Insulin activates nuclear factor kappa B in mammalian cells through a Raf-1-mediated pathway.

We examined the effect of insulin on nuclear factor κB (NF-κB) activity in Chinese hamster ovary (CHO) cells overexpressing wild-type (CHO-R cells) or -defective insulin receptors mutated at Tyr1162 and Tyr1163 autophosphorylation sites (CHO-Y2 cells). In CHO-R cells, insulin caused a specific, time-, and concentration-dependent activation of NF-κB. The insulin-induced DNA-binding complex was identified as the p50/p65 heterodimer. Insulin activation of NF-κB: 1) was related to insulin receptor number and tyrosine kinase activity since it was markedly reduced in parental CHO cells which proved to respond to insulin growth factor-1 and phorbol 12-myristate 13-acetate (PMA) activation, and was dramatically decreased in CHO-Y2 cells; 2) persisted in the presence of cycloheximide and was blocked by pyrrolidine dithiocarbamate, aspirin and sodium salicylate, three compounds interfering with IκB degradation and/or NF-κB•IκB complex dissociation; 3) was independent of both PMA-sensitive and atypical (ζ) protein kinases C; and 4) was dependent on Raf-1 kinase activity since insulin-stimulated NF-κB DNA binding activity was inhibited by 8-bromo-cAMP, a Raf-1 kinase inhibitor. Moreover, insulin activation of NF-κB-driven luciferase reporter gene expression was blocked in CHO-R cells expressing a Raf-1 dominant negative mutant. This is the first evidence that insulin activates NF-κB in mammalian cells through a post-translational mechanism requiring both insulin receptor tyrosine kinase and Raf-1 kinase activities.

Insulin exerts a wide array of biological effects including regulation of growth and gene expression. The cytoplasmic events implicated in the transduction of insulin signals from the membrane insulin receptor to the transcriptional machinery in the nucleus are beginning to be understood. Studies using transfected cells overexpressing human insulin receptors and/or various proteins of insulin signaling pathways have shown that, after binding to its receptor, insulin activates insulin receptor tyrosine kinase activity and triggers tyrosine phosphorylation of at least two major substrates, IRS-1 and Shc (1). Once phosphorylated, insulin receptor substrates interact with several proteins including the Grb2⅐Sos complex which activates the Ras-Raf-1-MAP kinase 1 pathway (1). In contrast, little is known at present as concerns the nuclear transcription factors which are specifically activated by insulin to regulate gene expression.
The nuclear factor-B (NF-B) was originally described as being present in the cytosol of most cell types as an inactive heterodimer composed of 50-kDa (p50, NF-B1) and 65-kDa (p65, Rel A) subunits and bound to one of the IB inhibitor proteins (2)(3)(4). Activation of NF-B involves phosphorylation and degradation of IB. This results in the dissociation of the NF-B⅐IB complex, a process which can be blocked by inhibitors such as pyrrolidine dithiocarbamate (PDTC), aspirin, and sodium salicylate (5,6). Thereafter, the active NF-B p50/p65 heterodimer translocates to the nucleus, where it directly binds to its cognate DNA sequences. NF-B is a pleiotropic activator which participates in the induction of a wide variety of cellular genes including genes encoding for signaling proteins. Activation of this factor can be achieved in many cell types by agonists of immune and inflammatory responses and also by mitogens (7)(8)(9). In this regard, a recent paper (10) reported that the sequence of events triggered by insulin to induce maturation of Xenopus oocytes involves NF-B activation. This finding prompted us to examine whether insulin was able to activate NF-B in mammalian cells and, if so, to examine the signal transduction pathway involved. To this end, we used parental Chinese hamster ovary (CHO) cells and CHO cells overexpressing either wild-type human insulin receptors or kinase-defective insulin receptors mutated at Tyr 1162 and Tyr 1163 , two autophosphorylation sites playing a crucial role in receptor activation (11)(12)(13)(14)(15). Our study demonstrates that insulin activates NF-B in mammalian cells through a pathway which requires insulin receptor tyrosine kinase and Raf-1 kinase activities. 32 36, Paris) used in this study have been previously described (12)(13)(14). These include the parental cell line (CHO), the CHO cell line transfected with a plasmid coding for the native form of the human insulin receptor (CHO-R), and the CHO cell line expressing human insulin receptors in which the 2 tyrosines at positions 1162 and 1163 have been replaced with phenylalanine residues by directed mutagenesis (CHO-Y2). Cells were grown in Ham's F-12 medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (FCS). Prior to stimulation with insulin, cells were growth arrested at confluence by a 24-h incubation with serum-free Ham's F-12 medium.

Reagents-[␥-
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared from parental and transfected CHO cells by the method of Dignam et al. (16) with minor modifications, after treatment with or without insulin, IGF-1, or PMA, in the absence or presence of various agents. After two washings in 5 ml of ice-cold phosphate-buffered saline (PBS), cells were harvested in a 1.5-ml tube and centrifuged at 2,000 ϫ g for 5 min in a microcentrifuge. The cell pellet was resuspended in 400 l of buffer A (10 mM HEPES/ KOH, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol (DTT), 0.1% Triton X-100), supplemented with the following protease inhibitors: 0.2 mM phenylmethylsulfonyl fluoride, 2 M aprotinin, and 1.0 g/ml of antipaïn, pepstatin, benzamidine, and leupeptin. After homogenization in a tight-fitting Dounce homogenizer, cell lysates were maintained 10 min on ice and centrifuged at 2,000 ϫ g for 10 min. The nuclear pellet was then washed in buffer A without Triton X-100, resuspended in 100 l of buffer B (20 mM HEPES/KOH, pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 25% glycerol, and the mixture of protease inhibitors described above). After 30 min at 4°C under constant agitation, nuclear debris were centrifuged at 13.000 ϫ g for 5 min, and the supernatant (nuclear extract) was distributed into 15-l aliquots which were stored at Ϫ80°C until analysis by EMSA. Protein was determined by using the Bio-Rad reagent according to the manufacturer's instructions. Where indicated, nuclear extracts were incubated for 30 min at 4°C with antibodies against p50 or p65 subunits of NF-B before the binding reaction. The double-stranded NF-B probe: 5Ј-AGCTTCAGAGGGGACTTTCCGAGAGG-3Ј; 3Ј-AGTCTC-CCCTGAAAGGCTCTCCAGCT-5Ј (17) was annealed and end-labeled by using the DNA polymerase 1 Klenow fragment in the presence of 50 Ci of [␣-32 P]dCTP and other unlabeled dNTPs at 20 M each in a 50-l reaction mixture containing 10 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 1 mM DTT, and 1 mM EDTA. Unincorporated nucleotides were removed by filtration through a Nuctrap column (Stratagène). Binding reactions were carried out in a 20-l binding reaction mixture (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.5 mM DTT, 10% glycerol, 0.2% Nonidet P-40, and 3 g of poly(dI-dC)) containing 5 g of nuclear proteins and 0.5-2 ng of the NF-B probe (40,000 -80,000 counts/min). In some experiments, the binding reaction mixture also contained a large excess of unlabeled oligonucleotides containing either the NF-B or the OCT-1 (immunoglobulin enhancer octanucleotide factor-1) consensus sequences. Samples were incubated at room temperature for 25 min and fractionated by electrophoresis on a 6% non-denaturing polyacrylamide gel in TAE buffer (7 mM Tris, pH 7.5, 3 mM sodium acetate, 1 mM EDTA), which had been pre-electrophoresed for 1 h at 80 V. Gels were run at 160 V for 2.5 h. Following electrophoresis, gels were transferred to 3MM paper (Whatman Ltd.), dried in a gel dryer under vacuum at 80°C, and exposed to x-ray Hyperfilm-MP at Ϫ20°C using an intensifying screen. Where indicated, results were quantified by laser scanning densitometry of the autoradiographs.
Down-regulation of Protein Kinase C (PKC)-CHO-R cells were incubated for 24 h in FCS-free Ham's F-12 medium in the absence or presence of 2.5 M PMA. Cells were then washed and assayed for specific binding of [ 3 H]PDBu as described previously (12).
Amplification of PKC-mRNA by Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)-Total RNA was isolated by the method of Chirgwin et al. (18) from CHO-R cells and both colonic Caco-2 cells and hepatoma BC1 cells as controls. Before reverse transcription, RNA was treated at 68°C for 10 min and cooled on ice. Two g of RNA was reverse transcribed for 1 h at 37°C to cDNA in a 20-l reaction mixture containing 500 M dNTPs, 10 mM DTT, RNasin at 1 unit/l, 5 M random hexamer, and reverse transcriptase at 10 units/l. Then 5 l of the reverse transcription reaction was used in a PCR reaction volume of 25 l containing PCR buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl 2 , 0.001% gelatin), 2 units of Taq DNA polymerase, and two pairs of primers, each at 0.5 M. PKC-primers [5Ј-TCCGTCAAAGC-CTCCCATGTT-3Ј (sense) and 5Ј-ACGGGCTCGCTGGTGAACTGT-3Ј (antisense)] were designed to generate a 228-base pair fragment (from nucleotides 1425-1653) in the catalytic domain. ␤-Actin-specific primers designed to amplify a 316-base pair fragment were chosen as reported elsewhere (19). To prevent evaporation, 70 l of mineral oil was layered on top of each sample. A thermocycler (Perkin-Elmer) was set to the following cycle parameters: 1 min at 94°C for denaturation, 30 s at 61°C for primer annealing, and 1.5 min at 72°C for elongation. This cycle was repeated 30 times. Following amplification, samples (10 l) were electrophoresed on 2% agarose gels, transferred to Hybond N ϩ membranes, and hybridized to a [␥-32 P]ATP-labeled PKC-specific probe (generous gift from Y. Nishizuka, Kobe University, Japan) or to a 30-mer oligonucleotide ␤-actin probe (19). Hybridization was carried out with buffer containing 6 ϫ SPE (0.2 M Na 2 HPO 4 , pH 7.6, 3.6 M NaCl, 20 mM) EDTA, 5 ϫ Denhardt's solution, 0.5 mg/ml heparin, and 0.2% SDS. Membranes were then washed with 6 ϫ SPE, 0.1% SDS at 55°C for 30 min, and autoradiographed (Hyperfilms-MP).
In Situ PKC-Activity Assay-The effect of insulin on PKC-activity was evaluated by using a recently described (20) in situ assay. CHO-R and Caco-2 cells were distributed to Eppendorf microcentrifuge tubes (20,000 cells/tube) in serum-free medium. After incubation in the presence or absence of insulin or PMA, the medium was removed and replaced with 40 l of permeabilization solution (137 mM NaCl, 5.4 mM KCl, 0.3 mM sodium phosphate, 0.4 mM potassium phosphate, 1 mg/ml glucose, 20 mM HEPES, pH 7.2, 50 g/ml digitonin, 10 mM magnesium chloride, 25 mM ␤-glycerophosphate) containing 150 M ⑀-peptide (pseudosubstrate site in PKC-⑀ substituting Ser for Ala 159 ) together with or without 300 M PKC-inhibitor peptide. The reactions were initiated by addition of 100 M [␥-32 P]ATP and were terminated after 10 min at 30°C with 15 l of 25% trichloroacetic acid (w/v). Samples were left for at least 10 min on ice and then centrifuged at 5,000 ϫ g for 5 min. Supernatants (50 l) were spotted on 2 ϫ 2-cm phosphocellulose squares (Whatman P-81) which were washed batchwise in three changes (500 ml each) of 75 mM phosphoric acid and one change of 75 mM sodium phosphate (pH 7.5). Once dried, the P-81 papers were counted for bound radioactivity. Assays were performed in duplicate and tubes containing no ⑀-peptide were included in each assay to estimate the "background" phosphorylation of basic cell components which were not precipitated by trichloroacetic acid and which adhered to P-81 paper. The same in situ assay was used to measure the activity of classical and novel PKCs, except that the permeabilization solution contained 2.5 mM CaCl 2 and that reactions were carried out with the [Ser 25 ]PKC-␣ substrate peptide.
MAP Kinase Western Blotting-Serum-deprived CHO-R cells were incubated with or without 10 Ϫ7 M insulin, in the absence or presence of 0.25 mM 8-bromo-cAMP. After three washings with ice-cold PBS, cells were lysed and nuclear fractions prepared as above were examined for lactate dehydrogenase activity which was used as a cytosolic marker. These fractions were then analyzed by Western blotting using an antirat MAP kinase R2 antibody (1:1000) and a goat anti-rabbit IgG conjugated to horseradish peroxidase. ECL detection was monitored with reagents from Amersham, as recommended by the manufacturer. Quantification was performed by scanning densitometry of the autoradiographs.
Transient Transfection-CHO-R cells were transfected using the calcium phosphate precipitation method. Briefly, CHO-R cells (5 ϫ 10 5 cells/60 mm-dish) were seeded 24 h before transfection in 5 ml of Dulbecco's modified Eagle's medium supplemented with 10% FCS. Medium was renewed 4 h prior to transfection. Cells were transfected with 5 g of either (Ig)3-conaluc, a reporter plasmid which contains three copies of the immunoglobulin chain enhancer B site upstream of the minimal conalbumin promoter fused to the luciferase reporter gene or its control counterpart conaluc (generous gift from Dr. A. Israel, Institut Pasteur, Paris, France). In some experiments, cells were co-transfected with 5 g of RSV-C4␤, a plasmid that expresses the Raf-1 dominant negative mutant Raf-C4 (generous gift from Dr. U. R. Rapp, Institute of Medical Radiobiology and Cell Biology, Wurzburg, Germany). One g of a pRSV-CAT was always included as a monitor for transfection efficiency. Transfections were carried out in the presence of a constant amount of DNA (11 g/60-mm dish). After transfection, cells were incubated for 36 h in Ham's F-12 medium containing 0.3% FCS, then for 12 h in the presence or absence of 10 Ϫ7 M insulin and harvested for luciferase and CAT assays.
Luciferase and CAT Assays-The cell monolayers were rinsed twice with Ca 2ϩ -and Mg 2ϩ -free PBS, covered with 0.7 ml of chilled lysis buffer (25 mM Tris, pH 7.8, with H 3 PO 4 , 10 mM MgCl 2 , 1 mM EDTA, 1 mM DTT, 1% Triton X-100, and 15% glycerol), maintained on ice for 30 min and scraped. The lysates were centrifuged for 10 min at 13,000 ϫ g. Luciferase activity was assayed in lysis buffer containing 50 mM ATP and 100 M luciferin, using a monolight 2010 luminometer (Berthold). CAT activity was measured by the diffusion-based CAT assay (21). Correction for differences in transfection efficiency between plates within an experiment was performed by normalizing the luciferase activity to the CAT activity in each extract. All transfections were performed three times in triplicate.

Insulin Activates NF-B in CHO-R Cells through Its Own
Receptors-Insulin activated NF-B in CHO-R cells in a timedependent manner. The effect of the hormone was detected at 1 h, reached a maximum at 6 h, and was no longer detected at 24 h (Fig. 1A). Similarly, Baldwin et al. (7) reported that serum induction of NF-B in BALB/c3T3 cells culminated at 6 h and was no longer detected at 24 h. The hormone increased the formation of two NF-B complexes with a major effect on the upper complex designated as band 1 and a minor effect on the lower complex designated as band 2 (Fig. 1A). Insulin activated NF-B in a concentration-dependent manner (Fig. 1B). The effect of the hormone regularly increased from 10 Ϫ9 M t0 10 Ϫ7 M (Fig. 1B). At the latter concentration, the effect of insulin was equivalent to that elicited by 10 Ϫ7 M IGF-1 (Fig. 1B). Two observations reported in our previous (14) and present papers argue for insulin activating NF-B through its own receptors. First, insulin was potent to activate NF-B at 10 Ϫ9 and 10 Ϫ8 M, two concentrations at which insulin was unable to displace the binding of 125 I-labeled IGF-1 to IGF-1 receptors in CHO-R cells (14). Second, insulin activation was related to insulin receptor number. This is supported by the finding that the sensitivity to insulin exhibited by CHO-R cells for NF-B activation (Fig. 1B) was markedly decreased in parental CHO cells (Fig. 2). In contrast, these cells proved to be fully responsive to IGF-1 as well as to PMA, a well-known inducer of NF-B activity in several cell types (2, 3, 5) (Fig. 2). The above results therefore indicate that both insulin and IGF-1 are good inducers of NF-B in CHO-R cells and that insulin activates this transcription factor in the insulin receptor overexpressing cell line through its own receptors.
Characterization of Insulin-induced NF-B⅐DNA Complexes-The specificity of bands 1 and 2 was assessed in competition experiments. Both bands disappeared in the presence of the unlabeled oligonucleotide containing the NF-B-binding site whereas they remained unchanged in the presence of the unlabeled oligonucleotide containing the unrelated OCT-1 consensus sequence (Fig. 3A). We further characterized the insu-lin-induced DNA⅐protein complexes by using antibodies directed against the p50 and p65 NF-B subunits. Incubation of nuclear extracts (30 min at 4°C) from insulin-treated cells (6 h, 10 Ϫ7 M) with anti-p50 antibody abolished band 2 with a concomitant supershift, and reduced the amount of band 1, suggesting the presence of p50 in the two protein complexes (Fig.  3B). In contrast, the anti-p65 antibody had no effect on band 2, suggesting that p65 was absent from the band 2 DNA⅐protein complex. However this antibody, like the p50 antibody, markedly reduced the intensity of band 1, indicating that band 1 corresponds to a DNA⅐protein complex composed of both p50 and p65 NF-B subunits (Fig. 3B). These experiments enabled us to identify the major and minor DNA⅐protein NF-B complexes induced by insulin as the p50/p65 heterodimer (band 1) and p50 homodimer (band 2), respectively. This finding is of particular interest since the p50/p65 heterodimer is known as the main effective and also the major inducible form of NF-B (22), whereas the p50 homodimer has been reported to occur as constitutive factor in nuclei of certain cell types (2,3,23).
Insulin Activates NF-B-dependent Luciferase Reporter Gene Expression-To determine whether insulin enhances NF-Bdriven reporter gene expression, CHO-R cells were transfected

FIG. 1. Activation of NF-B in CHO-R cells by insulin and IGF-1. Serum-starved CHO-R cells were incubated
in FCS-free medium with or without 10 Ϫ7 M insulin for the indicated times (A) or with or without the indicated concentrations of insulin or IGF-1 for 6 h (B). Nuclear extracts were prepared and assayed for NF-B activity, as described under "Experimental Procedures." Autoradiographs from the concentration experiments were quantified at the level of band 1 by laser scanning densitometry. The autoradiographs are representative of three (A and B for IGF-1) or seven (B for insulin) experiments.

FIG. 2. Activation of NF-B in CHO cells by insulin, IGF-1, and
PMA. Serum-deprived CHO cells were incubated for 6 h in FCS-free medium with or without the indicated concentrations of insulin, or IGF-1, or PMA. Nuclear extracts were prepared and assayed for NF-B activity, as described under "Experimental Procedures." This is a representative experiment independently performed three times.
transiently with (Ig)3-conaluc, a reporter plasmid in which the activity of the minimal conalbumin promoter fused to the luciferase reporter gene may be enhanced by three copies of the immunoglobulin chain enhancer B site (24). When transiently transfected CHO-R cells were treated for 12 h with 10 Ϫ7 M insulin, a 3.2 Ϯ 0.5-fold increase in normalized luciferase activity was observed as compared to the activity measured in unstimulated cells (Fig. 4B). Under identical conditions, insulin had no effect on normalized luciferase activity in CHO-R cells which had been transfected with the control conaluc plasmid (Fig. 4A). This argues for NF-B-binding sites being specifically activated by insulin treatment. Together, these experiments support the conclusion that insulin stimulation of NF-B DNA binding activity in CHO-R cells (Fig. 1) results in activation of NF-B-mediated gene transcription (Fig. 4B).
Insulin Activation of NF-B Requires Insulin Receptor Tyrosine Kinase Activity-To investigate the role of the insulin receptor tyrosine kinase in the activation of NF-B by insulin, we studied the effect of insulin in CHO-Y2 cells expressing insulin receptors made kinase defective by mutation at Tyr 1162 and Tyr 1163 , two major autophosphorylation sites of the insulin receptor tyrosine kinase domain playing a crucial role in receptor activation (25). Nuclear extracts from CHO-Y2 cells which had been treated with 10 Ϫ7 M insulin for various time periods (0 -6 h) or with graded concentrations (0 -10 Ϫ7 M) of insulin for 6 h were tested for NF-B activity. As shown in Fig. 5, the effect of insulin in CHO-Y2 cells was dramatically reduced as compared to that observed in CHO-R cells (Fig. 1). Similar reduction was previously reported for insulin receptor tyrosine kinase activity (14). This finding further argues for an effect of insulin being mediated by insulin and not IGF-1 receptors, since we previously reported that CHO-Y2 cells expressed the same number of IGF-1 receptors as CHO-R cells (13). Most importantly, this finding shows that insulin receptor autophosphorylation sites Tyr 1162 and Tyr 1163 play a pivotal role in insulin activation of NF-B and therefore indicate that this process takes place among the multiple insulin-stimulated pathways requiring the integrity of the insulin receptor tyrosine kinase activity (25). Otherwise, the finding that the activation of NF-B by insulin was lower in CHO-Y2 cells than in CHO cells most probably reflects a dominant negative effect exerted by the overexpressed mutated insulin receptors on endogenous insulin receptors or their downstream targets, as has been previously found for other insulin-responsive pathways (12,14).
Insulin Activation of NF-B Involves a Post-translational Mechanism-To approach the mechanism whereby insulin induced NF-B in CHO-R cells, nuclear extracts were prepared from CHO-R cells which had been treated for 6 h with 10 Ϫ7 M

FIG. 5. Loss of NF-B activation by insulin in CHO-Y2 cells.
Experiments described in Fig. 1, A and B, with insulin were performed in CHO-Y2-cells. Maximal insulin activation of NF-B in CHO-R cells is given as a control. This is a representative experiment independently performed three times. insulin in the presence or absence of the protein synthesis inhibitor cycloheximide (10 g/ml). This cycloheximide treatment was found to inhibit 90 -95% of basal and insulin-stimulated protein synthesis in CHO-R cells, in accordance with previous results (14). Cycloheximide alone increased NF-B activity in CHO-R cells (Fig. 6A), as previously reported in other cell types (5,7,26). The mechanism underlying this increase is unknown but may involve inhibition of IB synthesis, as has been suggested by others (5,8,26). Most importantly, we observed that activation of NF-B by insulin was preserved in the presence of cycloheximide, indicating that this process did not require de novo protein synthesis (Fig. 6A). This finding, together with the finding that NF-B activation by insulin occurred within the first hour of treatment, argue for a post-translational effect of insulin on NF-B activity. This may proceed from the ability of the hormone to promote dissociation of the NF-B⅐IB complex.
To test this hypothesis, we examined the effect of: 1) PDTC, a thiol compound scavenging reactive oxygen intermediates and, thereby, impeding the release of IB from NF-B (5); and 2) aspirin and sodium salicylate, two drugs interfering with a pathway that leads to IB phosphorylation and/or degradation or both (6). As shown in Fig. 6B, PDTC (0.1 mM) completely abolished maximal insulin activation of NF-B, as indicated by the complete disappearance of band 1 and band 2 DNA-binding complexes in nuclear extracts prepared from CHO-R cells treated with insulin (6 h, 10-7 M). PDTC was reported to cause similar inhibition in Jurkat cells exposed to various NF-B activators including lipopolysaccharide, tumor necrosis factor-␣, interleukin-1, and PMA (5). As shown in Fig. 6C, sodium salicylate (10 mM) and aspirin (5 mM) were also potent to inhibit insulin induction of NF-B in CHO-R cells. In accordance with these results, sodium salicylate and aspirin were recently shown to inhibit activation of NF-B by lipopolysaccharide, tumor necrosis factor-␣, and phytohemagglutinin plus PMA in Jurkat cells (6). The mechanisms involved in sodium salicylate or PDTC inhibitory effects in CHO-R cells could be identical to those previously described in the above studies, i.e. impairment of a pathway leading to IB degradation and/or NF-B⅐IB complex dissociation.
Insulin Activation of NF-B Appears to be Independent of PMA-sensitive and PMA-insensitive PKC Isoforms-Because PKC is a known inducer of NF-B (2-4) and PMA activates NF-B in CHO cells (Fig. 2), we next examined PKC involvement in insulin activation of NF-B. We thus tested the effect of insulin (6 h, 10 Ϫ7 M) on NF-B activity in CHO-R cells which had been treated for 24 h in the presence or absence of 2.5 M PMA. This treatment was previously found to produce efficient down-regulation of PMA-sensitive PKC isoforms, as evaluated by measuring the specific binding of [ 3 H]PDBu to whole cells (12). As shown in Fig. 7A, the activation of NF-B by insulin was similar in untreated and PMA-treated CHO-R cells, indicating that insulin signaling of NF-B activity was mediated by a pathway which was independent of PMA-sensitive PKCs, i.e. classical and novel PKC isoforms. However, PKC-, an atypical PMA-insensitive PKC isoform (27), was recently reported to FIG. 7

. Effect of a long-term exposure of CHO-R cells to PMA on NF-B activation by insulin (A)and PKC-mRNA expression in CHO-R cells (B).
A, nuclear extracts prepared from control or PMA-treated (24 h, 2.5 M) CHO-R cells which had been incubated for 6 h in the absence or presence of 10 Ϫ7 M insulin were assayed for NF-B activity as described under "Experimental Procedures." This is a representative experiment independently performed three times. B, total RNA (2 g) from CHO-R cells or from Caco-2 or BC1 cells was reverse transcribed to cDNA and amplified by PCR as described under "Experimental Procedures." Amplified PKC-cDNA was detected by hybridization to a specific PKC-cDNA probe. Level of ␤-actin mRNA was evaluated in these assays as a control. Ethidium bromide staining of PKCand ␤-actin bands is given in the bottom panel.

FIG. 6. Effect of cycloheximide and inhibitors of NF-B⅐IB complex dissociation on NF-B activation by insulin in CHO-R cells.
Nuclear extracts were prepared from CHO-R cells which had been incubated for 6 h with or without 10 Ϫ7 M insulin in the absence or presence of 10 g/ml cycloheximide (CHX) (A) or 0.1 mM PDTC (B) or 5 mM aspirin or 10 mM sodium salicylate (NaSal) (C). EMSA was performed as described under "Experimental Procedures." This is a representative experiment independently performed three times.
induce phosphorylation of IB in vitro (28) and also to be involved in insulin-induced maturation of Xenopus oocytes, a Ras-mediated process associated with NF-B activation (10). We therefore evaluated the level of expression of this PKC isoform in CHO-R cells by the RT-PCR technique. The data indicated that PKC-was detectable in CHO-R cells but at a very low level as compared to that observed in control Caco-2 and BC1 cells (Fig. 7B). Since this finding raised the possibility that PKC-may be involved in insulin stimulation of NF-B activity in CHO-R cells, we next examined whether insulin was able to increase PKC-activity under conditions where it activated NF-B in CHO-R cells. To this end, we took advantage of a recently described in situ PKC-activity assay (20). In this assay, the phosphorylation of the PKC-⑀ peptide, the most efficient substrate for PKC- (29), was measured in digitoninpermeabilized cells either in the presence or in the absence of the PKC-inhibitor pseudosubstrate peptide. As could be expected from the results presented in Fig. 7B, the level of PKCactivity in CHO-R cells (0.9 nmol of 32 P transferred to PKC-⑀ peptide/10 5 cells) was far lower than that measured in Caco-2 cells (3.5 nmol of 32 P transferred to PKC-⑀ peptide/10 5 cells). Insulin (10 Ϫ7 M) had no effect on PKC-activity (90 -100% of control) in CHO-R cells, whatever the time of incubation examined (30 min, 1, 2, and 6 h). Similarly, PMA (1.62 ϫ 10 Ϫ7 M, 15 min) was ineffective on PKC-activity in CHO-R cells (110% of the control value), as previously shown in other cell types (27). In contrast, PMA (1.62 ϫ 10 Ϫ7 M) caused a marked increase in PKC activity (130 and 200% of the control value at 5 and 15 min, respectively) when the same permeabilization procedure was run out with the PKC-␣ peptide, a good substrate for classical and novel PKCs (29). Taken as a whole, the above results argue against the involvement of PMA-sensitive PKC isoforms in insulin activation of NF-B in CHO-R cells and do not favor the hypothesis that the PMA-insensitive PKC-isoform may play a role in this process.
Insulin Activation of NF-B Is Inhibited by 8-Bromo-cAMP, an Agent Interfering with Growth Factor Activation of Raf-1 Kinase-The recent study of Li and Sedivy (30) reported the ability of Raf-1 kinase to activate NF-B by dissociating the NF-B⅐IB complex. Moreover, several papers provided evidence that, in mammalian cells, the Ras-Raf-1 pathway mediated the activation of NF-B by various stimuli or inducers (31)(32)(33). We judged these findings of particular interest since we (34) and others (35,36) previously reported the capacity of insulin to activate this pathway in CHO-R cells. We therefore investigated whether insulin activation of NF-B in these cells was affected when inhibiting Raf-1 kinase by 8-bromo-cAMP. This cell-permeable cAMP analogue is an activator of protein kinase A, a kinase which, in some cell types (37,38), inhibits Ras-dependent activation of Raf-1 kinase by phosphorylating the enzyme on its kinase domain (39). As a preliminary to this experiment, we examined insulin activation of MAP kinase, a downstream substrate of Raf-1, by evaluating MAP kinase nuclear translocation in CHO-R cells (40).
Insulin (6 h, 10 Ϫ7 M) increased the amount of immunoreactive MAP kinase associated with the nuclear fraction (225% of the value measured in the nuclear fraction from control cells), indicating the ability of the hormone to induce a long term MAP kinase translocation in the nucleus of CHO-R cells. Treatment of CHO-R cells with 8-bromo-cAMP (0.25 mM) prevented the insulin-induced increase in nuclear immunoreactive MAP kinase (92% of the control value). In view of these data, we investigated the effect of 8-bromo-cAMP on NF-B activation by insulin. Fig. 8 shows that 8-bromo-cAMP (0.25 mM) did not increase the DNA binding activity of NF-B, making unlikely in vivo activation of this transcription factor by protein kinase A in CHO-R cells, in contrast to what has been found in in vitro experiments using cytosolic fractions from 70 Z/3 cells (41). This compound inhibited insulin activation of NF-B in a concentration-dependent manner, with the maximal inhibition being observed at 0.25 mM (Fig. 8). In contrast, even at a 2-fold higher concentration (0.5 mM), 8-bromo-cGMP poorly modified insulin-stimulated NF-B DNA-binding activity in CHO-R cells (Fig. 8), demonstrating the specificity of 8-bromo-cAMP inhibitory effect. These experiments showed that: 1) under conditions where insulin induced the maximal stimulation of NF-B activity, it activated Raf-1 kinase, as assessed by MAPkinase nuclear translocation, a process shown to be associated with growth factor signaling (40); and 2) 8-bromo-cAMP activation of protein kinase A inhibited insulin stimulation of both Raf-1 kinase and NF-B DNA binding activity. Such findings argue for the notion that Raf-1 kinase mediates the activation of NF-B by insulin in CHO-R cells.
Insulin Activation of NF-B Is Blocked in CHO-R Cells Expressing a Raf-1 Dominant Negative Mutant-To strengthen the above hypothesis, we examined the activation by insulin of NF-B-mediated luciferase gene expression in CHO-R cells transfected with the Raf-1 dominant negative mutant Raf-C4. This mutant encodes the cysteine-rich amino-terminal regulatory domain of Raf-1 but lacks the kinase domain and a serine/ threonine region which is believed to function as a regulatory phosphorylation site (42). In these experiments, CHO-R cells were transfected with the (Ig)3-conaluc reporter plasmid (5 g), in the presence or absence of the expression plasmid for Raf-C4 (5 g). A pRSV-CAT (1 g) was included as a monitor for transfection efficiency. Co-transfection of Raf-C4 in CHO-R cells completely abolished the effect of insulin on the B enhancer activity: the significant insulin-induced increase (3.2 Ϯ 0.5-fold) in normalized luciferase activity measured in CHO-R cells transfected with (Ig)3-conaluc (Fig. 4B) was no longer observed in cells which had been co-transfected with Raf-C4 (0.90 Ϯ 0.07-fold) (Fig. 4B). These results provide clear evidence that the pathway whereby insulin activates NF-B-mediated gene expression in CHO-R cells involves Raf-1 kinase.
The question of whether insulin activation of NF-B in these cells is a MAP kinase-mediated process cannot be answered by the results presented here. A recent paper (33) studying activation of NF-B by hypoxia concluded to the involvement of Ras and Raf-1 but not of MAP kinase in this process. Whether or not this may be the case for insulin activation of NF-B remains to be determined. As well, it would be interesting to examine whether, in other cell lines, a pathway bifurcating from Ras and involving PKCmay, in parallel to the Raf-1 kinase pathway, participate in NF-B activation by insulin. In this regard, both Raf-1 and PKCwere reported to initiate parallel pathways downstream of Ras for regulation of mouse fibroblast proliferation (43).
In conclusion, our study provides the first evidence that insulin specifically activates NF-B in mammalian cells. This process involves a post-translational mechanism requiring both insulin receptor tyrosine kinase and Raf-1 kinase activities.