan Saltiel (Parke-Davis). Peroxovanadium bpV(phen) was synthesized and purified as reported previously (45). Collagenase was from Worthington Biochemical Corporation (Halls Mills Road, NJ). Cell culture medium and antibiotics were from Life Technologies, Inc., and Vitrogen-100 was from Collagen Corporation (Toronto, Ontario, Canada). [-³²P]ATP and [-³²P]dCTP were provided by ICN Biomedicals Canada Ltd. (Mississauga, Ontario, Canada). Most other reagents and chemicals were obtained from Sigma and were of the highest grade available.

Hepatocytes, isolated from 180-200 g male Sprague Dawley rats (Charles River, St. Constant, Quebec, Canada) by perfusion in situ with collagenase, were seeded on a collagen matrix (Vitrogen-100) and bathed for 24 h in Dulbecco's modified Eagle's medium/Ham's F-12 containing 10% fetal bovine serum, 10 mM Hepes, 20 mM NaHCO₃, 500 IU/ml penicillin, and 500 µg/ml streptomycin. Cells were serum-starved for 48 h in Dulbecco's modified Eagle's medium/Ham's F-12 (SF medium), which differed from the seeding medium in that it lacked serum and contained 1.25 µg/ml Fungizone, 0.4 mM ornithine, 2.25 µg/ml L-lactic acid, 2.5 × 10⁸ M selenium, and 1 × 10⁸ M ethanolamine (47). For mRNA studies, cells were seeded on 78-cm² Petri dishes (Starstedt Canada, St. Laurent, Quebec, Canada). For protein and enzyme activity assays, cells were seeded on 9.6-cm² multiwell plates (Corning Costar Corporation, Cambridge, MA). In all experiments, cell confluency was approximately 60% at the time of study. Hormone and drug treatments, carried out in SF medium, were initiated 72 h after plating for the times and at the concentrations indicated in the figure legends.

Primary rat hepatocytes were solubilized in guanidinium isothiocyanate (0.7% -mercaptoethanol), and material from two identical Petri dishes was combined prior to RNA extraction in phenol/chloroform (48). Dot blot analyses of 10 µg of total RNA were performed on Hybond-N nylon membranes in a dot-blot manifold (Bio-Rad), according to the manufacturer's protocol. RNA was fixed to the membranes by UV cross-linking and hybridized sequentially, with intermittent stripping, with IGFBP-1, IGFBP-4, and GAPDH cDNA probes (see below).

IGFBP-1 cDNA (1-kilobase fragment; Ref. 49) was kindly provided by Dr. Liam Murphy (Departments of Internal Medicine and Physiology, University of Manitoba, Winnipeg, Manitoba, Canada). The rat IGFBP-4 cDNA fragment (477 base pairs) (50) was a generous gift from Dr. Shunichi Shimasaki (The Whittier Institute for Diabetes and Endocrinology, Scripps Memorial Hospital, La Jolla, CA). The GAPDH cDNA consisted of a 750-base pair fragment. The cDNA probes were labeled with [-³²P]dCTP to a specific activity of 10⁹ dpm/µg DNA using the ^T7QuickPrime kit (Pharmacia Biotech), and hybridization followed by membrane washing and stripping procedures were performed as described elsewhere (47). The blots were exposed to Kodak X-AR film (Eastman Kodak Co.) at 70 °C for varying lengths of time, and mRNA was quantitated using an LKB Ultrascan XL enhanced laser densitometer (Pharmacia Biotech Inc.). Ratios of the amount of IGFBP mRNA and GAPDH mRNA in each dot blot were expressed as a percentage of their ratios in appropriate controls, which were normalized to 100% (see figure legends).

After treatment with the test agents described in the figure legends, rat hepatocytes were rinsed twice with cold phosphate-buffered saline, pH 7.4, and lysed at 4 °C by adding 1 ml/well of lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 10 mM sodium pyrophosphate, 100 mM sodium fluoride, 1.5 mM MgCl₂, 1 mM EGTA, 200 µM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, 10 µg/ml aprotinin, 10% glycerol, and 1% Triton X-100). Lysates were centrifuged at 10,000 × g for 20 min, and protein concentrations in the resulting supernatants were determined by the method of Bradford using bovine serum albumin as a standard (51). Immunoprecipitations were carried out on cell lysates containing 1 mg of protein/ml.

The activity of ERK-1 and ERK-2 was analyzed by an immune complex kinase assay using myelin basic protein (MBP) as substrate (52) with slight modifications. Cell lysates were incubated with mild agitation for 90 min at 4 °C with 5 µl of ERK-1 (C-16) or ERK-2 (C-14) (Santa Cruz Biotechnology, Inc.) antisera preadsorbed to protein A-Sepharose beads (Pharmacia Biotech Inc.). The beads were washed three times with lysis buffer and twice with MAP kinase assay buffer (50 mM Hepes, pH 7.4, 5 mM magnesium acetate, 2 mM dithiothreitol, 1 mM EGTA, 0.2 mM sodium orthovanadate). The phosphorylation of MBP was assayed by resuspending the beads in a total final volume of 100 µl of MAP kinase assay buffer containing 25 µg/ml MBP, 50 µM ATP, and 1 µCi [-³²P]ATP. Reactions, initiated upon the addition of ATP, were carried out at 30 °C for 30 min and terminated by the addition of 25 µl of 5 × Laemmli sample buffer and boiling for 5 min. They were subsequently subjected to SDS-polyacrylamide gel electrophoresis on 12.5% gels, after which gels were incubated for 3 h in 5% acetic acid/17% methanol/78%H₂O, dried under vacuum, and exposed to x-ray film. Quantitative assessment of ERK activity was achieved by scintillation counting of phosphorylated MBP bands excised from the gels.

IRS-1 was immunoprecipitated from cell lysates with rat IRS-1 antibody (Upstate Biotechnology Inc.) preadsorbed to protein A-Sepharose under conditions identical to those described for ERK-1 and ERK-2 (see above). IRS-1 immunoprecipitates were washed extensively (53), resuspended in 50 µl of PI3-kinase reaction buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 0.5 mM EGTA) containing 0.2 mg/ml L--phosphatidylinositol (Avanti Polar Lipids, Inc., Alabaster, AL), and assayed for PI3-kinase activity as described previously (53), with minor modifications. The reaction was started by adding 10 µCi [-³²P]ATP and 20 mM MgCl₂. Following a 3-min incubation at 25 °C, the reactions were terminated by the addition of 150 µl of chloroform:methanol:11.6 N HCl (100:200:2). Then, 100 µl of chloroform were added, and the organic phase was separated and washed twice with methanol:11.6 N HCl (1:1). The lipids were concentrated in vacuo, spotted onto Silica Gel 60 thin layer chromatography (TLC) plates (Merck), and developed in chloroform:methanol:28% ammonium hydroxide:H₂O (43:38:5:7). The phosphorylated products were visualized by autoradiography, and PI3-phosphate was identified as the species that comigrated with nonradioactive PI4-phosphate (Avanti Polar Lipids, Inc.), which was spotted on the TLC plates and revealed by reaction with potassium iodide vapor.

Hepatocyte lysates (1 mg of protein), to which were added 4 µg of specific p70^s6k antibody that does not recognize p90^s6k (Upstate Biotechnology, Inc.), preadsorbed to protein A-Sepharose beads, were gently agitated for 90 min at 4 °C. Immune complexes were washed three times with lysis buffer and twice with p70^s6k kinase assay buffer, the composition of which differed from the MAP kinase assay buffer only in that it contained protein kinase inhibitor (final concentration, 4 µM). The beads were resuspended in 50 µl of p70^s6k assay buffer containing 200 µM of S6 peptide KKRNRTLTK (Upstate Biotechnology, Inc.), 50 µM ATP, and 1 µCi [-³²P]ATP. Reactions, initiated upon addition of ATP, were carried out at 30 °C for 20 min and were terminated by the addition of 10 µl of 88% formic acid. The reaction products (30 µl) were applied to phosphocellulose P-81 filters (Whatman), which were washed four times for 15 min with 500 ml 1% phosphoric acid, twice with distilled water, once in ethanol, and counted in scintillation fluid (54).

RESULTS

Incubation of serum-deprived rat primary hepatocytes for 6 h in the presence of the glucose counterregulatory hormones Dex/Gluc markedly increased IGFBP-1 and IGFBP-4 mRNA levels (Fig. 1). Insulin completely prevented Dex/Gluc stimulation of IGFBP-1 mRNA but failed to suppress stimulated IGFBP-4 mRNA levels. In contrast, the peroxovanadium compound bpV(phen) inhibited the induction of both IGFBP mRNA species to a comparable degree (Fig. 1). These observations suggest that insulin and bpV(phen) activate a common signaling pathway leading to the inhibition of IGFBP-1 gene transcription but imply the activation, by bpV(phen), of a distinct pathway mediating IGFBP-4 mRNA inhibition.

The involvement of PI3-kinase and p70^s6k in the mediation of the inhibitory effect of insulin on the IGFBP-1 gene and of bpV(phen) on IGFBP-1 and -4 mRNA was investigated using the specific PI3-kinase inhibitor wortmannin (8, 9) and the potent inhibitor of p70^s6k activation, rapamycin (21, 22). Wortmannin reversed the ability of insulin to inhibit IGFBP-1 mRNA levels in Dex/Gluc-stimulated cells in a dose-dependent manner; partial inhibition (approximately 50%) was observed using a single wortmannin dose of 100 nM (data not shown), and almost complete inhibition of the insulin effect was seen when 100 nM was added twice (zero time and 3 h) during the incubation or with a single 500 nM wortmannin dose (Fig. 2A). Similarly, rapamycin reversed the inhibitory effect of insulin on IGFBP-1 mRNA and restored it to levels seen in dimethyl sulfoxide controls stimulated with Dex/Gluc (Fig. 2B). Neither inhibitor effected IGFBP-1 mRNA when added alone to cell cultures (data not shown). In contrast to the potent inhibitory effects of rapamycin and wortmannin on insulin-mediated inhibition of IGFBP-1 mRNA, neither of these agents reversed the inhibitory effects of bpV(phen) on Dex/Gluc-stimulated IGFBP-1 and IGFBP-4 mRNA levels (Fig. 2, C and D). We examined whether this could be explained by the failure of wortmannin and rapamycin to inhibit PI3-kinase and p70^s6k, respectively, in bpV(phen)-treated cells. As shown in Fig. 3A, IRS-1-associated PI3-kinase activity was stimulated by insulin and slightly more so by bpV(phen), correlating with the latter's greater ability to promote p85 association with IRS-1 in this assay (Fig. 3B). However, the greater magnitude of PI3-kinase activity seen in response to bpV(phen) cannot account for its persistent inhibitory effect on IGFBP mRNA in the presence of wortmannin, because PI3-kinase activity was attenuated to the same degree by wortmannin when stimulated with either insulin or bpV(phen) (Fig. 3A). Fig. 4 demonstrates that p70^s6k activity stimulated by insulin or bpV(phen) was inhibited to below basal levels by rapamycin. Rapamycin inhibits p70^s6k activation by insulin, although its detailed mechanism of action is not fully established (23). An inhibitory effect of wortmannin on insulin-mediated p70^s6k activation, suggesting that p70^s6k is a downstream component involved in PI3-kinase signaling, has been reported (13, 20). We also observed that wortmannin reversed insulin (and bpV(phen))-mediated increases in p70^s6k activity to basal levels or lower (Fig. 4). Taken together, these findings indicate that activation of PI3-kinase and p70^s6k is necessary for the mediation of the transcriptional inhibition of the IGFBP-1 gene by insulin but not by bpV(phen) and that an additional signal transduction pathway(s), independent of PI3-kinase and p70^s6k and activated by bpV(phen), can mediate IGFBP-1 as well as IGFBP-4 mRNA inhibition.

The observation of a complete reversal, by wortmannin, of the inhibitory effect of insulin on IGFBP-1 mRNA (Fig. 2A) suggested that activation of the PI3-kinase signaling pathway by insulin was an absolute requirement for the latter's inhibitory effect on IGFBP-1 gene transcription, without excluding the possible contribution of other signaling intermediates to the overall inhibitory effect of insulin. The wortmannin and rapamycin-insensitive effects of bpV(phen) on IGFBP-1 and -4 mRNA demonstrate the involvement of another signaling mechanism(s), which may or may not be activated by insulin. We hypothesized a role for the Ras signaling pathway in the mediation of the inhibitory actions of bpV(phen) on IGFBP mRNA and looked at ERK-1 and ERK-2 activities in response to treatment with insulin and bpV(phen). We reasoned that ERK-1 and-2 involvement in insulin-mediated inhibition of IGFBP-1 gene transcription could be excluded if their activities were high in the presence of wortmannin, a condition where the effect of insulin is abrogated. As shown in Fig. 5A, insulin-stimulated ERK-1 activity was largely inhibited (approximately 75%) by wortmannin. Qualitatively, identical results were obtained in ERK-2 activity assays (data not shown), but the signal intensities revealed by autoradiography were weaker, possibly reflecting a lower abundance of this ERK isoform in hepatocytes. Because insulin markedly activated ERKs, these studies could not rule out the possible involvement of ERKs in the negative regulation of IGFBP-1 mRNA by insulin. The inhibitory effect of wortmannin on insulin-stimulated ERK activity suggests that in primary rat hepatocytes, PI3-kinase activity is required for maximal activation of ERKs by insulin. Wortmannin, at concentrations up to 5 µM, did not inhibit ERK-1 activity when added directly to ERK-1 immunoprecipitates from insulin- and bpV(phen)-stimulated cells (data not shown). Shown in Fig. 5B is the stimulatory effect of bpV(phen) on ERK-1 activity and the failure of wortmannin, at doses of up to 5 µM, to inhibit this effect. Thus, bpV(phen) does not require PI3-kinase activity to stimulate ERK-1. Under conditions of PI3-kinase inhibition by wortmannin, ERK activity seen in response to bpV(phen) is high, thus correlating with its ability to inhibit IGFBP mRNA.

As evidenced in Fig. 6, Dex/Gluc inhibited insulin-mediated activation of ERK-1 to levels below those seen in untreated controls, despite the fact that under these conditions, insulin exerted a dominant inhibitory effect on IGFBP-1 mRNA (Fig. 1). In marked contrast, Dex/Gluc failed to antagonize bpV(phen) stimulation of ERK-1 activity. Similar results were obtained when ERK-2 activity was assessed. Although these data indicate that insulin inhibition of IGFBP-1 mRNA levels does not require ERK activation, they are consistent with a role for ERK activation in mediating the inhibitory effect of bpV(phen) on IGFBP-1 and -4 mRNA levels.

To confirm the involvement of ERK-1 in the mediation of the inhibitory effect of bpV(phen) on IGFBP mRNA levels, we attempted to block ERK-1 activation using the selective MEK inhibitor PD98059 (46, 55). Fig. 7 demonstrates that concentrations of up to 50 µM PD98059 failed to inhibit bpV(phen)-induced ERK-1 activation, but a lower dose (30 µM) completely reversed insulin-induced ERK-1 activation. Activation of MEK by insulin and growth factors depends on Ras activation and recruitment and activation of Raf leading to activation of MEK, which in turn phosphorylates and activates ERKs (27, 28). Ras signaling components, upstream of MEK, are not the only determinants of ERK activation, and other regulatory processes, independent of Ras, account for the activation of ERK-1 by bpV(phen).

DISCUSSION

We have used primary rat hepatocyte cultures, in which endogenous levels of IGFBP mRNA can be measured, to study the regulation of IGFBP-4 mRNA and compare it with that of IGFBP-1 mRNA. IGFBP-1 gene expression in liver and hepatoma cells is powerfully increased by dexamethasone and glucagon (cAMP), the effects of which are additive and mediated at the transcriptional level (39, 40, 41, 56). The hormone combination of Dex/Gluc comparably stimulated IGFBP-1 and -4 mRNA levels (Fig. 1), but in the case of the latter, this effect was almost entirely mediated by glucagon, mimicked by cAMP, and only slightly produced by dexamethasone (data not shown). These results are consistent with the presence of cAMP response elements and multiple putative AP-1 binding sites in the IGFBP-4 gene (57). A glucocorticoid response element(s), however, has not been mapped to the promoter region of the IGFBP-4 gene (57).

The dominant inhibitory effect of insulin over dexamethasone- and cAMP-induced IGFBP-1 gene expression, described previously (39, 41, 44), was observed on IGFBP-1 mRNA levels maximally stimulated by Dex/Gluc (Fig. 1). In contrast, insulin was without effect on IGFBP-4 mRNA in Dex/Gluc-stimulated cells (Fig. 1) or on basal IGFBP-4 mRNA levels (data not shown). These data provide the first account of the hormonal regulation of IGFBP-4 gene expression in hepatocytes, a major site of IGFBP synthesis. It is of particular interest that in a variety of other cell types that express IGFBP-4 mRNA, only stimulatory effects have been described, whereas inhibition has never been reported. In addition, in vitro assays relating the biological effects of IGFBP-4 have shown that this protein uniformly inhibits IGF-1 action (58). From a metabolic point of view, Dex/Gluc-mediated stimulation of IGFBP-1 and -4 mRNA levels, if paralleled by an increase in their respective protein levels, could lead to a reduction of free local IGF-1 levels and, therefore, prevent possible anabolic actions of IGF-1 mediated by hepatic IRs. Moreover, under physiologic circumstances of high circulating insulin, the impact of IGFBP-1 on IGF-1 levels would be reduced, whereas the influence of IGFBP-4 on IGF-1 levels would be maintained, or perhaps increased, depending on the ambient concentrations of glucose counterregulatory hormones, particularly glucagon. Our findings are consistent with a potential role of glucagon in augmenting circulating IGFBP-4 levels. This could play a part in countering hypoglycemia induced by circulating IGF-1, as already established for IGFBP-1 (59).

The peroxovanadium bpV(phen), a potent inhibitor of IR-associated tyrosine phosphatase activity involved in IRK dephosphorylation and inactivation (45), has insulin-like lipogenic and hypoglycemic effects in vitro (45) and in vivo, (60, 61), respectively. We tested its ability to mimic insulin suppression of IGFBP-1 gene transcription in Dex/Gluc-stimulated primary hepatocytes and also investigated its effect on IGFBP-4 mRNA. bpV(phen) inhibited IGFBP-1 mRNA levels as did insulin, and completely reversed stimulated levels of IGFBP-4 mRNA (Fig. 1). The inhibitory action of bpV(phen) on both IGFBP mRNA species contrasted with the ability of insulin to inhibit only IGFBP-1 mRNA and prompted us to study the signaling mechanisms that could account for these discrepant effects.

Insulin inhibition of IGFBP-1 mRNA levels was reversed by wortmannin and rapamycin (Fig. 2, A and B) at concentrations that effectively inhibited insulin-stimulated IRS-1-associated PI3-kinase activity and p70^s6k activity, respectively (Fig. 3A and 4). Thus, both enzymes are necessary for insulin-mediated transcriptional inhibition of the IGFBP-1 gene. Because wortmannin decreased insulin-stimulated p70^s6k activity to levels below those seen in control cells (Fig. 4), then PI3-kinase is upstream of p70^s6k with respect to insulin action, as has been reported by others (13, 20). Interestingly, p70^s6k phosphorylates and activates nuclear proteins directly involved in transcriptional regulation of specific genes (23, 62). The relevance of these nuclear proteins to IGFBP-1 gene transcription has yet to be established. Recently, the stimulatory effect of insulin on hexokinase II mRNA levels in skeletal muscle cells was shown to be inhibited by wortmannin and rapamycin (19), suggesting that hexokinase II and IGFBP-1 mRNA are coordinately regulated by the same signaling pathway. On the other hand, the inhibitory effect of insulin on phosphoenolpyruvate carboxykinase gene transcription in rat hepatoma cells (18, 63) was prevented by wortmannin (18) and LY294002 (63) but not by rapamycin (18), suggesting that PI3-kinase can mediate transcriptional effects of insulin independently of p70^s6k by as yet unknown signaling pathways.

The effect of bpV(phen) on IRS-1-associated PI3-kinase and p70^s6k activity was strikingly similar, both qualitatively and quantitatively, to that of insulin (Figs. 3 and 4). The dose of bpV(phen) used in our studies (0.1 mM) was selected based on its equipotence with insulin (100 nM) in terms of IRK activation in rat hepatoma cells.² PI3-kinase activity was inhibited by wortmannin (Fig. 3), and p70^s6k was inhibited by wortmannin and rapamycin in bpV(phen)-treated hepatocytes (Fig. 4) without affecting bpV(phen)-induced suppression of IGFBP-1 and -4 mRNA levels (Fig. 2, C and D). Our observations suggest that two inhibitory pathways signal to the IGFBP-1 gene. Alternatively, bpV(phen) might activate downstream effectors of p70^s6k. However, the signaling pathway(s) leading to the inhibition of IGFBP-4 mRNA is distinct from that inhibiting IGFBP-1 and does not involve PI3-kinase or p70^s6k because: 1) IGFBP-4 mRNA inhibition by bpV(phen) is insensitive to wortmannin and rapamycin; and 2) insulin activates PI3-kinase and p70^s6k but fails to inhibit IGFBP-4 mRNA.

To assess the possible involvement of the Ras signaling pathway in mediating IGFBP mRNA inhibition, we measured ERK-1 and ERK-2 (MAP kinase) activities in hepatocytes treated with insulin or bpV(phen). Both agents markedly increased MAP kinase activity (Fig. 5). Wortmannin prevented the stimulation of ERK-1 by insulin (Fig. 5A), demonstrating that in primary hepatocytes, PI3-kinase activity is upstream of MAP kinase. However, ERK activation is not necessary for the inhibitory effect of insulin because in the combined presence of Dex/Gluc and insulin, a condition where insulin exerts an inhibitory effect on IGFBP-1 mRNA (Fig. 1), ERK-1 activity was lower than in untreated hepatocytes (Fig. 6). The attenuation of insulin-stimulated ERK-1 activity was effected by glucagon (not dexamethasone) and was mimicked by cAMP,³ consistent with the known inhibitory effect of cAMP/protein kinase A on Raf activation and hence on MAP kinase activity (64, 65). Dex/Gluc did not affect insulin-stimulated PI3-kinase activity.³ Thus, in insulin-stimulated cells, PI3-kinase could not overcome glucagon suppression of ERK-1 activity, consistent with PI3-kinase feeding into the Ras signaling pathway at or upstream of Raf and with the recent finding that PI3-kinase can form a complex with Ras (66). In addition, expression of constitutively active p110 elevates GTP-bound Ras and stimulates Ras-dependent cellular processes (67).

In contrast to insulin, bpV(phen)-stimulated ERK-1 activity remained high in the presence of wortmannin (Fig. 5B) and Dex/Gluc (Fig. 6). Thus, MAP kinase activation may play a role in mediating IGFBP mRNA inhibition. This inference is strengthened by the fact that the IGFBP-4 gene is endowed with multiple putative binding sites for the AP-1 transcription factor (57), which is activated by MAP kinases (68). We suggest that physiologic agents that powerfully activate MAP kinases will suppress IGFBP mRNA levels.

In gel mobility retardation assays, Raf activity was stimulated by bpV(phen), demonstrating that the latter activates components of the Ras signaling pathway upstream of MAP kinase.³ PD98059 (46), a specific noncompetitive inhibitor of inactive, nonphosphorylated MEK (55), was preincubated for 60 min prior to stimulation with insulin or bpV(phen) to achieve similar levels of MEK inhibition. This MEK inhibitor completely abrogated insulin-mediated ERK-1 activation but had no effect on ERK-1 activation by bpV(phen), even at the high dose of 50 µM (Fig. 7). Thus, MAP kinase activation by bpV(phen) appears to be effected by a MEK-independent process. Our observation that cAMP fails to antagonize ERK-1 activation by bpV(phen)³ suggests that bpV(phen) also affects ERK-1 independently of Raf. It is possible that activation of another signaling pathway could lead to MAP kinase activation. It is unlikely that one such possibility is the protein kinase C pathway, because phorbol esters, if anything, augment, rather than suppress, IGFBP-1 mRNA levels (41, 44). In addition, protein kinase C activation influences MAP kinase activity at the level of Raf (69), whereas the bpV(phen) effect on MAP kinase is, as noted above, independent of Raf activity.

To our knowledge, this is the first report of a MEK-independent activation of MAP kinase. Given that peroxovanadium compounds are potent tyrosine phosphatase inhibitors (45), the effect of bpV(phen), in the presence of PD98059, could be explained by inhibition of a specific tyrosine/threonine phosphatase(s) involved in the negative regulation of MAP kinases (29, 30), such that MAP kinase activation by autophosphorylation (70) would be unopposed. This hypothesis warrants further investigation and suggests that peroxovanadium compounds could be of value in identifying phosphatases regulating MAP kinase activity.

The use of bpV(phen) has revealed a signaling pathway(s) not necessary for the effect of insulin on IGFBP-1 gene transcription. This may be of physiologic relevance in the action of other, as yet unidentified, regulators of IGFBP gene expression. The effect on IGFBP gene expression of powerful stimulators of hepatic MAP kinase activity, the effects of which can be abrogated by MEK inhibition, will be useful in confirming or refuting the role of the Ras signaling pathway in the regulation of IGFBP gene expression.