Effects of stimulation of AMP-activated protein kinase on insulin-like growth factor 1- and epidermal growth factor-dependent extracellular signal-regulated kinase pathway.

AMP-activated protein kinase (AMPK) is tightly regulated by the cellular AMP:ATP ratio and plays a central role in the regulation of energy homeostasis. Previously, AMPK was reported to phosphorylate serine 621 of Raf-1 in vitro. In the present study, we investigated a possible role of AMPK in extracellular signal-regulated kinase (Erk) cascades, using 5-aminoimidazole-4-carboxamide-1-beta-d-ribofuranoside (AICAR), a cell-permeable activator of AMPK and antisense RNA experiments. Activation of AMPK by AICAR in NIH-3T3 cells resulted in drastic inhibitions of Ras, Raf-1, and Erk activation induced by insulin-like growth factor 1 (IGF-1). Expression of an antisense RNA for the AMPK catalytic subunit decreased the AMPK activity and significantly diminished the AICAR effect on IGF-1-induced Ras activation and the subsequent Erk activation, indicating that its effect is indeed mediated by AMPK. Phosphorylation of Raf-1 serine 621, however, was not involved in AMPK-mediated inhibition of Erk cascades. In contrast to IGF-1, AICAR did not block epidermal growth factor (EGF)-dependent Raf-1 and Erk activation, but our results demonstrated that multiple Raf-1 upstream pathways induced by EGF were differentially affected by AICAR: inhibition of Ras activation and simultaneous induction of Ras-independent Raf activation. The activities of IGF-1 and EGF receptor were not affected by AICAR. Taken together, our results suggest that AMPK differentially regulate Erk cascades by inhibiting Ras activation or stimulating the Ras-independent pathway in response to the varying energy status of the cell.

key role in the regulation of energy homeostasis and is highly conserved among animals, plants, and fungi (reviewed in Refs. [1][2][3]. AMPK is a heterotrimeric enzyme consisting of a catalytic subunit (␣) and two regulatory subunits (␤ and ␥), and it is activated by the cellular stress causing ATP depletion, which in turn leads to elevation of the AMP:ATP ratio (reviewed in Refs. [1][2][3]. These stresses include heat shock, ischemia/hypoxia in cardiac muscle, and exercise in skeletal muscle. In addition to allosteric activation by AMP, AMPK is activated by phosphorylation by an upstream kinase termed AMPK kinase (4). Once activated, AMPK suppresses the key enzymes involved in ATPconsuming anabolic pathways such as fatty acid and cholesterol synthesis (5,6). Besides, AMPK initiates a series of compensatory changes that increase cellular ATP supply by activating the rate of fatty acid oxidation (7,8) and glucose uptake in cardiac and skeletal muscle (9,10). Thus, AMPK has been speculated to play a role as a "fuel gauge" that recognizes ATP depletion and maintains ATP level (reviewed in Refs. [1][2][3]. Most of the currently identified substrates of AMPK are metabolic enzymes. However, the recent implications of AMPK in transcriptional control (11,12), insulin secretion in pancreatic ␤ cell (13), and regulation of endothelial NO synthase (14) and Raf-1 kinase (15) suggest that it might be involved in the regulation of many cellular processes other than those that have been identified. Thus, it appears that numerous novel targets remain to be discovered.
Recently, Sprenkle et al. (15) identified the principal Raf-1 Ser 621 kinase activity present in cytosolic extracts of NIH-3T3 cells as AMPK by analyzing cytosolic fractions for Ser 621 peptide kinase activity. They also demonstrated that AMPK phosphorylated Ser 621 of Raf-1, which was expressed in Escherichia coli or Sf9 insect cell (15). The Ser/Thr Raf-1 kinase is a key intermediate in the transduction of growth factor signals from the cell membrane to the nucleus (16,17). Activation of receptor tyrosine kinases stimulates the small GTP-binding protein Ras, which interacts with cytoplasmic inactive Raf and recruits it to the plasma membrane where further activation steps occur (18,19). Activated Raf in turn phosphorylates and activates MAPK/Erk kinase 1 (MEK1), which phosphorylates and stimulates mitogen-activated protein kinases/extracellular signal-regulated kinases (MAPKs/Erks). This pathway (Ras 3 Raf 3 MEK 3 Erk) has been known to play a significant role in the transmission of cellular proliferation and developmental signals (20). Although the exact regulatory mechanisms for Raf-1 activity are not fully understood despite extensive investigations, they include phosphorylation of the enzyme (17). The role of Raf-1 Ser 621 phosphorylation has been controversial. Ser 621 was reported to be a constitutive phosphorylation site, and the mutation of this site to alanine resulted in a total loss of the kinase activity (21). On the other hand, it was also claimed that phosphorylation of Ser 621 by cAMP-dependent protein kinase (PKA) confers negative regulation (22).
In the present study, we have taken a pharmacological and molecular approach to determine whether AMPK plays a role in Raf-1-involved signaling pathways. Our results suggest that AMPK can attenuate Ras activation induced by IGF-1 or EGF and that AMPK can also further stimulate Ras-independent Raf activation induced by EGF. Phosphorylation of Raf-1 Ser 621 , however, was not involved in the AMPK-mediated regulation of Erk cascades. To our knowledge, this is the first report demonstrating that AMPK is involved in the regulation of growth factor-induced multiple signaling pathways.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium (DMEM), DMEM/ Ham's F-12 medium, and the other cell culture products were purchased from Life Technologies, Inc. Forskolin and PD098059 were obtained from Calbiochem. EGF and IGF-1 were from Calbiochem and Sigma, respectively. 5-Aminoimidazole-4-carboxamide-1-␤-D-ribofuranoside (AICAR) and other chemicals were from Sigma. [␥- 32  Antibodies and DNA Constructs-The anti-dual phospho-specific antibody that recognizes the active Erk1/2 was from New England Biolabs, Inc. Erk1/2 antibody that recognizes the total Erk1/2 regardless of their phosphorylation was from Transduction Laboratories. Antibodies for Raf-1 (C-12), EGF receptor (EGFR, 1005), IGF-1 receptor ␤ subunit (IGF-1R, C-20), Grb2 (C-23), phosphotyrosine (PY20), and c-Myc (9E10) were obtained from Santa Cruz Biotechnology Inc. Phosphospecific Raf Ser 621 antibody was kindly provided by Dr. Andrey S. Shaw (Center for Immunology and Department of Pathology, Washington University, St. Louis, MO). The AMPK ␣1-specific antibody and the AMPK pan-␣ antibody were kindly provided by Dr. Ian Salt  NIH-3T3 Cells Expressing an Antisense RNA for AMPK-cDNA corresponding to a portion of the coding region of AMPK ␣1 (amino acids 1-392) was amplified by polymerase chain reaction. It was subcloned into pcDNA 3 in such an orientation as to express an antisense RNA. The plasmid was transfected into NIH-3T3 cells using LipofectAMINE (Life Technologies, Inc.), and stable transfectants were obtained in the presence of G418. Individual clones were further isolated and examined for the effect of an antisense RNA expression.
AMPK Activity Assay-Cells were lysed with digitonin buffer (50 mM Tris-HCl, pH 7.3, 50 mM NaF, 30 mM glycerol phosphate, 250 mM sucrose, 1 mM sodium metavanadate, and 0.4 mg/ml digitonin) on ice for 2 min. AMPK was partially purified from cell lysates by adding saturated ammonium sulfate to final 35% (v/v) concentration on ice for 15 min. AMPK activity was determined as previously described (23)  without 250 M SAMS peptide (HMRSAMSGLHLVKRR) at 30°C for 10 min. The reaction was terminated by spotting the reaction mixture on phosphocellulose paper (P81), and the paper was extensively washed with 150 mM phosphoric acid. The radioactivity was measured with a scintillation counter.
Raf-1 Kinase Assay-Raf-1 kinase assay was performed essentially as described by Ziogas et al. (24) with the following modifications. After cell lysis, 500 g of protein extracts were subjected to immunoprecipitation with a Raf-1 antibody (C-12) or c-Myc antibody (9E10) coupled to protein G-agarose. After extensive washing of the immunoprecipitates, the kinase assay was performed at 30°C for 30 min in a kinase assay buffer containing the Raf-1 immune complex, 10 M ATP, 10 Ci of [␥-32 P]ATP, and 1 g of kinase defective GST-MEK (Ϫ) as a specific substrate of Raf-1. After reaction, samples were centrifuged, and the supernatants were separated on 10% SDS/polyacrylamide gel electrophoresis. Proteins were blotted on nitrocellulose membranes, and the radioactivity incorporated into the substrates was measured by Phos-phorImager. The resultant pellets were also resolved on the 10% SDS/ polyacrylamide gel electrophoresis for immunoblotting with a Raf-1 antibody to compare the amount of Raf-1 present in the immune complex.
Activated Ras Affinity Precipitation Assay-Raf Ras-binding domain (RBD) fragment (Raf-1 amino acid residues 1-149) fused to GST was purchased from Upstate Biotechnology. The fusion protein was immobilized on glutathione-agarose. The activated Ras affinity precipitation assay was performed as described according to the manufacturer's protocol. Briefly, 500 g of cell extracts were incubated with 5 g of GST-RBD complexes for 30 min at 4°C. After extensive washing of the agarose beads five times with immunoprecipitation washing buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 10 mM MgCl 2 , 1 mM EDTA, and 2% glycerol), the active Ras (Ras-GTP) bound to GST-RBD complexes was released by addition of 2ϫ SDS/polyacrylamide gel electrophoresis loading buffer. The amount of active Ras was determined by immunoblotting with an anti-pan Ras monoclonal antibody.

AMPK Inhibits Erk Activation Induced by IGF-1 but Not by EGF-To investigate a possible involvement of AMPK in the
Ras/Raf/MEK/Erk signaling pathway, dose-and time-dependent effects of AICAR on IGF-1-induced Erk1 and Erk2 activation were first examined (Fig. 1, A and B). AICAR becomes a potent activator of AMPK after its intracellular phosphorylation to AMP-mimetic AICA-ribotide (ZMP) (5). NIH-3T3 cells were serum-starved for 16 h, pretreated with AICAR, and then challenged with 50 nM IGF-1 for 10 min. In the absence of AICAR, IGF-1 rapidly stimulated phosphorylation of Erk1 (p44 MAPK) and Erk2 (p42 MAPK) when examined by immunoblotting with an antibody specific for the active biphosphorylated form of Erk1 and Erk2. The pretreatment with AICAR inhibited IGF-1-dependent Erk activation in a dose-and time-dependent manner, and a maximum effect was observed at 1 mM of AICAR for 1 h of preincubation (Fig. 1, A and B). The total amount of Erk proteins was essentially the same at each condition. The kinetics of Erk phosphorylation correlated well with the enzyme activity, which was directly measured by an immune complex using mylein basic protein as a substrate (data not shown). As expected, AMPK was activated by AICAR in a dose-and time-dependent manner, and the maximum 3-fold activation was observed in 1 h at 1 mM AICAR (Fig. 1, C and D). These results suggest that AMPK is probably involved in the negative regulation of Erk cascades. The AMPK activation conditions found in this study are consistent with other reports showing that the maximum effect of AICAR on AMPK activity or cellular processes regulated by AMPK was observed in the range of 0.5-1 mM (5)(6)(7)(8)(9)(10)(11)(12). Therefore, this pretreatment condition (1 mM AICAR/1 h of incubation) was used throughout the present study.
Because of the lack of molecular approaches for manipulating AMPK activity or a specific inhibitor, AICAR has been widely used to demonstrate a role of AMPK in various cell lines and tissues (5-12). However, AMPK-independent effects of AICAR were also reported in some cases (25,26). Thus, to substantially demonstrate the role of AMPK in Erk cascades, we established NIH-3T3 cells stably expressing an antisense RNA for the catalytic core region of AMPK catalytic ␣1 subunit, a predominantly expressed isoform in these cells, as described under "Experimental Procedures." Two isoforms of the AMPK catalytic subunit (␣1 and ␣2) were identified, and ␣1 is ubiquitously expressed, whereas ␣2 is highly expressed in skeletal muscle, heart, and liver (27,28). Both isoforms show an identical molecular mass (63 kDa) and have 90% amino acid sequence identity within the catalytic core (28). For simplicity, we use anti-␣1 to refer to NIH-3T3 cells expressing an antisense RNA. The immunoblot analysis using either an AMPK ␣1-specific antibody or an AMPK pan-␣ antibody that recognizes both ␣1 and ␣2 revealed that the expression level of AMPK ␣ was substantially reduced in anti-␣1 cells compared with the vectortransfected control cells ( Fig. 2A). The reduced expression level of the ␣ subunit resulted in a ϳ30 -40% decrease in the basal and the AICAR-stimulated AMPK activity (Fig. 2B). When anti-␣1 cells were pretreated with AICAR and then challenged with IGF-1, the inhibitory effect of AICAR on Erk activation was significantly diminished (Fig. 2C). These results indicate that the effect of AICAR is indeed mediated by AMPK.
In contrast to IGF-1, AICAR did not block EGF (100 ng/ml)induced Erk activation, whereas PD098059, a specific inhibitor of MEK, almost completely inhibited Erk activation by two growth factors (Fig. 3A). One explanation for the selective effects of AICAR is that EGF could suppress the AICAR-stimulated AMPK activity. To this end, the effects of IGF-1 and EGF on AMPK activity were examined (Fig. 3B). However, the results revealed that these two growth factors practically had no effect on the basal and the AICAR-stimulated AMPK activities (Fig. 3B), supporting the general perspective that AMPK system is stress-sensitive but not hormone-sensitive (1-3). Therefore, these results suggest that the differential effects of AICAR are due to the intrinsic differences between signal pathways leading to Erk activation by IGF-1 and EGF.
Next, to determine whether AMPK is involved in the regulation of Erk pathways in other cell types, we examined the AICAR effects in the COS-7, H9c2 cardiomyotube, and 3T3-L1 preadipocyte cell lines (Fig. 4). Although these cells showed slight differences, IGF-1-induced Erk activation was inhibited, whereas EGF-induced Erk activation was not affected by AICAR treatment in all cell lines tested (Fig. 4). Therefore, the effect of AMPK on Erk cascades is likely to be quite a widespread phenomena rather than certain cell type-specific.

DNA Synthesis Induced by IGF-1 Was Significantly Inhib-
ited by AICAR Pretreatment-Erk cascades are among the most intensively studied signal transduction systems, and they have shown to participate in a diverse array of cellular activities such as cell proliferation, development, cell survival, and death (20). To investigate a physiological significance of the AMPKmediated Erk regulation, the effect of AICAR on the growth factor-induced DNA synthesis was examined. IGF-1 and EGF induced DNA synthesis ϳ2-fold after 9 h of incubation and 4-fold after 19 h of incubation (Fig. 5A). In close correlation with the Erk phosphorylation level, AICAR treatment led to ϳ70% inhibition of IGF-induced DNA synthesis after 9 h and 50% inhibition after 19 h, whereas EGF-induced DNA synthesis was relatively less affected, resulting in about 25% reduction (Fig. 5A). Furthermore, the inhibitory effect of AICAR on IGF-1-dependent DNA synthesis was distinctively diminished in the anti-␣1 cell (Fig. 5B), in which IGF-1-induced Erk activation is quite resistant to the AICAR-dependent down-regulation (Fig. 2C). Under these conditions, no cytotoxic effects were observed (data not shown). A MEK inhibitor, 50 M PD098059, almost completely blocked DNA synthesis induced by both growth factors, indicating that Erk activation is critical for cell proliferation. As shown in Fig. 3A, AICAR exerted practically no effect on the basal Erk activity (first and fourth lanes) or EGF-induced Erk activation (third and sixth lanes). Nevertheless, this treatment resulted in statistically significant inhibition of the basal as well as EGF-induced DNA synthesis (Fig. 5). Thus, besides Erk cascades, some other mechanism(s) required for DNA synthesis seems to be affected by AICAR. This possibility is considered later under "Discussion." Activation of AMPK Did Not Induce Phosphorylation of Ser 621 of Raf-1-The previous report demonstrated that AMPK can phosphorylate Ser 621 of Raf-1 in vitro (15). The sequence around Raf-1 Ser 621 exactly matches the recognition motif of AMPK (29). Because it was also demonstrated that phosphorylation of this residue by PKA leads to inhibition of Raf-1 kinase (22), we next examined the possibility that phosphorylation of Raf-1 Ser 621 is a mechanism responsible for AMPKmediated inhibition of Erk activation induced by IGF-1.
To this end, we first analyzed the effect of AICAR on the phosphorylation level of Ser 621 of Raf-1 (Fig. 6). The Myctagged Raf-wild type (mycRaf-wt) plasmid was transiently transfected in NIH-3T3 cells, and in 48 h post-transfection, cells were treated for 1 h with 1 mM AICAR or 50 M forskolin, which enhances cAMP accumulation and in turn activates The phosphorylation level and the total protein amount of Erk1 and Erk2 were determined by immunoblotting with an anti-dual phospho-specific Erk antibody (P-ERK) and an antibody for Erk1 and Erk2 regardless of their phosphorylation (ERK), respectively. The effect of AICAR on AMPK activity was also examined from the dose-dependently (C) and timedependently (D) treated cells. Partially purified AMPK was subjected to the activity assay using SAMS peptide as a substrate in the presence of 200 M AMP. The data shown represent the means Ϯ S.E. for three separate experiments in duplicate.
PKA. The expressed mycRaf-wt protein was immunoprecipitated with a c-Myc epitope-specific monoclonal antibody 9E10 and then immunoblotted with a phosphospecific Raf Ser 621 antibody. The basal phosphorylation level of Ser 621 of my-cRaf-wt was hardly detectable under our experimental conditions, and AICAR did not exert any effect on this residue, whereas forskolin markedly increased the phosphorylation level of Ser 621 (Fig. 6, upper panel). Under the identical condition, forskolin did not exert any effect on mycRafS621A, in which Ser 621 is replaced with alanine. There was essentially no difference in the amount of the immunoprecipitated mycRaf-wt and mycRafS621A (Fig. 6, lower panel). This result supports the previous finding that Ser 621 of Raf-1 can be phosphorylated by PKA (22). However, in contrast to the report describing that AMPK in vitro phosphorylated Ser 621 of the overexpressed Raf-1 from bacteria and insect cell (15), our results argue that this may not be the case in intact cells.
The Mechanism by Which Raf-1 Is Regulated by AMPK Dif-fers from the PKA-mediated One-Besides Ser 621 , Raf-1 contains multiple phosphorylation sites critical for the kinase activity (17). Although phosphorylation of Ser 621 was not induced by AICAR, we next examined whether or not Raf-1 kinase activity is directly affected by AICAR treatment (Fig. 7). NIH-3T3 cells were pretreated with AICAR and then challenged with IGF-1 (50 nM) or EGF (100 ng/ml) for 5 min. Endogenous Raf-1 was immunoprecipitated with a Raf-1 antibody, and in vitro kinase assays were performed using the recombinant kinase-defective GST-MEK (Ϫ) as a Raf-1-specific substrate. IGF-1 and EGF stimulated Raf-1 activity ϳ3-fold in 5 min, and AICAR pretreatment almost completely blocked Raf-1 activation by IGF-1, but it did not interfere with EGF-induced Raf-1 activation (Fig. 7A). This result is consistent with the Erk phosphorylation profile presented in Fig. 3A, indicating that Raf-1 is a main activator of MEK and Erk in NIH-3T3 cells. Next, Myc-tagged Raf-wild type (mycRaf-wt) and FIG. 2. Expression of an antisense RNA for the AMPK catalytic subunit decreased the AMPK activity and diminished the inhibitory effect of AICAR on IGF-1-induced Erk activation. NIH-3T3 cells expressing an antisense RNA for AMPK ␣1 (Anti-␣1) were established as described under "Experimental Procedures." A, the expression level of the AMPK ␣ subunit of control and anti-␣1 cells was compared by immunoblot analysis with AMPK ␣1-specific antibody (upper panel) or with AMPK pan-␣ antibody that recognizes both the ␣1 and ␣2 isoforms (lower panel). B, the basal and AICAR-stimulated AMPK activities of the control and anti-␣1 cell. The data shown represent the means Ϯ S.E. for two independent experiments in duplicate. C, the control and anti-␣1 cells were pretreated with or without 1 mM AICAR and then stimulated with 50 nM IGF-1 for 10 min. The phosphorylation level (P-ERK) and the total protein amount of Erk (ERK) were compared by immunoblot analysis with each antibody.

FIG. 3. Selective effects of AICAR on Erk activation induced by IGF-1 and EGF.
A, NIH-3T3 cells were pretreated with or without 1 mM AICAR or 50 M PD098059 for 1 h and then stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 10 min. The phosphorylation level of active Erk (P-ERK) and the total amount of Erk (ERK) in each sample were measured. B, NIH-3T3 cells were pretreated for 1 h with or without 1 mM AICAR. After incubation with IGF-1 (50 nM) or EGF (100 ng/ml) for 10 min, AMPK activity was examined. The data shown represent the means Ϯ S.E. for three separate experiments in duplicate.

FIG. 4. Similar effects of AICAR on Erk activation were observed in several cell lines.
The effect of AICAR on Erk activation was examined in H9c2 cardiomyotubes, 3T3-L1 preadipocytes, and COS-7 cells. Each cell line was pretreated with or without 1 mM AICAR for 1 h and then stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 10 min. Erk activities were determined by immunoblotting with an antiphosphospecific Erk antibody (P-ERK). mycRafS621A plasmids were transiently transfected in NIH-3T3 cells, and the effects of AICAR and forskolin on each Raf kinase form were compared. Consistent with the results of the endogenous Raf-1 activity (Fig. 7A), AICAR pretreatment blocked IGF-induced Raf-wt activity but not EGF-induced Raf-wt activity (Fig. 7B, upper panels). In contrast, forskolin inhibited Raf-wt activation induced by EGF and IGF-1 (Fig. 7B,  upper panels, sixth and seventh lanes). RafS621A was activated about 1.7-fold by IGF-1 or EGF (Fig. 7B, lower panels), but the degree of activation was relatively mild compared with 3-fold activation of endogenous Raf or Raf-wt. Notably, RafS621A activation by IGF-1, but not by EGF, was blocked by AICAR (Fig. 7B, lower panels, fourth and fifth lanes), whereas activation of this mutant form by IGF or EGF was resistant to the PKA-mediated down-regulation (Fig. 7B, lower panels, sixth  and seventh lanes). The effects of forskolin on Raf-wt and RafS621A activation together with the results shown in Fig. 6 support the previous finding that Raf Ser 621 serves as a nega-tive regulation site by PKA phosphorylation (22). Therefore, these results indicate that the mechanism by which Raf-1 is regulated by AMPK should differ from the PKA-mediated one.
Under our experimental conditions, the basal activity of RafS621A was partially active showing about 30% of the Raf-wt activity in contrast to the previous report demonstrating that the RafS621A mutant is completely inactive as a kinase (21). Although we cannot completely rule out the possibility that the observed kinase activity of RafS621A is due to nonspecific protein kinases contaminated in an immune complex, this possibility seems unlikely because c-Myc antibody immunoprecipitates of vector-transfected cell lysates did not show any activity exceeding the background level (data not shown). Rather the discrepancy may be generated by experimental conditions because the other mutant form, RafS621D, was reported to retain a low inducible kinase activity when expressed in Sf9 insect cells (22).
Ras-independent Raf-1 Activity Induced by EGF Is Further Stimulated by AICAR Treatment-Ras is the best characterized upstream activator of Raf-1, but recent studies demonstrate that Raf-1 is also activated via Ras-independent mechanisms (18, 19, 24, 30 -32). Because it is also known that EGF-induced Erk activation can occur via a Ras-independent as well as a Ras-dependent pathway (22,33), we investigated the nature of the differential effects of AICAR on IGF-1-and EGF-induced Raf activation using mycRafR89L, which is defective for Ras binding (Fig. 8A). EGF indeed activated RafR89L about 2-fold (Fig. 8A, first and third lanes), whereas IGF showed very little effect (Fig. 8A, first and second lanes), indicating that EGF signal can be transduced in a Ras-independent pathway. In agreement with the earlier report (22), forskolin still blocked EGF-induced RafR89L activation (Fig.  8A, third and seventh lanes), indicating that Raf-1 is a direct target of PKA-mediated down-regulation. Unexpectedly, AICAR further stimulated the EGF-induced RafR89L activity by 50% (Fig. 8A, third and fifth lanes), suggesting that Rasindependent Raf-1 activation by EGF is positively regulated by AICAR treatment. Additional evidence supporting this suggestion was obtained by a cotransfection experiment with Raf-wt and a dominant negative RasS17N (Fig. 8B). RasS17N expression, which blocks Ras-mediated signal transmission, resulted in a drastic reduction of IGF-1-dependent Raf-wt activation (Fig. 8B, second and sixth lanes), but it exerted a relatively less effect on EGF-dependent Raf-wt activation (Fig. 8B, third and  seventh lanes). AICAR treatment clearly stimulated the EGFdependent Raf-wt activity that was partially blocked by RasS17N (Fig. 8B, seventh and ninth lanes). Taken together, these data demonstrate that IGF-1 transmits its signal to Raf-1 predominantly via Ras, whereas the EGF signal can be transduced in a Ras-independent as well as in a Ras-dependent manner. Furthermore, our data suggest that Ras-independent Raf-1 activation by EGF is up-regulated by AMPK. At this point, this result seems very contradictory because the endogenous Raf-1 or mycRaf-wt activation by EGF was not affected by AICAR (Fig. 7). Thus, we suspected that a compensatory mechanism such as attenuation by AMPK of Ras-dependent Raf-1 activation by EGF would exist.
AMPK Inhibits Ras Activation without Attenuating the Re-ceptor Tyrosine Kinase Activity-To investigate the effect of AICAR on Ras activity, we used the commercially available Ras activity assay kit (upstate biotechnology), which is based on the fact that only activated Ras (Ras-GTP) can bind to the RBD of Raf-1 (34,35). Active GTP-bound Ras was pulled down from cell lysates with the GST-Raf-RBD coupled to glutathione agarose, and the fraction of activated Ras was determined by immunoblotting with a Ras antibody (Fig. 9). EGF and IGF-1 rapidly activated Ras within 3 min; pretreatment of AICAR almost completely blocked this activation, and this effect of AICAR on Ras was distinctively diminished from the anti-␣1 cells, indicating again that its effect is mediated by AMPK (Fig.  9). This result together with the data in Fig. 8 illustrates that two distinct signaling pathways originated from an identical ligand (EGF) are differentially affected by AICAR. Therefore, the receptor itself is not likely to be directly affected by AICAR. Thus, to test this possibility, we examined the effect of AICAR on phosphorylation of EGFR and IGF-1 receptor ␤ subunit (IGF-1R) (Fig. 10); EGF and IGF-1 start transmitting signals by binding to their receptors, and this interaction causes autophosphorylation of receptors on tyrosine residues. These receptors were immunoprecipitated with each receptor antibody and immunoblotted with an antibody specific for phosphotyrosine. The results revealed that growth factor-enhanced phosphorylation of EGFR and IGF-1R on tyrosine residues was hardly affected by AICAR, and these results were reconfirmed by immunoblotting with the receptor antibody following immunoprecipitation with the phosphotyrosine-specific antibody (Fig.  10). Additionally, we examined the effect of AICAR on the interaction between the receptor and growth factor receptorbound protein 2 (Grb2) (Fig. 10). Grb2 forms a complex with the guanine nucleotide-exchanging factor Sos in the cytoplasm. Once the EGFR or IGF-1R is activated by autophosphorylation, Grb2-Sos complex is recruited to the receptor and then catalyzes the formation of the active GTP-bound form of Ras (36). Our result revealed that the receptor-Grb2 interaction stimu- FIG. 7. AICAR treatment blocked the Raf-1 activation induced by IGF-1 but not by EGF. A, NIH-3T3 cells pretreated with or without 1 mM AICAR for 1 h were stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 5 min. Then the endogenous Raf-1 kinase was immunoprecipitated with a Raf-1 antibody, and in vitro kinase activity was measured using kinase inactive recombinant GST-MEK (Ϫ) as a substrate. B, NIH-3T3 cells transfected with Myc-tagged Raf-wt (upper panels) or RafS621A plasmid (lower panels) were pretreated with or without 1 mM AICAR or 50 M forskolin for 1 h and then stimulated with IGF-1 (50 nM) or EGF (100 ng/ml) for 5 min. Each expressed Raf form was immunoprecipitated with a c-Myc antibody, and the Raf kinase activity assay was performed. The amounts of immunoprecipitated endogenous Raf-1 (A), mycRaf-wt, and mycRafS621A (B) were detected by immunoblotting with a Raf-1 antibody as indicated by an arrow. The kinase activity was normalized for the amount of Raf-1 protein present in each immune complex and expressed as a fold induction of the basal activity. Experiments were repeated 2-4 times with similar results, and a representative result is shown.

FIG. 8. Ras-independent Raf-1 activation by EGF was further induced by AICAR.
A, NIH-3T3 cells transfected with mycRafR89L plasmid were identically treated as described in Fig. 7B legend, and mycRafR89L kinase activity was compared at each condition. B, NIH-3T3 cells were cotransfected with mycRaf-wt and RasS17N at 1:3 ratio of DNA concentration. These cells were pretreated with or without 1 mM AICAR for 1 h and then stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 5 min, and Raf-wt activity was determined. The arrow in the lower panel indicates the amount of mycRafR89L (A) or Raf-wt (B) present in each immune complex. The experiments were repeated twice with similar results, and a representative result is shown. lated by each growth factor was not affected by AICAR, either (Fig. 10). Consequently, our data suggest that AMPK can down-regulate the growth factor-induced Erk pathway at the Ras level. In addition, AMPK is likely to up-regulate a Rasindependent pathway leading to Raf-1.

DISCUSSION
In the present study, we have investigated the effects of AMPK activation on the Erk pathway induced by two different growth factors, IGF-1 and EGF. Pretreatment of AICAR, an activator of AMPK, drastically inhibited the Ras activation induced by IGF-1 or EGF (Fig. 9). This inhibitory effect of AICAR was distinctively diminished in NIH-3T3 cells in which AMPK activity was suppressed by expression of an antisense RNA for the AMPK catalytic subunit (Figs. 2 and 9). Therefore, these observations illustrate that the observed effects of AICAR in the present work are indeed mediated by AMPK. Because IGF-1 or EGF receptor activity was not affected by AICAR (Fig. 10), our results suggest that AMPK can negatively regulate Erk pathway at Ras level. In contrast to the previous report (15), phosphorylation of Raf-1 Ser 621 was not involved in the AMPK-mediated regulation of the Erk pathway (Fig. 6) We have identified another signaling pathway that seems to be positively affected by AMPK. In case of IGF-1-induced Erk cascades, Ras-dependent pathway is predominant as shown in Fig. 8, and down-regulation of Ras activity was directly reflected to the downstream effector Raf-1 and Erk activity (Figs.  1 and 7). Although Ras activation induced by EGF was also inhibited by AICAR, EGF-dependent Raf-1 and Erk activation was not attenuated under this condition (Figs. 3A and 7). Our results revealed that EGF transmits its signal to Raf-1 kinase through Ras-independent as well as Ras-dependent pathways (Fig. 8), and we also demonstrated that the Ras-independent pathway leading to Raf-1 was further stimulated by AICAR (Fig. 8). As a consequence of these two compensatory effects, the endogenous Raf-1 and the subsequent Erk activation induced by EGF appear to be not affected by AICAR treatment (Figs. 3A and 7).
The activity of AMPK with the cell is tightly regulated by the AMP:ATP ratio. Because of adenylate kinase, which maintains its reaction (2ADP 7 ATP ϩ AMP) close to equilibrium, the AMP:ATP ratio varies as the square of the ADP:ATP ratio, and this makes the AMP:ATP ratio a very sensitive indicator of cellular energy charge (reviewed in Ref. 1). In cells with sufficient energy supply, the AMP:ATP ratio is very low, and AMPK is inactive. However, any cellular stress, which causes ATP depletion, can lead to a dramatic increase of the AMP:ATP ratio caused by adenylate kinase reaction and the subsequent activation of AMPK in even fully energized cells. Ras is a key regulator of cell growth in all eukaryotic cells. Thus, downregulation of Ras by AMPK could represent the mechanism for inhibiting cell proliferation when proliferating cells are exposed to stresses causing ATP depletion. The fact that the yeast homologue of AMPK is involved in regulating the gene expression and the cell cycle supports this possibility (2). To reveal the physiological significance of the AMPK-mediated down-regulation of Ras/Erk pathway, the effect of AMPK activation on DNA synthesis was only examined in the present study (Fig. 5). However, the role of AMPK may not be limited to the control of cell proliferation because Ras has been implicated in a variety of cellular activities including differentiation, development, and cell death (37). Furthermore, our data suggest that AMPK is likely to positively regulate the Ras-independent pathway, which also plays an important role in cells. For example, in mitotic cells, the Ras-independent mechanism results in a cytoplasmic active Raf-1 (24), and interferon activation of Raf-1 is mediated by Janus tyrosine kinase-1 independently of Ras (30). In case of EGF signaling, protein kinase C or calcium influx was implicated in Ras-independent pathway for Erk regulation (33); members of protein kinase C family have been also widely implicated in the regulation of cell growth and differentiation (38). Therefore, the possibility that AMPK is involved in the regulation of various cellular processes in addition to cell proliferation deserves further investigation.
Besides the known stress conditions, AMPK would be differentially regulated in a multitude of cellular processes such as proliferation, differentiation, and cell death because changes in the intracellular level of ATP and AMP occur in these processes as well. For example, an altered level of phosphometabolites including AMP and ATP has been constantly observed among differentiated cells, proliferating cells, and tumor cells (39). In addition, intracellular ATP levels are known to be a determinant of cell death fate by apoptosis or necrosis (40). Thus, FIG. 9. Effects of AICAR on Ras activation induced by IGF-1 and EGF. The control NIH-3T3 cells and anti-␣1 cells were incubated with or without 1 mM AICAR for 1 h and then stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 3 min, respectively. The fraction of active Ras was determined by affinity precipitation assay as described under "Experimental Procedures," and the Ras activity at each condition was expressed as a fold induction of the basal activity. Experiments were repeated 2-4 times with similar results, and a representative result is shown.
FIG. 10. IGF-1 and EGF receptor kinase activity was not affected by AICAR treatment. NIH-3T3 cells pretreated or untreated with AICAR were stimulated with 50 nM IGF-1 or 100 ng/ml EGF for 3 min, respectively. IGF-1 receptor ␤ subunit and EGF receptor were immunoprecipitated with an antibody for each receptor (IP: IGF-1R and IP: EGFR) followed by immunoblotting with an antibody specific for phosphotyrosine (IB: P-Y). The amount of each receptor in the immune complex was simultaneously determined by immunoblotting with each receptor antibody (IB: IGF-1R, EGFR). The experiment by an opposite way showed a similar result: immunoprecipitation with a phosphotyrosine antibody (IP: P-Y) followed by immunoblotting with each receptor antibody. Under the identical experimental condition, the association of Grb2 with each receptor was also examined by immunoprecipitation with Grb2 antibody (IP: Grb2) followed by immunoblotting with each receptor antibody. although further studies are required, the current implications of AMPK in the regulation of Ras activity and Ras-independent signaling pathways suggest that AMPK may play an important role in coupling the varying energy status of the cell to the regulation of diverse cellular processes.
As an attempt to reveal the underlying mechanisms for the AMPK-mediated regulation of signaling molecule(s), it would be of interest to determine whether AMPK is localized in plasmalemmal caveolae, small membrane invaginations present in most cells of higher eukaryotes. These membrane specializations are known to function as signal transduction organizing centers and compartmentalize a subset of signal transducing molecules including G-protein ␣ subunit, Ras, Src family tyrosine kinase, eNOS, EGFR, and protein kinase C isoforms (41). A portion of AMPK is membrane-bound probably because of the N-terminally myristoylated ␤1 subunit of AMPK (42), which serves as a scaffold protein for the formation of the trimeric complex with the ␣ and ␥ subunits. In addition, it was demonstrated that AMPK coimmunoprecipitated with eNOS (14), which is rich in caveolae. Thus, a signal module that couples the localized recognition of intracellular energy charge with the regulation of signaling molecules might exist in caveolae.
Although the antisense RNA experiments support the possibility that the AICAR effects are mediated by AMPK (Figs. 2, 5, and 9), it should be carefully examined whether or not the AMPK-independent effects of AICAR still interfere with and complicate the interpretation of our data. AICAR could affect de novo purine synthesis because ZMP is a natural intermediate in the pathway (26). In fact, AICAR was reported to inhibit the cell growth of Chinese hamster fibroblasts, and this was associated with a reduction of phosphoribosyl pyrophosphate and an increase in purine nucleotides (43). Thus, these subsidiary effects might be responsible for the partial inhibition by AICAR of the basal and EGF-dependent DNA synthesis in an Erk-independent manner (Fig. 5). However, the observation that IGF-1-induced DNA synthesis was more drastically inhibited under the identical condition cannot be explained solely by the derangement of purine synthesis caused by AICAR, and moreover, IGF-1-dependent DNA synthesis of anti-␣1 cells was hardly inhibited by AICAR (Fig. 5). Therefore, the AMPKmediated down-regulation of Ras/Erk pathway must be more critical for the AICAR-dependent inhibition of IGF-1-induced DNA synthesis as demonstrated in our study. Furthermore, similar effects of AICAR on EGF-and IGF-1-dependent Erk activation were observed from several different cell types (Fig.  4), which have the remarkable different rates of purine metabolism. This further supports the possibility that the AICAR effects observed in the present work are mediated by AMPK rather than caused by the secondary effects of AICAR on the purine biosynthesis pathway.
More relevant to our study, a recent report shows that a combinatorial action of 5Ј-AMP and small heat shock proteins mediate the inhibitory effect on Src kinase in terminally differentiated adult cardiomyocytes (44). Raf-1 activation can be mediated via tyrosine phosphorylation by Src (45) or family tyrosine kinase such as Lck (46). In addition, the oncogenic form of Src and Ras can synergize in phosphorylation and activation of Raf-1 (45,47). Because ZMP acts as 5Ј-AMP analogue, down-regulation by AICAR of IGF-1-dependent Raf activation could be interpreted as a result of ZMP-mediated inhibition of Src without involvement of AMPK. However, this assumption is not in line with our observation that EGF-dependent Raf-1 activation was not blocked by AICAR (Fig. 7). Src family tyrosine kinase can also interact with growth factor receptors including EGFR and IGF-1R (48). These interactions are known to be bi-directional; Src can bind, phosphorylate, and activate the receptor and vice versa (48). Under our experimental conditions, however, EGF and IGF-1 receptor activities were not affected by AICAR (Fig. 10). Therefore, inhibition of Src by AMP-mimetic ZMP is not likely to be engaged in our system. Moreover, the inhibitory effect of 5Ј-AMP on Src was predominantly observed in heart tissue (44).
For optimal AMPK activity, the formation of a trimeric subunit complex is necessary (49), and this has in part hindered the current progress in transfection studies aimed at revealing the downstream effects of AMPK. However, the molecular approaches that overcome this technical difficulty are currently under development (50,51). Thus, we expect that these approaches will markedly enhance our understanding of the physiological roles of AMPK within a few years.