Activation of AMP-activated Protein Kinase Induces p53-dependent Apoptotic Cell Death in Response to Energetic Stress*

Tumor suppressor p53-dependent stress response pathways play an important role in cell fate determination. In this study, we have found that glucose depletion promotes the phosphorylation of AMP-activated protein kinase catalytic subunit α (AMPKα) in association with a significant up-regulation of p53, thereby inducing p53-dependent apoptosis in vivo and in vitro. Thymocytes prepared from glucose-depleted wild-type mice but not from p53-deficient mice underwent apoptosis, which was accompanied by a remarkable phosphorylation of AMPKα and a significant induction of p53 as well as pro-apoptotic Bax. Similar results were also obtained in human osteosarcoma-derived U2OS cells bearing wild-type p53 following glucose starvation. Of note, glucose deprivation led to a significant accumulation of p53 phosphorylated at Ser-46, but not at Ser-15 and Ser-20, and a transcriptional induction of p53 as well as proapoptotic p53 AIP1. Small interference RNA-mediated knockdown of p53 caused an inhibition of apoptosis following glucose depletion. Additionally, apoptosis triggered by glucose deprivation was markedly impaired by small interference RNA-mediated depletion of AMPKα. Under our experimental conditions, down-regulation of AMPKα caused an attenuation of p53 accumulation and its phosphorylation at Ser-46. In support of these observations, enforced expression of AMPKα led to apoptosis and resulted in an induction of p53 at protein and mRNA levels. Furthermore, p53 promoter region responded to AMPKα and glucose deprivation as judged by luciferase reporter assay. Taken together, our present findings suggest that AMPK-dependent transcriptional induction and phosphorylation of p53 at Ser-46 play a crucial role in the induction of apoptosis under carbon source depletion.

AMP-activated protein kinase (AMPK) 3 was originally identified as an enzyme that has an ability to inhibit hydroxymethylglutaryl-CoA reductase (1) and also regulate acetyl-CoA carboxylase by reversible phosphorylation (2). Subsequent studies demonstrated that AMPK is widely expressed and exists as a heterotrimeric complex, which consists of a catalytic subunit (␣) and two regulatory subunits (␤ and ␥). The mammalian genome contains seven AMPK genes encoding two ␣ (␣1 and ␣2), two ␤ (␤1 and ␤2), and three ␥ (␥1, ␥2, and ␥3) isoforms (3)(4)(5). The catalytic ␣ subunit is composed of three functional domains, including an NH 2 -terminal Ser/Thr protein kinase domain, a central auto-inhibitory region, and a COOH-terminal regulatory subunit-binding domain. AMPK acts as an intracellular energy sensor by monitoring cellular energy levels. For example, AMPK becomes activated by the tumor suppressor LKB1 complex-mediated phosphorylation at Thr-172 in response to certain energy-depleting stresses such as glucose deprivation, hypoxia, and oxidative stress, which increase the intracellular AMP:ATP ratio (6 -10). AMPK can also be activated allosterically in the AMP:ATP ratio (11). Upon activation, AMPK down-regulates the ATP consuming metabolic pathways and activates the energy-generating processes through phosphorylating the primary targets involved in energy metabolism, thereby maintaining energy balance within cells (4).
Pro-apoptotic p53 is a founding member of the p53 tumor suppressor family and acts as a critical regulator of many cellular processes such as cell cycle arrest and apoptosis (12). p53 is frequently mutated in over 50% of human tumors (13)(14)(15), and p53-deficient mice developed spontaneous tumors (16). p53 acts as a nuclear sequence-specific transcription factor and transactivates its numerous target genes implicated in cell cycle arrest and apoptotic cell death, including p21 WAF1 , Bax, Puma, Noxa, and p53 AIP1. Its pro-apoptotic function is closely linked to its DNA binding activity. Indeed, over 90% of the p53 mutation is detected within its sequence-specific DNA-binding domain (13)(14)(15). Intracellular p53 protein levels are tightly regulated predominantly through the ubiquitin-proteasome protein degradation pathway. MDM2, which interacts with the NH 2 -terminal transactivation domain of p53 and inhibits its transcriptional activity, is one of the ubiquitin-protein isopeptide ligases for p53 (17,18). In response to various types of cellular stress, including DNA damage, hypoxia, nucleotide pool reduction, and thermal shock, p53 is induced to be stabilized as well as activated in the cell nucleus, and thereby plays a key role in the regulation of cell fate determination (19 -21). Recently, it has been shown that cells treated with low glucose arrest in the G 1 phase of the cell cycle in association with a significant activation of AMPK (22). According to their results, AMPK-mediated cell cycle arrest in response to low glucose required the phosphorylation of tumor suppressor p53 at Ser-15. Because stress-induced phosphorylation of p53 at Ser-15, which disrupts the p53-MDM2 interaction, enhances its activity as well as stability (19 -21), it is likely that p53 plays an important role in the regulation of cell cycle arrest caused by glucose limitation. Indeed, p53-deficient cells failed to arrest under low glucose conditions (22). Consistent with these results, activation of AMPK in human hepatocellular carcinoma-derived HepG2 cells bearing wild-type p53 resulted in G 1 cell cycle arrest through stabilization of p53 (23). Previously, Stefanelli et al. (24) described that AMPK has a protective role against thymocyte apoptosis in response to dexamethasone treatment.
In this study, we have found that glucose deprivation induces phosphorylation of AMPK␣ and promotes p53-dependent apoptotic cell death in vivo and in vitro. Under our experimental conditions, p53 was induced in response to glucose depletion at mRNA and at protein levels. Our present findings suggest that AMPK acts as a metabolic sensor to determine cell fate through the activation of pro-apoptotic p53.

EXPERIMENTAL PROCEDURES
Mice-Six-week-old male c57BL/6 mice (23-24 g), which were purchased from Charles River Laboratories (Tokyo, Japan), were housed in an animal facility maintained on a 12-h light/dark cycle at a constant temperature of 22 Ϯ 1°C and given free access to food and water ad libitum. For starvation, food was withdrawn from the cages at the onset of the dark cycle for the indicated times, whereas access to water was allowed. All experiments with these mice were carried out with the approval of the Chiba Cancer Center Experimental Animal Care and Use Committee.
Cell Lines and Transfection-Human osteosarcoma-derived U2OS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum (FBS; Invitrogen), 50 g/ml penicillin, and 50 g/ml streptomycin (Invitrogen). Human lung carcinoma H1299 cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated FBS plus antibiotics mixture. These cells were cultured in a humidified atmosphere of 5% CO 2 , 95% air at 37°C. Where indicated, U2OS cells were cultured in glucosefree Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% dialyzed FBS. For transient transfection, U2OS and H1299 cells were transfected with the indicated combinations of the expression plasmids using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturer's instructions. pcDNA3 (Invitrogen) was used as a blank plasmid to balance the amount of DNA introduced in transient transfection.
Immunoblotting-Cells were scraped off the plates and transferred into the microcentrifuge tubes. The cells were then lysed in lysis buffer containing 10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM EGTA, 50 mM ␤-mercaptoethanol ,1% Triton X-100, a commercial protease inhibitor mixture (Sigma) and phosphatase inhibitor mixture (Sigma) for 30 min on ice, and subjected to a brief sonication for 10 s at 4°C followed by centrifugation at 15,000 rpm at 4°C for 10 min to remove insoluble materials. The protein concentrations were measured using the Bradford protein assay according to the manufacturer's instructions (Bio-Rad). The equal amounts of protein (20 g) were separated by 10% SDS-PAGE and electrophoretically transferred onto polyvinylidene difluoride membranes (Immobilon-P, Millipore, Bedford, MA). The transferred membranes were blocked with Tris-buffered saline containing 5% nonfat dry milk and 0.1% Tween 20 at 4°C overnight. After blocking, the membranes were incubated with monoclonal anti-p53 (DO-1; Oncogene Research Products, Cambridge, MA), monoclonal anti-Bax (6A7; eBioscience, San Diego, CA), polyclonal anti-phospho-p53 at Ser-15 (Cell Signaling, Beverly, MA), polyclonal anti-phospho-p53 at Ser-20 (Cell Signaling), polyclonal anti-phospho-p53 at Ser-46 (Cell Signaling), polyclonal anti-p21 WAF1 (H-164; Santa Cruz Biotechnology), polyclonal anti-AMPK␣ (Cell Signaling), polyclonal anti-phospho-AMPK␣ (Cell Signaling), polyclonal anti-PARP (Cell Signaling), or with polyclonal anti-actin (20 -33; Sigma) antibody for 1 h at room temperature. After incubation with primary antibodies, the membranes were incubated with horseradish peroxidase-coupled goat anti-mouse or anti-rabbit IgG secondary antibody (Cell Signaling) for 1 h at room temperature. Immunoblots were visualized by ECL detection reagents according to the manufacturer's instructions (Amersham Biosciences).
Immunoprecipitation-At the indicated time points after glucose depletion, whole cell lysates prepared from U2OS cells were pre-cleared with 30 l of protein G-Sepharose beads for 90 min at 4°C, and supernatants were incubated with the indicated antibodies overnight at 4°C. After incubation, the reaction mixtures were mixed with 30 l of protein G-Sepharose beads and incubated for 1 h at 4°C. The immune complexes were eluted with SDS-sample buffer, and separated by 10% SDS-PAGE followed by immunoblotting with the indicated antibodies.
Indirect Immunofluorescence Microscopy-U2OS cells were cultured in the absence of glucose. At the indicated times after glucose deprivation, cells were fixed in 3.7% formaldehyde for 30 min at room temperature, permeabilized in 0.2% Triton X-100 for 5 min at room temperature, and then blocked with 3% bovine serum albumin in phosphate-buffered saline (PBS) for 1 h at room temperature. After blocking, cells were washed in PBS and incubated with polyclonal anti-phospho-AMPK␣ and monoclonal anti-p53 or with polyclonal anti-phospho-AMPK␣ and monoclonal anti-nucleolin antibodies (StressGen Biotechnologies, Cambridge, UK) for 1 h at room temperature, followed by the incubation with fluorescein isothiocyanateconjugated anti-rabbit IgG and rhodamine-conjugated antimouse IgG (Invitrogen) for 1 h at room temperature. Cell nuclei were stained with DAPI. The specific fluorescence was observed by using a confocal laser scanning microscope (Olympus, Tokyo, Japan).
Flow Cytometry-After glucose deprivation, both floating and attached cells were collected by low speed centrifugation and washed in PBS. The cells were treated with 500 g/ml of RNase A (Sigma) and subsequently stained with 50 g/ml of propidium iodide (Sigma) for 30 min at room temperature. Then the DNA content indicated by propidium iodide staining was analyzed by FACSCalibur flow cytometer (BD Biosciences).
Luciferase Reporter Assay-p53-deficient H1299 cells were plated in 12-well plates at a density of 50,000 cells/well and transiently co-transfected with a constant amount of a luciferase reporter construct driven by the p53 promoter (100 ng) and 10 ng of Renilla luciferase expression plasmid (pRL-TK) together with or without the increasing amounts of the expression plasmids for AMPK␣1 plus AMPK ␣2 (50, 100, or 200 ng). For all transfections, the total DNA amounts were kept constant (510 ng) using empty parental plasmid. Forty eight hours after transfection, cells were lysed, and their luciferase activities were measured by dual luciferase reporter assay system (Promega, Madison, WI). Results represent an average firefly luciferase value after normalization to Renilla luciferase signal. Each experiment was performed at least three times by triplicates.
Construction of the Expression Plasmids for AMPK␣1 and AMPK␣2-To generate the expression plasmids for AMPK␣1 and AMPK␣2, we employed PCR-based amplification using cDNA prepared from U2OS cells as a template. For AMPK␣1, the 5Ј-part of the entire coding region was amplified by PCR using the following primer sequences: 5Ј-GGAATTCCATGC-GCAGACTCAGTTCCTG-3Ј and 5Ј-CTGCAGCATATGTT-TCAAAAG-3Ј, which include EcoRI and PstI restriction sites, respectively. The 3Ј-part of the entire coding region was also amplified by PCR using the following primer sequences: 5Ј-CTGCAGGTGGATCCCATGAAG-3Ј and 5Ј-CCGCTCGAGCGGTTATTGTGCAAGAATTT-3Ј, which contain PstI and XhoI restriction sites, respectively. The resultant 5Ј-and 3Ј-parts of the entire coding regions were digested completely with EcoRI and PstI or with PstI and XhoI, respectively, and subcloned into EcoRI and XhoI restriction sites of pcDNA3 (Invitrogen) to give pcDNA3-AMPK␣1. For AMPK␣2, the 5Ј-part of the entire coding region was amplified by PCR using the following primer sequences: 5Ј-CGGGATC-CCGATGGCTGAGAGAAGCAGAAGC-3Ј and 5Ј-ACTAG-TTCTCAGAAATTCAC-3Ј, which include BamHI and SpeI restriction sites, respectively. The 3Ј-part of the entire coding region was also amplified by PCR using the following primer sequences: 5Ј-ACTAGTTGCGGATCTCCAAATTATAC-3Ј and 5Ј-GGAATTCCTCAACGGGCTAAAGTAGTAGTAATC-3Ј, which contain SpeI and EcoRI restriction sites, respectively. The amplified PCR products corresponding to 5Ј-or 3Ј-part of the entire coding region were treated with BamHI and SpeI or with SpeI and EcoRI, respectively, and introduced into BamHI and EcoRI sites of pcDNA3 (Invitrogen) to give pcDNA3-AMPK␣2. Nucleotide sequences of the PCR products were determined to verify the absence of random mutations.
RNA Interference-To knock down the endogenous AMPK␣, U2OS cells were transiently transfected with 10 nM of the chemically synthesized siRNAs targeting AMPK␣1 and AMPK␣2 or with the nonsilencing control siRNA (Invitrogen) using Lipofectamine TM RNAiMAX (Invitrogen) according to the manufacturer's recommendations. Total RNA and whole cell lysates were prepared 48 h after transfection. siRNA sequences used in the present study are available upon request.

Glucose Depletion Induces Apoptotic Cell Death in Vivo and
in Vitro-To examine whether glucose depletion could induce apoptotic cell death in vivo, we have prepared thymocytes from wild-type and p53-deficient mice that were maintained in the absence of glucose for 24 h, and their cell cycle distributions were analyzed by FACS. As shown in Fig. 1A, glucose deprivation resulted in a significant increase in number of thymocytes with sub-G 1 DNA content in wild-type mice, whereas glucose depletion had undetectable effects on thymocytes derived from p53-deficient mice, indicating that glucose deprivation-mediated apoptotic cell death might be regulated in a p53-dependent manner. Consistent with the previous observations (22), glucose depletion promoted the extensive phosphorylation of AMPK␣ in wild-type thymocytes (Fig. 1B). Under our experimental conditions, p53 accumulated in response to glucose starvation. Intriguingly, the accumulation of p53 was clearly associated with a significant induction of AMPK␣ phosphorylation as well as pro-apoptotic Bax, which is one of the direct targets of p53 (19 -21). Additionally, proteolytic cleavage of PARP, which is one of the substrates of the activated caspase-3, was induced in response to glucose deprivation. In contrast, the expression levels of p21 WAF1 , which have been shown to be implicated in p53-dependent cell cycle arrest (19 -21), remained almost unchanged regardless of glucose deprivation.
To further confirm this issue in vitro, human osteosarcomaderived U2OS cells bearing wild-type p53 were cultured in medium completely deficient in glucose for the indicated times, and we examined their cell cycle distributions by FACS. As shown in Fig. 2A, glucose depletion led to a massive apoptotic cell death. RT-PCR analysis revealed that the expression levels of AMPK␣1 and AMPK␣2 remain unchanged (Fig. 2B). Unexpectedly, p53 was transcriptionally induced upon removal of glucose, which was associated with the up-regulation of pro-apoptotic Bax and p53 AIP1 (25,26), although the expression levels of p21 WAF1 remained almost constant during the glucose deprivation-mediated apoptotic cell death. Immunoblot analysis demonstrated that AMPK␣ is induced to be phosphorylated following glucose starvation (Fig. 2C). In contrast, total amounts of AMPK␣ remained unchanged even in the absence of glucose. Additionally, glucose deprivation led to an accumulation of p53 as well as an induction of p53 phosphorylation at Ser-46 but not at Ser-15 and Ser-20. As expected, the expression levels of pro-apoptotic Bax but not of p21 WAF1 increased, and proteolytic cleavage of PARP was detectable in response to glucose deprivation.
Effects of p53 Knockdown on Glucose Deprivation-mediated Apoptotic Cell Death-To ask whether p53 could be involved in apoptotic cell death in response to glucose removal, U2OS cells were transiently transfected with control siRNA or with siRNA against p53. Twenty four hours after transfection, cells were switched into fresh medium lacking glucose. At the indicated time points after glucose starvation, whole cell lysates were prepared and analyzed for the expression levels of p53 by immunoblotting. As shown in Fig.  3A, p53 was induced to accumulate in cells transfected with control siRNA, whereas the amounts of p53 were kept at extremely low levels in cells transfected with siRNA against p53. FACS analysis revealed that siRNA-mediated knockdown of p53 strongly reduces the number of cells with sub-G 1 DNA content in response to glucose deprivation as compared with that of control cells (Fig. 3B). Similarly, p53-deficient human osteosarcoma-derived SAOS-2 cells underwent apoptotic cell death in response to glucose deprivation to a lesser degree, suggesting that p53 contributes at least in part to the induction of apoptotic cell death following glucose depletion.
Because glucose deprivation resulted in the strong induction of AMPK␣ phosphorylation, we examined the effects of  AMPK␣ on glucose deprivationmediated apoptotic cell death. To this end, siRNA-mediated knockdown of AMPK␣1 and AMPK␣2 was performed. Twenty four hours after transfection, cells were maintained in the absence of glucose. At the indicated time points after glucose starvation, cells were harvested, and their cell cycle distributions were analyzed by FACS. As shown in Fig. 4A, siRNA-mediated knockdown remarkably inhibited apoptotic cell death caused by glucose removal relative to control cells. Immunoblot analysis demonstrated that simultaneous knockdown of AMPK␣1 and AMPK␣2 does not induce the phosphorylation of AMPK␣ as well as the accumulation of p53 in response to glucose deprivation (Fig. 4B). Intriguingly, induction of p53 phosphorylation at Ser-46 was not detectable in cells where AMPK␣1 and AMPK␣2 were knocked down. Additionally, RT-PCR analysis showed that knockdown of AMPK␣1 and AMPK␣2 significantly inhibits the transcriptional up-regulation of p53 as well as p53 AIP1 (Fig. 4C). Under our experimental conditions, these results suggest that AMPK␣ regulates the glucose deprivation-mediated apoptotic cell death through the induction of p53 at mRNA and protein levels.
AMPK␣ Has an Ability to Transactivate p53-Although accumulating evidence suggests that p53 is stabilized at a protein level in response to a variety of cellular stresses (19 -21), our present results suggest that activation of AMPK␣ induces the transcriptional activation of p53. To address whether the p53 promoter could respond to AMPK␣, we have generated a luciferase reporter construct carrying a 5Ј-upstream region that encompasses nucleotide sequences from Ϫ2000 to ϩ22 relative to the first transcriptional initiation site of the human p53 gene termed p53-luc (upper panel of Fig.  5). U2OS cells were transiently transfected with the constant amount of p53-luc and pRL-TK   together with or without the increasing amounts of the expression plasmids for AMPK␣1 and AMPK␣2. Forty eight hours after transfection, cells were lysed, and their luciferase activities were measured. As shown in lower left panel of Fig. 5, simultaneous expression of AMPK␣1 and AMPK␣2 increased the luciferase activities driven by p53 promoter in a dose-dependent manner. Furthermore, luciferase activities driven by the p53 promoter were significantly increased in response to glucose starvation (lower right panel of Fig. 5). These results were con-sistent with our present observations showing that p53 is regulated at least in part at a transcriptional level following glucose depletion.
However, the precise molecular mechanisms behind the AMPK␣mediated transcriptional activation of p53 promoter remain unclear. Further studies are necessary to address this issue.
Interaction between AMPK␣ and p53 in Cells-Next, we examined whether AMPK␣ could associate with p53 in cells. For this purpose, we performed the indirect immunofluorescence staining experiments. At the indicated times after glucose deprivation, U2OS cells were double-stained with anti-p53 and antiphospho-AMPK␣ antibodies. Cell nuclei were stained with DAPI. Merged images clearly showed that p53 co-localizes with phospho-AMPK␣ within the cell nucleus, and their staining patterns are punctate in response to glucose deprivation (left panels of Fig. 6A). Because phospho-AMPK␣ co-localized with nucleolin, which is one of the convenient markers for nucleoli (right panels of Fig. 6A), it is likely that p53 co-localizes with phospho-AMPK␣ in nucleoli. Under our experimental conditions, any significant signals were undetectable in the absence of secondary antibody, and glucose deprivation-mediated accumulation of phospho-AMPK␣ in nucleoli was not observed in cells transfected with siRNA against AMPK␣ (supplemental Fig. S1). To further confirm the complex formation of p53 with AMPK␣, we performed immunoprecipitation experiments. U2OS cells were cultured in the absence of glucose. At the indicated time periods after glucose removal, whole cell lysates were prepared and immunoprecipitated with monoclonal anti-p53 antibody followed by immunoblotting with polyclonal anti-phospho-AMPK␣ or with polyclonal anti-ANPK␣ antibody. Under our experimental conditions, phospho-AMPK␣ and AMPK␣ were not co-immunoprecipitated with normal mouse serum. As shown in Fig. 6B, the anti-p53 immunoprecipitates contained phospho-AMPK␣ and AMPK␣, suggesting that p53 might be associated with both phosphorylated and unphosphorylated forms of AMPK␣. Our in vitro pulldown assays showed that the radiolabeled AMPK␣2  . Interaction between AMPK␣ and p53. A, indirect immunofluorescence staining of p53 and phospho-AMPK␣ in response to glucose depletion. At the indicated times after glucose deprivation, U2OS cells were fixed, permeabilized, and double-stained for p53 (green) and phospho-AMPK␣ (red). Cell nuclei were stained with DAPI (blue) (left panels). Similarly, U2OS cells were double-stained for nucleolin (green) and phospho-AMPK␣ (red). Cell nuclei were stained with DAPI (blue) (right panels). B, immunoprecipitation. U2OS cells were maintained in the absence of glucose. At the indicated times after starvation, whole cell lysates were prepared and immunoprecipitated (IP) with normal mouse serum (NMS) or with monoclonal anti-p53 antibody followed by IB with polyclonal anti-phospho-AMPK␣ (1st panel) or with polyclonal anti-AMPK␣ antibody (2nd panel). The right panels show the expression of phospho-AMPK␣ (1st panel), AMPK␣ (2nd panel), and actin (3rd panel) in response to glucose deprivation.
Enforced Expression of AMPK␣ Induces Apoptotic Cell Death in Association with the Up-regulation of p53-To confirm whether AMPK␣ could promote apoptotic cell death through the induction of p53, U2OS cells were transiently transfected with or without the increasing amounts of the expression plasmids for AMPK␣1 and AMPK␣2. Forty eight hours after transfection, cells were harvested, and their cell cycle distributions were analyzed by FACS. As shown in Fig. 7A, cells underwent apoptotic cell death in a dose-dependent manner. Immunoblot analysis demonstrated that the exogenously expressed AMPK␣ is phosphorylated in a dose-dependent fashion (Fig. 7B). As expected, levels of total p53 and its phosphorylation at Ser-46 were elevated in the presence of exogenously expressed AMPK␣1 and AMPK␣2. Consistent with our present results, enforced expression of AMPK␣1 and AMPK␣2 resulted in the transcriptional up-regulation of p53 as well as pro-apoptotic p53 AIP1 (Fig. 7C). In addition, simultaneous expression of AMPK␣1 and AMPK␣2 led to an induction of p53 and p53 AIP1 in a time-dependent manner (Fig. 7D). Taken together, our present findings suggest that AMPK-mediated induction of p53 plays an important role in the regulation of cell fate determination in response to glucose deprivation.

DISCUSSION
Recently, it has been shown that, under low glucose conditions, AMPK-dependent activation of tumor suppressor p53 causes cell cycle arrest, suggesting that p53 might act as a metabolic sensor in response to glucose limitation (22). According to their results, AMPK-dependent phosphorylation of p53 at Ser-15 plays an essential role in mediating the effects of activated AMPK on p53-dependent cell cycle arrest. In contrast, glucose depletion led to apoptotic cell death through the activation of pro-apoptotic Bax (27). In this study, we have found that glucose deprivation enhances the phosphorylation of AMPK␣ and induces p53-dependent apoptotic cell death in vivo and in vitro. Intriguingly, glucose depletion-mediated apoptotic cell death was associated with a significant transcriptional up-regulation of p53 and specific induction of p53 phosphorylation at Ser-46.
In response to various cellular stresses such as DNA damage, p53 is extensively phosphorylated at several amino acid residues, including Ser-15, Ser-20, and Ser-46 (21). Induction of cell cycle arrest and apoptotic cell death has been considered to be the most important biological functions of p53 in cells exposed to various cellular stresses (19 -21); however, the precise molecular mechanisms responsible for the choice between p53-dependent cell cycle arrest and apoptotic cell death in response to cellular stresses have been elusive. Previously, Oda et al. (25) found an important clue to understand this issue. According to their results, p53 was extensively phosphorylated at Ser-46 following severe DNA damage, and thereby cells with damaged DNA underwent apoptotic cell death. Thus, DNA damage-induced phosphorylation of p53 at Ser-46 might trigger the p53-dependent apoptotic program mediated by proapoptotic p53 AIP1. p53 AIP1, which is localized in mitochondria, is one of direct transcriptional targets of p53 and has an ability to down-regulate mitochondrial membrane potential and thereby release cytochrome c from the mitochondria to the cytoplasm (25,26). Therefore, it is likely that the stress-induced phosphorylation of p53 at Ser-46 is one of the critical events for commitment of cell fate into apoptotic cell death. Based on our present results, metabolic stress elicited by glucose deprivation promoted apoptotic cell death in association with a significant phosphorylation of AMPK␣ as well as p53 at Ser-46 but not at Ser-15 and Ser-20. As expected, phosphorylation of p53 at Ser-46 was significantly associated with the transcriptional upregulation of pro-apoptotic Bax and p53 AIP1. In support of these results, siRNA-mediated knockdown of the endogenous AMPK␣ markedly inhibited the glucose depletion-mediated apoptotic cell death and phosphorylation of p53 at Ser-46 as well as the transcriptional up-regulation of p53 AIP1. Indeed, enforced expression of AMPK␣ promoted apoptotic cell death and p53 phosphorylation at Ser-46 in association with the transcriptional activation of p53 AIP1. Collectively, our present results suggest that the activation of the AMPK complex is required for the metabolic stress-induced phosphorylation of p53 at Ser-46, thereby inducing apoptotic cell death. As described (28 -30), HIPK2 (homeodomain-interacting protein kinase-2) associates with p53 and mediates ionizing radiationinduced p53 phosphorylation at Ser-46. In addition to HIPK2, Yoshida et al. (31) found that protein kinase C␦ mediates phosphorylation of p53 at Ser-46 in response to genotoxic stress, indicating that protein kinase C␦ is a novel candidate for Ser-46 kinase.
Another new finding of this study was that glucose deprivation-mediated accumulation of p53 might be attributed at least in part to the transcriptional up-regulation of p53, and AMPK complex might play an important role in the transcriptional regulation of p53. Accumulating evidence strongly suggests that p53 is induced to accumulate through the post-translational modifications such as phosphorylation and acetylation in response to a variety of cellular stresses (19 -21). On the other hand, the regulatory mechanisms of p53 gene transcription have been poorly understood. Previously, Reich and Levine (32) described that p53 is transcriptionally regulated in response to mitogen stimulation as well as serum starvation. Noda et al. (33) identified the promoter element termed PE21 at a nucleotide position from Ϫ79 to Ϫ60 (relative to the first transcriptional initiation site) within the p53 promoter region responsible for p53 gene basal expression and ultraviolet response. In addition, it has been shown that homeobox protein HOXA5 has an ability to transactivate the p53 gene (34,35). It is worth noting that there exists a putative p53-responsive element within the p53 promoter region, thereby forming a novel positive feedback loop regulating p53 expression (36). According to our present results, siRNA-mediated knockdown of the endogenous AMPK␣ significantly inhibited the transcriptional induction of p53 in response to glucose deprivation. Consistent with these observations, glucose depletion enhanced the promoter activity of p53 as examined by luciferase reporter analysis. Furthermore, enforced expression of AMPK␣ resulted in a dramatic increase in luciferase activities driven by the p53 promoter in the absence of metabolic stress. Because the AMPK complex is involved in the regulation of gene expression (5), it is possible that the activated form of AMPK complex might induce the transcription of p53 through as yet unknown direct or indirect mechanisms. Further studies are required to address this issue.
Immunoprecipitation experiments indicated that p53 is associated with both unphosphorylated and phosphorylated forms of AMPK␣ during glucose deprivation-mediated apoptosis. Considering that glucose depletion-mediated phosphorylation of AMPK␣ is significantly associated with the up-regulation of p53, it is likely that phosphorylated forms of AMPK␣ might contribute to glucose starvation-mediated induction of p53 through the interaction with it. Our indirect immunofluorescence staining experiments clearly demonstrated that, upon glucose starvation, phosphorylated forms of AMPK␣ co-localize with p53 within the nucleolus, a subnuclear compartment that is a site of ribosomal assembly (37). In addition to the production of ribosomes, it has been postulated that the nucleolus might be involved in numerous cellular processes, including cell cycle control, DNA damage repair, and tRNA processing (38). Recently, Karni-Schmidt et al. (39) reported that nuclear localization signal I and the extreme COOH-terminal region of p53 are required for its nucleolar distribution. Although the functional consequences of the nucleolar localization of p53 are largely unknown, Wesierska-Gadek et al. (40) reported that the nucleolar distribution of p53 might contribute to its reactivation in response to genotoxic stress caused by cisplatin treatment. Thus, it is possible that the nucleolus might provide an important subnuclear locale for p53 function under specific conditions. According to our in vitro pulldown assay, AMPK␣2 but not AMPK␣1 was co-immunoprecipitated with endogenous p53. AMPK is a heterotrimeric protein kinase consisting of ␣, ␤, and ␥ subunits. In mammalian cells, each subunit exists in different isoforms, which might give rise to 12 distinct heterotrimeric isoform-subunit combinations. Recently, it has been shown that two wild-type AMPK complexes containing AMPK␣1 or AMPK␣2 display similar catalytic activities and equal response to AMP (41). Based on our present results, it is likely that AMPK␣2-containing AMPK complexes might have a distinct role in the regulation of apoptotic cell death through the functional interaction with p53, although further experiments are required to address this issue.