Cyclic AMP inhibits extracellular signal-regulated kinase and phosphatidylinositol 3-kinase/Akt pathways by inhibiting Rap1.

Cyclic AMP inhibited both ERK and Akt activities in rat C6 glioma cells. A constitutively active form of phosphatidylinositol 3-kinase (PI3K) prevented cAMP from inhibiting Akt, suggesting that the inactivation of Akt by cAMP is a consequence of PI3K inhibition. Neither protein kinase A nor Epac (Exchange protein directly activated by cAMP), two known direct effectors of cAMP, mediated the cAMP-induced inhibition of ERK and Akt phosphorylation. Cyclic AMP inhibited Rap1 activation in C6 cells. Moreover, inhibition of Rap1 by a Rap1 GTPase-activating protein-1 also resulted in a decrease in ERK and Akt phosphorylation, which was not further decreased by cAMP, suggesting that cAMP inhibits ERK and Akt by inhibiting Rap1. The role of Rap1 in ERK and Akt activity was further demonstrated by our observation that an active form of Epac, which activated Rap1 in the absence of cAMP, increased ERK and Akt phosphorylation. Inhibition of ERK and/or PI3K pathways mediated the inhibitory effects of cAMP on insulin-like growth factor-I (IGF-I) and IGF-binding protein-3 gene expression. Moreover, cAMP, as well as ERK and PI3K inhibitors produced equivalent stimulation and inhibition, respectively, of p27(Kip1) and cyclin D2 protein levels, potentially explaining the observation that cAMP prevented C6 cells from entering S phase.

. The growth-inhibitory effects of cAMP are proposed to be mediated at least in part through cAMP-dependent inactivation of MAPK, also known as ERK1/2 (6 -9). ERK1/2 are phosphorylated and activated by MEK-1 and -2, which are phosphorylated and activated by members of the Raf family of protein kinases (Raf-1, A-Raf, and B-Raf) (10). The activities of Raf kinases are regulated by the Ras family of small GTP-binding proteins, including Ras and Rap1 (11). Three different pathways have been suggested to be involved in the inhibition of ERK by cAMP: 1) cAMP-activated PKA phosphorylates Rap1, which induces Rap1 GTP-binding by an unknown mechanism, resulting in competition with Ras for the binding of Raf-1 and thereby blocking Ras-induced Raf-1 activation (9,12); 2) phosphorylation of Raf-1 directly by PKA at two serine residues inhibits both Ras binding and Raf-1 activity (13)(14)(15); 3) cAMP inhibits ERK by inhibiting B-Raf using an unknown mechanism (16), although in other studies, cAMP is reported to activate B-Raf through PKA/Rap1, which leads to ERK activation (12,17).
Cyclic AMP and trophic hormones in which the second messenger is cAMP activate PI3K, Akt, and/or p70 S6 kinase in thyroid cells and ovarian granulosa cells, which may mediate the cAMP-induced growth of these cells (18 -20). Cyclic AMP activates Akt through a PI3K-dependent mechanism (20,21). Moreover, activation of Rap1 is suggested to mediate cAMPinduced Akt activation (21). Inhibition of PI3K/Akt pathways by cAMP has been reported in Swiss 3T3, HEK293, COS, and Rat2 cells, although the mechanism(s) by which this occurs is largely uncharacterized (22).
Rat C6 glioma cells are one of the well established glioma cell lines used for a variety of studies related to glioma cell biology (23). Our previous study showed that cAMP inhibits both C6 cell growth and expression of the autocrine growth factor IGF-I (24). The addition of exogenous IGF-I peptide to cAMP-treated C6 cells only partially prevents the decrease in cell growth caused by cAMP treatment, suggesting that cAMP inhibits C6 cell growth by another mechanism(s) in addition to down-regulation of IGF-I gene expression. It has been reported that cAMP inhibits ERK activity in both primary astrocytes and C6 cells (6,16,25). This inactivation of ERK has been proposed to mediate the inhibitory effect of cAMP on astrocyte or C6 cell growth, although as described above, the mechanism(s) by which cAMP inhibits ERK remains to be clarified. In this study, we showed that cAMP inhibited ERK and PI3K/Akt pathways by inhibiting Rap1 activity in C6 cells. Neither PKA nor Epac (Exchange protein directly activated by cAMP) was involved in the cAMP-dependent inactivation of ERK and Akt. Moreover, the inhibition of these two pathways contributes to cAMP effects on cellular gene expression and growth.

EXPERIMENTAL PROCEDURES
Cell Culture-Rat C6 glioma cells were obtained from ATCC (Manassas, VA) and maintained in Ham's F12 medium (Life Technologies, * This study was supported by Grant DK-47357 from NIDDK, National Institutes of Health, Grant AQ-1385 from the Robert A. Welch Foundation, and Grant 07 from the Children's Cancer Research Center of University of Texas Health Science Center at San Antonio (to M. L. A.) and by Grant DK-56166 (to F. L.) from NIDDK, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Plasmids-The pcDNAIII KZ-HA and pcDNAIII KZ-HA ERK2 were kind gifts of Dr. J. Silvio Gutkind (National Institutes of Health, Bethesda, MD). pcDNA3.1/Myc-His(ϩ)A was purchased from Invitrogen (Carlsbad, CA), and pcDNA3.1/Myc-Akt was described previously (26). A vector containing p110*, a constitutively active form of PI3K, was provided by Drs. Anke Klippel and Lewis T. Williams (Chiron Corp., Emeryville, CA). The empty vector for p110* (p110 vector) was generated by releasing the p110* insert from the XbaI and BamHI sites of the vector. PKA-REVab, a dominant negative form of PKA, PKA-CEV␣, an active form of PKA, and their parental vector Zem3 were kind gifts from Dr. G. Stanley McKnight (University of Washington, Seattle, WA). An active form of Epac (Epac⌬cAMP), its parental vector pmt2-HA, and RalGDS GST-RBD constructs were kindly provided by Dr. Johannes L. Bos (University Medical Center, Utrecht, Netherlands). The pcDNA3 was from Invitrogen, and Flag-Rap1 and Flag-Rap1GAP1 were generous gifts from Dr. Philip J. Stork (Oregon Health Sciences University, Portland, OR).
Transient Transfection-C6 cells were plated in 60-mm plates (Corning, Corning, NY) at 5 ϫ 10 5 cells/60-mm plate and were grown for 3 days in complete medium. Transient transfection was then performed using 2 g of the expression vector(s) or its respective parental vector, with the LipofectAMINE Plus system (Life Technologies, Inc.) in Opti-MEM medium (Life Technologies, Inc.). Three hours after transfection, Opti-MEM medium was replaced with Ham's F-12 medium containing 1% FBS. After 24 h, 100 M 8-CPT-cAMP was added into the conditioned medium. Cells transfected with HA-ERK2 were then harvested after 30 min, and cells transfected with Myc-Akt were harvested after 3 h.
Cell Extract, Immunoprecipitation, and Western Immunoblot-Cells were lysed in lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 50 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 25 g/ml aprotinin, 25 g/ml trypsin inhibitor, 25 g/ml leupeptin, and 2 mM ␤-glycerol-phosphate for 30 min on ice. Lysates were then centrifuged at 14,000 ϫ g for 20 min, and supernatants were used for immunoprecipitation or Western immunoblot. For immunoprecipitation studies, cell lysates were incubated with specific antibodies precoupled to protein A-Sepharose overnight at 4°C with gentle rotation. After incubation, immunoprecipitates were collected and washed twice with wash buffer containing 50 mM Hepes, pH 7.6, 150 mM NaCl, and 0.1% Triton X-100. Proteins bound to the beads were eluted by heating at 95°C for 5 min in Laemmli buffer. Immunoprecipitates or cell lysate proteins were separated by SDS-PAGE and transferred electrophoretically to Immobilon-P membrane (Millipore, Bedford, MA). The membrane was blocked in TTBS buffer (20 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.1% Tween 20) with 5% nonfat dry milk. The blots were then incubated with the appropriate primary and secondary antibody, with each incubation followed by washing in TTBS. Proteins in the membrane were detected by exposure to film after incubation with Supersignal West Pico chemiluminescent substrate (Pierce).
Akt in Vitro Kinase Assay-Two hundred g of cell lysate proteins were immunoprecipiated with anti-Akt antibody precoupled to protein A-Sepharose overnight at 4°C with gentle rotation and then washed twice with wash buffer and once with reaction buffer containing 20 mM Hepes, pH 7.4, 10 mM MgCl 2 , 10 mM MnCl 2 , and 1 mM dithiothreitol. The precipitates were resuspended in 30 l of reaction buffer with 3 g of histone H2B, 2 M ATP, and 5 Ci of [␥-32 P]ATP (6000 Ci/mmol). The reaction was carried out for 15 min at room temperature and stopped by the addition of Laemmli buffer. The reaction products were heated at 95°C for 5 min and resolved by SDS-PAGE, followed by autoradiographic visualization of 32 P-labeled H2B.
In Vivo Rap1 Activation Assays-Activated Rap1, i.e. Rap1GTP, was isolated from cell lysate proteins using a protocol provided by Dr. Johannes L. Bos, which was adapted from Ref. 27. The RalGDS GST-RBD fusion protein, which contains a binding site (RBD) for Rap1GTP, Phospho-ERK1/2 (pERK1 and pERK2) or total ERK1/2 protein levels were measured by Western immunoblot using anti-phospho-ERK antibody with 20 g of cell lysate proteins or anti-ERK1 antibody, which has a low affinity for ERK2, with 2 g of cell lysate proteins, respectively. All of the experiments were repeated at least three times with similar results.

FIG. 2. Cyclic AMP inhibits Akt phosphorylation in C6 cells.
A, confluent C6 cells were treated with (ϩ) or without (Ϫ) 100 M 8-CPT-cAMP for the indicated times. 0 represents the level before treatment. B, confluent C6 cells were treated with (ϩ) or without (Ϫ) 100 M 8-CPT-cAMP for the indicated times. C, confluent C6 cells were treated without (Control) or with 100 M 8-CPT-cAMP, 10 M forskolin, or 1 M isoproterenol for 3 h. Levels of Akt phosphorylated (pAkt) at Ser-473 or total Akt protein levels were measured by Western immunoblot using anti-phospho-Akt(Ser-473) antibody or anti-Akt antibody, respectively, with 20 g of cell lysate proteins. Levels of Akt phosphorylated at Thr-308 were measured by immunoprecipitating 600 g of cell lysate proteins with anti-Akt antibody followed by Western immunoblot using anti-phospho-Akt(Thr-308) antibody. Akt kinase activity was determined using H2B as substrate as described under "Experimental Procedures"; the results are shown on the bottom autoradiograph in C. All of the experiments were repeated at least three times with similar results. was expressed in bacteria by growing them to an A 600 between 0.6 and 1 followed by induction of protein expression with 1 mM isopropyl-1thio-␤-D-galactopyranoside for 3 h. The bacteria were spun down and lysed in bacteria lysis buffer (9/400 of the original culture volume) containing 50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 2 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 1 g/ml leupeptin, 10 g/ml trypsin inhibitor, 10% glycerol, and 1% Nonidet P-40 with sonication 8 to 10 times for 30 s each. The bacterial lysates were then spun down, and 150 l of the supernatant was incubated with 20 l of glutathione-agarose beads at 4°C for 1 h with gentle rotation. Beads containing bound RalGDS GST-RBD were then pelleted and washed four times with ice-cold phosphate-buffered saline. C6 cell lysate proteins were then incubated with these glutathione-agarose beads coupled to RalGDS GST-RBD fusion protein for 1 h at 4°C with gentle rotation. Beads were pelleted and washed four times with lysis buffer, and proteins were eluted from the beads with Laemmli buffer and detected by Western immunoblot using anti-Flag or anti-Rap1 antibodies for transfected Flag-Rap1 or endogenous Rap1 proteins, respectively.
Total RNA Extraction and RNase Protection Assays-Total RNA was prepared using the Ultraspec reagent (Tel-Test, Inc., Friendswood, TX). RNA concentrations were determined using the absorbance at 260 nm. Antisense RNA probes were labeled and synthesized using the protocol described previously (28) with reagents from Promega (Madison, WI) and Ambion (Austin, TX), and [␣-32 P]UTP (800 Ci/mmol). Solution hybridization/RNase protection assays were conducted using the protocol described previously (28) with reagents supplied by Ambion. The antisense RNA probes used in this study to measure levels of rat IGF-I, IGFBP-3, and ␤-actin mRNAs were described previously (24,29).
DNA Synthesis-C6 cells were plated in 48-well cell culture clusters (Corning) at 5 ϫ 10 4 cells per well in complete Ham's F-12 medium and were cultured for 24 h followed by another 24-h incubation in Ham's F-12 medium with 1% FBS in place of 10% FBS. Cells were then treated with 8-CPT-cAMP, PD98059, or LY294002 in Ham's F-12 medium containing 1% FBS. Twenty-four hours later, 1 Ci of [methyl-3 H]thymidine was added into each well. After a 4-h incubation, C6 cells were harvested for scintillation counting as described previously (24).
Flow Cytometry-C6 cells were plated in 100-mm plates (Corning) at 1.4 ϫ 10 6 cells/100-mm plate and grown for 24 h in complete medium followed by a 24-h incubation in Ham's F-12 medium with 1% FBS in place of 10% FBS. Cells were then treated with or without 100 M 8-CPT-cAMP. Twenty-four hours later, cells were trypsinized, washed twice with ice-cold phosphate-buffered saline, and fixed in 70% ethanol overnight at Ϫ20°C. Cells were then pelleted, washed twice with icecold phosphate-buffered saline, and treated with 250 g/l RNase A (DNase-free, Sigma) for 30 min at room temperature. Cells were stained with propidium iodide and analyzed using a fluorescence-activated cell sorting flow cytometer (FACSCalibur, Becton Dickinson Immunocytometry Systems, Inc., San Jose, CA). Cells were illuminated with 15 milliwatts of 488 nm argon-ion laser light. Propidium iodide fluorescence was read through a 625/35 nm band-pass filter. Typically, 40,000 cells were counted for each sample, and data were analyzed using the ModFit LT V2.0 DNA analysis program from Verity Software House, Inc. (Topsham, ME).

RESULTS
Cyclic AMP Inhibits Both ERK and Akt Activities-When C6 cells were treated with 100 M 8-CPT-cAMP, a synthetic cell-permeable cAMP analogue, the phosphorylation levels of ERK1/2 were reduced at 30 min to 6 h after cAMP addition but returned back to near control levels at 12 and 24 h (Fig. 1A). Total ERK1/2 protein levels were not altered by cAMP (Fig.  1A). Forskolin, an adenylate cyclase activator, and isoproterenol, a ␤-adrenergic receptor agonist, both of which should increase intracellular cAMP levels, also inhibited ERK1/2 phosphorylation levels at 30 min after treatment (Fig. 1B). Because phosphorylation levels of ERKs are an indicator of ERK activation, these results strongly suggest that a cAMP-dependent signaling pathway inhibits ERK activity in C6 glioma cells, consistent with previous reports (6,16).
Cyclic AMP exerts differential effects on Akt activity depending on the specific cell types (19 -22). Treatment of C6 cells with 100 M 8-CPT-cAMP caused a reduction in levels of Akt phosphorylated at the Ser-473 site in a time-dependent manner ( Fig. 2A). The inhibition was maximal at 3 h after treatment and was sustained for at least 24 h after treatment ( Fig. 2A). Akt phosphorylation at Thr-308 was detectable only using much higher amounts of cell lysate proteins requiring immunoprecipitation with anti-Akt antibody prior to immunoblot. Phosphorylation of Akt at Thr-308 was inhibited by 8-CPT-cAMP in a time-dependent pattern similar to that at Ser-473 (Fig. 2B). Phosphorylation of both Thr-308 and Ser-473 is essential for Akt activation (30). Total Akt protein levels were not altered by 8-CPT-cAMP treatment (Fig. 2, A and B). Increasing intracellular cAMP levels by treatment with forskolin or isoproterenol also reduced Akt phosphorylation at both the Ser-473 and Thr-308 sites (Fig. 2C). Neither of these agents altered total Akt levels. The inhibition of Thr-308 phosphorylation by cAMP appeared to be more potent than that of Ser-473. This inhibitory effect of cAMP on Akt activation was confirmed by in vitro Akt kinase assay using histone H2B as substrate. Akt activity was potently decreased by cAMP, forskolin, or isoproterenol (Fig. 2C), showing that Akt phosphorylation levels in- deed reflect its activation state.
In order to use transfected ERK or Akt to study the signaling pathways by which cAMP inhibits ERK and Akt activation in C6 cells, we first determined the effect of cAMP on transfected HA-tagged ERK2 or Myc-tagged Akt phosphorylation. The time of cAMP treatment was chosen based on our observation that maximal inhibition of endogenous ERK and Akt phosphorylation by cAMP occurred at 30 min (Fig. 1A) and 3 h (Fig. 2, A and  B), respectively. Phosphorylation of transfected ERK was inhibited by treatment with 8-CPT-cAMP for 30 min (Fig. 3A), which is consistent with the changes in endogenous ERK phosphorylation in response to cAMP. Phosphorylation of transfected Akt at both Thr-308 and Ser-473 was inhibited by cAMP after 3 h, and phosphorylation of Thr-308 appeared to be more susceptible to inhibition by cAMP (Fig. 3B) similar to what was observed for endogenous Akt.
Cyclic AMP Inhibits PI3K/Akt Pathway-A previous study demonstrated that cAMP inhibits Akt activity in a number of cell lines by altering the subcellular localization of phosphoinositide-dependent kinase-1 as a consequence of PI3K inactivation (22). To confirm that inhibition of Akt by cAMP is a result of PI3K inhibition in C6 cells, a constitutively active form of the PI3K catalytic subunit (p110*) was co-transfected with the Akt expression vector in C6 cells. Akt phosphorylation levels were elevated by p110* in the absence of cAMP (Fig. 4), presumably as a result of increasing phosphatidylinositol 3Ј-phosphate production. Moreover, p110* prevented the decrease in Akt phosphorylation levels caused by cAMP treatment (Fig. 4). These results support the hypothesis that cAMP inhibits Akt activity by inhibiting PI3K.
Cyclic AMP Inhibits ERK and Akt in a PKA-independent Manner-cAMP-mediated effects in eukaryotes have traditionally been considered the result of cAMP binding to the regulatory subunits of the PKA tetramer, leading to the activation of the PKA catalytic subunits and subsequent phosphorylation of cellular substrates (31). To determine whether the inhibition of ERK and Akt by cAMP is mediated by PKA, C6 cells were pre-incubated with or without H89, a specific PKA inhibitor, prior to treatment with 8-CPT-cAMP. We have previously shown that H89 effectively blocks cAMP-induced PKA activation in C6 cells (24). H89 did not alter the ERK1/2 phosphorylation either in the presence or in the absence of cAMP (Fig. 5A), suggesting that PKA is not involved in the inhibition of ERK phosphorylation by cAMP. This conclusion is supported by the findings that a dominant negative form of PKA (PKA-REVab) did not block the cAMP inhibitory effect on ERK phosphorylation and that overexpression of an active form of PKA (PKA-CEV␣) had a minor effect on ERK phosphorylation (Fig. 5B).
With respect to Akt, H89 alone did not alter the phosphorylation at Ser-473, whereas it actually reduced phosphorylation at Thr-308 (Fig. 5C). Nevertheless, Akt phosphorylation at both Ser-473 and Thr-308 was inhibited by cAMP even in the presence of H89 (Fig. 5C). Expression of the dominant negative PKA neither affected basal Akt phosphorylation nor did it prevent the cAMP from inhibiting Akt (Fig. 5D). Moreover, the active form of PKA did not inhibit Akt (Fig. 5D). Thus, cAMP appears to inhibit both ERK and Akt phosphorylation in a PKA-independent manner.
Epac Is Not Involved in the Inhibitory Effect of cAMP on ERK and Akt-Because PKA was not involved in the inhibition of ERK and Akt by cAMP (Fig. 5), we assessed the possible involvement of another cAMP effector, Epac, a recently discovered cAMP-dependent guanine nucleotide exchange factor, which is activated upon cAMP binding and then stimulates Rap1 (32,33). A plasmid encoding an Epac protein lacking the cAMP-binding domain (Epac⌬cAMP), which represents an activated version of this exchange factor (32), was co-transfected with Rap1, ERK, or Akt expression vector in C6 cells. As shown in Fig. 6A, Epac⌬cAMP expression increased Flag-Rap1 GTP binding but did not alter total transfected Flag-Rap1 protein levels, indicating that Epac⌬cAMP is indeed an active form of Epac. If cAMP inhibits ERK and Akt by activating Epac, Epac⌬cAMP should mimic cAMP effects. However, we found that Epac⌬cAMP increased rather than decreased phosphorylation of ERK and Akt in the absence of cAMP (Fig. 6, B and C), suggesting that activation of Epac may increase ERK and Akt phosphorylation. Moreover, in the presence of Epac⌬cAMP, the phosphorylation levels of ERK and Akt were still inhibited by cAMP (Fig. 6, B and C). These results suggest that Epac is unlikely to mediate the inhibitory signal from cAMP to ERK or Akt that we have observed in C6 cells.
Inhibition of Rap1 Mediates the Decrease in ERK and Akt Phosphorylation by cAMP-Rap1 has been shown to function as a regulator of ERK and Akt pathways in response to cAMP (9,12,21). In C6 cells, it has been reported that cAMP increases GTP-bound Rap1, which is the active form of Rap1 (6,16), and it was suggested that activation of Rap1 leads to inhibition of ERK activation (6). In contrast to these results, we found that cAMP reduced Rap1 activation in C6 cells (Fig. 7A). Total Rap1 protein levels were not altered. To determine the role of Rap1 in cAMP-induced ERK inactivation, we measured ERK phosphorylation in the absence or presence of a specific Rap1 inhibitor, Rap1GAP1 (Rap1 GTPase activating protein-1). Expression of Rap1GAP1 decreased ERK phosphorylation in the absence of cAMP (Fig. 7B), suggesting that inhibition of Rap1 led to the inhibition of ERK. This result is consistent with the finding that activation of Rap1 by an active form of Epac increased ERK phosphorylation (Fig. 6B). Moreover, in the presence of Rap1GAP1, cAMP did not cause a further decrease in ERK phosphorylation (Fig. 7B). These data suggest that Rap1 is an upstream activator of ERK and that cAMP may inhibit ERK by inhibiting Rap1. Similarly, Akt phosphoryla-FIG. 7. Cyclic AMP inhibits ERK and Akt phosphorylation by inhibiting Rap1. A, confluent C6 cells were treated without (Control) or with 100 M 8-CPT-cAMP for 10 min. 600 g of cell lysate proteins were used for Rap1 activation assay as described under "Experimental Procedures." Rap1GTP is the active form of Rap1. Total levels of Rap1 were determined by Western immunoblot with anti-Rap1 antibody on 50 g of cell lysate proteins. B and C, C6 cells were transiently transfected with HA-ERK2 (B) or Myc-Akt (C) with (ϩ) or without (Ϫ) Flag-Rap1GAP1, a Rap1GTPase-activating protein, or its empty vector (pcDNA3). Cells were then treated with (ϩ) or without (Ϫ) 100 M 8-CPT-cAMP for 30 min (B) or 3 h (C). Phospho-ERK, total ERK, phospho-Akt(Thr-308), and total Akt protein levels were measured as described in the legend to Fig. 3. Total Flag-Rap1GAP1 protein levels were measured using anti-Flag antibody with 3% of cell lysate proteins. All of the experiments were repeated at least three times with similar results.
FIG. 8. The inhibitory effects of cAMP on ERK1/2 and Akt are independent. Confluent C6 cells were treated with (ϩ) or without (Ϫ) 30 M PD98059 in the presence (ϩ) or absence (Ϫ) of 20 M LY294002 for the indicated times. Fresh PD98059 and LY294002 were added into the conditioned medium after 6 h. Phospho-ERK1/2, total ERK1/2, phospho-Akt(Ser-473), and total Akt protein levels were determined as described in the legends to Figs. 1 and 2. All of the experiments were repeated at least three times with similar results. tion was also decreased by Rap1GAP1 and cAMP was unable to decrease Akt phosphorylation in the presence of Rap1GAP1 (Fig. 7C), suggesting that the inactivation of Rap1 also mediates the cAMP inhibitory effect on Akt phosphorylation in C6 cells.
The Inhibitory Effects of cAMP on ERK and PI3K/Akt Are Not Sequential-To determine whether the inhibition of ERK resulted from the inhibition of PI3K/Akt or vice versa, C6 cells were treated with PD98059, a MEK-1 inhibitor, and/or LY294002, a PI3K inhibitor. PD98059 transiently reduced ERK1/2 phosphorylation at 3 and 6 h after treatment, and phosphorylation was returned to normal at 12 h (Fig. 8). In contrast, the inhibition of Akt phosphorylation at Ser-473 by LY294002 was sustained throughout the 12-h treatment (Fig.  8). The influence of PD98059 and LY294002 on ERK1/2 and Akt, respectively, is quite similar to that of cAMP in C6 cells (compare Fig. 8 with Figs. 1 and 2). However, PD98059 and LY294002 had minor effects on Akt or ERK1/2 phosphoryla- tion, respectively (Fig. 8), suggesting that the inhibitory effects of cAMP on ERK and PI3K/Akt pathways are likely to be parallel instead of sequential effects. (24,29). The finding that the inhibition of ERK and PI3K/Akt pathways by cAMP is also PKA-independent in C6 cells (Fig. 5) led us to hypothesize that ERK and PI3K/Akt may regulate IGF-I and IGFBP-3 gene expression in these cells. Indeed, IGF-I mRNA was reduced by 70, 56, or 84% in response to treatment with PD98059, LY294002, or their combination, respectively (Fig. 9), suggesting that that the inhibition of both ERK and PI3K inhibits IGF-I gene expression. In contrast, IGFBP-3 mRNA was reduced by LY294002 but not by PD98059 (Fig. 9), which indicates that PI3K, but not ERK, is upstream of IGFBP-3 expression. ␤-Actin mRNA levels were not altered by either compound (Fig. 9).

Inhibition of ERK and PI3K/Akt Affects Growth Factor Gene Expression and Cell Growth-Previously we have reported that cAMP inhibits IGF-I and IGFBP-3 gene expression in C6 cells in a PKA-independent manner
We and others have previously shown that cAMP inhibits C6 cell growth (6,24). As shown in Fig. 10A, the inhibition of ERK or PI3K by PD98059 or LY294002, respectively, caused a profound reduction of new DNA synthesis in C6 cells. These results are consistent with the observations that cAMP inhibited ERK, PI3K, as well as DNA synthesis in C6 cells (Figs. 1, 2, and  10A). The percentage of C6 cells in G 0 /G 1 phase increased after the treatment with cAMP, whereas the percentage of C6 cells in S phase decreased (Fig. 10, B and C), suggesting that cAMP prevents S phase entry. G 1 /S transition is regulated by cyclin D/Cdk (cyclin-dependent kinase) 4/6 and cyclin E/Cdk2 complexes (34). Protein levels of Cdk2, Cdk4, and Cdk6 were not altered by cAMP (data not shown). We did not detect cyclin D1 protein in C6 cells using Western immunoblot (data not shown). However, protein levels of another D cyclin, cyclin D2, were inhibited by cAMP, PD98059, or LY294002 (Fig. 10D), supporting the hypothesis that cAMP inhibits cyclin D2 expression by inhibiting ERK and PI3K pathways. Cyclin E protein levels were not affected by cAMP, PD98059, or LY294002 (Fig.  10D). We did not detect p21 Cip1 protein, a Cdk inhibitor, in C6 cells (data not shown). Another Cdk inhibitor, p27 Kip1 , was induced by cAMP, as well as by PD98059 and LY294002 (Fig.  10D). These results suggest that the inhibition of ERK and PI3K pathways by cAMP may contribute to the stimulation of p27 Kip1 expression as well.

DISCUSSION
The Mechanisms by Which cAMP Inhibits ERK and Akt-We have shown here that cAMP inhibits ERK activity in C6 cells using a novel mechanism, which does not require PKA and Epac, two known cAMP effectors, but does require Rap1 inactivation. This is the first demonstration that cAMP can inhibit Rap1. The positive effect of Rap1 on ERK activation was supported by two findings that the inhibition of Rap1 by Rap1GAP1 inhibited ERK phosphorylation (Fig. 7) and that the activation of Rap1 by an active form of Epac increased ERK phosphorylation (Fig. 6). There are two possible mechanisms by which cAMP can inhibit Rap1 activity, either by activating a Rap1 GTPase-activating protein (GAP) to induce intrinsic Rap1 GTPase activity or by inhibiting Rap1 guanine nucleotide exchange factor (GEF) activity to decrease Rap1 GTP loading. There are 10 Rap1 GEFs identified thus far (35,36). Among these proteins, only Epac and its closely related protein, Epac-2, are known to be regulated by cAMP (32,33). However, upon cAMP binding, Epac and Epac-2 are activated and initiate an exchange of GTP for GDP on Rap1 protein. Therefore, it is unlikely that Epac proteins are involved in decreasing the levels of Rap1GTP in cAMP-treated C6 cells. Several Rap1specific GAPs have been identified and cloned (11). However, the regulation of these Rap1 GAPs is largely uncharacterized (37)(38)(39), and it is not known whether cAMP can regulate these Rap1 GAPs. To determine the nature of Rap1-specific GEF(s) or GAP(s) involved in the inhibition of Rap1 in response to cAMP will be one of the future goals of our study.
It has been proposed that in cells expressing B-Raf, activation of Rap1 stimulates ERK, whereas in cells lacking B-Raf, ERK is inhibited by Rap1 activation (6,12). Active Rap1 has been shown to associate directly with and activate B-Raf (12). We easily detected B-Raf protein in C6 cells. 2 Thus, we propose that cAMP inhibits ERK by inhibiting a Rap1/B-Raf/MEK-1/ ERK signaling cascade in C6 cells.
Compared with the inhibition of ERK activity by cAMP, the mechanism by which cAMP inhibits Akt activity is much less well understood. One report has shown that cAMP inhibits Akt activity by blocking phosphoinositide-dependent kinase-1 membrane localization as a result of PI3K inactivation (22). Our results also suggest that the inhibition of Akt activity is a result of PI3K inactivation in C6 cells (Fig. 4). Moreover, we 2 L. Wang and M. L. Adamo, unpublished results.

Cyclic AMP Inhibits Rap1
showed that neither PKA nor Epac is involved in the cAMP inhibitory effect on Akt activity in C6 cells (Figs. 5 and 6). We found that inactivation of Rap1 by cAMP probably mediates the inhibitory signaling from cAMP to PI3K/Akt (Fig. 7). This is consistent with the report that activation of Rap1 leads to activation of PI3K/Akt in thyroid cells (21). PI3K is a direct downstream effector of Ras as a result of the interaction between Ras and the PI3K catalytic subunit p110 (40). Because the effector domain region of Rap1 is almost indistinguishable from that of Ras (11), it is possible that Rap1 activates the PI3K pathway by a direct interaction, which is supported by the observation that Rap1 binds PI3K in vitro (41), and cAMP inhibits PI3K by disrupting this interaction. However, this model still remains to be tested.
The Mechanisms by Which cAMP Regulates C6 Cell Growth-The results presented here clearly show that the inhibition of both ERK and PI3K led to the inhibition of IGF-I gene expression, whereas only the inhibition of PI3K decreased IGFBP-3 gene expression (Fig. 9). These results strongly suggest that cAMP can influence cellular gene expression by inhibiting ERK and PI3K pathways in C6 cells. To our knowledge, this is the first report showing that ERK and/or PI3K are upstream of IGF-I or IGFBP-3 gene expression in tumor cells. Moreover, the inhibition of either ERK or PI3K led to a decreased rate of DNA synthesis (Fig. 10), suggesting that both pathways provide critical cell-proliferative signals in C6 cells.
Our results suggest that cAMP blocked G 1 /S transition in C6 cells (Fig. 10) consistent with the effect of cAMP in other cell types (42,43). Cell cycle progression is controlled by cyclin-dependent kinases (Cdks), in which activities are regulated by a series of cyclins and Cdk inhibitors (44). Cyclin D2 and cyclin E mRNA levels have been reported to peak in late G 1 phase (45,46) suggesting an important role for cyclin D2 and cyclin E in the G 1 /S transition. Cyclin D2 and cyclin E can heterodimerize with Cdk4/6 and Cdk2, respectively, to phosphorylate pRb, which regulates the G 1 /S transition (34). p27 Kip1 inhibits cyclin E/Cdk2 activity upon binding, whereas p27 Kip1 may not inhibit kinase activity of cyclin D/Cdk4 complexes (44). We have shown here that cyclin D2 and p27 Kip1 protein levels were reduced and stimulated, respectively, by cAMP in C6 cells, whereas cAMP did not alter the abundance of cyclin E and Cdk2/4/6 proteins (Fig. 10, data not shown). Thus, we propose that cAMP inhibits Cdk2 activity by increasing the concentration of its inhibitor, p27 Kip1 and inhibits Cdk4/6 activity by decreasing the concentration of its activator, cyclin D2. A progressive loss of p27 Kip1 accompanies increasing malignancy in the progression from low-to high-grade astrocytoma; low-grade astrocytomas show abundant p27 Kip1 staining, and the poorly differentiated malignant gliomas show marked reductions in p27 Kip1 staining (47,48). Thus, restoration of p27 Kip1 through cAMP treatment may represent a useful therapy for gliomas.
Summary-We propose that cAMP inhibits ERK and PI3K/ Akt pathways in C6 cells using a novel mechanism involving Rap1 inactivation (Fig. 11). Two known cAMP effectors, PKA and Epac, are not involved in the decrease of ERK and PI3K/ Akt activities in response to cAMP treatment. Cyclic AMP down-regulates IGF-I gene expression by inhibiting ERK and PI3K. Moreover, cAMP causes G 1 phase arrest in C6 cells, probably as a result of up-regulation of p27 Kip1 and downregulation of cyclin D2. Inhibition of both ERK and PI3K strongly contributes to these regulatory events.