Stimulatory Heterotrimeric GTP-binding Protein Inhibits Hydrogen Peroxide-induced Apoptosis by Repressing BAK Induction in SH-SY5Y Human Neuroblastoma Cells*

Heterotrimeric stimulatory GTP-binding protein (Gs) stimulates adenylate cyclases to activate the cAMP signaling pathway. Although the cAMP pathway has been reported to be involved in apoptosis, the role of the Gs-cAMP signaling pathway during reactive oxygen species (ROS)-mediated apoptosis, which is involved in the resistance of cancer cells to chemotherapy and radiation, is not clearly understood. Thus, in this study we aimed to investigate the role of the α subunit of Gs (Gαs) in the ROS-induced apoptosis of cancer cells. The stable expression of constitutively active Gαs (GαsQL) inhibited the hydrogen peroxide-induced apoptosis of SH-SY5Y human neuroblastoma cells and reduced the hydrogen peroxide-induced increase in Bak and the decrease in Bcl-xL protein expression. Exogenous Bak expression abolished these inhibitory effects of GαsQL, but Bak small interfering RNA decreased hydrogen peroxide-induced apoptosis. Gαs repressed hydrogen peroxide-induced Bak expression by inhibiting the transcription of Bak mRNA, which resulted from the inhibition of the hydrogen peroxide-induced activation of transcription factors such as AP1, NF-κB, and NFAT. Moreover, Gαs also inhibited the hydrogen peroxide-induced binding of AP1, NF-κB, and NFAT to the Bak promoter. Furthermore, hydrogen peroxide-induced apoptosis was reduced by treating cells with prostaglandin E2, which activates Gαs, but this was augmented by CCPA, which activates Gαi causing a decrease in cAMP levels. From the results, we conclude that Gαs protects neuroblastoma cells from hydrogen peroxide-induced apoptosis by repressing Bak induction, which is mediated by the inhibition of the hydrogen peroxide-induced activations of AP1, NF-κB, and NFAT through cAMP-PKA-CREB signaling system.

The heterotrimeric GTP-binding proteins (G proteins) 2 are composed of ␣, ␤, and ␥ subunits, and the ␣ subunits of G protein (G␣) are classified into four main families: G␣ s , G␣ i , G␣ q , and G␣ 12 . When a signaling molecule, such as a hormone or a neurotransmitter binds to a G protein-coupled receptor (GPCR), the receptor stimulates the replacement of GDP with GTP on the G␣. GTP-bound activated heterotrimer dissociates into an ␣ subunit (G␣-GTP) and a ␤␥ dimer (G␤␥), which can independently regulate effectors including adenylate cyclases, phospholipases, phosphodiesterases, and ion channels. The hydrolysis of GTP to GDP by intrinsic GTPase, a process that is regulated by RGS (regulator of G-protein signaling) proteins, leads to the reassociation of the heterotrimer and the termination of the activation cycle (1)(2)(3). G protein signaling systems are involved in regulation of a variety of cellular responses, which include metabolism, neurotransmission, proliferation, differentiation, and apoptosis. Although signaling for cellular proliferation and apoptosis has generally been attributed to growth factor receptors that possess ligand-regulated protein-tyrosine kinase activity, growing evidence now indicates that GPCRs also regulate cellular growth and apoptosis (4 -6).
Reactive oxygen species (ROS) are constantly generated under normal conditions as a consequence of aerobic metabolism, and the most common ROS types are superoxide anions (O . 2 ), hydrogen peroxide (H 2 O 2 ), and hydroxyl radicals (HO Ϫ ) (7). ROS can react with DNA, proteins, carbohydrates, and lipids in a destructive manner due to their high levels of chemical reactivity. Thus, ROS are considered DNA-damaging agents that increase mutation rates and promote oncogenic transformation, and are implicated in the development of cancer and metastases (8,9). Moreover, ROS participate in the modulation of apoptosis following treatment with various agents including Fas, ultraviolet, and chemotherapeutic drugs (10 -12). Therefore, the ability of a cancer cell to defend itself against ROS is associated with resistance to chemo-and radiotherapy. Consequently, a more detailed understanding of the mechanisms that regulate ROS-induced apoptosis would contribute to our abilities to develop novel therapeutic strategies directed toward killing resistant cancer cells (13).
The ␣ subunit of the stimulatory G protein (G␣ s ) activates adenylate cyclases to increase cellular cAMP, which regulates various cellular responses by activating cAMP-dependent protein kinase (PKA), guanine nucleotide exchange factor activated by cAMP, and cyclic nucleotide gated ion channels. Moreover, increased G␣ s signaling was found to lead to the formation of tumors in endocrine cells (14), and thus, the activated mutant of G␣ s is known as gsp oncogene. Increased G␣ s signaling also leads to the apoptosis of cardiac myocytes (4) but protects neuronal apoptosis (15). However, the effect of G␣ s signaling on ROS-mediated death of cancer cells is not understood. Thus, we undertook this study to examine whether G␣ s can modulate hydrogen peroxide-induced cancer cell death, and if so, to elucidate the molecular mechanism responsible.
Flow Cytometric Analysis of Annexin V and Propidium Iodide-stained Cells-Apoptosis was quantified using Annexin V-FITC apoptosis kits (BD Biosciences) according to the manufacturer's instructions. Stained cells were analyzed using a FACSCalibur flow cytometer and the CellQuest analysis program (BD Biosciences NorthRyde, Australia).
Luciferase Activity Assays-SH-SY5Y cells were transfected with plasmids containing luciferase reporter gene by electroporation. Luciferase activities were assayed using the Bioluminescent Reporter Gene Assay System (Tropix, Bedford, MA) according to the manufacturer's instructions. At least three independent experiments in duplicate were performed, and promoter activities were normalized versus ␤-galactosidase activity.
Measurement of cAMP Accumulation-cAMP contents were determined by measuring the competitive binding of [ 3 H]cAMP to a cAMP-binding protein, the regulatory subunit RI␣ of the cAMP-dependent protein kinase was expressed in Escherichia coli. In brief, the culture medium was removed from a 12-well plate containing the SH-SY5Y neuroblastoma cells, and cAMP was extracted by immediate addition of 2.0 M perchloric acid. After the acid extract was neutralized with 4.0 M KOH, cAMP levels were determined and normalized to the amount of acid-insoluble protein.
Data Analysis-At least three independent experiments were conducted in all cases. Data are presented as mean Ϯ S.E.
The nonparametric Mann-Whitney U test was used to analyze mean values, and p values Ͻ 0.05 were considered statistically significant.

G␣ s Inhibited Hydrogen Peroxide-induced Apoptosis of SH-SY5Y Neuroblastoma
Cells-To investigate the role of G␣ s in hydrogen peroxide-induced apoptosis, a constitutively active mutant G␣ s , G␣ s Q227L (G␣ s QL), was stably expressed in SH-SY5Y human neuroblastoma cells; this was confirmed by Western blotting using antibodies against G␣ s and hemagglutinin tag in G␣ s QL (Fig. 1A). The expression of G␣ s QL increased the basal cAMP level to 28.5 Ϯ 0.5 pmol/mg protein (p Ͻ 0.05) from 3.4 Ϯ 0.6 pmol/mg of protein in vector-transfected cells, and the CREB phosphorylation. Twenty-four hours after treatment with 100 M hydrogen peroxide for 30 min, cells were stained with Annexin V to detect apoptosis. Stable expression of G␣ s QL reduced Annexin V positive and propidium iodide-negative cells to 9.6 Ϯ 1.7 from 45.8  (Fig. 1D). Transfection of G␣ s siRNA increased hydrogen peroxide-induced caspase-3 cleavage to 193.3 Ϯ 2.3% (p Ͻ 0.05) and PARP to 146.3 Ϯ 4.4% (p Ͻ 0.05) in vectortransfected cells, and it also increased hydrogen peroxide-induced caspase-3 cleavage from 36.6 Ϯ 4.9 to 162.6 Ϯ 4.6% (p Ͻ 0.05) and PARP from 24.0 Ϯ 2.7 to 129.8 Ϯ 4.7% (p Ͻ 0.05) in G␣ s QL expressing cells versus the hydrogen peroxide-treated vector-transfected control (Fig. 1D), indicating that G␣ s protected SH-SY5Y neuroblastoma cells from hydrogen peroxideinduced apoptosis.

G␣ s Inhibited Hydrogen Peroxide-induced Apoptosis by Repressing Increases in Bak Protein in SH-SY5Y Neuroblastoma
Cells-To probe the mechanism whereby G␣ s inhibits hydrogen peroxide-induced apoptosis of the neuroblastoma cells, we examined the effects of G␣ s on the expression of Bcl2 family proteins. Treatment with hydrogen peroxide induced an increase in pro-apoptotic Bak protein, but decreases in Bax and Bad proteins. Moreover, the stable expression of G␣ s QL inhibited up-regulation of Bak and down-regulation of Bax and Bad induced by hydrogen peroxide (Fig. 2A). However, hydrogen peroxide treatment did not decrease anti-apoptotic Bclexpression, and the overexpression of G␣ s QL did not change Bcl2 expression. The expression of Bcl-x L , another anti-apoptotic Bcl2 family protein, was reduced by hydrogen peroxide treatment, and the overexpression of G␣ s QL blocked the hydrogen peroxide-induced reduction in Bcl-x L expression ( Fig. 2A). Moreover, treatment with hydrogen peroxide increased mitochondrial Bak protein to 3.1 Ϯ 0.3-fold (p Ͻ 0.02) in vector-transfected cells. The stable expression of G␣ s QL resulted in an increase in the basal level of mitochondrial Bak by 1.5 Ϯ 0.2-fold (p Ͻ 0.02), but reduced the hydrogen peroxide-induced expression of Bak to 1.7 Ϯ 0.1-fold (p Ͻ 0.02) of the untreated vector control level (Fig. 2B).
To confirm the participation of Bak in the anti-apoptotic effect of G␣ s , the effects of the transient expressions of Bak and Bak siRNA were analyzed. Overexpression of Bak caused an increase in the hydrogen peroxide-induced cleavages of PARP and caspase-3 in both vector-transfected and G␣ s QL-transfected neuroblastoma cells (Fig. 2C). On the other hand, the expression of Bak siRNA decreased basal and hydrogen peroxide-induced Bak expressions in both G␣ s QL expressing cells and vector-transfected cells, and inhibited hydrogen peroxideinduced caspase-3 cleavage to 61.6 Ϯ 17.4% (p Ͻ 0.02) and PARP cleavage to 51.8 Ϯ 15.8% (p Ͻ 0.02) of the vector transfected control level (Fig. 2D).

G␣ s Repressed Hydrogen Peroxide-induced Bak Expression by
Inhibiting the Transcription of Bak mRNA-To investigate the mechanism whereby G␣ s represses the hydrogen peroxide-induced expression of Bak protein, we examined the degradation rate of Bak protein after blocking protein synthesis with cycloheximide. It was found that the degradation rate of Bak protein in G␣ s QL expressing cells was not significantly higher than that in vector-transfected cells (Fig. 3A). On the other hand, hydrogen peroxide treatment markedly increased Bak mRNA in vector-transfected cells, but only slightly in G␣ s QLtransfected cells (Fig. 3B). This finding was confirmed by real time RT-PCR, which showed a 29.0 Ϯ 4.5-fold (p Ͻ 0.02) increase in Bak mRNA in hydrogen peroxidetreated vector-transfected control cells, and a 6.5 Ϯ 2.3-fold increase (p Ͻ 0.02) in G␣ s QL expressing cells versus the untreated vector-transfected control cells. In addition, expression of G␣ s QL increased the basal Bak mRNA level to 3.2 Ϯ 0.7fold (p Ͻ 0.02) versus the untreated vector-transfected control cells (Fig. 3C). Furthermore, hydrogen peroxide treatment increased luciferase activity under the control of the Bak promoter to 12.6 Ϯ 0.3-fold (p Ͻ 0.02) compared with untreated vector-transfected cells. The expression of G␣ s QL increased basal Bak luciferase activity to 1.8 Ϯ 0.5-fold (p Ͻ 0.02), but reduced hydrogen peroxide-stimulated luciferase activity to 3.1 Ϯ 0.5-fold of the untreated vector-transfected cells (p Ͻ 0.05).
G␣ s Inhibited the Transcription of Bak mRNA via cAMP-PKA-CREB-dependent Pathways-In a study to probe the signaling pathway mediating the repression of Bak induction by G␣ s , we treated SH-SY5Y neuroblastoma cells with forskolin. Forskolin treatment reduced the hydrogen peroxide-induced cleavages of PARP and caspase-3 and the release of cytochrome c to the cytosol (Fig. 4A). Similar protective effects were exhibited by treatment of the cells with dibutyryl cAMP (data not shown). Moreover, forskolin treatment blocked the increase in Bak and the decrease in Bad and Bax induced by hydrogen peroxide. Forskolin also reduced the observed down-regulation of Bcl-x L by hydrogen peroxide, but did not change Bcl2 expression (Fig. 4A). Treatment of cells with H89, a cAMP dependent protein kinase (PKA) inhibitor, abolished the anti-apoptotic effects of G␣ s QL and the inhibitory effect of G␣ s QL on Bak induction, as evidenced by PARP and caspase-3 cleavage (Fig. 4B). Furthermore, transfection of decoy CRE abolished the anti-apoptotic effects of G␣ s QL and its inhibition of Bak induction by hydrogen peroxide (Fig. 4C).

G␣ s Repressed Hydrogen Peroxide-induced Bak Expression by
Inhibiting the Binding of Transcription Factors on Bak Promoter-To localize the region of the Bak promoter responsible for hydrogen peroxide induction, the effects of serial deletion on Bak luciferase activity were analyzed in SH-SY5Y neuroblastoma cells. Treatment with hydrogen peroxide induced Bak luciferase activity to 12.9 Ϯ 0.2fold under the control of the Ϫ3500-bp untranslated region, to 9.5 Ϯ 0.7-fold under the control of Ϫ2500-bp untranslated region, and 6.1 Ϯ 0.1-fold under the control of Ϫ1500-bp untranslated region to p Ͻ 0.02 versus the vector-transfected control, indicating that induction-fold decreased in proportion to the 5Ј deletion extent in the 5-untranslated region of the Bak promoter (Fig. 6A). To study the mechanism of G␣ s to inhibit hydrogen peroxide-induced Bak transcription mediated by transcription factors, the effects of G␣ s on the binding of AP1, CREB, NFAT, and NF-B to the Bak promoters were analyzed by electrophoretic mobility shift assay in SH-SY5Y cells. The expression of G␣ s QL inhibited the hydrogen peroxide-induced binding of AP1 probe from 3.2 Ϯ 4.3-to 1.6 Ϯ 2.3-fold (p Ͻ 0.05), NFAT probe from 2.2 Ϯ 4.1-to 1.7 Ϯ 1.8-fold (p Ͻ 0.05), and NF-B probe from 2.6 Ϯ 4.4-to 1.7 Ϯ 2.9-fold (p Ͻ 0.05) to the Bak promoter. However, G␣ s QL
G␣ s Inhibited Hydrogen Peroxide-induced Activation of NF-B and NFAT in a PKA-dependent Pathway-To probe the mechanism for G␣ s -cAMP signaling to inhibit hydrogen peroxide-induced AP1 activation, the effects of G␣ s on the phosphorylation and expression of AP1 components: c-Jun and c-Fos. Treatment with hydrogen peroxide increased phosphorylation of c-Jun at Ser 63 and Ser 73 , and expression of c-Jun and c-Fos. G␣ s QL expression blocked the hydrogen peroxide-induced phosphorylation of c-Jun, and inhibited the induction of c-Jun to 40.0 Ϯ 7.9% (p Ͻ 0.02) and c-Fos to 60.0 Ϯ 8.1% (p Ͻ 0.02) of that of the vector-transfected control (Fig. 6D). The phosphorylation of JNK, a major kinase that activates c-Jun was increased by hydrogen peroxide treatment. The expression of G␣ s QL also increased the basal phosphorylation of JNK as high as that of hydrogen peroxide-treated vector-transfected cells, without resulting in an increase in the phosphorylation and expression of c-Jun and c-Fos (Fig. 6D). Treatment of G␣ s QLexpressing cells with hydrogen peroxide did not cause a significant change in JNK phosphorylation. The expression of G␣ s QL also increased the phosphorylation of p38 and ERK mitogen-activated protein kinase as high as that of hydrogen peroxide-treated vector-transfected cells (data not shown), suggesting that G␣ s inhibits hydrogen peroxide-induced AP1 activation by inhibiting other enzymes. Thus, the involvement of glycogen synthase kinase-3␤ (GSK-3␤) was examined, because GSK-3␤ was reported to involve in c-Jun induction. Treatment of SH-SY5Y cells with hydrogen peroxide decreased the phosphorylation of GSK-3␤ at serine 9 to activate the enzyme (Fig. 6E). G␣ s QL expression increased the basal phosphorylation level of GSK-3␤ (Ser 9 ) (2.3 Ϯ 0.1-fold, p Ͻ 0.05) and decreased the basal c-Jun level (0.8 Ϯ 0.1-fold, p Ͻ 0.05) versus vector-transfected cells. G␣ s QL expression repressed the decrease in GSK-3␤ (Ser 9 ) phosphorylation induced by hydrogen peroxide (2.1 Ϯ 0.1-from 0.5 Ϯ 0.2-fold in vectortransfected cells, p Ͻ 0.05) and in expression of c-Jun (0.8 Ϯ 0.1-fold from 1.9 Ϯ 0.1-fold, p Ͻ 0.05) (Fig. 6E). Treatment with lithium chloride (selectively inhibit GSK-3) increased the phosphorylation of GSK-3␤, and decreased hydrogen peroxide-induced phosphorylation and induction of c-Jun in both vectorand G␣ s QL-transfected cells. Moreover, treatment with H89 also blocked the increase in GSK-3␤ phosphorylation (Ser 9 ) induced by G␣ s QL (data not shown). This result suggests that G␣ s repress hydrogen peroxide-induced AP1 activation by inhibition of GSK-3␤ in a PKA-dependent pathway.
Next, the effects of G␣ s on the degradation of IB␣ following hydrogen peroxide treatment were examined to probe the mechanism for G␣ s to inhibit hydrogen peroxide-induced activation of NF-B. Treatment of SH-SY5Y cells with hydrogen peroxide decreased the IB␣ level to activate NF-B. G␣ s QL expression increased the basal expression

. G␣ s inhibited the hydrogen peroxide-induced transcription of Bak in SH-SY5Y cells.
A, effects of G␣ s QL on the degradation rate of Bak protein. SH-SY5Y cells were treated with 10 g/ml cycloheximide (CHX), and Bak protein levels were quantified at various times by Western blotting analysis of total cell lysates. The graph shows the average densities of Bak bands. B and C, effects of G␣ s QL on Bak mRNA levels and Bak-luciferase activity. Bak mRNA expressions were measured by RT-PCR (B) and quantitative real-time RT-PCR (C). Amounts of Bak mRNA were normalized versus glyceraldehyde-3phosphate dehydrogenase (GAPDH), and presented as ratio versus vector transfected controls. For luciferase assay, SH-SY5Y cells were transfected with 2.5 g of (Ϫ3500 bp)Bak-pLuc and 2.5 g of a ␤-galactosidase construct. After 24 h, cells were treated with 100 M hydrogen peroxide for 24 h, and luciferase activities were measured and normalized versus ␤-galactosidase activity and are presented as ratio versus vector-transfected controls. Asterisks indicate significant differences (p Ͻ 0.05 versus vector-transfected cells, Mann-Whitney U test).

Repression of Hydrogen Peroxide-induced Bak Expression by G␣ s
JANUARY 18, 2008 • VOLUME 283 • NUMBER 3 level of IB␣ versus vector-transfected cells (1.9 Ϯ 0.1-fold, p Ͻ 0.05), and repressed the hydrogen peroxide-induced decrease in IB␣ levels from 0.4 Ϯ 0.1-to 1.7 Ϯ 0.1-fold (p Ͻ 0.05) (Fig. 6F). Treatment with H89 slightly decreased the basal level of IB␣, but augmented the hydrogen peroxideinduced decrease in the IB␣ level, suggesting G␣ s inhibit hydrogen peroxide-induced activation of NF-B by PKA-dependent blocking of the IB␣ degradation in SH-SY5Y cells. Finally, to probe the mechanism for G␣ s -cAMP signaling to inhibit hydrogen peroxide-induced NFAT activation, the effects of G␣ s on the nuclear translocation of NFAT following hydrogen peroxide treatment was analyzed, because NFATs are known to translocate into the nucleus after activation. Most of NFAT was localized in the cytosol fraction in SH-SY5Y cells before treatment, but hydrogen peroxide treatment induced translocation of most of NFAT from the cytosol to nuclear fraction (Fig. 6F). However, G␣ s QL expression decreased the hydrogen peroxide-induced translocation of   NFAT to the nuclear fraction (0.2 Ϯ 0.1-from 0.8 Ϯ 0.2-fold, p Ͻ 0.05), without significant change in its distribution before the treatment. Treatment with H89 caused a decrease in cytosolic distribution and concomitant increase in nuclear distribution of NFAT in vector-transfected cells and G␣ s -transfected cells, suggesting that G␣ s inhibits hydrogen peroxide-induced NFAT activation by blocking its nuclear translocation in a PKAdependent pathway.

GPCR Ligands Modulated the Hydrogen Peroxide-induced Apoptosis of SH-SY5Y Neuroblastoma
Cells-Because G␣ s was found to inhibit the hydrogen peroxide-induced apoptosis of SH-SY5Y neuroblastoma cells, we examined the effects of PGE 2 , which binds to the G␣ s -coupled receptor to stimulate adenylate cyclases, on hydrogen peroxide-induced Bak expression and apoptosis. Treatment with PGE 2 reduced hydrogen peroxideinduced CREB dephosphorylation at Ser 133 , Bak expression, and the cleavages of PARP and caspase-3. The cleavage of caspase-3 was also decreased by PGE 2 from 5.5 Ϯ 0.4to 2.7 Ϯ 0.7-fold (p Ͻ 0.02), and that of PARP was from 6.4 Ϯ 0.5to 3.7 Ϯ 0.3-fold (p Ͻ 0.05) (Fig.  7A). Because PGE showed anti-apoptotic effects similar to the forskolin effect, we examined the temporal patterns of cAMP concentration and CREB phosphorylation at serine 133 following PGE 2 and forskolin treatments. Treatment of SH-SY5Y cells with forskolin resulted in a rapid increase in cAMP concentration that reached a peak at 30 min (110.4 Ϯ 3.2 pmol/mg of protein), and CREB phosphorylation that reached a peak at 3 h (38.9 Ϯ 2.5-fold). Both the elevated cAMP level and CREB phosphorylation by forskolin decreased so slowly that they remained at elevated levels until 24 h after treatment (31.6 Ϯ 4.6 pmol/mg protein and 15.5 Ϯ 3.7fold, respectively). In contrast, PGE 2 treatment caused a rapid increase in the cAMP level with a peak at 30 min (78.3 Ϯ 3.9 pmol/mg of protein), a sharp decrease to about half the peak value at 1 h, and then a slow decrease to reach the basal level at 6 h after PGE 2 treatment. In contrast to the cAMP level, CREB phosphorylation induced by PGE 2 reached a peak (31.6 Ϯ 4.2-fold) at 1 h, and then slowly decreased to maintain elevated levels of phosphorylation (10.8 Ϯ 4.3-fold p Ͻ 0.05) until 24 h, showing a similar temporal pattern of CREB phosphorylation by forskolin. This result suggests that PGE 2 can maintain CREB activation for several hours after cAMP concentration is elevated for a short period in SH-SY5Y cells (Fig. 7B). Furthermore, the expression of constitutively active G␣ i2 (G␣ i2 QL), which antagonizes G␣ s effects, augmented hydrogen peroxide-induced Bak expression and the cleavages of caspase-3 and PARP, and partially eliminated the inhibitory effect of G␣ s QL on hydrogen peroxide-induced Bak expression and cleavages of caspase-3 and PARP (Fig. 7C). However, G␣ i2 QL expression also increased the basal expression level of the Bak protein in vector-transfected cells and G␣ s QL-transfected cells, implying that G␣ i2 may increase basal Bak expression by a CREB-independent pathway because G␣ i2 was shown to decrease CREB phosphorylation in SH-SY5Y cells (data not shown). Treatment with CCPA, which (binding of CCPA to adenosine A1 receptor activates G␣ i2 to inhibit adenylate cyclases) augments hydrogen peroxide-induced Bak expression, CREB dephosphorylation, and cleavages of PARP and caspase-3. The hydrogen peroxide-induced cleavage of caspase-3 was increased by CCPA from 5.8 Ϯ 0.3-to 7.9 Ϯ 0.7-fold (p Ͻ 0.02), and of PARP from 6.8 Ϯ 0.1-to 9.9 Ϯ 0.1-fold (p Ͻ 0.05) (Fig. 7D).

DISCUSSION
This study was performed to determine whether the ␣ subunit of stimulatory G protein, G␣ s , can modulate hydrogen peroxide-induced apoptosis and if so, to elucidate the molecular mechanism responsible. We found that G␣ s inhibits hydrogen peroxide-induced apoptosis by repressing Bak induction in SH-SY5Y human neuroblastoma cells. This finding is supported by the following results. First, the overexpression of constitutively active G␣ s QL was found to protect SH-SY5Y cells from hydrogen peroxide-induced apoptosis, and G␣ s siRNA augmented apoptosis. Second, G␣ s QL overexpression repressed the hydrogen peroxide-induced up-regulation of Bak protein by inhibiting transcription of Bak. Third, Bak overexpression abolished the protective effect of G␣ s on hydrogen peroxide-induced apoptosis, and knockdown of Bak (using siRNA) inhibited the apoptosis. We also found that the repression of Bak induction by G␣ s is mediated by cAMP, PKA, and CREB. This was evidenced by the observations that Bt 2 cAMP and forskolin repressed Bak induction and protected cells against hydrogen peroxide-induced apoptosis; moreover, treatment with the PKA inhibitor H89 or transfection with decoy CRE restored Bak promoter activity, augmented hydrogen peroxide-induced apoptosis, and abolished the protective effect of G␣ s QL. In addition, the protective effect of G␣ s was abolished by co-expressing constitutively active mutant G␣ i2 , which antagonizes G␣ s by inhibiting adenylate cyclase.
The G␣ s -cAMP signaling pathway has been reported to regulate cell death in various ways. Treatment with cholera toxin (which activates G␣ s ) was found to protect neuronal cells from glutamate-induced ROS-mediated apoptosis by up-regulating the anti-apoptotic protein Bcl2 in a cAMP-dependent manner (15), and forskolin has been reported to prevent the hydrogen peroxide-induced apoptosis of PC12 pheochromocytoma cells by increasing the level of antioxidant glutathione (21). Bt 2 cAMP inhibited apoptosis by blocking the induction of Bax mRNA (22). Moreover, PKA was reported to protect cells from apoptosis by inducing Bad phosphorylation (23), whereas CREB induces the transcription of the anti-apoptotic gene Bcl2.
However, it has not been previously reported that the G␣ s -cAMP signaling pathway modulates apoptosis by regulating Bak expression. Thus, this study presents the first evidence, to the best of our knowledge, that the G␣ s -cAMP signaling pathway modulates cancer cell apoptosis by repressing Bak induction.
Bak protein is a multidomain pro-apoptotic protein of the Bcl2 family, and acts as an essential gateway to intrinsic cell death pathways by facilitating the release of cytochrome c from mitochondria (24,25). Moreover, Bax Ϫ/Ϫ Bak Ϫ/Ϫ double knock-out cells have been reported to be resistant to anticancer drug-induced apoptosis in many human cancer cells (26). In addition, Bak has been reported to be up-regulated and to act as a downstream mediator of an oxidative stress pathway that leads to the apoptosis of SH-SY5Y neuroblastoma cells in response to fenretinide (27). These reports demonstrate that Bak mediates hydrogen peroxide-induced apoptosis, and support our finding that G␣ s can exert a protective effect against hydrogen peroxide-induced apoptosis by repressing Bak expression in SH-SY5Y neuroblastoma cells.
In our study to investigate the mechanism underlying the repression of hydrogen peroxide-induced Bak expression by G␣ s , we found that G␣ s represses Bak expression by inhibition of activation and binding of several transcription factors, i.e. AP1, NF-B, and NFAT, to Bak promoter following hydrogen peroxide treatment. This conclusion was reached because the transcription factors AP1, NFAT, and NF-B, but not CREB, were found to be activated by hydrogen peroxide when assessed by luciferase assay; specific inhibitors of the transcription factors blocked Bak transcription induced by hydrogen peroxide; and G␣ s or forskolin were found to inhibit the hydrogen peroxide-induced activations of AP1, NFAT, and NF-B, and thus to repress Bak gene transcription. In addition, G␣ s also inhibited the binding of AP1, NF-B, and NFAT to the Bak promoter induced by hydrogen peroxide treatment. The binding sites for these transcription factors seems to be scattered widely in the Bak promoter region (Ϫ3500 bp), because Bak luciferase was found to be increased in proportion to the extent of deletion in this region. The nuclear transcription factor AP1 is composed of dimers of fos and jun proto-oncogenes, and has been linked to a variety of cellular events including proliferation, differentiation, transformation, and apoptosis. Moreover, oxidative stress is known to induce the expressions and phosphorylations of Fos and Jun (28,29), and the DNA binding activity of AP1 in various cell types is known as a prelude to cell death (30,31). This study indicates that AP1 is involved in the repression of hydrogen peroxide-induced Bak expression by G␣ s . This is based on the results that the deletion of the two AP1 binding sites in the Bak promoter blocked Bak induction by hydrogen peroxide, and that G␣ s repressed the expression and phosphorylation of both c-Fos and c-Jun. Recently, it was reported that nitric oxide activates an apoptotic cascade in human brain tumor cells, involving sustained JNK activation to activate c-Jun and AP1 DNA binding activity and subsequent Bak induction (32). These reports show that AP1 mediates the apoptosis induced by oxidative stress, and support our finding that the repression of AP1 activation by the G␣ s -cAMP signaling system protects cells from apoptosis. We found that G␣ s represses hydrogen peroxide-induced activation of GSK-3␤ by phosphorylation of GSK-3␤ at serine 9, which can result in inhibition of the phosphorylation and induction of c-Jun to repress AP1 activation. Furthermore, a GSK-3 inhibitor, lithium, was found to block the hydrogen peroxide-induced phosphorylation and induction of c-Jun. The phosphorylation of GSK-3␤ in G␣ s QLexpressing cells were blocked by the PKA inhibitor, as suggested by the report that PKA can directly phosphorylate GSK-3␤ at serine 9 and inhibit its apoptotic activity in neurons (33). GSK-3 was reported to act in tandem with JNK to coordinate the full execution of the c-Jun stress response and AP-1induced death (34). Together with these reports, our finding suggests that G␣ s -cAMP signaling inhibits hydrogen peroxideinduced AP1 activation by PKA-dependent phosphorylation and inhibition of GSK-3␤. However, G␣ s QL expression increased the basal and hydrogen peroxide-stimulated phosphorylation of JNK, p38, and ERK without a significant increase in the phosphorylation and expression of c-Jun and c-Fos, suggesting that G␣ s repressed hydrogen peroxide-induced AP1 activation in a mechanism that is independent of JNK, p38, and ERK. NF-B-induced transcription has been reported to play a central role in host defense and inflammatory response, and to prevent apoptosis by increasing the expressions of anti-apoptotic genes including Bcl2, and by decreasing the expression of pro-apoptotic Bax protein in many human tumors. Thus, NF-B may be a major factor that controls the ability of cells to resist apoptosis-based tumor surveillance systems (35). However, NF-B has also been reported to stimulate apoptosis by inducing the expressions of death-promoting genes including p53, the death receptor Fas, and its ligand, TNF-␣, and Bak (36 -39) and by linking fenretinide-induced ROS production to apoptosis. NF-B dimers are held in an inactive cytoplasmic complex with a family of inhibitory proteins, the IBs, including IB␣. Phosphorylation of IBs at two serine residues in their N-terminal regulatory domain by the IB kinase complex targets them for rapid ubiquitin-mediated proteasomal degradation, which induces the nuclear translocation of NF-B to function (40). Vasoactive intestinal peptide and pituitary adenylate cyclase-activating polypeptide were reported to inhibit the activation-induced cell death in T cells by repressing expression of Fas ligand through stabilization of IB␣ via the cAMP signaling system (41). cAMP was found to decrease NF-B activation in response to the superoxide anion (42,43), and to increase the IB␣ levels to inhibit lipopolysaccharideinduced NF-B activation in THP-1 cells (44). Furthermore, activation of the PKA pathway is suggested to inhibit NF-B transcription by phosphorylating CREB, which competes with p65 for limiting amounts of CBP, because both NF-B and NFAT require the coactivator CBP for transcriptional activity (45). These reports support our finding that G␣ s increases the IB␣ level in a PKA-dependent pathway to inhibit hydrogen peroxide-induced activation of NF-B in neuroblastoma cells. This finding indicates that the G␣ s -cAMP signaling pathway protects cells from apoptosis by inhibiting the hydrogen peroxide-induced activation of NF-B, and implies that NF-B mediates apoptosis by stimulating Bak expression. NFATs constitute a family of transcription factors that transduce calcium signals and coordinate the expressions of proteins involved in cell growth and apoptosis. NFAT is known to be activated by activation of various receptors inducing calcium mobilization, and the receptor activation and calcium mobilization result in activation of many intracellular enzymes including the calcium-and calmodulin-dependent phosphatase calcineurin. Calcineurin is a major upstream regulator of NFAT proteins, and causes the dephosphorylation and translocation of NFAT proteins from the cytoplasm to the nucleus of activated cells (46,47). NFAT has been reported to mediate oxidative stress-induced apoptosis by inducing Fas ligand expression. Moreover, Bcl2 and Bcl-x L were found to mediate anti-apoptotic effects by inhibiting NFAT activation and translocation to the nucleus through the sequestration of calcineurin (48). cAMP signaling has been reported to inhibit NFAT activity by direct phosphorylation of NFAT by PKA resulting in inhibition of nuclear entry and transcriptional activity (41,49). Together with these reports, our finding that G␣ s inhibits hydrogen peroxide-induced nuclear translocation of NFAT, which was inhibited by a PKA inhibitor, suggest that G␣ s -cAMP signaling inhibits hydrogen peroxide-induced activation of NFAT by a PKA-dependent phosphorylation. This study shows that the G␣ s -cAMP signaling system might act upstream of transcription factors such as AP1, NF-B, and NFAT to block their activation induced by hydrogen peroxide, the resulting Bak expression, and then apoptosis of SH-SY5Y neuroblastoma cells. The present study also shows that G␣ s QL expression increases the basal level of the Bak protein and CREB binding to the Bak promoter, but that G␣ s QL expression inhibits Bak induction by hydrogen peroxide. Furthermore, it was also shown that hydrogen peroxide reduced CREB binding to the Bak promoter, and that the transfection of decoy CREB increased hydrogen peroxide-induced Bak luciferase activity. Thus, to interpret this seemingly paradoxical phenomenon, we speculate that the G␣ s -cAMP signaling system stimulates basal CREB binding to the Bak promoter to increase the basal level of transcription, which is reduced by hydrogen peroxide, but that it represses hydrogen peroxide-induced Bak expression by inhibiting activation of transcription factors mediating Bak expression in a CREB-dependent manner. In addition to Bak expression, G␣ s was found to represses the hydrogen peroxide-induced down-regulation of Bcl-x L in SH-SY5Y human neuroblastoma cells, and the resulting increase in Bcl-x L expression shifted the ratio of antito pro-apoptotic Bcl2 family proteins to favor cell survival (50). This finding suggests that the G␣ s -cAMP signaling system maintains mitochondrial integrity and protects cells from apoptosis not only by repressing the induction of pro-apoptotic Bak but also by maintaining the levels of anti-apoptotic Bcl-x L proteins, which implies that G␣ s -cAMP affects multiple proand anti-apoptotic Bcl2 family proteins.
Because G␣ s was found to protect cells from hydrogen peroxide-induced apoptosis, we examined whether PGE 2 , an agonist of the receptors that couple with G␣ s , protects SH-SY5Y cells from hydrogen peroxide-induced apoptosis, and found this to be the case. This finding is similar to reports that PGE 2 protects human colon cancer cells from apoptosis (51,52). In fact, EP receptors have been identified as potential targets for the treatment and prevention of colorectal cancer (53). PGE 2 binds to four different receptors termed EP1, EP2, EP3, and EP4, and of these EP2 and EP4 couple to G␣ s (54). PGE 2 induced a transient increase in cAMP levels, but maintained CREB phosphorylation for several hours, which might be involved in the anti-apoptotic effect of the PGE 2 -G␣ s signaling system. On the other hand, treatment with CCPA, an adenosine A1 receptor agonist, enhanced hydrogen peroxide-induced apoptosis in SH-SY5Y neuroblastoma cells. Adenosine A1 receptor interacts with the pertussis toxin-sensitive G i protein that inhibits adenylate cyclase, and is considered a promising therapeutic target in cancer (55). The anti-apoptotic effect of PGE 2 and the pro-apoptotic effect of CCPA were accompanied by corresponding changes in Bak induction after hydrogen peroxide treatment, indicating that both PGE 2 and CCPA influence apoptosis by modulating the G␣ s /G␣ i -cAMP signaling pathway that regulates Bak expression. This result suggests that apoptosis induced by various stimuli including ␥-radiation and chemotherapeutic agents, thus their therapeutic efficiency, can be improved by co-treatment with specific GPCR ligands that modulate sensitivity or resistance to apoptosis.
Based on the findings of the present study, we concluded that G␣ s can protect neuroblastoma cells from hydrogen peroxideinduced apoptosis by repressing Bak induction, and that this is mediated by the inhibition of hydrogen peroxide-induced transcription factor activation (e.g. AP1, NF-B, and NFAT) by the cAMP-PKA-CREB signaling pathway. We also conclude that hydrogen peroxide-induced apoptosis can be modulated by the administration of receptor agonists that activate G␣ s or G␣ i . These findings provide a better understanding of how G␣ s signaling pathways modulate ROS-induced apoptosis, and have important clinical implications, namely, that the apoptosis of cancer cells mediated by hydrogen peroxide following chemotherapy or radiation can be enhanced by administering agonists or antagonists of cancer cell-specific receptors to improve the therapeutic efficiency.